**Contents**


**John F. Alderete**

Advancing Prevention of STIs by Developing Specific Serodiagnostic Targets: *Trichomonas vaginalis* as a Model

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 5783, doi:10.3390/ijerph17165783 . . **87**

### **Mar´ıa Dolores Pozo-Cano, Adelina Mart´ın-Salvador, Mar´ıa Angeles ´ Perez-Morente, ´ Encarnacion´ Mart´ınez-Garc´ıa, Juan de Dios Luna del Castillo, Mar´ıa Gazquez-L ´ opez, ´ Rafael Fernandez-Castillo ´ and Inmaculada Garc´ıa-Garc´ıa**

Validation of the Women's Views of Birth Labor Satisfaction Questionnaire (WOMBLSQ4) in the Spanish Population

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 5582, doi:10.3390/ijerph17155582 . . . **103**

### **Antonio Mundo-Lopez, ´ Olga Ocon-Hern ´ andez, ´ Ainhoa P. San-Sebastian, ´ Noelia Galiano-Castillo, Olga Rodr´ıguez-Perez, ´ Mar´ıa S. Arroyo-Luque, Manuel Arroyo-Morales, Irene Cantarero-Villanueva, Carolina Fernandez-Lao ´ and Francisco Artacho-Cordon´** Contribution of Chronic Fatigue to Psychosocial Status and Quality of Life in Spanish Women

Diagnosed with Endometriosis

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3831, doi:10.3390/ijerph17113831 . . . **119**

### **Roman Pabayo, Amy Ehntholt, Daniel M. Cook, Megan Reynolds, Peter Muennig and Sze Y. Liu**

Laws Restricting Access to Abortion Services and Infant Mortality Risk in the United States Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3773, doi:10.3390/ijerph17113773 . . . **135**

### **Vanessa Sanchez-Mendoza, Encarnacion Soriano-Ayala and Pablo Vallejo-Medina**

Psychometric Properties of the Condom Use Self-Efficacy Scale among Young Colombians Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3762, doi:10.3390/ijerph17113762 . . **149**

### **Xiaojie Wang, Wenjie Nie and Pengcheng Liu**

Son Preference and the Reproductive Behavior of Rural-Urban Migrant Women of Childbearing Age in China: Empirical Evidence from a Cross-Sectional Data Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3221, doi:10.3390/ijerph17093221 . . **165**

### **Rugsapon Sanitya, Aniqa Islam Marshall, Nithiwat Saengruang, Sataporn Julchoo, Pigunkaew Sinam, Rapeepong Suphanchaimat, Mathudara Phaiyarom, Viroj Tangcharoensathien, Nongluk Boonthai and Kamheang Chaturachinda**

Healthcare Providers' Knowledge and Attitude Towards Abortions in Thailand: A Pre-Post Evaluation of Trainings on Safe Abortion

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3198, doi:10.3390/ijerph17093198 . . . **177**

### **Leticia Avila-Burgos, Julio C´esar Montanez-Hern´andez, ˜ Lucero Cahuana-Hurtado, Aremis Villalobos, Patricia Hern´andez-Pena˜ and Ileana Heredia-Pi**

Government Expenditure on Maternal Health and Family Planning Services for Adolescents in Mexico, 2003–2015

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 3097, doi:10.3390/ijerph17093097 . . . **189**

**Mar´ıaAngeles ´ P´erez-Morente, Adelina Mart´ın-Salvador, Mar´ıa G´azquez-Lopez, ´ Pedro Femia-Marzo, Mar´ıa Dolores Pozo-Cano, C´esar Hueso-Montoro and Encarnacion´ Mart´ınez-Garc´ıa** Economic Crisis and Sexually Transmitted Infections: A Comparison Between Native and Immigrant Populations in a Specialised Centre in Granada, Spain Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 2480, doi:10.3390/ijerph17072480 . . . **207**

### **Sabera Turkmani, Caroline S. E. Homer and Angela J. Dawson**

Understanding the Experiences and Needs of Migrant Women Affected by Female Genital Mutilation Using Maternity Services in Australia Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 1491, doi:10.3390/ijerph17051491 . . **217**

### **Rosario M. Rom´an-G´alvez, Sandra Mart´ın-Pel´aez, Juan Miguel Mart´ınez-Galiano, Khalid Saeed Khan and Aurora Bueno-Cavanillas**

Prevalence of Intimate Partner Violence in Pregnancy: An Umbrella Review Reprinted from: *Int. J. Environ. Res. Public Health* **2021**, *18*, 707, doi:10.3390/ijerph18020707 . . . **235**

### **Nuria Infante-Torres, Milagros Molina-Alarcon, ´ Angel Arias-Arias, Juli´an Rodr´ıguez-Almagro and Antonio Hern´andez-Mart´ınez**

Relationship Between Prolonged Second Stage of Labor and Short-Term Neonatal Morbidity: A Systematic Review and Meta-Analysis

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 7762, doi:10.3390/ijerph17217762 . . . **249**

### **Miguel Garc´ıa-Mart´ın, Carmen Amezcua-Prieto, Bassel H Al Wattar, Jan Stener Jørgensen, Aurora Bueno-Cavanillas and Khalid Saeed Khan**

Patient and Public Involvement in Sexual and Reproductive Health: Time to Properly Integrate Citizen's Input into Science

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 8048, doi:10.3390/ijerph17218048 . . **275**

**Raquel Casado Santa-B´arbara, C´esar Hueso-Montoro, Adelina Mart´ın-Salvador,**

**Mar´ıa AdelaidaAlvarez-Serrano, ´ Mar´ıa G´azquez-Lopez ´ and Mar´ıaAngeles ´ P´erez-Morente** Association between Sexual Habits and Sexually Transmitted Infections at a Specialised Centre in Granada (Spain)

Reprinted from: *Int. J. Environ. Res. Public Health* **2020**, *17*, 6881, doi:10.3390/ijerph17186881 . . . **287**

## **About the Editors**

**Juan Miguel Mart´ınez Galiano** Ph.D.) is currently a midwife at a public hospital and Professor at the University of Jaen (Spain). He holds a degree in nursing and midwifery, master's degree in ´ Research and Advances in Preventive Medicine and Public Health, and Ph.D. degree, received in 2012 from the University of Granada for research on the impact of prenatal education programs on mother and newborn (cum laude). Galiano's research expertise is in sexual and reproductive health. He is a member of the Consortium for Biomedical Research in Epidemiology and Public Health (CIBERESP) and the group for Research, Development and Innovation in Epidemiology, Preventive Medicine and Surgery (CTS-435). He is also the principal investigator of numerous research projects with competitive financing.

**Miguel Delgado-Rodr´ıguez** (Ph.D.) is Professor of Preventive Medicine and Public Health at the University of Jaen and a medical specialist in preventive medicine and public health. His Master ´ of Public Health in Epidemiology was completed at UCLA. He has supervised more than 50 doctoral theses and published more than 400 articles and 200 books and book chapters. He has participated in more than 30 national and international projects with competitive financing. He has served as Scientific Director of the CIBERSP and coordinator of Public Health of the Health Research Fund (FIS) of the Carlos III Health Institute.

International Journal of *Environmental Research and Public Health*

### *Editorial* **The Relegated Goal of Health Institutions: Sexual and Reproductive Health**

**Juan Miguel Martínez-Galiano 1,2,\* and Miguel Delgado-Rodríguez 2,3**


Academic Editor: Paul B. Tchounwou Received: 5 February 2021; Accepted: 9 February 2021; Published: 11 February 2021

Sexual and reproductive health does not always receive the attention it deserves and frequently is not supported with the necessary resources to guarantee its maintenance. The World Health Organization (WHO) is aware of its importance and in its structure has specific programs focused on this health aspect.

Although conceptually sexual and reproductive health is seen as a united whole, this does not occur in some parts of the world. Sexual health, according to the WHO, is a state of physical, mental, and social well-being concerning sexuality. It requires a positive and respectful approach to sexuality and sexual relations, as well as the possibility of having pleasant and safe sexual experiences that are free from all coercion, discrimination, and violence. Reproductive health focuses on all aspects of the reproductive system, its functions, and its processes. [1]

The WHO aims to give a joint approach to sexual and reproductive health with which it intends to support the development of research and learning policies in this field. Even the United Nations General Assembly has established the specific goal of guaranteeing universal access to sexual and reproductive health services by 2030.

One of the biggest challenges in sexual and reproductive health is sexually transmitted infections (STI). For the development of well-oriented health programs on this issue, it is necessary to know the attitude of health professionals towards patients with STIs. One of the most relevant STIs, due to the absence of curative treatments and vaccines, is human immunodeficiency virus (HIV); the results of a study carried out in Spain with nursing students show that they have a positive attitude towards HIV patients, and this may improve their future healthcare practice in tackling this condition [2]. The identification of groups vulnerable to STIs has a special relevance for prevention programs; Perez-Morente et al. have found that immigrant women in periods of crisis are more susceptible to acquiring an STI [3,4]. Screening and diagnostic methods, improved treatments, as well as the availability of more effective vaccines also deserve special attention. Alderete has identified a new *Trichomonas vaginalis* epitope chain protein and the epitope chain proteins of infectious agents that may have potential applicability for a possible vaccine against this parasite [5]. Health education plays a special role in the implementation of STI prevention programs, for which it is necessary to identify risky practices. In this issue, Santa-Barbara et al. show in their results that the practice of oral sex, the practice of anal sex, and the non-use of condoms are clear risk factors. [6] Valid instruments are available, as shown by Sánchez-Mendoza et al., to determine the self-efficacy of condom use in adolescents, [7] one of the social strata with the highest risk of acquiring STIs. This favours the design of adequate strategies designed to better promote the use of condoms as a method of protection against STIs.

Family planning is another basic component of sexual and reproductive health. Contraceptive methods allow people to enjoy sexual acts without the worry of unwanted pregnancy and help in family planning. Adolescents are the most susceptible group to unwanted pregnancies, and this occurs in people residing in higher-risk environments or more disadvantaged places, as shown by the results of Ávila-Burgos et al. in Mexico. [8] Accessibility to contraceptive methods is critical for providing the correct medication and giving adequate instructions to avoid unwanted pregnancies. This is displayed in this issue by Langer et al. in a study carried out in pharmacies in Germany. [9]. In the case of having to perform an abortion due to an unwanted pregnancy or for any other reason, it must be carried out in safe conditions for the woman, and therefore it is necessary to implement training and awareness programs [10]. Besides this, it is a priority to have adequate legislation to guarantee the safety of women. Legal restrictions on abortion can increase maternal and infant morbidity and mortality. This is what Pabayo et al. conclude in a study performed in the United States in which infant mortality in states with restrictive abortion laws are compared with that of states with more permissive laws [11].

Women are the main recipients of health programs on sexual and reproductive health. They may suffer from chronic diseases, such as endometriosis, which causes chronic fatigue, a worse quality of life, greater anxiety, and worse sleep, among other consequences, as Mundo-López et al. [12] show. On the other hand, clinical and educational practices during pregnancy and childbirth influence the health status of women. For example, Cano-Ibáñez et al. [13] show that correct nutrition and promoting a dietary pattern based on the Mediterranean diet improve the outcome of childbirth. Another example is provided by Infante-Torres et al. in a systematic review with meta-analysis [14] in which they show that care without prolonged expulsive periods, since an excessive duration of this labour stage leads to greater neonatal morbidity, improves newborns' health. Postpartum post-traumatic stress disorder is an increasingly attention-grabbing disorder. For this reason, predictive models are useful, such as the one developed by Hernandez-Martinez et al. that helps to predict the risk of this disease [15]. Prenatal care is essential, but it is not practised in the same way in all areas or by all women. Thus, for example, in China women belonging to ethnic minorities and living in rural areas initiate pregnancy control later [16]. This necessitates the development of special programs for these situations. Not only must the woman be cared for, but her satisfaction with the healthcare process must also be taken into account. Instruments are available to gain knowledge of this aspect concerning attendance at delivery, such as the satisfaction scale with the assistance received, as validated by Pozo-Cano et al. [17].

Violence must be taken into account in reproductive health. The WHO in 2014 made a declaration for the prevention and eradication of disrespect and mistreatment during childbirth care in health centres; however, the figures of obstetric violence a few years later remain high [18]. Gender-based violence is a serious public health problem; it is exercised in various ways, with pregnancy being a clear risk factor. The umbrella review carried out by Roman-Galvez et al. [19] makes the magnitude of this problem visible and shows the high prevalence figures around the world. Another form of violence against women is female genital mutilation. Turkmani et al. present a conceptual model, based on cultural values and the physical and emotional needs of women, as a framework to guide maternity services to better address this problem [20]. The same authors conclude that empowering women and increasing their awareness of their health care rights can help involve women as active partners in the design and delivery of health information.

To progress in research and therefore in its application to the protection of sexual and reproductive health, it is necessary to take into account the social and health professionals, scientific societies, and institutions involved. García-Martin et al. stress the importance of cooperation between health professionals, academic and health institutions, and the community as an essential point to improve research quality and make significant advances in the field of women's health [21].

Every health program needs to address popular beliefs that do not have any scientific proof. For instance, some Chinese women maintain a preference for male children, which can cause a greater number of pregnancies with their associated risks [22].

In summary, it is necessary to develop policies and strategies tackling sexual and reproductive health in collaboration with the different agents involved to address its different components, such as family planning, STIs, women's health, pregnancy, and violence, among others. These policies must be translated into primary prevention, screening, early treatment, and rehabilitation programs. For the development of these programs, it is necessary to identify the most vulnerable populations, as well as their risk factors. They should also be based on scientific evidence and apply validated tools. Research in sexual and reproductive health should be promoted so that its results are the basis for the design of appropriate policies and strategies in sexual and reproductive health.

**Author Contributions:** Both coauthors contributed equally to this work. Both authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


Abortions in Thailand: A Pre-Post Evaluation of Trainings on Safe Abortion. *Int. J. Environ. Res. Public Health* **2020**, *17*, 3198. [CrossRef] [PubMed]


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Elaboration and Validation of Two Predictive Models of Postpartum Traumatic Stress Disorder Risk Formed by Variables Related to the Birth Process: A Retrospective Cohort Study**

**Antonio Hernández-Martínez 1, Sergio Martínez-Vazquez 2, Julián Rodríguez-Almagro 1,\*, Miguel Delgado-Rodríguez 3,4 and Juan Miguel Martínez-Galiano 2,4**


Received: 31 October 2020; Accepted: 21 December 2020; Published: 24 December 2020

**Abstract:** This study aimed to develop and validate two predictive models of postpartum post-traumatic stress disorder (PTSD) risk using a retrospective cohort study of women who gave birth between 2018 and 2019 in Spain. The predictive models were developed using a referral cohort of 1752 women (2/3) and were validated on a cohort of 875 women (1/3). The predictive factors in model A were delivery type, skin-to-skin contact, admission of newborn to care unit, presence of a severe tear, type of infant feeding at discharge, postpartum hospital readmission. The area under curve (AUC) of the receiver operating characteristic (ROC) in the referral cohort was 0.70 (95% CI: 0.67–0.74), while in the validation cohort, it was 0.69 (95% CI: 0.63–0.75). The predictive factors in model B were delivery type, admission of newborn to care unit, type of infant feeding at discharge, postpartum hospital readmission, partner support, and the perception of adequate respect from health professionals. The predictive capacity of model B in both the referral cohort and the validation cohort was superior to model A with an AUC-ROC of 0.82 (95% CI: 0.79–0.85) and 0.83 (95% CI: 0.78–0.87), respectively. A predictive model (model B) formed by clinical variables and the perception of partner support and appropriate treatment by health professionals had a good predictive capacity in both the referral and validation cohorts. This model is preferred over the model (model A) that was formed exclusively by clinical variables.

**Keywords:** post-traumatic stress disorder; predictive model; validation

### **1. Introduction**

Post-traumatic stress disorder (PTSD) has been described as "the complex somatic, cognitive, affective, and behavioral effects of psychological trauma" [1]. PTSD affects the newborn, the mother–child relationship, and the mother's health and quality of life [2–4]. PTSD prevalence can vary considerably, and in a systematic review including 28 studies the average prevalence was 4.0% in the general perinatal population and 18.5% in women at risk [5].

Several factors have been associated with the development of postpartum PTSD [6]. However, research into the obstetric factors involved in PTSD development is scarce. Most studies focus on the influence of birth type [7–15] and newborn variables such as prematurity [16,17], low weight [16], low Apgar scores [16,18], type of newborn feeding [7], and newborn hospitalization [9]. Several studies have recently observed a relationship between certain obstetric practices and the risk of PTSD [2,15].

One of the biggest challenges in health is creating tools for predicting the risk of particular health problems. The purpose of these tools is the early identification of those most susceptible to developing the problem or anticipating the appearance of the first symptoms. Several published prediction models exist for PTSD after childbirth, using different variables and populations [9,18–27]. Although several models exist, only one studied the prediction capacity with ROC curves (19), and none were validated in populations other than those used to create the model. No prediction tool has been made based on the factors related to either the birth process or obstetric practices, and that can be easily used as a screening tool by health professionals. One of the difficulties in creating these prediction models is the choice of initial predictive factors, which may be purely objective or also subjective. We consider that predictors of an objective clinical nature, such as the clinical data of problems and interventions during pregnancy and childbirth, are easily obtainable but may be insufficient to predict complex phenomena such as PTSD. Conversely, the use of variables of a more subjective nature such as anxiety, emotions, depression, among others, although used in various predictive models [19,20,22–24,26], is more complex to use. Many of these subjective variables are assessed through scales and questionnaires, some of them long and complex, limiting their application in clinical practice.

Therefore, the objective of this study was to develop and validate two postpartum PTSD risk prediction models, one based exclusively on objective clinical predictive factors and another including subjective factors such as perceptions of partner support and treatment received by professionals.

### **2. Materials and Methods**

An observational study was conducted using a retrospective cohort of women who gave birth between 2018 and 2019. The Research Ethics Committee of the province of Jaen approved this study with reference number TD-VCDEPP-2019/1417-N-19. The main tool employed to collect the relevant data for this study was medical records. The women were required to read an information sheet about the study and its objectives and check a box in which they showed their consent to participate in it; that is, they signed an ad hoc digital informed consent. The STROBE statement has been followed in the reporting of this study. [28]

### *2.1. Design and Participants*

This analytical and observational study used a retrospective cohort of women who gave birth between 2018 and 2019. A total of 1752 women were included in the referral cohort for the predictive models, which were subsequently validated with a cohort of 875 women. The Clinical Research Ethics Committees of Universidad de Jaen gave ethical approval prior to the start of the study. All of the participants received written information on the study, including the fact that participation was completely voluntary and anonymous.

We used the maximum modeling principle to estimate the sample size. We needed 10 events (women at risk of PTSD) per each incorporated variable [29]. If we consider that our initial model may contain 15 variables, we would need 150 women at risk of PTSD. Taking into account that the prevalence of PTSD in other Spanish studies is reported as 10.6% [2], we would need a minimum of 1415 women for the derivation cohort and at least half as many for validation, about 708 women. Nevertheless, the researchers' team opted to recruit the maximum number of women.

### *2.2. Data Collection and Information Sources*

The main tool employed to collect the relevant data for this study was medical records. The primary outcome variable—risk of PTSD—and the following objective independent variables were collected from the medical records.

Independent variables were:



**Table 1.** Bivariate analysis of potential predictive factors of post-traumatic stress disorder (PTSD) risk.

**Table 1.** *Cont.*


The primary outcome variable, risk of PTSD, was determined using the modified Perinatal Post-Traumatic Stress Disorder Questionnaire (PPQ) [30] (Appendix A: Spanish version). The PPQ is a 14-item measure assessing post-traumatic symptoms related to the childbirth experience, including intrusiveness or re-experiencing, avoidance behaviors, and hyperarousal or numbing of responsiveness. The PPQ also contains one item about feelings of guilt. Response options were modified from the original dichotomous scale to a five-level Likert scale (scored 0 to 4). The total possible score on the modified PPQ ranged from 0 to 56. In the current study, internal consistency was higher than in previous investigations using the dichotomous scaling, with an α = 0.90 [30].

### *2.3. Statistical Analysis*

A descriptive statistical analysis using absolute and relative frequencies for qualitative variables and means and standard deviation (SD) for quantitative variables was performed.

The analysis of potential predictive factors, which have been previously identified in the literature as risk factors of delayed onset breastfeeding, was carried out in a bivariate analysis using the chi-square and Student's *t*-test to estimate qualitative and quantitative variables, respectively. Of these variables, and following Lemeshow's statistical criteria, associations with *p*-values < 0.25 were selected for inclusion in the multivariate binary logistic regression model [31,32] (Table 1). These analyses were performed in the derivation cohort.

Then, two models were created (Table 2): model A based on exclusively clinical criteria and model B based on clinical criteria plus maternal perceptions of the degree of partner support and the treatment received by healthcare professionals. These models were constructed using backward elimination (RV in SPSS) with the derivation of cohort women's data. To assess the prediction qualitatively, we used Swets's criteria, which uses the following category values: 0.5–0.6 (bad), 0.6–0.7 (poor), 0.7–0.8 (satisfactory), 0.8–0.9 (good), and 0.9–1.0 (excellent) [33]. In addition, the Nagerlkerkes R-square and the calibration were determined using the Hosmer–Lemeshow test *p*-value of both models.


**Table 2.** Predictive models of PTSD risk during the postpartum.

OR: odds ratio; Bold: statistically significant differences.

The derivation and validation cohorts were compared after using chi-square and Student's *t*-test for qualitative and quantitative variables, respectively (Table 3). Finally, the AUC-ROC in the validation cohort was estimated for the predictive model that we created (Table 2). In this case, the probabilities used proceed from applying the predictive model created with the derivation cohort using the data of

the women in the validation cohort. SPSS 20.0. (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.


**Table 3.** Comparison of characteristics between the derivation and validation cohort.


**Table 3.** *Cont.*

\* Student–Fisher *t*-test.

### **3. Results**

### *Characteristics of Participants*

The derivation cohort consisted of 1752 women and the validation cohort 875 women, with a prevalence of PTSD risk of 14.2% (248) and 10.9% (95), respectively. First, we built predictive models using the derivation cohort. The variables associated with the risk of PTSD (screening criterion *p*-value < 0.25) selected for the multivariate analysis were: maternal age, parity, live birth, place of delivery, induced delivery, use of natural methods for pain, regional analgesia, general anesthesia, type of delivery, perineal tear, skin-to-skin contact, breastfeeding the first hour of life, admission of the newborn to care unit, hospital stay, breastfeeding at discharge, postpartum surgical intervention, postpartum readmission, degree of partner support during pregnancy, delivery and postpartum, and degree of respect received from professionals during pregnancy, delivery and postpartum. (Table 1).

Then, two predictive models were created. Model A was based exclusively on clinical variables, and model B consisted of clinical variables plus subjective variables on support received from their partner and treatment received from healthcare professionals (see Table 2). The variables to be included in the final predictive models were selected automatically by the SPSS program, through backward step instruction.

When performing the multivariate analysis, model A included the following variables: type of delivery, skin-to-skin contact, admission of the newborn to care unit, perineal tear, type of infant feeding at discharge, and postpartum hospital readmission. The predictive capacity (AUC-ROC) in the referral cohort was 0.70 (95% CI: 0.67–0.74) (Figure 1), while in the validation cohort it was 0.69 (95% CI: 0.63–0.75) (Figure 2), which is considered as satisfactory in Swets's criteria.

**Figure 1.** Predictive capacity of model A and model B in the derivation cohort. Area under the ROC curve to determine the predictive ability of the model in the validation cohort, representing the sensitivity on the *y* axis and 1- specificity in the *x* axis.

**Figure 2.** Predictive capacity of model A and model B in the validation cohort. Area under the ROC curve to determine the predictive ability of the model in the validation cohort, representing the sensitivity on the *y* axis and 1-specificity on the *x* axis.

The predictive factors in the final model B were: type of delivery, admission of the newborn to care unit, type of infant feeding at discharge, postpartum hospital readmission, support received by the partner, and the perception of respect from healthcare professionals. The predictive capacity (AUC-ROC) in the derivation cohort was 0.82 (95% CI: 0.79–0.85) (Figure 1), while in the validation cohort it was 0.83 (95% CI: 0.78–0.87) (Figure 2). This predictive capacity is considered good per Swets's criteria. Finally, we examined comparability issues in both cohorts, and found no statistically significant differences with any variable except for the risk of PTSD (*p* = 0.018), which was 14.2% (248) in the referral cohort and 10.9% (95) in the validation cohort (see Table 3).

### **4. Discussion**

This study presents the main results of the development of two postpartum PTSD risk prediction models. Model A, constructed using only clinical variables, presented a satisfactory predictive capacity (AUC-ROC = 0.70), while model B, constructed with clinical variables and subjective patient perceptions, presented a good predictive capacity (AUC-ROC = 0.82). The predictive variables common to both models were: type of delivery, admission of the newborn to care unit, type of infant feeding at discharge, and postpartum hospital readmission. However, model A also included skin-to-skin contact and the presence of a severe tear as exclusive variables. In contrast, model B included variables relating to the support received from the partner and the perception of respect from healthcare professionals during childbirth.

Currently, there are several studies published with PTSD risk prediction models [9,19–27]. However, only the study by Van Heumen et al. studied the prediction capacity with ROC curves [19], presenting an AUC-ROC of 0.795, lower than our best model. Moreover, none have been validated in populations other than those used to create the model, which is an important limitation. The sample sizes were also all smaller than ours, and only one exceeded 1000 subjects [19], and only three studies exceeded 500

subjects [9,19,20]. Some of these studies have been carried out on very specific population groups, such as the study by López et al. [21], who used a sample of women who had a cesarean delivery, excluding most women who give birth vaginally. Other models have included other scales and assessments based on questionnaires of anxiety, emotions, depression, among others, as predictive factors [19,20,22–24,26]. The use of multiple questionnaires and scales could hinder their application due to the complexity in obtaining this information and does not allow universal use because these scales and questionnaires were designed and validated to be used in specific populations. In terms of obstetric predictors, only three authors have included this type of clinical variable. Concerning these models, we agree on the inclusion of the variable "type of delivery" as a predictive factor of PTSD risk [9,18,23]; specifically, instrumental delivery [9] and emergency cesarean section [9,18] as they present forms of childbirth with the greatest risks. Although multiple studies associate the type of delivery with the risk of PTSD [7–15], and the presence of perineal tears [13,18,34], no predictive models have been developed that include these.

Regarding neonatal variables as factors that influence the risk of PTSD, various variables related to the newborn were identified, including the newborn's hospital admission [9]. Along similar lines, the risk was also related to the lack of skin-to-skin contact and formula feeding. Although these variables have not been included in other models, they have been related to an increased risk of PTSD in other studies [2,7,15,35]. Although these three variables are related to each other, the authors believe that they also have a partially independent effect. In the first place, not all hospitalized children stop skin-to-skin contact, as in many cases, admission occurs several hours after birth. Second, many women whose children are hospitalized continue to breastfeed despite the great obstacle it poses. The predictive model of Fairtbrother et al. [18] also includes low birth Apgar scores as a factor. In our sample, this variable was not assessed.

Another variable included in our model B was the perception of respectful treatment by healthcare professionals toward women. This aspect is closely related to the concept of obstetric violence and has not been evaluated in other predictive models, despite the existence of publications that identify a relationship between the treatment received by healthcare professionals during childbirth care and the presence of PTSD [36,37]. This aspect takes on particular relevance as the World Health Organization [38] and the United Nations [39,40] report an upward trend in women who perceive inadequate treatment during childbirth care.

Finally, the support provided by the partner plays a relevant role in the risk of PTSD, in such a way that women who perceived that their partners supported them during pregnancy, childbirth and the postpartum period had a lower risk of PTSD, coinciding with the model of Czarnocka and Slade based on a study carried out with 264 women [27].

### *Strengths and Limits*

One of the potential limitations of this study was that the observed prevalence of PTSD risk was high compared to other studies. In a systematic review with a meta-analysis carried out by Yildiz et al., average rates of 4.0% were found overall (95% CI: 2.77–5.71), and 18.5% (95% CI: 10.6–30.38) in women at risk [5]. The higher prevalence of our sample can be attributed to the use of a screening tool (PPQ) as we did not diagnose PTSD; instead, the risk of presenting PTSD was estimated.

Another limitation of the study was that it was carried out in a population residing in Spain, and even though the validation results were good, they need validation in other countries and cultural contexts. Regarding strengths, in addition to satisfactory predictive capacity, these models have other positive characteristics, such as including only five variables (parsimony principle), using variables that are usually recorded in medical records, and having a justified relationship with the risk of PTSD. We should also highlight that the model was validated in a population different than the one used to create the models, and they also had different prevalences for PTSD risk. These differences are interesting for the extrapolation of

results; this validation cohort could almost be considered external validation. Additionally, creating two predictive models may be useful for clinicians because it will expand application possibilities. For example, in situations where verbal contact with patients is not possible, model A, based on clinical variables recorded in medical records, could be used. While when verbal contact with the patient is possible, model B would be the instrument of choice. In particular, this tool is especially useful for professionals who have initial contact with women after childbirth. These professionals can use this tool as screening to identify patients who require further evaluation by more specialized professionals in the field of postpartum PTSD, such as psychologists and psychiatrists.

### **5. Conclusions**

In short, two predictive models formed by clinical variables and perceptions of support from their partner and the care received from health professionals presented adequate predictive capacities to predict the risk of postpartum PTSD both in the referral cohort and in the validation cohort. The model of choice includes the woman's perceptions of support received from her partner and the relationship with healthcare professionals. These models can help identify women at increased risk for postpartum PTSD, increasing the early detection of this increasingly prevalent problem. On the other hand, they can also be useful in primary prevention if health policies are applied that reduce risk factors such as cesarean delivery and inadequate treatment by health professionals and encourage other factors such as skin-to-skin contact and breastfeeding.

**Author Contributions:** Conceptualization, S.M.-V. and J.R.-A.; methodology, M.D.-R. and J.M.M.-G.; formal analysis, A.H.-M.; writing—original draft preparation, S.M.-V. and J.M.M.-G.; writing—review and editing, S.M.-V., M.D.-R. and J.M.M.-G.; supervision, J.R.-A., J.M.M.-G. and A.H.-M.; project administration, A.H.-M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Appendix A**

**Table A1.** Modified perinatal post-traumatic-stress questionnaire symptoms (modified PPQ).


<sup>\*</sup> Notes: the response and weight of scores for each question: (0) nothing; (1) once or twice; (2) sometimes; (3) often, but less than 1 month; (4) often for more than a month. The clinical range for high-risk mothers is established at 19 or more points.

### **References**




**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Spanish Nursing Students' Attitudes toward People Living with HIV**/**AIDS: A Cross-Sectional Survey**

**María Adelaida Álvarez-Serrano 1, Encarnación Martínez-García 2,3, Adelina Martín-Salvador 4,\*, María Gázquez-López 1,\*, María Dolores Pozo-Cano 2, Rafael A. Caparrós-González 2,5 and María Ángeles Pérez-Morente <sup>6</sup>**


Received: 25 August 2020; Accepted: 19 November 2020; Published: 22 November 2020

**Abstract:** Human immunodeficiency virus (HIV) infection is still a public health issue. Human immunodeficiency virus/acquired immune deficiency syndrome (HIV/AIDS) creates, in society, stigmatizing attitudes, fear, and discrimination against infected people; even health professionals do not feel trained enough to adequately take care of these patients, which affects the quality of care provided to such patients. The purpose of this study was to explore nursing students' attitudes and other related factors toward people with HIV/AIDS, as well as their evolution in subsequent academic years. A cross-sectional study was performed with students in four academic years from four Spanish health sciences institutions (*n* = 384). Data were collected voluntarily and on an anonymous basis, utilizing the "Nursing students' attitudes toward AIDS" (EASE) validated scale. The students' attitudes toward people with HIV/AIDS were relatively positive, with a total mean EASE value of 85.25 ± 9.80. Statistically significant differences were observed according to the academic year (*p* = 0.041), in 4 out of 21 items of the scale and among students with no religious beliefs. By adjusting every variable, only the weak association with religion was maintained (*p* = 0.045).

**Keywords:** attitudes; HIV/AIDS; students; nursing

### **1. Introduction**

Human immunodeficiency virus (HIV) infection is still an international public health problem [1,2]. In 2018, an estimated 37.9 million people worldwide were infected, of which 24.5 million people were treated with antiretroviral medications by mid-2019 [3]. In this sense, the World Health Organization, within the Sustainable Development Goals for the year 2030, included as a global goal to reduce HIV infection to zero contagions, zero discrimination, and zero deaths caused by the virus [2,4].

Today, particularly in developed countries, people with HIV live longer and experience the disease as a chronic condition, the most common comorbidities being cardiovascular diseases, kidney diseases, psychiatric disorders, and neoplasias, mainly associated with aging [3,5,6]. Currently, having HIV/AIDS still has a significant impact on the economic, political, legal, and social levels [1,2]. Regarding the latter, it has been documented that HIV/AIDS generates, among people, stigmatizing attitudes, fear, and discrimination against infected persons [7].

Furthermore, with the passing of time, as the disease settles in society, knowledge and education about HIV/AIDS have been reduced [8]. In the field of health, several authors state that even healthcare workers who are more exposed to work accidents with biological materials [9], such as nursing students and professionals, do not feel prepared enough to treat such patients adequately [6,8,10,11]. Variables such as ideology, age, stress, and negative opinions on homosexuality affect healthcare professionals' attitudes toward people with HIV/AIDS [1]. In addition, the lack of knowledge and the existence of erroneous concepts as regards transmission of HIV not only foster stigmatizing attitudes and fear of contagion, but they also impact clinical practices related to HIV/AIDS and the quality of the care received by seropositive patients [6,10,12–14]. In relation to these attitudes, in a study carried out by Valencia et al., it was reflected that women with HIV felt that they were mistreated by health professionals and that their rights to confidentiality and privacy were violated [15], which undoubtedly undermines legislation and ethics of care [12,13]

With respect to the assessments of HIV/AIDS patients performed by nursing students, there is research that affirms that these negative evaluations are related to stigmas related to promiscuity, drug addiction, and homosexuality [8,16,17]. In order to change such perceptions, many research contributions have shown that an increase in knowledge in this area reduces fear and anxiety [6,11].

Although it is reasonable to think that nursing students' negative attitudes toward people with HIV/AIDS will diminish as their academic level increases, there is a discrepancy in scientific literature regarding such a hypothesis [1,18,19]. Therefore, the objective of this study is to explore nursing students' attitudes and other related factors toward HIV/AIDS and their evolution in subsequent academic years.

### **2. Materials and Methods**

### *2.1. Data Collection*

An online, cross-sectional survey was conducted anonymously by self-administration. The participants were nursing students from four Spanish health sciences, institutions with a four-year-long study program of 240 credits. Although these four institutions are located in different geographical areas, student´s sociocultural levels of similar. The collection of the questionnaires was administered simultaneously in the four institutions during one month of the 2019–2020 academic year. It was sent through the teaching support platform, and participation was voluntary outside of class hours through the teaching support platform. The members of the research group invited all students enrolled across the four academic years to participate. After the first dissemination of the questionnaire, a reminder was sent to those students who had not yet answered the survey.

### *2.2. Measurement Tool*

For data collection, the "Nursing students' attitudes toward AIDS" (EASE) one-dimensional scale was employed, the Spanish version of which was initially created and validated by Tomás-Sábado (Appendix A) [20]. This tool showed good internal validity in the identification of nursing students' attitudes toward HIV/AIDS in Spain (Cronbach's alpha coefficient = 0.779). This scale comprises 21 items with Likert-type responses ranging from 1 = totally agree to 5 = totally disagree. The score for each item depends on its directionality. Items 3, 5, 7, 8, 11, 14, 15, and 21 are scored from 5 to 1, 5 being totally agree and 1 being totally disagree, and items 1, 2, 4, 6, 9, 10, 12, 13, and 16–20 are scored from 1 to 5, 1 being totally agree and 5 being totally disagree. The scale's maximum score is 105, which indicates the most positive attitudes, and the minimum score is 21, which indicates the most negative and preconceived attitudes. For each participant, the scale's global value was calculated by adding each item's score (with a total of 21 scores). The original scale was used in most of the studies analyzed [1,21,22]. Additionally, the following sociodemographic variables were considered:

age, sex (male or female), academic year, marital status (in a relationship or not in a relationship), sexual orientation (heterosexual or non-heterosexual), and religion (with religious beliefs or with no religious beliefs).

### *2.3. Data Analysis*

Prior to data analysis, the internal consistency of the scale was assessed by determining Cronbach's alpha, with a satisfactory result of 0.776. Next, a descriptive analysis of the entire sample was carried out by means of absolute and relative frequencies and the calculation of measures of central tendency (i.e., median and standard deviation) according to the quantitative or qualitative nature of the study's variables. The scale's normality of values was explored, and as a result of not being possible to verify their parametricity, for the inferential statistics, a nonparametric analysis was adopted. Differences among the scale's values according to the sociodemographic variables and academic year categories were analyzed through the Mann–Whitney *U* test, and for variables with more than two factors, the Kruskal–Wallis *H* test was employed. In cases of significant differences among academic years, post-hoc multiple comparisons were carried out through Bonferroni correction. The effect size was calculated using Kerby´s formula [23]. To establish a correlation between age and the final score of the EASE scale, Spearman's Rho was used. Finally, the influence of the sociodemographic variables and academic year on the EASE scale's final score was explored by using a multiple linear regression model through the "introduce" method, in which the academic year, re-categorized into three dummy variables, was included, taking the first year as reference. Data were treated with the Statistical Package for the Social Sciences (SPSS) program, version 25, (IBM, New York, NY, USA, for Windows).

### *2.4. Ethical Considerations*

This study complies with the good clinical practice regulations, as stated in the European Directive 2001/20/CE and Law 14/2007, of 3 July, on biomedical research. Treatment of personal data in health research is governed by Organic Law 3/2018, of 5 December, on Data Protection and Guarantee of Digital Rights. Every participant checked a box indicating consent to participate in the study.

### **3. Results**

The sample included 384 students, with the response rate being 18%. Twenty-nine percent of the sample (*n* = 115) were in their first year of study. With regards to sex, 84% were women, and the total mean age was 23 ± 6.7 years. In terms of academic year, the mean age was significantly higher for those in their fourth year of study (*p* < 0.001). With respect to students' sexual orientation, 337 (87.8%) were heterosexual, a higher proportion of which was observed in the third year of study (*p* = 0.012). Regarding relationship status, 207 (53.9%) stated they were not in a relationship, the fourth year of study being the year with a higher number of students in a relationship (*p* = 0.008). Of the total number of students, 265 (69%) considered themselves as having religious beliefs (Table 1).

The total EASE score in the analyzed sample was 85.25 ± 9.80. Table 2 shows such values according to sociodemographic variables and academic year. Among the second-year students and among those with no religious beliefs, the highest statistically significant scores were observed (*p* = 0.041, and *p* = 0.006, with an effect size of 0.17, respectively) (Table 2). Significant causes according to academic year were identified in the second and first year of study (*p* = 0.050 with an effect size of −0.22) (results not shown).

The scale items that reached a higher mean score were P1 ("AIDS does not affect heterosexual couples") with 4.62 ± 1.01 and P11 ("being an AIDS carrier should not be a barrier to accessing education and employment") with 4.68 ± 0.87. Such items show the highest percentages of "disagree" and "agree" answers, with 83.1% and 83.9%, respectively (results not shown).


**Table 1.** Sociodemographic variables of the sample according to academic year.

M = mean; SD = standard deviation; ns = *p* > 0.05; \* = *p* ≤ 0.05; \*\* = *p* ≤ 0.01.

**Table 2.** Descriptive statistics of the EASE scale according to the sociodemographic variables and academic year.


M = mean; SD = standard deviation; ns = *p* > 0.05; \* = *p* ≤ 0.05; \*\* = *p* ≤ 0.01.

With respect to the questions asked in the scale, there were significant differences among academic years in 4 out of the 21 items included, namely in item 2 ("fetuses infected with AIDS should be aborted"), 3 ("there is no risk arising from AIDS carriers' use of restaurants and pubs"), 4 ("seropositive women should not be allowed to get pregnant"), and 16 ("in hospitals, AIDS carriers should not share a room with non-infected persons") (Table 3). The causes for the significance among academic years in item 2 were observed between the first and second years (*p* = 0.041), between the first and third years (*p* = 0.0001), and between the first and fourth years (*p* = 0.021); in item 3, between the first and second years (*p* = 0.022); in item 4, between the first and third years (*p* = 0.018); and in item 16, between the first and second years (*p* = 0.014) (results not shown). In all cases, the first-year score was lower than that obtained in the subsequent years.


**Table 3.** Descriptive statistics of the EASE scale items, according to academic year.

M = mean; SD = standard deviation; ns = *p* > 0.05; \* = *p* ≤ 0.05; \*\* = *p* ≤ 0.01. Definitions of P1–21 can be found in Appendix A.

In the multiple linear regression model, it was observed that by adjusting all variables, only religion was associated with the scale values, with those stating no religious beliefs showing more favorable attitudes toward HIV/AIDS ((95% CI: 0.05–4.26); *p* = 0.045) (Table 4).


**Table 4.** Multiple linear regression model for the EASE scale.

ß = Beta coefficient; \* = *p* ≤ 0.05; CI: confidence interval.

### **4. Discussion**

The attitudes toward people with HIV/AIDS of the students that participated in the study were relatively positive or favorable, since they indicated a mean EASE score above 85 points, as recommended by Tomás-Sábado and Aradilla-Herrero [1]. This result coincides with that reported in several similarly designed studies [1,21,24], reinforcing the idea that nursing, as a profession, through its humanistic approach, generates among future professionals positive attitudes toward an infection such as HIV or a disease such as AIDS [25].

According to the academic level, some differences were found in the EASE scores, the first academic year's score being lower than those obtained in subsequent years. Such a result coincides with that published by Leyva-Moral et al., where the lowest percentage of positive attitudes was observed among first-year students [25]. Nevertheless, differences in scores were found between the second- and first-years, but not in subsequent years with respect to the first one, which contradicts previous studies regarding an increase in favorable attitudes as the academic level increases [1,18].

Some authors have stated that when facing problems that are especially sensitive at a social level and that are ideologically charged, such as HIV and AIDS, an increase in knowledge alone is not enough to change people's beliefs or attitudes [22,24–27]. Furthermore, contact with actual patients occurring in subsequent years during clinical practice, on the contrary, could increase fear of contagion among students, thus obstructing an advance toward higher levels of disease tolerance, which is an issue that requires further investigation [22].

The median age of the analyzed sample was in accordance with those published in similar studies [1,9,21,22,24,25,28], corresponding to young adults, although this variable was not associated with attitudes toward HIV/AIDS. However, the observed tendency, that as age increased, attitudes became more negative, coincides with the results of other studies that confirmed such an association [25,29–31].

Despite the fact that the analysis was reconducted to assess the EASE scale's internal consistency by obtaining a Cronbach's alpha satisfactory result, some items showed a change opposite to the one expected. That is, it was expected that in cases of positive assertions the level of agreement would increase and in cases of negative assertions, the level of disagreement would increase. However, in terms of item 14, "persons infected with AIDS should be considered victims of the social system," 66.4% of students disagreed or totally disagreed, and for item 10, "seropositive persons should be identified as such," the opposite occurred, i.e., 30% of students agreed or totally agreed. Serrano-Gallardo and Giménez-Maroto already identified such contradictions due to a certain degree of ambiguity in the wording of these items [9]. Although some of the items on the EASE scale could be considered outdated at present due to clinical and cultural changes, we believe that the scale allows general and non-specific attitudes toward HIV/AIDS among nursing students to be explored [25].

Having no religious beliefs was the only analyzed variable that showed an association with better attitudes toward HIV/AIDS, regardless of the influence of the other variables. This result coincides with the findings of other studies, where nursing professionals who practiced a religion, or students who believed religion had a significant role in their lives or firmly thought that the disease etiology was due to divine retribution, showed more stigmatizing attitudes toward their HIV/AIDS patients [20,32–35].

### *Limitations*

This study has some limitations. Although the number of students who agreed to answer the questionnaire was higher than the figures published in other studies [1,9,21,22,24,25,28], we believe it could be improved, since students who are more aware of the subject being analyzed could be overrepresented. It is also possible that given the sensitive nature of the subject addressed, when asking for personal opinions, there could have been a social desirability bias that was resolved by guaranteeing the anonymous nature of all participants. Finally, and as we have commented previously, the clinical and cultural changes that have occurred in the last decade with respect to HIV/AIDS could mean that some items on the scale are out of date, although this possible limitation is minimized by establishing a general analysis of attitudes toward HIV/AIDS.

Despite the limitations referred to above, this study contributes current information on the persistence of false opinions regarding HIV/AIDS in a varied sample of nursing students, since four health sciences institutions were analyzed, which were located in different geographic areas but shared the same curriculum. Results may be useful to rethink the type of training our students receive and about the need, therefore, to establish educational content that fosters closer humanization of care and the reduction of HIV/AIDS-associated stigma.

### **5. Conclusions**

Nursing students' attitudes toward HIV/AIDS are relatively positive and get better following the first academic year. When considering all variables of the study, the influence of the academic level waned and the association with religion was maintained. Thus, students with no religious beliefs showed more positive attitudes toward this problem. An improvement in the level of knowledge and experiences among nursing students does not seem to be enough to cause a change of attitude, which may reflect the strong ideological charge that still leads to this health issue today.

It is therefore necessary to carry out a critical analysis of teaching strategies in current nursing programs in order to achieve not only better and more in-depth knowledge of the disease and its modes of transmission, but also a change of attitude, free of negative preconceptions toward HIV/AIDS. This approach would result in a better quality and humanization of the care provided to people with HIV/AIDS. It would also be pertinent in future research to reformulate some of the items on the EASE scale, adapting them to the current reality of people with HIV/AIDS.

**Author Contributions:** Conceptualization, M.G.-L.; methodology, M.A.Á.-S., E.M.-G., and A.M.-S.; investigation, M.A.Á.-S., E.M.-G., A.M.-S., M.G.-L., M.D.P.-C., R.A.C.-G., and M.Á.P.-M.; data curation, M.A.Á.-S. and E.M.-G.; visualization, A.M.-S.; writing—original draft preparation, M.A.Á.-S., E.M.-G., A.M.-S., M.G.-L., M.D.P.-C., R.A.C.-G., and M.Á.P.-M.; writing—review and editing, M.A.Á.-S., E.M.-G., A.M.-S., M.G.-L., M.D.P.-C., R.A.C.-G., and M.Á.P.-M.; supervision, E.M.-G. and M.Á.P.-M. All authors read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** To all students who participated in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Appendix A**

Scale of attitudes toward AIDS for nursing.


### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Ethnic Disparities in Utilization of Maternal and Child Health Services in Rural Southwest China**

### **Chaofang Yan 1, Charuwan Tadadej 1,\*, Kanittha Chamroonsawasdi 2, Natkamol Chansatitporn <sup>3</sup> and John FC Sung <sup>4</sup>**


Received: 4 November 2020; Accepted: 17 November 2020; Published: 19 November 2020

**Abstract:** Background: Studies in China on ethnic disparities in access to health care in remote and rural population remain insufficient. This study aimed to assess the disparities in utilization of maternal and child health (MCH) services, including antenatal care (ANC), hospital birth, child growth monitoring, and immunization compliance between Han and ethnic minority women in Yunnan Province. Methods: A multi-stage sampling scheme was used to randomly recruit women from 40 townships in 14 remote prefectures of extremely remote areas in Yunnan. From birth records, we identified and recruited 303 Han women and 222 ethnic minority women who had given birth to a child within 3 years for an interview. Results: Overall, 96% of women used the ANC checkups and more than 95% had infants born in hospitals. However, the proportion of women compliant with early ANC visits (having antenatal care in the first trimester) was 22.5% lower in minority women than in Han women (61.3% vs. 83.8%, *p* < 0.001) with an adjusted odds ratio (aOR) of 2.04 (95% confidence interval (CI) of 1.13–3.66) for the minority group. The proportion of children under one year old with immunizations completed in a timely manner was also lower in minority families than in Han families (80.2% vs. 86.8%, *p* < 0.05) with an aOR of 1.99 (95% CI = 1.16–3.40). Conclusions: Ethnic disparities remain in utilization of early ANC visits and timely immunization completion for newborns. Ethnic minority women tended to lag behind for both. Further intervention should focus on assisting minority women living in extremely rural areas to comply with the MCH policy. Culturally-sensitive policies and skills are needed, and priority should be given to improve utilization of early ANC and timely immunization completion.

**Keywords:** ethnic disparity; utilization; maternal and child health services; China

### **1. Introduction**

Women and children are at a crucial position in social development. The United Nations (UN) has highlighted strategies to promote the health of women and children in the Millennium Development Goals (MDG) 4 and 5 from 2000 to 2015, and in Sustainable Development Goal (SDG) 3 from 2016 to 2030. Substantial progress in maternal and child survivals has been made since the 1990s [1]. Timely maternal and child health (MCH) services utilization is vital to reduce both the maternal and child deaths [2–5]. The mediation of the MDG has increased utilization of antenatal care (ANC) services to 44% and assisted delivery to 12%. The UN efforts have thus reduced maternal deaths by 44% from

283.2 per 100,000 live births in 1990 [4,6]. The coverage for priority interventions among disadvantaged populations in a country is an indication of the strength of the health system [7].

Reducing MCH disparity has been a crucial issue across the whole nation in China, a developing country with the largest population in the world [8–11]. The Chinese government has taken a series of actions in order to increase the coverage of the interventions of MCH among disadvantaged populations in less developed regions [12–14]. For example, from 1996 to 1999 the government started a World Bank loan program, the Poverty Alleviation Fund for Maternal and Child Health. The program aimed to reduce maternal mortality rate (MMR) and infant mortality rate (IMR) by assisting 5% of poor families with their payment of the MCH services in five western provinces [12,15]. From 2000 to 2008, the program of Reducing Maternal Mortality and Eliminating Neonatal Tetanus was launched to reduce MMR, mainly targeting 378 counties in rural areas [14]. The Basic Public Health Service (BPHS) project, launched in 2009, is one of the most effective policies for providing and promoting residents with equal access to basic public health services. This government-funded project includes nine service categories, including one reform to improve MCH services [13,16]. After years of effort, the urban–rural disparity of MMR in China has been greatly narrowed. The urban-to-rural ratio of MMR reduced from 1:2.2 in 1990 to 1:1.3 in 2018 (15.5 vs. 19.9 deaths per 100,000 livebirths) [10]. The western part of China is less developed than the eastern part of China; the eastern-to-western region ratio of MMR reduced from 1:4.7 in 1996 to 1:2.3 (10.9 vs. 25.2 per 100,000 livebirths) in 2018. The project also reduced the urban–rural ratio of the under-5 mortality from 1:3.4 in 1991 to 1:2.3 in 2018 (4.4 vs. 10.2 per 1000 livebirths). The gap in under-5 mortality rates between east and west regions was reduced to 8.5‰ in 2018 from 66.5‰ in 1991 [10].

The MCH disparities between east and west China, or between urban and rural, have been explored and well documented. However, studies evaluating the ethnic disparities with robust methods remain severely insufficient in China [17]. The Lancet–Lowitja Institute Global Collaboration study has shown that Yunnan province and Tibet Autonomous Region in China have not yet reached the UN SDG for underserved populations [18]. Some studies related to ethnic populations have reported health coverage and health outcomes based on the whole population from the Autonomous Region rather than classified by ethnic type [8,19]. No study has compared differences of MCH between Han and ethnic minorities at a provincial level, because the medical records and vital statistics were rarely classified by ethnicity in the official health statistics system. Most studies have emphasized the improvement of health data collection for ethnic minorities in China [17,18,20].

China is populated with multiple ethnic groups with high cultural and language diversities; these consist of the Han majority accounting for 91.5% of the total population and 55 minority groups accounting for the rest of the population (8.5%). The 2010 census estimated that 114 million people in China were ethnic minorities. Three-quarters (71.4%) of all ethnic minorities live in the remote southern and western regions of China, including Yunnan Province [21].

Yunnan Province is located in Southwestern China, bordering Myanmar in the west, and Laos and Vietnam in the south. It is one of the poorest remote provinces and has the highest ethnic diversity in China (Figure 1), including 8 ethnic autonomous prefectures and 29 ethnic autonomous counties. Among 56 recognized ethnic groups in China, Yunnan Province has recognized at least 25 ethnic minority groups, making up 33.6% of the 48 million population in the province in 2018. Yunnan is situated in mountainous land, with only 6% of plain areas suitable for cultivation [22]. The provincial authority has endeavored to carry out the ambitious provincial goal since 2010 for reducing MMR to 25/100,000 and IMR to 11 per 1000 live births by 2020, then reducing the MMR to 12/100,000 and the IMR to 5 per 1000 live births by 2030, based on the MMR 37.3/100,000 and the IMR 15.2 per 1000 live births in 2010 [23,24]. For better utilization of MCH services, pregnant women are officially encouraged to complete at least five ANC with the first ANC in the first trimester, hospital delivery, having their children under one year old checked for growth monitoring at least four times, and getting their children vaccinated in a timely manner according to the National Immunization Program [25].

**Figure 1.** Ethnic diversity of Yunnan Province, China.

As a province with typical ethnic minorities, no study has yet compared the compliance of MCH care services between Hans and ethnic minorities in Yunnan. Therefore, we conducted a survey in remote rural areas in the province to assess the disparities in utilization of MCH services between Han and the ethnic minority mothers.

### **2. Materials and Methods**

### *2.1. Study Design and Participants Recruitment*

This study was a cross-sectional survey with a comparison between Han and ethnic minorities. Women who had given birth within the past three years at the time of the survey and had been residing in the selected township in Yunnan for at least three years were included in the study. This inclusion criteria ensured that these women could provide the health service information for both mothers and children, such as the information of growth monitoring and immunization. Women who worked and used health services out of the selected township were excluded to show the effect of ethnicity in the same context. A conceptual framework, adapted from the model of access to medical care by Andersen [26], was used to examine the access to MCH services and to analyze the factors affecting access to MCH services.

The sample size was calculated based on detecting a significant proportional difference of hospital births between ethnic minorities and the Han group at a 5% level with 80% power of a two-sided test. Previous studies in Yunnan Province have reported that the hospital birth rates were 81% (512/631) in minority women and 91% (476,132/523,223) in Han women [27,28]. These data were used to calculate the minimum sample size being at least 188 subjects in each group for this study.

A multi-stage sampling method was used to select the study sample. Firstly, we excluded the provincial capital Kunming to eliminate the effects of economic inequality and Nujiang prefectures that are difficult to reach by automobile transportation. For the rest of the 14 prefectures, 1 county was randomly selected from each prefecture. In each selected county, after excluding the township in which the county government was located, we randomly selected a township. Based on the birth

records from the township hospital in the selected township, 40 women with a birth within the past three years at the time of the survey were randomly selected from each township, resulting in a total of 560 women selected as potential participants. For participants with more than one pregnancy and delivery, information from the last pregnancy was collected and used.

This study was approved by the ethics committee of the Faculty of Public Health, Mahidol University (COA No. MUPH 2014-214; 24 November 2014). All participants were clearly informed of their rights and any risks associated with participation. Verbal consent was obtained from all interview participants.

### *2.2. Study Procedure and Data Collection*

We applied the Andersen health behavior model to establish a structured questionnaire for data collection of the survey to investigate MCH services associated with predisposing (demographic and social) factors, enabling (economic) factors, and need (health outcomes) factors [26]. The questionnaire was reviewed by experts in MCH, social medicine, and health management.

The questionnaire asked for four types of information. The first type regarded predisposing factors including the general characteristics of respondents, comprising age of the woman, age of the child, sex of the child, parity, ethnicity, language ability, education, and knowledge on MCH. The second part regarded enabling factors: the family income, family size and health insurance, the usual health facility, and travel time to the nearest health facility for childbirth. The third part regarded need factors: the woman's health status during the perinatal period, previous obstetric problems, child health status, preterm delivery, and birth weight of the child. The fourth part asked for information about MCH utilization, including ANC utilization, timing for ANC visits, hospital birth, delivery type, child growth monitoring services, timing for child growth monitoring visits, and immunization services.

Before the actual data collection, the questionnaire was pre-tested on 20 women from villages of the study area. The reliability of scaling variables was assessed by calculating Cronbach's coefficient alpha (α = 0.716), which indicated that the measures had fairly high levels of reliability. The questionnaire was administered in person by trained interviewers at each participant's household. We trained 32 medical students who could speak the local dialog as interviewers. A two-day interview workshop was given to them before conducting the survey.

After excluding 22 women who refused interviews and 13 women who could not complete the questionnaires, information was obtained from 525 households in total. The ethnic minorities included 15 ethnicity groups of Yi, Hani, Thai, Zhuang, Hmong, Bai, Yao, Hui, Lisu, Va, Lahu, Nu, Jingpo, Achang, and Paijiao.

### *2.3. Statistical Analysis*

Eligible questionnaires were edited and coded, and the data were entered and processed using SPSS version 19. Data analysis first compared distributions of participants' characteristics in predisposing factors, enabling factors, and need factors between Han and ethnic minority participants. Utilization of MCH care (ANC visit, infant delivery, growth monitoring, and immunization) was presented by numbers and proportions. Chi-square test was used to examine the difference between the two groups. Logistic regression analysis was further used to calculate odds ratio (OR) and 95% confidence interval (CI) of inadequate utilization of MCH care for the ethnic minority women, compared to Han women. The adjusted odds ratio (aOR) was estimated with a multivariable by three models: after controlling for predisposing factors, after controlling for predisposing and enabling factors, and after controlling for predisposing, enabling, and need factors.

### **3. Results**

### *3.1. Socio-Demographic Profile*

With response rates of 97.1% and 89.5% in Han and ethnic minorities, respectively, 303 Han women and 222 minority women completed the questionnaire interviews. The average ages of participants in both Han and ethnic minority groups were approximately similar at 26 years old. The Han group had higher education and better Mandarin language ability, with higher income, than the ethnic minority group. Ethnic minority women were more likely (69.4%) to be married to ethnic minority husbands (refer to Husband's Ethnicity type) than were Han women (8.3%), and they had a larger family size than Han women did. Ethnic minority women needed more travel time to reach their nearest health care facility and perceived that their children had a poor health status. (Table 1).


**Table 1.** Sociodemographic characteristics of study subjects in Yunnan.

<sup>1</sup> CNY, Chinese yuan, 1 USD = 6.2 CNY; <sup>2</sup> NCMS, New rural cooperating medical scheme.

### *3.2. Comparison of MCH Utilization*

More than 90% of women had at least one ANC visit and nearly 60% made five visits during their pregnancy (Table 2). The Han women had a higher adherence to the recommended 5-visit schedules than did ethnic minority women by making their first ANC visit in the first trimester (58.1% vs. 43.2%, *p* < 0.001). However, making an early ANC visit was 22.5% lower in minority women than in Han

women (61.3 vs. 83.8%). Over 95% of participants had their infants delivered at health institutions, with a higher incidence of Caesarean delivery in Han women (26.1% vs. 20.3%, p 0.12). However, a lower proportion of ethnic minority women completed the required vaccinations for children within the first year of age than Han women (80.2% vs. 86.8%, *p* < 0.05).

**Table 2.** Comparison on utilization of maternal and child health care between Han and ethnic minority women.


<sup>1</sup> MCH, maternal and child care; <sup>2</sup> ANC, antenatal care; <sup>3</sup> Early ANC visit, having antenatal care in the first trimester.

### *3.3. Odds Ratio of MCH Services Utilization*

Table 3 shows that, compared with the Han women, the minority women were at higher risks for inadequate antenatal care, with significant unadjusted ORs of 3.59 (95% CI = 1.37–9.40) for failing to make at least one ANC visit during the pregnancy, of 3.28 (95% CI = 2.00–4.75) for failing to make an early ANC visit in the first trimester, and of 1.82 (95% CI = 1.28–2.58) for failing to make at least 5 ANC visits with an early ANC visit in the first trimester. However, only the risk of failing to make an early ANC visit remained significant after controlling for predisposing, enabling, and need factors, with an aOR of 2.04 (95% CI = 1.13–3.66). The ethnic minorities were also at a higher risk of not receiving vaccinations for children with all doses in a timely manner, with an aOR of 1.99 (95% CI = 1.16–3.40) after controlling for all covariables.


**Table 3.** Ethnic minority women compared to Han women odds ratio (OR) of inadequate MCH utilization and adjusted odds ratio (aOR) controlling for predisposing,enabling,andneedfactors.

husband education, Mandarin language ability, income, travel time, and perceived child health, which were significant predictors. \* *p* < 0.05.

### **4. Discussion**

This study aimed to assess the disparities in the utilization of MCH services between Han and ethnic minority mothers in remote rural areas. Theoretically, minority women and Han women could receive similar care even in the remote rural areas. We found that gaps of using MCH services still existed in ANC utilizations, significant for the early ANC visit in the mountainous rural region of Yunnan. The pregnant women from minority families were less likely than Han women to make the first ANC within the first 12 gestational weeks. Previous studies have rarely measured the disparities between the Han and ethnic minority populations in pregnancies, although some studies have reported the rates of the early ANC visits related to ethnic minority population in other provinces [29–33]. The timing of the first antenatal care visit is an important compliance measure to ensure optimal pregnancy outcomes for women and children [4,34]. The early antenatal care visit also has been used as an indicator to measure the quality of MCH services since launching the health care reform in 2009 in China [25]. This finding might imply the need of exploring the cultural factors for early ANC visits among ethnic minority women. Our study also showed the need to develop additional culture-sensitive policy to eliminate inequality and deliver cost effective MCH interventions.

Further, compared to Han mothers, the ethnic minority mothers were also less likely to have their children immunized in a timely manner, which is crucial for preventing childhood diseases. Studies have consistently reported a poor compliance with immunization services for children of ethnic minorities in China [17,35–37]. With an investment of USD \$120 billion, the universal health insurance coverage increased to 95% of the population in 2019 from 30% in 2003 [38]. The economic barriers to accessing ANC, hospital births, and immunization services for rural populations have been removed, yet disparities were still observed.

This study found that hospital birth and caesarean delivery between Han and ethnic minority women were not significantly different. This was likely because of the compliance to the UN's MDG and SDG strategies to promote the health of women and children. In China, like many developing countries, the economic level plays an important role in health service accessibility, especially in remote rural areas where many ethnic minorities live. Over the past 20 years, the Chinese Government has introduced several strategies aimed specifically at reaching underserved populations in rural areas. In 2003, a voluntary health insurance program, New Cooperative Medical Scheme (NCMS), was launched for rural residents. This heavily subsidized program has increased outpatient and inpatient services utilization by reducing the health service cost with a sliding fee discount of 30% to 80%, varying by region annually [39,40]. In addition to the major support from the central authority, some of the local health administrations also received support from international organizations such as the United Nations International Children's Emergency Fund (UNICEF) [12,15,41]. The administrations were thus capable of conducting pilot studies using poverty alleviation funds to help low income pregnant women adhere to MCH services in a timely manner in some identified poverty-stricken counties.

Similarly, there were also no significant differences in using service of growth monitoring for children under 1 year old between Han and ethnic minority families in this study. The BPHS project of new medical reform policies has made a great contribution to promoting equal access to basic public health services since it was launched in 2009 [42]. This government-funded project included nine service categories, one of which was to improve infant cares. The project encouraged all families having their children under 1 year old to complete "four times-routine examination at the 3, 6, 8, and 12 months of age" to monitor growth and development free of charge. Further, all MCH physicians are required to have standardized training and to deliver the same maternal services for rural and urban populations [25].

As a whole, we demonstrated a substantial improvement in utilization of MCH services among all women in this study compared with the utilizations in the 1990s, providing valid information on the improvement of MCH services in remote rural areas. Overall, 96% of pregnant women used the antenatal checkups at least once in the present study, compared to 58.07% in 1992 in the whole Yunnan province. The compliance with at least five antenatal checkups improved to 60.6% from 23.78% in 1992. The greatest improvement is that the hospital birth increased to 95.8% from only 30.33% in 1990. More importantly, vital statistics data showed the MMR had a great decline from 115.29 per 100,000 livebirths in 1992 to 17.72 per 100,000 livebirths in 2018, and the infant mortality rates declined from 46.38‰ in 1992 to 5.85‰ in 2018 in Yunnan Province [43,44]. All the policies such as the NCMs and BPHS contribute to this great progress. The Targeted Poverty Alleviation Project was recently launched to further increase accessibility to local health care without charge for families officially registered as Poverty-Stricken Households [45]. We anticipate this effort will further reduce the ethnic disparities in MCH services utilization in rural China. However, most current policies tend to invest more money to reduce health inequality but have less emphasis on cultural barriers, transportation barriers, and development of the health workforce in extremely rural areas [46,47]. Minority women living in these areas deserve further intervention if the highest ethnic diversity province wants to reach the goals of MMR of 12/100,000 and the IMR of 5 per 100 live births by 2030 [24].

This study has a few limitations. This study excluded women who had experienced an abortion or had children who died under the age of 3 years. Their utilization of MCH services is not clear. Among the 560 women randomly invited to participate in this study, 35 women were not interviewed. These women were less likely to be compliant with required MCH checkups, and their information was not included in this study. Furthermore, information on MCH services was collected from women within 3 years of the childbirth at the time of survey, the self-reported data throughout the period might vary slightly among women. However, interviewers conducted the survey following strict protocols. The response bias was likely unintentional and would not adversely affect the survey outcome.

### **5. Conclusions**

This study found that gaps of using MCH services between ethnic minorities and Han women exist in ANC utilization and timely immunization completion in China. Ethnic minority women tended to lag behind in utilization of early ANC, and in timely immunization completion for their children under one year old. However, there were no significant differences in hospital birth, caesarean delivery, and growth monitoring for children under 1 year old between Han and ethnic minority women in this study. China's political commitment to poverty reduction for compliance with the UN MDG and SDG might have contributed to the improvement. A heavily subsidized NCMS and the BPHS of new medical reforms might also play an important role in promoting equal access to MCH services. Minority women living in extremely rural areas deserve further intervention. Culturally-sensitive policies and skills are needed to target ethnic populations to improve their access to MCH services in order to continue decreasing the urban–rural and ethnic disparities. Priority should be given to improve utilization of early ANC and timely immunization completion.

**Author Contributions:** C.Y. designed the study, analyzed data, interpreted findings, and drafted the manuscript. C.T. provided supervision in the development of the study concept and design, advised on the scope of the paper, and provided critical revision of the manuscript. K.C. advised on the study theory and revised the manuscript critically. N.C. advised on sampling and data analysis methods and revised the manuscript critically. J.F.S. assisted in study design and data presentation and revised critically the manuscript for important intellectual content. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We are thankful to health officers, rural physicians and medical students from Kunming Medical University in Yunnan, China, for their great assistance in this study. This article is a fulfillment of the requirements for a Doctoral study in Public Health, Mahidol University, Bangkok, Thailand, supported by the grant of International Fellowships Program of Ford Foundation (Grant ID:15090343).

**Conflicts of Interest:** The authors declare that they have no competing interests.

### **Abbreviations**


### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Maternal Dietary Patterns during Pregnancy and Their Association with Gestational Weight Gain and Nutrient Adequacy**

### **Naomi Cano-Ibáñez 1,2,3, Juan Miguel Martínez-Galiano 2,4,\*, Miguel Angel Luque-Fernández 2,3,5, Sandra Martín-Peláez 1,2, Aurora Bueno-Cavanillas 1,2,3 and Miguel Delgado-Rodríguez 2,6**


Received: 2 October 2020; Accepted: 25 October 2020; Published: 28 October 2020

**Abstract:** Several epidemiologic studies have shown an association between Gestational Weight Gain (GWG) and offspring complications. The GWG is directly linked to maternal dietary intake and women's nutritional status during pregnancy. The aim of this study was (1) to assess, in a sample of Spanish pregnant women, the association between maternal dietary patterns and GWG and (2) to assess maternal dietary patterns and nutrient adequate intake according to GWG. A retrospective study was conducted in a sample of 503 adult pregnant women in five hospitals in Eastern Andalusia (Spain). Data on demographic characteristics, anthropometric values, and dietary intake were collected from clinical records by trained midwives. Usual food intake was gathered through a validated Food Frequency Questionnaire (FFQ), and dietary patterns were obtained by principal component analysis. Nutrient adequacy was defined according to European dietary intake recommendations for pregnant women. Regression models adjusted by confounding factors were constructed to study the association between maternal dietary pattern and GWG, and maternal dietary patterns and nutritional adequacy. A negative association was found between GWG and the Mediterranean dietary pattern (crude β = −0.06, 95% CI: −0.11, −0.04). Independent of maternal dietary pattern, nutrient adequacy of dietary fiber, vitamin B9, D, E, and iodine was related to a Mediterranean dietary pattern (*p* < 0.05). A Mediterranean dietary pattern is related to lower GWG and better nutrient adequacy. The promotion of healthy dietary behavior consistent with the general advice promoted by the Mediterranean Diet (based on legumes, vegetables, nuts, olive oil, and whole cereals) will offer healthful, sustainable, and practical strategies to control GWG and ensure adequate nutrient intake during pregnancy.

**Keywords:** maternal dietary patterns; pregnancy; gestational gain weight; offspring

### **1. Introduction**

Excess body weight during pregnancy is a public health concern owing to its high prevalence; increased risk of maternal diseases, such as gestational diabetes; and delivery complications [1]. Even though the current maternal guidelines stress the need for Gestational Weight Gain (GWG) control [2], the 2011 US pregnancy nutrition surveillance system has shown that up to 47% of pregnant women had excessive GWG and 23% had inadequate GWG according to the recommendations provided by the Institute of Medicine (IOM) [3].

GWG, or the total amount of weight gained in pregnancy, is a complex physiologic response to accommodate the natural responses to pregnancy, such as gestational fat deposition and fetal growth [4]. Several epidemiological studies have shown an association between GWG and offspring complications. Women who gain excessive weight during pregnancy (i.e., more than the amount recommended in guidelines) are more likely to have infants with high birth weight, premature delivery, and infants with an increased risk of developing childhood obesity [5]. Furthermore, excessive GWG is associated with an increased risk of maternal diseases, such as gestational diabetes mellitus and preeclampsia [6]. On the other hand, women with insufficient GWG are more likely to have infants with low birth weight and intrauterine growth retardation [7].

Nevertheless, GWG is a potentially modifiable risk factor because it is directly linked to maternal nutritional habits during pregnancy. Current research evaluating GWG according to dietary intake has focused on isolated foods [8,9] or nutrients [10,11] instead of the dietary pattern of food consumption. The analysis of dietary patterns could provide a better understanding of maternal dietary food intake and thus of women's nutritional status during pregnancy. Although a correct diet is essential to maintain an adequate nutritional status at all stages of life, during pregnancy, nutritional needs are increased in order to meet fetal requirements, especially for some micronutrients. In particular, essential micronutrients are necessary to prevent maternal and perinatal adverse health conditions. However, a deficient intake of essential micronutrients is commonly reported [12] and is associated with an increased nutritional vulnerability in pregnant women [13,14], specifically among those women with worse dietary food intake. Usually, this deficient intake is linked to occidental dietary patterns characterized by a high intake of meat or meat products; snacks; baked desserts; and sugar-sweetened beverages, providing large amounts of saturated fatty acids and simple carbohydrates as added sugars [15].

There is a scarcity of findings regarding dietary factors, nutritional adequacy, and GWG. Furthermore, inadequate GWG and nutrient intake during pregnancy has a negative impact on maternal, perinatal, and fetal health [16]. Therefore, we aimed (1) to assess the association of maternal dietary patterns with GWG and (2) to analyze the association between maternal dietary patterns and nutrient adequate intake according to GWG among Spanish pregnant women.

### **2. Methods**

### *2.1. Study Design, Settings, and Participants*

We designed a retrospective, observational study. Data were obtained from a cohort of pregnant women attending five hospitals in Eastern Andalusia (Spain). Women were recruited from 15 May 2012 through 15 July 2015. Eligible participants were women who resided in the referral area of the five hospitals located in the provinces of Jaen (2 hospitals), Granada (2 hospitals), and Almeria (1 hospital), who understood the Spanish language, gave birth to a single live newborn, and agreed to complete and return the Food Frequency Questionnaire (FFQ) after delivery assessing their dietary intake during pregnancy. After applying the inclusion criteria, 533 women were included in the study. Fifteen women refused participation, and 15 women were excluded because they presented energy intakes outside of predefined limits [17]. A sample of 503 women was analyzed in the current analysis (Figure 1).

**Figure 1.** Study Flow-Chart.

The ethics committees from the hospitals approved the study protocol. All women included in this study filled out informed consent and data treatment forms to enroll in the study, following the ethical standards of institutions where they were identified.

### *2.2. Data Collection and Outcomes*

Data were collected retrospectively on anthropometric measures and dietary food and energy intake. Deficient dietary patterns of food consumption and excessive GWG were identified based on IOM guidelines. Next, we evaluated the quality of diet, based on nutrient adequate intake according to recommendations for pregnant women. The primary outcome of the study was maternal GWG and adequacy of nutrient intake, with maternal dietary patterns as the main exposure.

### 2.2.1. Dietary Assessment

Trained midwives collected dietary intake information using a 137 item semiquantitative FFQ, which was given to women after delivery. This FFQ has been previously translated, adapted, and validated in Spanish women 18 to 74 years of age [18,19]. The FFQ provided a list of foods commonly used by the Spanish population and inquired about the consumption of these foods during the previous year. For each food item in the FFQ, women were asked to report the frequency of consumption and portion size. The FFQ included nine response options (never or almost never, 1–3 times a month, once a week, 2–4 times a week, 5–6 times a week, once a day, 2–3 times a weekday, 4–6 times a day, and more than six times a day). The dietary intake in grams per day was estimated using the indicated frequencies of consumption that were converted to intakes per day and multiplied by the weight of the standard serving size. Nutrient information, as well as total energy intake, was derived from Spanish food composition tables [20,21]. After computing total energy intake, 15 women were excluded because of implausible extreme energy intakes (<500 kcal/day and >3500 kcal/day) [17], leaving 503 women for analysis. Finally, food intake was adjusted for total energy intake using the residual method proposed by Willet et al. [22].

### 2.2.2. Dietary Pattern Construction

Factor analysis has been extensively used to detect common patterns among highly correlated variables through the use of FFQ [23]. This methodology is a tool commonly employed to extract posteriori dietary patterns [24]. We applied factor analysis with the Principal Components Method (PCA). First, the food items of the FFQ were combined into 16 groups by similar nutrient profile and culinary usage. A detailed description of each food group is reported in Table 1. The daily intake (in grams) for each food group, adjusted by total energy intake, was used in the factor analysis to identify maternal dietary patterns. On the basis of the values of the factor loadings, two main dietary patterns were defined, characterized by high factor loadings of specific food groups.


**Table 1.** Food groupings used in factor analysis.

### 2.2.3. Anthropometry

Pre-pregnancy weight was self-reported by pregnant women during the first appointment with the midwife. During each antenatal appointment, maternal weight (in kg) and height (in cm) were measured at the antenatal outpatient clinic, and maternal weight (in kg) before delivery was measured at delivery admission by a midwife. Total GWG (in kg) was obtained as the difference in maternal weight between pre-pregnancy weight and weight at delivery admission, as other authors have done previously [25]. Finally, according to the IOM guidelines, we defined the GWG as 12.5–18 kg for underweight, 11.5–16 kg for normal or adequate weight, 7–11.5 kg for overweight, and 5–9 kg for obese women. Weight gain below or above the recommended range was considered as inadequate or excessive GWG.

### 2.2.4. Diet Quality: Nutrient Adequate Intake

The dietary intake of a selection of nutrients, including dietary fiber, vitamins A, B9, B12, D, and E, and minerals such as calcium, phosphorus, magnesium, iron, iodine, potassium, selenium, and zinc, was compared with the recommended intakes for these nutrients in pregnant women according to the criteria established by the European Food Safety Agency (EFSA) [26]. Considering the Average Requirements (AR) and/or Average Intake (AI) and Upper-Level intake (UL), we built three categories: (i) deficit intake (intake below AR/AI), (ii) adequate intake (intake between AR/AI and UL), and (iii) excessive intake (intake higher UL). To decrease potential measurement errors derived from the use of the FFQ (overestimation bias), we calculated the proportion of women with intakes below two

thirds (2/3) of the Dietary Reference Intakes (DRIs), as other authors have reported previously [14,27]. Results were based on dietary intake data only, excluding supplements.

### 2.2.5. Other Maternal Variables Related to Patient Characteristics

The assessment of data was obtained via three different sources: (i) personal interviews carried out by trained midwives in the hospital within the two days after giving birth, (ii) clinical records, and (iii) prenatal care records. Information was obtained on general sociodemographic characteristics, including maternal age at pregnancy, lifestyle habits, and personal characteristics. Social class ranged from I (the highest) to V (the lowest), coded according to the classification of the Spanish Society of Epidemiology [28], which is similar to that reported in the Black Report [29]. We categorized smoking status during pregnancy as a current smoker and non-smoker (including previous smokers). Finally, we also computed the number of prenatal appointment and the date of the first appointment. Prenatal care utilization was measured by using the Kessner index. This index considers the timing of entry in prenatal care, number of prenatal appointments, and gestational age at delivery [30].

### *2.3. Statistical Analysis*

We described the study variables using proportions for qualitative variables and means and standard deviations (SD) for quantitative variables. We used the Pearson Chi-square and Kruskal–Wallis tests to assess differences in the distribution of means and percentages.

We conducted a PCA analysis to ascertain dietary group patterns [24]. According to the similarity of food items, we created 16 food groups (Table 1).

We used the Scree plot (Figure 2) and the eigenvalues >1 of the principal components to decide the number of factors to retain. We retained the factors with loadings showing an absolute value ≥0.3. We used these loadings to define the food groups that characterized each dietary pattern.

**Figure 2.** Scree plot of eigenvalues after Principal Components Method (PCA).

To explain sampling adequacy and inter-correlation of variables, we used the Kaiser–Meyer–Olkin value and Bartlett's test of sphericity. Finally, we identified three different dietary patterns explaining 30.6% of the total variance among the food groups included (Table 2).

We used multiple linear regression models to investigate the association between the different dietary patterns (independent variable) with GWG (dependent variable). Crude β-coefficients and adjusted β-coefficients, with their respective 95% confidence intervals (CI), were derived from the fitted univariate and multivariate models. Finally, we used logistic regression models to assess the association between dietary patterns (independent variable) and nutrient adequacy (dependent variable). To control for potential confounding factors in each of the models mentioned previously, multiple logistic and linear models were adjusted for maternal age at pregnancy, social class, Kessner index, and smoking status. Statistical analysis was performed using Stata (15.0, StataCorp L.P. College Station, TX, USA).


**Table 2.** Factor loadings for two main dietary patterns derived from a principal component analysis.

The cumulative variance contribution rate is 30.6%. Values > 0.3 are factor loading of significant relevance.

### **3. Results**

### *3.1. Characteristic of the Study Population*

Based on pre-gestational BMI, the overall percentage of women with underweight, normal weight, overweight, and obesity were 12.1%, 57.3%, 22.7% and 8.0%, respectively. Regarding the GWG, 170 (33.8%) women had a reduced GWG. Meanwhile, 128 (25.5%) presented excessive GWG. Table 3 presents maternal sociodemographic, anthropometric, and lifestyle variables stratified by GWG. Women with an excessive GWG had a higher pre-pregnancy BMI (*p* < 0.001), higher mean birth weight, and length of gestation (<0.05) than women with a reduced and adequate GWG.

**Table 3.** Description of the study population characteristics in the study (*n* = 503).


Abbreviations: (SD): standard deviation; BMI: body mass index. Pearson chi-square test and Kruskal–Wallis test were performed for evaluating differences in categorical and continuous variables, respectively.

### *3.2. Dietary Patterns and GWG*

Table 4 presents the crude and adjusted beta (β) coefficients from the univariate and multivariate linear models evaluating the association between each of the two different dietary patterns with an increasing GWG. There was a negative association between GWG and Mediterranean dietary pattern (crude β = −0.06, 95% CI: −0.11, −0.04), whereas a positive association with GWG was found for the Occidental dietary pattern; nevertheless, this association is lost when adjusted for confounders.

**Table 4.** Multivariable regression models for the association between dietary patterns and gestational weight gain (GWG) (*n* = 503).


Crude β-coefficients: crude β-coefficients, adjusted β-coefficients a: adjusted β-coefficients. <sup>a</sup> Adjusted for age of parity, social class, Kessner index, and smoking habits.

### *3.3. Prevalence of Participants with Adequate, Deficient, or Excessive Nutrient Intake According to GWG*

Table 5 shows the prevalence of participants with adequate, deficient, or excessive nutrient intake according to their GWG. The vitamins that exhibited the highest deficient intake for all the study participants were vitamins B9 and D. Women with a reduced GWG showed a lower prevalence of vitamin B9 deficient intake than women with an adequate or excessive GWG. In contrast, women with adequate GWG exhibited a higher intake deficit of Vitamin D than women with inadequate or excessive GWG (*p* < 0.05).

**Table 5.** Prevalence of participants with an adequate, deficient, or excessive intake of nutrients according to 2/3 EFSA DRIs stratified by GWG.


Intake (I): <sup>a</sup> deficient intake (DI), <sup>b</sup> adequate intake (AI), and <sup>c</sup> excessive intake (EI). Values are % unless otherwise indicated. Pearson chi-square test was used in order to ascertain differences between groups. There is no Upper-Level intake (UL) in the micronutrient assessed.

### *3.4. Association between Maternal Dietary Patterns and Nutrient Adequate Intake According to GWG*

Table 6 shows the association between nutritional adequacy and dietary patterns according to the GWG status. Independent of GWG, a Mediterranean dietary pattern showed moderate evidence of a higher probability of meeting an adequate intake for dietary fiber; vitamins B9, D, and E; and iodine (*p* < 0.05).


**Table 6.** Multivariate logistic regression of association between nutrient adequacy and dietary patterns according to GWG.

The multivariable model was adjusted for age of parity, social class, Kessner index, and smoking habits.

### **4. Discussion**

This is the first study that evaluates the association of maternal dietary patterns and GWG and their association with adequacy nutrient intake in Spanish pregnant women. Independent of maternal GWG, we found that an adequate intake of dietary fiber; vitamins B9, D, and E; calcium; and iodine nutrients was directly related to a classical Mediterranean dietary pattern characterized by a high content of vegetables, olive oil, whole cereals, and nuts. Our findings showed moderate evidence for an association between this healthy dietary pattern and lower GWG trajectories.

Diet quality has been neglected as a risk factor and potential intervention target for inappropriate GWG. Traditionally, the classical nutritional epidemiological approach focused on isolated food groups and/or macronutrient intake instead of the assessment of dietary patterns [31]. However, the role of the overall diet versus individual foods or nutrients provides a more intuitive and objective holistic interpretation of the quality of a woman's dietary pattern [24].

Taking into account that women with healthy dietary intake also have a healthy lifestyle contributing to reducing excess GWG, the position of American Dietetic Association for pregnant women emphasized the adoption of healthy dietary patterns rich in fish and seafood, vegetables, legumes, and vegetable oils, along with engaging in physical activity [32], to prevent inadequate GWG. Accordingly, our study's main findings support the previous recommendation as we found an association between the classically traditional healthy Mediterranean dietary pattern richer in vegetables, olive oil, and nuts, and lower GWG. In line with our findings, Bassel et al. reported that a Mediterranean-style diet characterized by a high intake of extra virgin olive oil, vegetables, and legumes had a potential role in reducing GWG in British pregnant women [33]. Furthermore, the scientific literature has shown that an increased intake of whole-grain cereal has a positive effect on reducing weight gain not only in a general adult [34] but also in the pregnant women population [35]. Our findings highlight that the Mediterranean dietary pattern, which includes this food group, displays a similar effect. Surprisingly, in our study, dietary patterns dense in energy, such as processed meat and ready-made meals or sweets and dessert patterns, did not show any effect on weight gain. This result contradicts those of other authors that report that a higher energy intake pattern is associated with higher GWG in European pregnant women [36]. Among the reasons that could explain these inconsistent results could be that women who presented a higher GWG (overweight/obese women) reported a healthier dietary intake, overreporting food groups considered as "healthy", resulting in an information bias, as other authors have extensively communicated [35,36]. Another explanation that could explain the discrepancy in research findings could be that the pregnant women included in our study reported a low intake of these non-healthy products, thus making the identification of this association more difficult.

Pre-pregnancy and pregnancy dietary patterns influence fetal health and the risk of fetal diseases, not only during intrauterine life but also into adulthood. Many nutrients, specifically vitamins and minerals such as vitamin B9 and iron, have been extensively investigated due to their relationship with the development of maternal morbidities and neurodevelopmental disease in babies [37,38]. Even though traditional pregnant counseling has emphasized the consumption of complex prenatal vitamins to provide the necessary amount of some micronutrients [39], Cano-Ibáñez et al. found a positive association between a healthy and diverse dietary pattern with nutrient adequacy in Spanish pregnant women [40]. Accordingly, we found a positive effect of the Mediterranean dietary pattern on nutrient adequacy intake in the present study, independent of GWG. These food groups are considered as "healthy food groups", typical of dietary patterns such as the Mediterranean diet [41] or Prudent diet [42]. In line with our results, several authors have demonstrated that adherence to both patterns might provide a balanced intake of micronutrients [43–45] instead of low energy quantity. The notion is that a food group provides the intake of several nutrients. For example, vitamin B9 is present in leafy vegetables, and dietary fiber can be found in legumes and vegetables.

### *Strengths and Limitations*

We assessed food intake during pregnancy within 48 h after giving birth; thus, dietary intake during the last trimester may be remembered better than earlier dietary intake during the first trimester. Nevertheless, it has been reported that dietary patterns are stable across pregnancy despite an increased energy intake, with the exception of alcohol intake [46], although recent studies associate very moderate alcohol consumption with lower indices of small for gestational age (SGA) newborns [47]. For this reason, we decided not to include this group in our PCA analysis, as other authors have reported previously [48]. The semiquantitative FFQ has been validated for the Spanish population [18], but self-reporting questionnaires and memory problems could lead to information bias. However, this would more likely cause a non-differential misclassification bias and estimations would tend toward the null. To correct possible errors derived from the FFQ, we excluded participants with energy intakes outside of predefined limits [17], and we used the residual method to adjust for food intake for energy intake. However, although the FFQ specifies the usual portion size as part of the question on frequency, it might not be the ideal tool to measure micronutrient intake and is not validated for this specific population group, although it has been used previously in pregnant women [14]. For this reason, we considered that intake was adequate only when the intake reached at least 2/3 of the recommendations proposed by EFSA for pregnant women, correcting the possible bias introduced by the FFQ and assuming, in any case, that the inadequate micronutrient intake (deficit intake or excessive intake) would be higher than the estimated figures [27].

Furthermore, we cannot exclude a potential reverse causation bias in the association between the independent variable with the assessed outcomes due to the study design. Finally, residual confounding that might affect GWG and nutrient intake, such as physical activity or other unmeasured socioeconomic factors, cannot be discarded, even if we have gathered data on the relevant confounders in nutritional epidemiology and adjusted for them in the multivariate analyses.

Notwithstanding the above limitations, our study includes several strengths reinforcing the validity and consistency of the findings obtained. The inclusion of a large representative sample (533 healthy pregnant women), from a reference population of around 120,000 healthy pregnant women providing exhaustive and specific information on dietary intake, is a strength of this study. The use of dietary patterns instead of single food or nutrients is another strength. Dietary patterns are more exhaustive as specific potential factors related to GWG than just single isolated foods. Dietary patterns are more intuitive and objective to determine women's overall dietary intake during pregnancy [24].

Furthermore, our PCA results explained the 30.6% of the total variance among food groups, and the analysis to derive dietary patterns was based on well-established criteria. We also included a considerable amount of information collected using a standardized protocol that reduces information bias regarding reported food intakes, sociodemographic characteristics, and lifestyles. Finally, to the extent of our knowledge, this report is the first study in Spain assessing the association between GWG with maternal dietary patterns and adequate micronutrient intake.

### **5. Conclusions**

In summary, we found that a healthy dietary food intake pattern, compatible with a Mediterranean diet, is associated with an adequate GWG during pregnancy and better nutrient adequacy. Pregnancy can be considered a window of opportunity for promoting healthy habits, as women are more willing to adopt healthier dietary habits during this time. Therefore, our findings support the promotion of healthy dietary habits based on a Mediterranean diet (characterized by a high intake of vegetables, nuts, whole cereals, and olive oil) during pregnancy. Furthermore, counseling and promoting the Mediterranean diet during antenatal visits could offer a sustainable and practical strategy in order to control GWG and ensure adequate nutrient intake during this critical fetal developmental period. Moreover, it could also be an important public health measure with implications that might span over a woman's life course.

**Author Contributions:** Conceptualization, N.C.-I., A.B.-C., J.M.M.-G., and M.D.-R.; methodology, N.C.-I. and M.A.L.-F.; data curation, M.D.-R., J.M.M.-G., and N.C.-I.; writing—original draft preparation, N.C.-I., S.M.-P., J.M.M.-G. and A.B.-C. writing—review and editing, all the authors; supervision, A.B.-C. and M.D.-R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant from the National Institute of Health Carlos III (PI11/02199).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Sexually Transmitted Infections and Associated Factors in Southeast Spain: A Retrospective Study from 2000 to 2014**

**María Ángeles Pérez-Morente 1, María Gázquez-López 2, María Adelaida Álvarez-Serrano 2,\*, Encarnación Martínez-García 3,4,\*, Pedro Femia-Marzo 5, María Dolores Pozo-Cano <sup>3</sup> and Adelina Martín-Salvador <sup>6</sup>**


Received: 19 September 2020; Accepted: 12 October 2020; Published: 13 October 2020

**Abstract:** The World Health Organization estimates that more than one million people acquire a Sexually Transmitted Infection (STI) every day, compromising quality of life, sexual and reproductive health, and the health of newborns and children. It is an objective of this study to identify the factors related to a Sexually Transmitted Infection diagnosis in the province of Granada (Spain), as well as those better predicting the risk of acquiring such infections. In this study, 678 cases were analyzed on a retrospective basis, which were treated at the Centre for Sexually Transmitted Diseases and Sexual Orientation in Granada, between 2000–2014. Descriptive statistics were applied, and by means of binary logistic regression, employing the forward stepwise-likelihood ratio, a predictive model was estimated for the risk of acquiring an STI. Sex, age, occupation, economic crisis period, drug use, number of days in which no condoms were used, number of sexual partners in the last month and in the last year, and number of subsequent visits and new subsequent episodes were associated with an STI diagnosis (*p* < 0.05). The risk of being diagnosed with an STI increased during the economic crisis period (OR: 1.88; 95%-CI: 1.28–2.76); during the economic crisis and if they were women (OR:2.35, 95%- CI: 1.24–4.44); and if they were women and immigrants (OR: 2.09; 95%- CI:1.22–3.57), while it decreased with age (OR: 0.97, 95%-CI: 0.95–0.98). Identification of the group comprised of immigrant women as an especially vulnerable group regarding the acquisition of an STI in our province reflects the need to incorporate the gender perspective into preventive strategies and STI primary health care.

**Keywords:** sexually transmitted diseases; public health; risk groups

### **1. Introduction**

Sexually Transmitted Infections (STIs) constitute a significant public health issue on a worldwide basis. The World Health Organization (WHO) estimates that more than one million people acquire an STI every day, compromising quality of life, sexual and reproductive health, and newborns' and children's health [1].

Such infections are caused by more than 30 different bacteria, viruses, and parasites, the most frequent being syphilis, gonorrhea, chlamydia, human papillomavirus (HPV), hepatitis B and C, and human immunodeficiency virus (HIV), the first three of which being curable and the last four being incurable viral infections, although treatments exist to attenuate or modify symptoms and the disease [1].

STIs are more frequent in people with risky sexual behaviors, such as not using condoms, having multiple sexual partners, having a highly risky partner (individual with many sexual partners or other risk factors), having anal sex or a partner who practices it, or having sexual intercourse with a partner who injects or has injected drugs before [2–4]. Teenagers, immigrants, sex workers, men who have sex with men (MSM), and bisexual individuals are particularly vulnerable groups regarding the acquisition of STIs [5,6].

Furthermore, the current financial situation impacts on the incidence of STIs, mainly in the most vulnerable population [1,7–10]. In the recent economic crisis period (2008–2014) [11–14], syphilis and gonorrhoea reappeared in Spain, and the incidence of HIV, hepatitis, and HPV increased. During this period, the probability of the appearance of an STI was higher than in the non-crisis period [15], due to the financial difficulties to access contraceptive methods, thus increasing the absence of protection against STIs [15–17].

At the world level, the prevalence of STIs in men and women is similar, with some regional differences. Nevertheless, complications in these processes do affect disproportionately women and generate a strong impact, mainly on their sexual and reproductive health. The WHO points out that the lack of data on STIs at the local level and, in particular, of data classified by sex, compromises the solution to this problem [18]. Moreover, today, despite the outstanding advance in medical knowledge and development of primary health care, sexual prevention programs seem to continue being inefficient, since the number of STIs continues to increase [1]. In this sense, an increase in STIs may be observed in Spain in general [19] and in the province of Granada in particular [10,15], which becomes an especially alarming issue.

The objective of this study was to identify any factors associated with STI diagnoses in the province of Granada (Spain), as well as those that better predict the risk of acquiring such infections.

### **2. Materials and Methods**

The group of cases involving users attended at the Center for Sexually Transmitted Diseases and Sexual Orientation in Granada (Spain) with a new STI diagnosis from 2000–2014 was analyzed on a retrospective basis. This center is attached to the Andalusian Health Service (AHS), which offers universal and free healthcare to people residing in the province regardless of their nationality and income level, providing healthcare to 730,000 people older than 15 years old [20].

The main source of information was the patient´s medical history in which health professionals completed sociodemographic data, symptoms, and medicals signs, results of diagnostic tests, therapeutic evolution, and final diagnosis. The study population consists of users with a positive or negative STI diagnosis clearly recorded on the medical histories contained in a database of 1437 histories created in a previous project from which this study is derived, and whose details on the sampling process are described in another paper [15]. Exclusion criteria included being younger than 18 years old and having a cognitive deficit. Investigators collected the information and attended a training course to reduce possible variations among them.

STI diagnosis was included as the dependent variable and coded as binary (yes/no), following the pattern established by other studies in this line of research [21]. Independent variables were classified in three groups: 1) Sociodemographic characteristics such as sex (female/male), age, nationality (Spanish, foreigner), occupation (sex worker/former worker/others), working status (active/non-active), education level (with no studies or primary/secondary/higher-level education); crisis period, with the year 2008 being considered the commencement of the financial crisis in Spain [11–14] (yes/no); 2) characteristics regarding the healthcare received, such as reason for consultation (HIV/another); previous treatment

(yes/no), number of subsequent consultations, number of new subsequent episodes, and 3) STI risk indicators such as sexual behaviour (heterosexual, homosexual, bisexual); stable relationship (yes/no); stable relationship with symptoms (yes/no); days elapsed since the last contact without using a condom; number of partners in the last month; number of partners in the last year; sex life understood as the total number of partners in their whole life; drug use (yes/no); previous STI (yes/no) and age of first sexual intercourse. Information was recorded on an ad hoc designed computerized data collection sheet, which was then exported to the statistical program for analysis.

Variables were analyzed on a descriptive basis calculating measures of central tendency (mean and standard deviation) for quantitative variables and absolute and relative frequencies for categorical variables. Comparison among groups was carried out using the chi-square test and the *t*-test according to the variable nature. The level of statistical significance was established at 0.05.

Finally, through binary logistic regression, the optimal model that best predicted a positive STI diagnosis was selected, employing the forward stepwise-likelihood ratio. At first, a model with no explanatory variable (only the constant) was used, and, in every step, a variable including a minor *p* was introduced. The potential predictors group comprised first-order interactions among explanatory variables. For each variable included in the model, the odds ratio (OR) and its confidence interval (CI) were calculated at 95%. The validity of the model was verified by means of the Hosmer-Lemeshow goodness-of-fit test. Analyses were performed using the Statistical Package for the Social Sciences (SPSS) program, version 24, (IBM, New York, NY, USA).

Before this study was carried out, approval was obtained from the Biomedical Research Ethics Committee of the province of Granada and from the Management Directorate of the Granada-Metropolitan Health District, which is responsible for the STI clinic where the research was approved. Patient data were handled with the utmost confidentiality and in compliance with the Spanish Organic Law 15/1999, of December 13, on Personal Data Protection, and the Spanish Organic Law 3/2018, of December 5, on Personal Data Protection and a guarantee of digital rights.

### **3. Results**

Inclusion criteria were fulfilled by 678 cases, out of which 440 (65.6%) had a positive STI diagnosis and 230 (34.2%), a negative one. Results show that STIs spread homogeneously across the population in terms of nationality, working status, and education level, establishing, with statistically significant differences, that persons diagnosed with an STI are younger (*p* < 0.001), mostly men (*p* < 0.001), who have not been or are not sex workers (*p* < 0.001), and who were diagnosed during the economic crisis period (2008–2014) (*p* < 0.001) (Table 1).

With respect to the healthcare received, differences were found in the number of subsequent visits (*p* = 0.001) and of new subsequent episodes (*p* < 0.001), being more frequent in cases diagnosed with STIs (Table 2).

As regards risk indicators, the group diagnosed with STIs used drugs more frequently, mentioned a higher number of days elapsed since the last unprotected sexual intercourse, and had more sexual partners in the last month and in the last year (Table 3).

Table 4 shows the logistic regression model adjusted that better predicts a positive STI diagnosis. It is noted that the probability of acquiring an STI is higher in a crisis period, if it is to be a woman in the said period, and if it is to be a woman and an immigrant. In addition, this probability diminishes drastically with age (Table 4).


**Table 1.** STI diagnosis vs. sociodemographic characteristics.

Note: STI—Sexually Transmitted Infection; *n*—sample size; SD—standard deviation; *p—p* value.

**Table 2.** STI diagnosis vs. healthcare received.


Note: *n*—sample size; SD—standard deviation; *p*—*p* value.


**Table 3.** STI diagnosis vs. risk indicators.

Note: *n*—sample size; SD—standard deviation; *p*—*p* value.



Note: †—Hosmer-Lemeshow goodness-of-fit test: χ<sup>2</sup> (8) = 3.89; *p* = 0.867. OR—odds ratio; CI: confidence interval.

### **4. Discussion**

In relation to the sociodemographic characteristics of the sample analyzed, it is noted that the mean age of individuals diagnosed with an STI was lower than in the case of those who did not have them. This result coincides with the publications in the scientific literature, considering the young population as one of the groups most exposed to STIs [22–24] along with other vulnerable groups such as gay men [5,25–28], men who have sex with men (MSM), transgender people, injecting drug users, women, sex workers, and immigrants, especially irregular immigrants [18,29–33]. Likewise, in a recent study on HIV and other STI epidemiology, it is observed that in 2006 the higher number of cases of some STIs such as syphilis, gonorrhea, chlamydia trachomatis, and lymphogranuloma venereum occurred in young adults between 25–44 years old [34], the age range in which the mean age of persons positively diagnosed with STIs is found in our study.

We have found no relationship between sex work and a positive STI diagnosis, despite the already-known vulnerability of such a group [35–37]. Although sexual orientation plays a relevant role, gay men sex workers assume more risky practices compared to heterosexuals [38], they were not well represented in our sample. In addition, safe sexual practices are often more present in commercial sexual relations, while they are relaxed in non-commercial sexual relations [38,39], which may have an impact on this study. However, our results coincide with the ones published regarding Spain by other authors, who found a low seroprevalence of some STIs [40], such as HIV, among sex workers [41–43].

The association found between the period of economic crisis and the increased risk of being diagnosed with STIs is in line with the fact that a crisis debilitates educational and health systems as well as prevention and promotion measures related to sexual health [44]. Therefore, such a situation may lead to an increase in the prevalence of infectious diseases like STIs, in particular to an increase in new HIV diagnoses [45,46].

Upon analyzing the healthcare received, the significant association found between the number of subsequent visits and of new episodes, among people who were positively diagnosed with an STI, could be justified by the need for a more intense follow-up of such new diagnoses.

In relation to risk indicators, we have found that drug use becomes a risky practice for transmission of STIs, due to the limitation imposed on the individual's decision-making capability, making them more vulnerable, as pointed out by other authors [33,47,48]. Evidence associates the low use of condoms with a higher risk of STI contagion [49–52], in line with our results, despite some nuances. We have analyzed the time period elapsed since the last time a condom was used and not its frequency of use, but literature identifies both factors as strongly associated with a significant risk of acquiring an STI [22,24,53]. Moreover, upon evaluating jointly two of the abovementioned variables, namely, the time elapsed since the last sexual intercourse without using a condom and the economic crisis period, it is observed that, in Spain, as from 2007, there has been a decrease in the use of contraceptive methods and STI prevention measures in one-fifth of casual or sporadic relationships [45]. Therefore, it should be noted that, in the crisis period, financial problems pose obstacles to access to contraception [16], with the consequent risk not only of unintended pregnancy but also of contagion of venereal diseases. Regarding the number of sexual partners in the last month and in the last year, people having a positive diagnosis had more partners, which finding is in line with the literature consulted [22,31].

In the predictive model designed for the diagnosis of STIs in our province, the variables of the economic crisis period and interactions between crisis and women, and between women and non-Spanish nationality remained, regardless of the other variables, showing an increase in risk, while age acted as a protective factor (Table 4). That is, structural determinants of health inequalities are the ones that have had a direct impact on positive STI diagnoses, reinforcing the knowledge that health is not just an individual issue, but it depends, to a great extent, on the surrounding environment [54,55].

As previously stated, this study is within the group of studies finding an association between an adverse financial context and bad health results, which becomes worse in women. The fact that women suffer in a different way than men do, the impacts of every financial crisis, has been widely reported [56–61]. And this is so because of the different and unequal opportunities men and women have regarding access to powerful, prestigious positions and to available resources, where women are usually at a disadvantage, even in developed countries like ours [62]. Furthermore, it should be pointed out that gender inequality occurring in the context of relationships affects the sexual and reproductive health of women, exposing them to a higher number of risks [45].

If to the fact of being a woman, the fact of being a foreigner is added, the likelihood of being diagnosed with a STI in our study increased even more. This result confirms the findings in previous Spanish studies where immigrant women were more exposed to HIV infection [32,33]. In this sense, the last report on Epidemiologic Surveillance of HIV and AIDS in Spain states that among the foreign population with new HIV diagnoses, 56% were women, had a worse immunological response to antiretroviral therapy, showed less follow-up, and less time for therapeutic failure [63]. Although the fact of being a foreigner, especially if coming from low-income countries, usually entails some disadvantages in the destination country, in financial, working, administrative and legal terms, that affect their health in the medium and long term, the austerity and exclusion policies carried out

during the economic crisis seem to have sped up the process [64,65], particularly in relation to women's sexual health.

Finally, the predictive model results confirm the protective effect of age in relation to the appearance of STIs. The fact that only a small group of European young people state they have access to information on STI prevention, while most of them have erroneous concepts on it and are in favor of causal sexual intercourse and with multiple partners, shows the existing lack of knowledge on these issues and the underestimation of the risk of acquiring STIs to which they are exposed [22].

### *Limitations*

This study presents some limitations and strengths. On the one hand, due to its cross-sectional nature, associations detected shall not be interpreted as causal relations, but, in any case, they allow for hypotheses that shall require confirmation in subsequent research with more complex designs. It should also be taken into account that the sample of the population analyzed comes from a single center and may not be fully representative of subjects vulnerable to an STI diagnosis in the province, which affects the external validity of the study. Specifically, it is fairly likely that the immigrant population is overrepresented, since, for instance, the native population may have more access to private healthcare for the treatment of these infections, as it occurs with other health/disease processes [66]. Such a fact would minimize the magnitude of the associations found. Nevertheless, the WHO recommendations are in line with the promotion of knowledge at the local level in order to address STIs, which justifies the need for this type of analysis [18]. On the other hand, although we may think people coming to the center may have had risky sexual behavior, information taken from clinical interviews in-person may be influenced by a social desirability bias, which, even in such a case, would not be discriminatory among the groups analyzed.

Finally, we believe the power of the study to evaluate factors related to STI diagnosis is high for several reasons. A long time period was evaluated; cases with an accurate (yes/no) STI diagnosis were selected, and individual and clinical data taken from medical histories were analyzed which, to the best of our knowledge, contribute more information and of better quality than any population database available in our context.

### **5. Conclusions**

Our findings have identified immigrant women as a risk group regarding the acquisition of STIs in the province of Granada, and especially during an economic crisis. However, there is an evident need for further research, with a gender-oriented perspective, of sexual behavior within such a group, as well as any potential limitations to access the health system, in order to understand differences in behavior and, therefore, in health results between men and women.

Public health policies should be aimed at a re-evaluation of prevention strategies present in our province, as well as of services available, reduced during the crisis. Designing and implementing specific, more effective preventive measures that shall prepare immigrant women to face these infections, as well as detecting non-reported cases and controlling and treating existing cases at a sustainable human and economic cost, entail the first-level challenge in the context of STIs.

**Author Contributions:** Conceptualization, A.M.-S; Methodology, E.M.-G. and P.F.-M.; Investigation, M.Á.P.-M., M.G.-L, M.A.Á.-S., E.M.-G., P.F.-M., M.D.P.-C, and A.M.-S.; Data curation, E.M.-G. and P.F.-M.; Writing—original draft preparation, M.Á.P.-M., M.G.-L, M.A.Á.-S., E.M.-G., P.F.-M., M.D.P.-C, and A.M.-S.; Writing—review and editing, M.Á.P.-M., M.G.-L, M.A.Á.-S., E.M.-G., P.F.-M., M.D.P.-C, and A.M.-S.; Supervision, M.Á.P.-M., E.M.-G., and A.M.-S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **The Quality of Counselling for Oral Emergency Contraceptive Pills—A Simulated Patient Study in German Community Pharmacies**

### **Bernhard Langer \*, Sophia Grimm, Gwenda Lungfiel, Franca Mandlmeier and Vanessa Wenig**

Department of Health, Nursing, Management, University of Applied Sciences Neubrandenburg, 17033 Neubrandenburg, Germany; sg2807@t-online.de (S.G.); Gwenda.lungfiel@gmx.de (G.L.); franca\_m@web.de (F.M.); vanessawenig@gmx.de (V.W.)

**\*** Correspondence: langer@hs-nb.de

Received: 8 August 2020; Accepted: 11 September 2020; Published: 15 September 2020

**Abstract:** Background: In Germany, there are two different active substances, levonorgestrel (LNG) and ulipristal acetate (UPA), available as emergency contraception (the "morning after pill") with UPA still effective even 72 to 120 h after unprotected sexual intercourse, unlike LNG. Emergency contraceptive pills have been available without a medical prescription since March 2015 but are still only dispensed by community pharmacies. The aim of this study was to determine the counselling and dispensing behaviour of pharmacy staff and the factors that may influence this behaviour in a scenario that intends that only the emergency contraceptive pill containing the active substance UPA is dispensed (appropriate outcome). Methods: A cross-sectional study was carried out in the form of a covert simulated patient study in a random sample of community pharmacies stratified by location in the German state of Mecklenburg-Vorpommern and reported in accordance with the STROBE statement. Each pharmacy was visited once at random by one of four trained test buyers. They simulated a product-based request for an emergency contraceptive pill, stating contraceptive failure 3.5 days prior as the reason. The test scenario and the evaluation forms are based on the recommended actions, including the checklist from the Federal Chamber of Pharmacies. Results: All 199 planned pharmacy visits were carried out. The appropriate outcome (dispensing of UPA) was achieved in 78.9% of the test purchases (157/199). A significant correlation was identified between the use of the counselling room and the use of a checklist (*p* < 0.001). The use of a checklist led to a significantly higher questioning score (*p* < 0.001). In a multivariate binary logistic regression analysis, a higher questioning score (adjusted odds ratio [AOR] = 1.41; 95% CI = 1.22–1.63; *p* < 0.001) and a time between 12:01 and 4:00 p.m. (AOR = 2.54; 95% CI = 1.13–5.73; *p* = 0.024) compared to 8:00 to 12:00 a.m. were significantly associated with achieving the appropriate outcome. Conclusions: In a little over one-fifth of all test purchases, the required dispensing of UPA did not occur. The use of a counselling room and a checklist, the use of a checklist and the questioning score as well as the questioning score and achieving the appropriate outcome are all significantly correlated. A target regulation for the use of a counselling room, an explicit guideline recommendation about the use of a checklist, an obligation for keeping UPA in stock and appropriate mandatory continuing education programmes should be considered.

**Keywords:** non-prescription drugs; community pharmacies; consultation; patient simulation; emergency contraception; ulipristal acetate; Germany

### **1. Introduction**

Sexual intercourse without contraception, an improperly used or broken condom, improperly applied regular contraception or sexual assaults can lead to unintended pregnancies [1]. The consequences of unintended pregnancies or the need to terminate a pregnancy affect not only the physical but also the social and emotional health of the women concerned and often do so over the long term [2]. It is estimated that in Germany about 34% [3] of all pregnancies are unintended (globally, about 44% [4]), of which about 43% [3] (globally, about 56% [4]) end in termination of a pregnancy. Emergency contraception (EC) plays an important role in preventing this situation. EC can be divided into copper intrauterine devices (Cu-IUD) and oral hormonal methods [5].

The Cu-IUD is the most reliable EC method with a failure rate of <0.1%, and it is also effective after ovulation has occurred [6]. However, access to this form of EC is made difficult in that a gynaecologist must insert the Cu-IUD, meaning that this is not the preferred method for many users [5]. In Germany, it is only used in isolated cases [7]. Of the oral hormonal methods, the Yuzpe method (oestrogen–progesterone combination comprising two doses of 50 μg ethinylestradiol and 0.25 mg levonorgestrel) is also no longer recommended in Germany for post-coital contraception [7] because it is less effective than other EC methods and is also associated with more adverse events [1,6,7]. Because mifepristone is not available in Germany for this indication [7,8]—unlike some other countries [5,9]—the oral hormonal methods are concentrated on emergency contraceptive pills (or "morning after pills") containing the active substances levonorgestrel (LNG) and ulipristal acetate (UPA).

LNG is a synthetic gestagen that reduces the surge in luteinising hormone (LH), thus delaying ovulation. However, it must be taken before the surge in LH. UPA is a selective progesterone receptor modulator that also inhibits the LH surge but, unlike LNG, is still effective during the LH surge up to the LH peak [5,6,10]. Therefore, UPA is effective for up to 120 h after unprotected sexual intercourse or contraceptive failure. In contrast to UPA, LNG is effective up to 72 h after unprotected sexual intercourse or contraceptive failure, even though a moderate efficacy of up to 120 h is discussed in the international literature [5,6]. The superior effectiveness of UPA compared to LNG is also apparent in the first 24 to 72 h [11–13]. There are differences in terms of the statistical significance, however [11–13], which in Germany has led to different recommendations from gynaecology associations (generally UPA is considered superior) and the Federal Chamber of Pharmacies (BAK) (less than 72 h LNG or UPA) [14,15]. Both substances are more effective the sooner they are taken after unprotected sexual intercourse [7], which is why prompt access to the medication is of great importance. Analogous to many other countries [16], it has therefore been possible to obtain such preparations in Germany without a prescription since March 2015 [17], but the dispensing of these medications is still restricted to community pharmacies. Gynaecology associations and the German Medical Association (BÄK) argued against no-prescription dispensing of emergency contraceptive pills, citing the need for a medical consultation [18,19]. Not least, given this background [20], the BAK published corresponding recommendations including a checklist for quality assurance of the counselling provided when dispensing emergency contraceptive pills, whereby the use of this checklist is not explicitly advised in the recommendations [15]. The recommendations, which were developed with input from many experts across a range of organisations in various disciplines (e.g., BÄK, professional organisations and associations of gynaecologists, pharmaceutical OTC industry, government-controlled, private and church-based organisations and centres providing advice on sex education and family planning) [20], include the prerequisites for dispensing emergency contraception containing the active substances LNG or UPA for self-medication. Information about counselling and dispensing are also provided as well as the criteria regarding the limits of self-medication and referral to a doctor. Not only pharmacists but pharmacy technicians and pharmaceutical technical assistants are authorised to provide counselling for dispensing medicinal products such as the emergency contraceptive pill in Germany. Considering that the issue is still a sensitive one for many consumers, maintaining the privacy of the customer for this indication is a particularly important counselling criterion [21–24]. Nevertheless, both the Ordinance on the Operation of Pharmacies in Germany and the BAK recommendations expect that the counselling is conducted in private to maintain confidentiality and prevent other customers overhearing the counselling as far as possible [15,25]. There is as yet no legal obligation or a recommendation based on BAK guidelines to keep a separate counselling room, however [15,25].

The BAK recommendations stipulate the dispensing of only UPA 72 to 120 h after unprotected sexual intercourse or contraceptive failure, but it is not currently known to what extent pharmacies comply with these recommendations. The three studies available to date in Germany on the counselling and dispensing behaviour for emergency contraception have also not investigated this issue [26–28]. The study results are also based on self-assessment by pharmacies and may therefore be distorted, for example, due to social desirability [29,30]. Applying the simulated patient method (SPM), which is considered to be the "gold standard" [31] even if the relatively high administrative and financial costs and the comparably smaller sample sizes [31] as well as any intra- and inter-observer variabilities [32] are taken into account, can avoid such weaknesses because a realistic (counselling) situation can be simulated [33]. Unlike the many studies conducted in other countries [21,23,30,34–45], in Germany, there have not been any investigations to date on the counselling and dispensing of an emergency contraceptive pill using an SPM approach because the few scientific SP studies available for Germany on the counselling quality in pharmacies are always based on indications other than emergency contraception. Clear deficits in the quality of counselling have been identified in these studies [46–51].

The aim of this study was to identify the counselling and dispensing behaviour of pharmacy staff and the factors that may influence this behaviour in a scenario that intends that the emergency contraceptive pill containing only the active substance UPA is dispensed (appropriate outcome).

### **2. Methods**

### *2.1. Design*

A cross-sectional study design was chosen in accordance with the "STROBE Statement—Checklist of items that should be included in reports of cross-sectional studies" [52]. The quality of the counselling provided in the pharmacies visited is determined using the SPM, which has often been applied in international studies in the community pharmacy (CP) setting [33,53,54]. The SPM is a covert participatory observation [55] by a person, who in an ideal case cannot be differentiated from a real customer (simulated patient, SP), who visits a CP to simulate a real-life counselling situation based on a previously defined scenario. The data are then collected according to previously defined criteria using an assessment form and the CP is provided with performance feedback, if applicable [33].

### *2.2. Setting and Participation*

The test purchases took place between 1 July and 30 September, 2019 in the German state of Mecklenburg-Vorpommern (31 December, 2019: approx. 1.61 million residents, 23,216 km<sup>2</sup> area, low population density of 69.3 residents/km<sup>2</sup> [56]). Upon request by telephone, the Pharmacy Association of Mecklenburg-Vorpommern did not want to provide a list of all registered community pharmacies in the state. Consequently, the list of these pharmacies was identified using the pharmacy finder of the website Apotheken-Umschau.de [57]. All community pharmacies that had a postcode in the state of Mecklenburg-Vorpommern on the reference date of 1 June, 2019 using the postcode search of the pharmacy finder were included in the study. These hits were validated with a corresponding Google search. As a result, a basic population of N = 395 pharmacies was formed. A comparison with the most recently available information from the German Federal Chamber of Pharmacies (ABDA; Bundesvereinigung Deutscher Apothekerverbände) for the end of 2018 regarding the total number of pharmacies in Mecklenburg-Vorpommern [58] showed a 98% agreement.

In Germany there have been no studies conducted to date on the dispensing of emergency contraceptive pills for a scenario in which dispensing UPA is mandatory. The degree of variability is therefore unknown. The minimum necessary sample size (n) was determined for a population size (N) of 395 and an error margin (e) of 0.05 using the following formula, which is based on a degree of variability of *p* = 0.5 and a 95% confidence interval [59]:

$$m = \frac{N}{1 + N(e)^2} = \frac{395}{1 + 395(0.05)^2} = \frac{395}{1.9875} = 198.74\\\text{Population size} = N \left| \text{ Margin of error} = e \right. \tag{1}$$

The assumed degree of variability of *p* = 0.5 maximises the required sample size. The 395 CPs were stratified by location to indicate whether they are urban or rural. Using the MS Excel random number generator, they were each then assigned a random number and the 199 participating CPs were then selected by simple random sampling from each stratum.

### *2.3. Scenario and Assessment*

The recommendations, including the checklist published by the BAK, formed the basis of the test scenario that was developed (see Table 1) and the evaluation form used (see Table 2). The request for the "morning after pill" because of a broken condom that was simulated in the scenario is realistic because this reason was cited most frequently in requests for emergency contraception in pharmacies in a recent German study [26]. Dispensing the emergency contraception ellaOne®, the only preparation with the active substance UPA available on the German market [60], was defined as the appropriate outcome. According to the recommendation from the BAK, unprotected sexual intercourse 3.5 days ago must be considered the criterion for dispensing UPA instead of LNG [15].

The evaluation form includes 14 items, whereby the first nine evaluate whether appropriate questions were asked analogous to the BAK checklist. Based on the form, the decision was subsequently made by pharmacy staff whether the emergency contraceptive pill ellaOne® with the active substance UPA is dispensed (item 10). It was also recorded whether LNG was erroneously dispensed instead of UPA (item 11). For the situation in which a medicinal product was dispensed, it was also evaluated according to the BAK recommendations whether the test buyers were informed about common adverse events such as headaches, nausea, dizziness, abdominal pain and cramps, dysmenorrhoea, vomiting, tiredness or breast tenderness [15] (item 12). Finally, the overall conditions of the counselling were also recorded, that is, whether the counselling was conducted in a separate room (item 13) and whether a checklist was visibly used to ensure the quality of the counselling (item 14).

For the evaluation form, only objective items were used to avoid a subjective assessment and thus latitude in the evaluation by the SPs (for example, on the friendliness of the pharmacy staff). Therefore, to complete the individual items, only dichotomous scales were used (closed yes/no questions).


#### **Scenario**

The test buyers enter the pharmacy and ask for oral emergency contraception without having a specific product in mind (product-based query).

When questioned by the pharmacy staff, the following information is provided:


**Table 2.** Evaluation form.


### *2.4. Data Collection*

Four female Master's students from the Department of Health, Nursing, Management of the Neubrandenburg University of Applied Sciences acted as test buyers. They were selected on the basis of their participation in a 3-semester student research project. The students are all in the age group of 18 to 29 years, which corresponds to the age group of the primary users of emergency contraception [61]. Each pharmacy was visited once (a total of 199 test purchases), whereby the pharmacies were distributed randomly across the four SPs.

Before the data collection was started, the test buyers underwent comprehensive training. Each test buyer first familiarised herself with the theoretical principles of the SP method, the test scenario itself and the contents of the evaluation form. The test buyers then each carried out four pre-tests for the scenario to confirm the functionality of the evaluation form and the test scenario and to train in the practical application of the SP method. The total of 16 pre-tests were carried out in pharmacies in Mecklenburg-Vorpommern that were not included in the stratified random sample. The pre-tests indicated that no changes to the test scenario or the evaluation form were required (see Figure 1).

The test purchases were carried out on different days of the week and at different times of the day. The SPs made their request to the pharmacy staff who first approached them. The SPs only provided additional information if they were then asked by the pharmacy staff to ensure that the information provided is consistent.

Along with the items on the evaluation form, the SPs planned, analogous to the international literature (see Table 3), to also collect a number of variables during and after the test purchases that may possibly affect the appropriate outcome. Unlike other international SP studies, the variable "pharmacy type" (chain vs. independent) was not recorded [35,37,44] because in Germany a pharmacy may only be operated by one pharmacist as the owner and in addition ownership is restricted to 1 primary pharmacy with up to 3 subsidiary pharmacies in the immediate neighbourhood [62].

**Figure 1.** The design and flow of the study.


Note: \* The possible influencing factors were taken from the specific literature sources.

So that, on the one hand, the most realistic counselling situation possible could be simulated and to identify any planned dispensing of medicinal products while at the same time avoiding unnecessary purchasing of medicinal products and thus waste (because the test buyers do not actually need the emergency contraceptive pill), in the cases in which a medicinal product was going to be dispensed, the test buyers informed the pharmacy staff just before making payment that they had left their wallet at home.

In the literature, audio recordings are recommended for quality assurance of the test purchases [70]. For data privacy reasons, however, this was omitted because otherwise the pharmacies must be informed beforehand about the audio recordings and thus the study design is no longer covert. However, the corresponding evaluation form was completed by the test buyers immediately after visiting the pharmacy so that any distortions in the study results due to faulty recall by the test buyers could be minimised.

After evaluating the data collection, each pharmacy received a written performance feedback specific for each pharmacy including graphically edited benchmarking, whereby for each pharmacy any improvement or deterioration regarding the individual criteria was shown compared to the other anonymised pharmacies in the stratified random sample. In this way, the pharmacies were informed about the position of their competitors, so that ideally appropriate optimisation processes by the pharmacies investigated could be initiated based on the feedback with the aim of sustainably improving the quality of counselling provided.

### *2.5. Ethical Approval*

According to the "Guideline for the use of mystery research in market and social research" [71], the data collected were anonymised and recorded in such a way that the pharmacies or the personnel involved could not be identified. To avoid a possible Hawthorne effect [72] and also a possible selection bias [73], the test purchases were carried out covertly—that is, without informing the pharmacies in advance—analogous to some other national [48–51] and international studies [21,33,38] and therefore pharmacies were not asked for consent to participate in advance. The lack of informed consent in advance was—analogous to the international literature [40,74,75]—resolved in that all pharmacies were informed about the procedure and the background of the study upon completion of the study. Recruited students provided their written informed consent to act as SPs. The study protocol was approved by the institutional ethics committee of the Neubrandenburg University of Applied Sciences (registration number: HSNB/KHM/152/19).

### *2.6. Data Analysis*

The data were entered using the four-eyes principle and analysed with SPSS Version 25 for Windows (IBM, Armonk, NY, USA). As part of the descriptive statistics, frequencies and percentages were determined. 95% confidence intervals for categorical data using bootstrapping were also reported. A Pearson's chi-square test was performed to determine if interactions involving "counselling room used" were more likely to result in "visible use of a checklist". Cramer's V was reported as a measure of effect size. Applying the Kolmogorov–Smirnov test as well as the Shapiro–Wilk test indicated that the data did not have a normal distribution. Therefore, the median, interquartile range (IQR), minimum and maximum were calculated for continuous variables. With the help of the non-parametric Mann–Whitney U test, it was analysed whether "visible use of a checklist" led to a significantly higher median questioning score. Pearson's r was reported as a measure of the effect size. A binomial logistic regression model was also developed to identify the influence of various independent variables (see Table 3) on the appropriate outcome (dispensing of the emergency contraceptive pill with the active substance UPA). All independent variables were checked for outliers and multicollinearity. Variables with a *p*-value less than 0.05 in the univariate analysis were included in the multivariate analysis. Odds ratios (OR), 95% confidence intervals, *p* values and as a measure of the effect size Cohen's f<sup>2</sup> were reported. A *p* value of less than 0.05 was considered to be significant in all analyses.

### **3. Results**

All 199 planned test purchases could be carried out (visit completion rate: 100%). Due to the exit strategy developed that resulted in the ending of the purchase, there were no direct costs. The total of 5080 km between the individual test purchases were driven in private vehicles and led to indirect costs totalling €1150 that were financed from the primary author's own resources. Socio-demographic data for the pharmacies or the advising pharmacy staff are shown in Table 4. Most of the pharmacies tested were in local competition with one another and often did not have a quality certificate (62.8%, 125/199). The advising pharmacy staff were in most cases female (87.4%, 174/199) and aged between 30 and 49 years (57.3%, 114/199). In the cases in which it could be determined, the proportion of pharmacists (37.7%, 75/199) compared to non-pharmacists (pharmacy technicians and pharmaceutical technical assistants) (40.7%, 81/199) was almost equal.


**Table 4.** Socio-demographic data for the pharmacies or the advising pharmacy staff.

The appropriate outcome—dispensing of UPA—was achieved in 78.9% of the test purchases (157/199) (see Table 5). In 3.0% of the test purchases LNG (6/199) was dispensed while in 18.1% (36/199) no preparation was dispensed. There was a median questioning score of 5.0 (IQR 2.0–9.0) with a minimum score of 0 in 2.5% (5/199) of test purchases and a maximum score of 9 in 27.6% (55/199) of test purchases. The question about the time since the unprotected sexual intercourse took place was asked most frequently (93.5%, 186/199). Regarding the test purchases in which a medication was dispensed, information was provided about possible side effects by the pharmacy staff in 59.5% (97/163) of test purchases.


**Table 5.** Assessment items (*n* = 199).

Counselling was provided in a counselling room in slightly less than half of all the test purchases (44.2%, 88/199). In just over half of all test purchases, a checklist was visibly used for the counselling (53.8%, 107/199). A significant correlation could be found between the use of a counselling room and the use of a checklist (Pearson's chi-square test; χ2(1) = 104.355, *p* < 0.001, V = 0.724) whereby the effect size V according to Cohen [76] corresponded to a 'large' effect. In 43.7% (87/199) of the test purchases, the counselling was not carried out in a counselling room nor was a checklist used. In contrast to this, in 41.7% (83/199) of the test purchases, a counselling room was sought out, and a checklist was used for the counselling.

In addition, the use of a checklist led to a significantly higher questioning score (Mann–Whitney U test; U = 385.500, *p* < 0.001, r = 0.806), whereby the effect size r according to Cohen [76] corresponded to a 'large' effect. If a checklist was visibly used, this led to a median questioning score of 9.0 (IQR 7.0–9.0) with a minimum score of 2 in 0.9% (1/107) and a maximum score of 9 in 51.4% (55/107) of test

purchases. If a checklist was not visibly used, the median questioning score was 2.0 (IQR 1.0–3.0) with a minimum score of 0 in 5.4% (5/92) and a maximum score of 7 in 2.2% (2/92) of test purchases.

Table 6 shows the binary logistic regression model. As part of the bivariate analysis, three (age of the pharmacy staff, time of the test purchase, questioning score) of nine predictor variables had a *p*-value of <0.05 and were included in the multivariate logistic regression model. A time between 12:01 and 4:00 p.m. (AOR = 2.54; 95% CI = 1.13–5.73; *p* = 0.024) compared to 8:00 to 12:00 a.m. and a higher questioning score (AOR = 1.41; 95% CI = 1.22–1.63; *p* < 0.001) were significantly associated with dispensing of UPA (appropriate outcome). The location of the pharmacy, the presence of a quality certificate, a queue and the age, gender and professional group of the pharmacy staff as well as the SP number did not have any significant effect on the appropriate outcome. The model yielded a Nagelkerke R2 value of 0.255, which corresponds to a Cohen's f2 of 0.342 and thus to a 'medium' effect size [76].



### **4. Discussion**

The appropriate outcome, that is, the dispensing of UPA, was achieved in almost 80% of all test purchases. International SP studies on the quality of counselling provided for "morning after pills" in the Democratic Republic of Congo, Kenya and India show a similarly high case resolution of 74% [38], 82% [34] and 86% [30], while in Australia and Switzerland values of 95% [21] and even 100% [23] were achieved. In contrast, in one Australian SP study, this value turned out to be considerably lower with 24% [35]. The highly divergent values in some cases must be viewed and interpreted in light of the very different scenarios used.

The fact that in slightly more than 20% of all test purchases, a medicinal product was not dispensed, or the wrong medicinal product was dispensed is problematic [77]. This situation is all the more significant for such an indication in which the consequence of providing the wrong advice or not dispensing the correct medicinal product—possibly an unwanted pregnancy—must be considered to be very high.

Possible reasons for not dispensing UPA that were also seen in the international SP studies of oral emergency contraceptive pills is the unavailability of the preparation UPA [78,79] and a lack of knowledge on behalf of the pharmacy staff [80]. Appropriate mandatory continuing education programs could reduce the issue of a lack of knowledge [81]. The unavailability of the UPA preparation could be minimised by mandating keeping the preparation in stock, especially as UPA is the only recommended active substance on the market for a specific time window and also because it should be taken as soon as possible.

In contrast to a Saudi Arabian study [64], the time proved to be a significant influencing factor on the appropriate outcome even though the type of outcome between the two studies is not comparable (UPA dispensed vs. antibiotics dispensed without a prescription). This creates a need for further research about the possible reasons, especially as the best counselling was provided not in the morning as expected on the basis of the model of the performance curve [82] but instead from midday to the late afternoon.

Analogous to the international literature [67,83] the questioning score had a significant effect on the appropriate outcome. Regarding the individual questions, the question about when the unprotected sexual intercourse took place proved to be the "master question" because only then could the correct decision—dispensing UPA—be made at all. The SP studies conducted in Switzerland, Scotland and Australia showed similarly high values to those in this investigation with values of 100% [23,45], 93% [35] and 88% [36] respectively. In contrast, in one SP study in the Democratic Republic of Congo, this question was only asked in 7% of the test purchases [38], and in a Turkish SP study it was asked in only one of 155 test purchases [41]. In an Indian and a Brazilian SP study this question was not asked once in 70 [30] and 122 [44] test purchases respectively.

A significant correlation between the questioning score and the visible use of a checklist was apparent, analogous to the international literature [23,35]. However, a checklist was only used in a little over half of all test purchases. In contrast, in a qualitative German study, it was reported that almost all pharmacies stated that they used a checklist [27]. Such discrepancies—worse results in SP studies than in self-reported studies of the same issue—are also seen in the international mixed-methods literature for requests for medicinal products other than oral emergency contraceptive pills [84,85]. Internationally, a checklist was only used in 10% of all test purchases in an Australian SP study [21] whereas this value increased to 83% in another Australian SP study that was conducted because a checklist had since been developed and recommended by the Pharmaceutical Society of Australia [35]. However, in a more recent Australian SP study, a checklist was not used in a single test purchase [36]. The use of a checklist is also recommended in the guidelines for Switzerland, which led to an application rate of 99% in a recent Swiss SP study [23]. In contrast to an Australian SP study [35], our study showed not only a significant correlation between "checklist" and "questioning score" but the questioning score also has a significant effect on the dispensing of UPA. It, therefore, appears to be advisable—despite criticism from pharmacy staff in a qualitative German study that the BAK checklist, in particular, is overly detailed and thus requires considerable time [27]—to introduce an explicit guideline recommendation about the use of a checklist in Germany as well. However, optimising or shortening the checklist should be considered to increase its acceptance.

Although there is a significant correlation between the use of a checklist and the use of a counselling room, in this study, a counselling room was only used in a little less than half of the test purchases. In contrast, two-thirds of the interviewed pharmacy staff in the qualitative German study indicated using a counselling room [27]. Internationally, there is an inconsistent picture regarding the use of a counselling room. For example, in an Australian SP study, only 9% of the test purchases were conducted in a counselling room [36] while in an Indian SP study this value was as high as 47% [30]. In contrast to this, an extra counselling room or a quiet area in the pharmacy was used in 90% of the consultations in another Australian SP study [21]. In Switzerland, pharmacists are supposed to carry out the consultation for the "morning after pill" in a separate counselling room, which was again implemented in 94% of test purchases in a corresponding SP study—analogous to the recommended use of a checklist [23]. To protect the privacy of patients for this sensitive issue, such a target regulation should therefore be considered for Germany as well. Furthermore, the flow of information between the consumer and the pharmacy staff could also be improved as a result, which was established in a qualitative Australian study [22]. Due to the correlation between the use of a counselling room and the use of a checklist, it is also likely that the use of a checklist could be increased in this way [21].

There is a need for further research in Germany regarding access to and the quality of counselling provided for emergency contraceptive pills for minors [30], male consumers [39] and victims of sexual assault [43]. It should also be analysed in future whether pharmacy staff provide additional important information—about sexually transmissible infections, for example—during the counselling [35]. Future studies should also analyse the size of any differences between SP results and self-reported results by pharmacy staff by using a mixed-methods approach regarding the counselling provided for oral emergency contraception.

### *Strengths and Limitations*

This study is the first in Germany that investigated the counselling and dispensing behaviour of pharmacies for the "morning after pill" using an SP method. The SP study design used is particularly advantageous because unlike other data collection methods, both the effect of social desirability and the Hawthorne effect can be avoided. The results—at least for one German state—can also be generalised because of the stratified random sampling used. There is also no selection bias because there is no option for opting out. The study could also be fully implemented as planned due to the 100% visit completion rate.

Because the study results only refer to one German state, future studies should be expanded to additional states or all of Germany. However, this would be an ambitious undertaking in light of the relatively high data collection costs. Furthermore, only one specific counselling and dispensing scenario (mandatory UPA dispensing) for oral emergency contraceptive pills were used for the data collection. Other scenarios could therefore lead to different results [48,67]. Because only students in a certain age range were used as SPs, it also cannot be ruled out that the counselling and dispensing behaviour would differ for other age groups (such as adolescents) or women with different levels of education (such as customers with secondary school education only). The study also did not differentiate between non-dispensing of UPA due to a lack of availability and non-dispensing due to incorrect counselling. Not dispensing UPA due to incorrect counselling would be classified as particularly serious because the test buyers would not receive UPA as a result, even though it would be the suitable active substance. However, not dispensing UPA due to a lack of availability would be problematic because the medication is supposed to be taken as soon as possible. Future studies should, therefore, record this differentiation because it would provide important information about possible optimisation measures. Furthermore, this is a cross-sectional study, and therefore, no causal relationship can be created between the variables studied. Because the study design (no audiotaping and no second observer) meant that the test buyers had to remember the recorded items until they left the pharmacy and many items had to be recorded (14 items for counselling as well as several influencing factor items), collecting additional important items (such as information about taking the medication as soon as possible or about taking the medication again in case of vomiting) was also omitted to prevent any recall bias. Immediate feedback after the test purchase would also have been desirable because then the pharmacy staff's memory of a specific counselling situation would have still been fresh. However, there is a risk that the pharmacy staff inform their colleagues in the neighbourhood about the test purchases. As a result, it was also not possible to ask the counselling pharmacy staff about their age, meaning that there may be certain distortions in the estimations of the age variable.

### **5. Conclusions**

In almost four-fifths of all test purchases, the appropriate outcome was achieved. This meant, however, that in a little over one-fifth of all test purchases, the required dispensing of UPA did not occur, which could have led to serious consequences such as an unwanted pregnancy. The use of a counselling room and a checklist, the use of a checklist and the questioning score as well as the questioning score and achieving the appropriate outcome are all significantly correlated. A target regulation for the use of a counselling room, an explicit guideline recommendation about the use of a checklist, an obligation for keeping UPA in stock and appropriate mandatory continuing education programs should be considered. Surprisingly, the time of day proved to be a significant influencing factor on the appropriate outcome, which requires further investigation.

**Author Contributions:** Conceptualisation: B.L., S.G., G.L., F.M. and V.W.; data curation, B.L., S.G., G.L., F.M. and V.W.; formal analysis, B.L., S.G., G.L., F.M. and V.W.; methodology, B.L., S.G., G.L., F.M. and V.W.; project administration, B.L.; validation, B.L., S.G., G.L., F.M. and V.W.; visualisation, B.L., S.G., G.L., F.M. and V.W.; writing—original draft preparation, B.L., S.G., G.L., F.M. and V.W.; writing—review and editing, B.L., S.G., G.L., F.M. and V.W.; All authors have read and agreed to the published version of the manuscript.

**Funding:** We acknowledge support for the article processing charge from the Open Access Publication Fund of the Hochschule Neubrandenburg (Neubrandenburg University of Applied Sciences).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Advancing Prevention of STIs by Developing Specific Serodiagnostic Targets:** *Trichomonas vaginalis* **as a Model**

### **John F. Alderete**

School of Molecular Biosciences, College of Veterinary Medicine, Washington State University, Pullman, WA 99164, USA; alderete@wsu.edu

Received: 22 June 2020; Accepted: 6 August 2020; Published: 10 August 2020

**Abstract:** Point-of-Care (POC) serum antibody screening of large cohorts of women and men at risk for the sexually transmitted infection (STI) caused by *Trichomonas vaginalis* requires the availability of targets with high specificity. Such targets should comprise epitopes unique to *T. vaginalis* immunogenic proteins detected by sera of women and men patients with trichomonosis but not uninfected controls. Three enzymes to which patients make serum IgG antibody were identified as fructose-1,6-bisphosphate aldolase (A), α-enolase (E), and glyceraldehyde-3-phosphate dehydrogenase (G). Epitopes within these proteins were identified that had no sequence identity to enzymes of humans and other pathogens. Therefore, I constructed a chimeric recombinant String-Of-Epitopes (SOE) protein consisting of 15-mer peptides, within which are the epitopes of A, E, and G. This novel protein of ~36-kD is comprised of two epitopes of A, ten epitopes of E, and seven epitopes of G (AEG::SOE2). The AEG::SOE2 protein was detected both by immunoblot and by enzyme-linked immunosorbent assay (ELISA) using highly reactive sera of women and men but not negative serum unreactive to *T. vaginalis* proteins. Finally, AEG::SOE2 was found to be immunogenic, as evidenced by serum IgG from immunized mice. I discuss how this approach is important in relation to infectious disease diagnostic targets for detection of serum IgG antibody in exposed and/or infected individuals and how such novel targets may have potential as subunit vaccine candidates against microbial pathogens.

**Keywords:** diagnostic; diagnostic targets; ELISA-enzyme linked immunosorbent assay; epitopes; immunogens; sera; serodiagnosis; sexually transmitted infections; *Trichomonas vaginalis*

### **1. Introduction**

*Trichomonas vaginalis* causes a non-viral sexually transmitted infection (STI) with adverse outcomes to infected women [1,2]. This STI is highly prevalent [3–5], and persistence within individuals may be due to the asymptomatic nature of infection. It is accepted that male partners of infected women with trichomonosis become infected. The organism and *T. vaginalis* DNA have been detected in hyperplastic prostate tissue [6,7], and there remains the possibility of a link between seropositivity to *T. vaginalis* in relation to prostate cancer (PCa) development [8–10]. More recently, a gene-expression model for *T. vaginalis*-mediated PCa was proposed [11], and other studies lend support to this hypothesis [6,7,12–15].

A rapid, inexpensive and specific serodiagnostic that could be used for screening large cohorts of at-risk individuals is desirable. A lateral flow, immunochromatographic Point-of-Care (POC) diagnostic (OSOMTM Trichomonas Rapid Test, Sekisui Diagnostics, Lexington, MA, USA) for rapid detection of active trichomonosis in women was invented in my laboratory [16]. Although the test meets criteria of being inexpensive, simple, rapid, and highly sensitive and specific, drawbacks of this test are that it is invasive for women, requiring a vaginal swab for obtaining sample, and the POC test

fails to detect the specific parasite protein in the urine of male patients. Although there are numerous reports of accurate nucleic acid amplification-based tests [17–19], these tests are neither compatible for large scale screening in non-sterile settings nor are suitable for use in community-based clinics and at under-developed countries.

It is acknowledged that advancing the prevention of STIs in general will require specific and sensitive POC tests [20]. Such POC tests should be rapid, inexpensive, and highly dependable for serum IgG antibody detection that can be employed for broad screening of populations regardless of geographic setting. POC diagnostics are needed for surveillance of the global burden of STIs in both developed and undeveloped countries. In the case of *T. vaginalis*, surveillance is necessary among sexually active populations [20], reinforcing the view that development of a serum-antibody POC test would advance the reproductive health of at-risk women and men. Control and even elimination of *T. vaginalis* and other STIs requires an approach and method for the development of highly specific serodiagnostic targets. In this report, I provide an approach for the identification and development of serodiagnostic targets using *Trichomonas vaginalis* as a model.

As infection by *T. vaginalis* results in an IgG response [8–11,21]; I hypothesize that an approach can be developed that will lead to the synthesis of a protein for detection of serum IgG to *T. vaginalis*. Using *T. vaginalis* as a model, I present the concept that a novel, chimeric protein comprised of a String-Of-Epitopes (SOE) can be synthesized as a serodiagnostic target. My laboratory has previously determined that women and men patients make serum IgG antibody to numerous *T. vaginalis* immunogenic proteins, including the enzymes fructose-1,6-bisphosphate aldolase (referred to as A), α-enolase (E), and glyceraldehyde-3-phosphate dehydrogenase (G) [21–23]. Epitope mapping of these proteins with women and men patient sera identified epitopes unique to the trichomonad proteins [21]. This earlier report showed a proof-of-principle for the construction of a novel recombinant chimeric protein, called AEG::SOE, with two each of the A, E, and G epitopes of the three enzymes. This earlier construct, however, failed to detect some positive sera when compared with the gold standard immunogenic truncated version of α-actinin called ACTP2 [8–10,24,25]. In this report I test this hypothesis and develop a stepwise approach to show that a new recombinant protein, two epitopes of A, ten epitopes of E, and seven epitopes of G (AEG::SOE2), is a serodiagnostic target equal to ACTP2. I discuss how the approach used here may advance the development of serodiagnostic targets for this and other STIs. Finally, I show that AEG::SOE2 is immunogenic in immunized mice.

### **2. Materials and Methods**

### *2.1. Epitopes Unique to the T. vaginalis A, E and G Proteins*

The identification of immunogenic epitopes reactive to women and men patient sera was done using oligopeptides (Custom Peptide Arrays) immobilized on membranes (SPOTs system; Sigma-Aldrich Corp, St Louis, MO, USA) as recently detailed [21,24]. As before, oligopeptides of fructose-1,6-bisphosphate aldolase (A), α-enolase (E), and glyceraldehyde-3-phosphate dehydrogenase (G) were derived from GenBank® accession numbers AAW78351 (A), AAK73099 (E), and AAA30325 (G). The protocols for probing of SPOTs membranes with sera were as detailed before [21]. The epitopes of A, E, and G reactive with women and men positive control sera were presented in an earlier publication [21]. Further, sequence identity analysis was performed with the enzyme homologs of human and other eukaryote and bacterial pathogens [21]. Finally, each 15-mer peptide within which the epitopes resided (Figure 1) was analyzed using the Immune Epitope Database and Analysis Resource (www.iedb.org) to show the linear nature of the peptide–epitope sequence.

**Figure 1.** Linear amino acid sequence of the chimeric, recombinant AEG::SOE2 protein. The peptide-epitope sequences linked by EE amino acids are underlined. Epitopes within the protein–epitope sequences are colored red. The letters A, E, and G refer to amino acid sequences derived from respective proteins. The epitopes were detected by either or both pooled women (W) or men (M) sera, as indicated, and the number next to W and M represents the order the epitope was identified within the protein during epitope mapping on the SPOTs system.

### *2.2. ACT-P2, a Truncated* α*-Actinin Protein Used for Screening of Patient Sera*

The α-actinin protein is one of the most immunogenic proteins of *T. vaginalis* [26,27] and has been shown to be a serodiagnostic target because it has no sequence identity to other known proteins in databases [24]. The truncated α-actinin protein called ACT-P2 of 558-amino acids (64.1-kDa) has been previously described [25]. ACT-P2 has been used to screen both male and female patient sera [24] and more recently to examine the relation of serum antibody in men and prostate cancer [8–10]. Thirteen epitopes were identified as reactive with women sera. In men, 5 of the 13 reactive epitopes were detected [24].

### *2.3. Plasmid Encoding AEG::SOE2 and ACT-P2*

The DNA coding sequence for the chimeric, recombinant AEG::SOE2 protein was synthesized by GenWay Biotech, Inc (San Diego, CA, USA). The pET-23a (+) plasmid that was prepared was encoded for ampicillin (Amp <sup>+</sup>) and chloramphenicol (Cam <sup>+</sup>) genes. The recombinant *E. coli* BL21DE3 cells were used to synthesize the protein. The description of ACT-P2 in *E. coli* has been described [24]. Both recombinant proteins have hexa-histidine at the carboxy terminus for purification [21,25].

### *2.4. Recombinant Proteins*

Detailed protocols for growing bacteria on Luria Broth (LB) agar plates with either 25 μg/mL Kanamycin (ACT-P2) or 100 μg/mL Ampicillin (AEG::SOE2) have been published [25]. A starter culture of LB medium with antibiotics was inoculated with recombinant *E. coli* and grown as before [21–24] prior to inoculation of medium and induction for expression with isopropyl β-D-1-thiogalactopyranoside (IPTG) for synthesis of recombinant proteins. Purification of the recombinant proteins was as before [21,24,25]. Purified proteins were obtained by Ni2<sup>+</sup>-NTA superflow affinity column chromatography (Qiagen Inc., Valencia, CA, USA), and soluble AEG::SOE2 was recovered from *E. coli* lysates using the Ni2+-NTA column, as evidenced by Figure 2. It is noteworthy that protein was also detected within inclusion bodies, and GenWay, Inc. indicates it is able to purify AEG::SOE2 from these inclusion bodies.

**Figure 2.** Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in 8% acrylamide of AEG::SOE2 from two different experiments (lanes 2 and 3) after purification by Ni2<sup>+</sup> NTA affinity chromatography. Lane 1 is of molecular weight (MW) standards, and numbers refer to daltons (×1000). Lane 4 is of 1μg bovine serum albumin (BSA) electrophoresed for comparative purposes.

### *2.5. Sera for Enzyme-Linked Immunosorbent Assay (ELISA) and Derivation of Positive*/*Negative (P*/*N) Scores*

Numerous reports have described the sera from women and men used for detection of ACTP2 [21,24]. Sera were also derived from research conducted at Washington University as reported earlier [8–10]. Institutional Review Board (IRB) approvals for collecting and using the sera were obtained at Washington University-Saint Louis, MO as well as by the IRB at Washington State University (approval number 01058) [8–10]. Equally importantly, the use of sera of patients has been reported before [28–30], and patients were diagnosed by microscopy and positive cultures.

My laboratory has published detailed descriptions for the use of sera by ELISA for detection of proteins, and identical materials and methods published earlier were used in this study to compare reactivities with ACTP2 and AEG::SOE2 [8–10,21,25]. Briefly, wells of microtiter plates prepared as before were stored at 4 ◦C prior to use. The processing of plates has been detailed before [25]. All washes of wells used phosphate-buffered saline (PBS), pH 7.4 containing 0.05% Tween-20 (PBS-Tween). Importantly, blocking of wells prior to addition of sera and dilutions of sera were done using a solution of 2% ELISA-grade bovine serum albumin (eBSA) (Sigma Chemical Co., St. Louis, MO, USA) prepared in PBS (eBSA-PBS). Where indicated, a 50 μL volume of undiluted hybridoma supernatant of a monoclonal antibody (MAb) or a cocktail of 25 μL for each of 4 MAbs (Figure 1) were used. Unless where indicated, all sera used for ELISA were diluted 1:25 (v/v) with eBSA-PBS.

In order to perform comparative analyses of ELISA using different targets, it was necessary to have serum standards as negative (scored as 0, 1+ and 2+) and positive (scored as 3+ and 4+) controls derived from testing ACTP2 as the gold standard for screening [8–10,24,25]. My laboratory previously determined the range of ELISA values for the different scores. Positive 3+ and 4+ sera detected trichomonad proteins by immunoblot, and 0, 1+ and 2+ sera did not detect proteins under the same

conditions [24]. For these experiments obtaining P/N values during ELISAs to provide scores 0 to 4+, the absorbance mean ± standard deviations were as follows: the blank control with eBSA-PBS was 0.050 ± 0.002, the 0 (zero) score was 0.131 ± 0.012, the 1+ score was 0.187 ± 0.010, and the 2+ score was 0.233 ± 0.023. The scores for 3+ and 4+ for obtaining absorbance values were 0.311 ± 0.025 and 0.441 ± 0.20, respectively. All scores were derived by subtracting the average blank optical density (OD) reading.

### *2.6. Mouse Anti-AEG::SOE2 Serum and Anti-T. vaginalis Serum and ELISA for Detecting Antibody to T. vaginalis*

Mouse anti-*T. vaginalis* serum was obtained by immunizing BALB/c mice as previously described [31]. Mouse anti-AEG::SOE2 serum (IMS) was derived by immunizing mice using the Washington State University Antibody Core Facility of the College of Veterinary Medicine. In this case, mice were immunized subcutaneously twice with 50 μg of purified protein (Figure 2) at 2-week intervals followed by the last booster injected into the tail vein. Prebleed normal mouse serum (NMS) was obtained prior to immunization and used as a negative control. All animals were treated humanely as governed by the Institutional Animal Use and Care Committee (IACUC number 6317) and National Institutes of Health protocols.

We also performed a whole cell ELISA for detecting antibody using microtiter wells coated with trichomonads. Parasites at logarithmic growth were washed three times with PBS, and a 50 μL suspension containing 1.25 <sup>×</sup> 105 organisms was added to individual wells of microtiter plates. After drying at 37 ◦C, 50 μL of 95% ethanol was added as fixative and wells were allowed to dry. Both protein-and trichomonad-coated wells were washed three times with PBS-Tween followed by blocking with 200 μL of eBSA-PBS. The remaining standard protocol is as previously reported elsewhere [8–10,21,24,25].

### *2.7. Reproducibility*

All experiments were performed at least three times under identical conditions. ELISAs on microtiter plates coated with proteins or trichomonads were done in quadruplicate unless otherwise indicated, and means and standard deviations were calculated. All statistical analyses were conducted with RStudio (Version 1.2.5033; RStudio, Inc.: Vienna, Austria), and figures were made in Prism (Version 8.4.3; GraphPad, LLC: San Diego, CA, USA). T-tests were used to examine differences in absorbance levels for ACTP2 and AEG::SOE2 for mouse sera, negative and positive human sera, and monoclonal antibodies. Statistical significance was defined as a *p*-value less than 0.05.

### **3. Results**

### *3.1. Epitopes of A, E, and G Unique to T. vaginalis and the AEG::SOE2 Protein Sequence*

Epitope mapping revealed that pooled women and men positive serum recognized a total of 12 epitopes of A, 18 epitopes of E, and 19 epitopes of G, for a total of 49 epitopes for the three proteins [21]. Table 1 shows there were only 2 of 12, 9 of 18, and 7 of 19 epitopes unique to the *T. vaginalis* A, E, and G proteins, respectively, that had no sequence identity with other bacterial, fungal, parasite, and human sequences [21]. The red underlined amino acid sequences are the epitopes detected by IgG antibody in the SPOTS system used for screening. Not surprisingly, the epitopes that were not unique to *T. vaginalis* proteins had identity, albeit to different degrees, with protein sequences of enzymes of other bacterial, fungal, parasite, and human proteins (data not shown).

Figure 1 presents the recombinant protein amino acid sequence representing the peptides containing the unique epitopes shown above in Table 1, and each peptide sequence was linked with two glutamic acid residues. The reactivity with women (W) and/or men (M) sera is shown above the red epitope amino acid sequences. The protein has a Mr of 35,896.31 daltons and pI of 5.05. Overall, there are 8 and 11 epitopes detected by positive women and men sera, respectively. The four

monoclonal antibodies (MAbs), labeled ALD 13, ALD 55, ALD 32, and ALD 30, are reactive with the epitopes, as indicated, and these MAbs have been previously characterized [21].


**Table 1.** The amino acid sequences containing the immunogenic epitopes (red underlined) of two epitopes of A, ten epitopes of E, and seven epitopes of G (AEG::SOE2) protein ‡ .

‡ AEG::SOE2 refers to the chimeric protein derived from epitopes unique to the *T. vaginalis* proteins fructose-1,6-bisphosphate aldolase (A), α-enolase (E), and glyceraldehyde-3-phosphate dehydrogenase (G), as described in Figure 1.

### *3.2. Purification of Recombinant AEG::SOE2*

Figure 2 shows the Coomassie-blue-stained gel after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of purified AEG::SOE2 from two different experiments (lanes 2 and 3). The relative mobility of the protein was compared with molecular weight standards (lane 1) and is consistent with the expected size of ~35.9-kDa. Lane 4 shows the stained band of 1 μg of BSA for comparison.

### *3.3. ELISA Using Di*ff*erent Amounts of AEG::SOE2-Coated Wells Probed with Di*ff*erent Antibodies*

ELISA was then performed to determine the amount of AEG::SOE2 immobilized onto microtiter wells detected by pooled positive patient sera, mouse anti-*T. vaginalis* serum [31], and a cocktail of MAbs reactive to the protein (Figure 1). As seen in Figure 3, 1μg (**A**) and 5 μg (**B**) of protein on wells were detected by the mouse antiserum at a 1:100 dilution (wells numbered 2), the cocktail of IgG1 monoclonal antibodies (MAbs) (wells numbered 3, 25 μL of hybridoma supernatant of each MAb), and pooled patient sera at a 1:25 dilution (wells numbered 4). Not surprisingly, the protein (Figure 2) was also detected by immunoblot with these antibody reagents. There was no detection of AEG::SOE2 even at 10 μg amounts using negative pooled sera of women and men (wells number 1). An irrelevant MAb to the actinin protein called HA423 [24], as shown also in Table 2, was unreactive to AEG::SOE2. One microgram amounts of AEG::SOE2 were chosen as the standard amount for ELISA as less than 1μg coated onto wells gave mixed results with patient sera and MAbs. The 1:25 dilution for patient sera was shown previously to be ideal for ELISA with 1μg of protein [8–10,21,24,25].

### *3.4. ELISA Comparing ACT-P2 and AEG::SOE2 with Di*ff*erent Antibodies*

We now compared AEG::SOE2 by ELISA with the gold standard ACTP2 for detection of IgG. Figure 4 compares ELISA values for negative human sera (NHS) versus positive women and men sera (PHS) characterized previously [21,24,25]. The pooled PHS known to have antibody to trichomonad proteins were five each of women sera (PHS-1), men sera (PHS-2), or combined women and men sera (PHS-3). ELISA results show statistically significant higher absorbance values for PHS to both ACTP2 and AEG::SOE2 when compared with NHS of five different women and five different men sera that lacked reactivity to trichomonad proteins by immunoblot. Additionally, sera of mice immunized with *T. vaginalis* (IMS) gave statistically significant higher values to both ACTP2 and AEG::SOE2 when compared with both prebleed, normal mouse sera (NMS), and secondary peroxidase-conjugated

goat anti-human IgG alone (P-G anti-H IgG). Finally, and not unexpectedly, the MAbs to AEG::SOE2 (Figure 2) gave statistically significant higher values with AEG::SOE2 but not ACTP2, and the MAb HA423 to ACTP2 was unreactive with AEG::SOE2 compared to statistically significant values with ACTP2. MAb L64 is a trichomonad cytoplasmic protein and was unreactive to both ACTP2 and AEG::SOE2.

**Table 2.** Testing by ELISA of women and men sera of different positive to negative (P/N) ‡‡ 0 to 4+ scores.


‡‡ Serum designations refer to the 0–4+ scoring of the sera following ELISA using standards as controls as described above. This scoring enables the examination of the relative reactivities against the ACT-P2 and ACT::SOE2 proteins. The red scores represent sera with reactivities to trichomonad proteins by immunoblot.

**Figure 3.** ELISA for detection of AEG::SOE2 at 1μg (**A**) and 5μg (**B**) AEG::SOE2 immobilized onto individual wells of 96-well microtiter plates. The negative controls were wells incubated with five different pooled negative sera each from women and men. This sera were shown previously to have no reactivity with any *T. vaginalis* proteins by immunoblot [24] (wells numbered 1). Wells with the different concentrations of AEG::SOE2 protein were incubated with mouse anti-*T. vaginalis* serum [31] (wells numbered 2), a cocktail of MAbs reactive with AEG::SOE2 epitopes as shown in Figure 1 (wells numbered 3), and with five different pooled positive women and men sera (wells numbered 4). The negative and positive pooled women and men sera have been previously reported [26]. Values were obtained by absorbance at 405 nm. As expected, a negative control irrelevant MAb to α-actinin called HA423 [24] was unreactive to AEG::SOE2 by ELISA, as shown in Figure 4. The ELISA was repeated on four different times with similar results.

**Figure 4.** ELISA comparing negative (NHS) and positive women and men sera (PHS) and prebleed, normal mouse serum (NMS), and immunized mouse anti-*T. vaginalis* serum (IMS), for detection of IgG antibody to ACT-P2 (dark blue) and AEG::SOE2 (light blue). Bars represent means and standard deviations that were calculated for the average of all ELISA performed (n = 8). The secondary peroxidase-conjugated goat anti-human IgG (Fc fraction; labeled P-G anti-H IgG) is the secondary antibody used for ELISA for detecting human antibody and gave values equal to the use of 2% eBSA-PBS alone as a negative control. NMS sera gave values equivalent to secondary antibody alone. The pooled negative human sera (NHS) lacking reactivity to trichomonad proteins [24,25] represented five different women and five different men sera. The three pooled positive human sera (PHS) that detect trichomonad proteins represented either five women sera (number 1), five men sera (number 2), or five combined women and men sera (number 3). Independent t-tests were used to compare the mean absorbance levels for each target protein to its corresponding control. This included IMS vs. P-G anti-H IgG (a), IMS vs NMS (b), PHS-1 vs NHS, PHS-2 vs NHS, and PHS-3 vs NHS. The MAb cocktail mix is comprised of equal volumes (25 μL) of hybridoma supernatants of the four MAbs to the epitopes of the recombinant AEG::SOE2 (Figure 2). HA423 is an MAb directed to ACTP2 and is reactive with ACTP2 but not AEG::SOE2 [21,24,25]. MAb L64 is an irrelevant control antibody that reacts with a cytoplasmic protein of *T. vaginalis* [22]. All monoclonal antibodies are of the IgG1 isotype. Lastly, differences in mean absorbance levels for each monoclonal antibody (i.e., cocktail MAb AEG, MAb HA423, and MAb L64) by protein type (i.e., ACTP2, AEG::SOE2) were examined. Absorbance values were obtained at 405nm. The means and standard deviations statistical significance is denoted as asterisks (\*) and found to be *p* < 0.001.

### *3.5. Di*ff*erent ELISA Experiments Testing Women and Men Sera of Di*ff*erent 0 to 4*+ *Scores*

Table 2 shows results of quadruplicate testing using up to ten individual women and men sera scored from 0 to 4+ based on reactivity to ACT-P2. Only 3+ and 4+ sera have been shown to have IgG antibody to trichomonad proteins [21,24,25]. Not unexpectedly, the negative women (W) and men (M) sera remained 0 to 2+ for ACT-P2 and AEG::SOE2. In contrast, and as shown within the red boxes, all of the 3+ and 4+ positive sera gave reproducible high reactions for both proteins.

For clarification of what the scores signify in terms of the mean and standard deviation (SD) absorbance readings for Table 2 and for Figure 5 and Table 3 below, the blank control with eBSA-PBS only was 0.050 ± 0.002. The mean and SD for sera giving different scores were as follows: sera with a 0 (zero) score was 0.131 ± 0.012, sera with a 1+ score was 0.187 ± 0.010, and the sera with a 2+ score was 0.233 ± 0.023. The sera with scores of 3+ and 4+ were 0.311 ± 0.025 and 0.441 ± 0.20, respectively. All scores were derived by subtracting from the average blank OD reading of 0.050.

### *3.6. ELISA of Women and Men Sera of 3*+ *and 4*+ *Scores at Di*ff*erent Dilutions*

Figure 5 compares the positive 3+ and 4+ sera reactivities toward ACTP2 (dark blue) and AEG::SOE2 (light blue) using women and men sera at different dilutions. Not shown is that the 0 to 2+ sera remained negative at all dilutions for both target proteins. For both ACTP2 and AEG::SOE2, the 3+ women (part A) and men (part B) sera became negative (2+ or lower) at dilutions of 1:50. Part A shows that the three replicates of the 4+ women sera were positive for ACTP2 at 1:100 dilutions. Only the first and second replicates were positive for AEG::SOE2 at 1:100 dilutions, but the third replicate sera were positive only up to the 1:50 dilution. Interestingly, the first replicate 4+ sera remained positive for AEG::SOE2 at 1:200 dilution. Part B shows that the first and second replicates of 4+ men sera were positive for ACTP2 at the 1:50 dilution, and only the first replicate sera was positive at the 1:100 dilution. For AEG::SOE2, only the first and second replicates remained positive at the 1:50 dilution. Overall, the data suggest that the 4+ sera of both women and men have higher titers of IgG than the 3+ sera for both target proteins.



‡ Refers to the number of the serum sample used for the ELISA comparisons of AEG::SOE2 and ACTP2. # All experiments were performed using quadruplicate wells. Agreement for 3+ and 4+ sera for both ACTP2 and AEG::SOE2 is denoted in yellow. The four 3+ sera that were reactive with ACTP2 but not AEG::SOE2 are denoted in green, and the four 3+ sera positive for AEG::SOE2 but not ACTP2 are denoted in blue.

**Figure 5.** ELISA of women and men sera of 3+ and 4+ scores at different dilutions. The sera designations for 3+ and 4+ scores following ELISA using standards as controls is as described above for Table 2. This scoring enables the comparative examination of dilutions with eBSA-PBS on the relative reactivities against the target ACT-P2 and ACT::SOE2 proteins. Each replicate consists of pooled different positive sera (n = 10) of women and men.

### *3.7. ELISA Using ACT-P2 and AEG::SOE2 as Targets with 0 to 4*+ *Individual Sera*

Finally, I randomly selected 42 each of sera of women and men with different scores for side-by-side evaluation. Table 4 presents results showing that, except for a few samples, there was almost 100% agreement for 3+ and 4+ positive sera for both ACT-P2 and AEG::SOE2. The score of 3++ had average and standard deviations much higher than the value for 3+ given above for Table 2. Likewise, there was almost 100% agreement for the 0 to 2+ negative sera. Interestingly, there were four 3+ sera that were reactive with ACT-P2 but not AEG::SOE2. Similarly, there were four 3+ sera that detected AEG::SOE2 but not ACT-P2. One possible explanation for these latter results is that the 3+ sera may be borderline negative, and this will require testing to determine whether or not there is IgG antibody that detects trichomonad proteins, such as by immunoblot, as before [21,24].

**Table 4.** Whole cell ELISA for detection of anti-*T. vaginalis* mouse serum IgG antibody and monoclonal antibody (MAb).


**\*** The peroxidase-conjugated goat anti-mouse IgG (Fc fraction) was used as a control and is the secondary antibody used for ELISA detection of mouse IgG antibody. It gave values equal to the use of 2% eBSA-PBS alone. Anti-*T. vaginalis* mouse serum has been described previously [29]. The normal mouse serum (NMS) and anti-AEG::SOE2 mouse serum are as described in Figure 4. MAb ALD30A and the MAb cocktail are to the epitopes of the recombinant AEG::SOE2 (Figure 1). MAb L64 is an irrelevant control antibody that reacts with a cytoplasmic protein of *T. vaginalis* and has been used previously [22,26]. All MAbs are the same IgG1 isotype. ‡ Absorbance values were obtained at 405nm. The mean and standard deviation (SD) were calculated for the average of ELISA readings, and all experiments were performed using quadruplicate wells. ‡‡ ND, not done.

### *3.8. AEG::SOE2 is Immunogenic, and Anti-AEG::SOE2 Serum IgG Antibody Detects T. vaginalis Organisms Immobilized onto Microtiter Wells by ELISA*

Finally, we wanted to examine whether mice immunized with AEG::SOE2 produced IgG antibodies. We compared mouse anti-AEG::SOE2 serum with mouse anti-*T vaginalis* serum [29] in individual wells of microtiter plates coated with fixed trichomonads. As shown in Table 4, both antisera gave ELISA readings greater than secondary peroxidase-conjugated goat anti-mouse IgG alone and prebleed normal mouse serum (NMS). Likewise, a cocktail of hybridoma supernatants of the MAbs reactive with AEG::SOE2 epitopes and MAb ALD30A alone (Figure 1) gave higher values compared to controls. The MAb L64 of the same isotype as the MAbs to AEG::SOE2 was used as another negative control. Furthermore, that MAbs to E and G react with trichomonads in this whole cell ELISA further supports earlier work that these metabolic enzymes are on the surface of *T. vaginalis* [23,24].

### **4. Discussion**

In this study, I use an approach to extend an earlier published work [21] and show the synthesis of a larger novel, chimeric String-Of-Epitopes (SOE) protein called AEG::SOE2 with additional epitopes of A, E, and G. This AEG::SOE2 protein possesses the same high specificity and sensitivity as ACT-P2, the gold-standard target for *T. vaginalis* seropositivity (Figure 4 and Table 2; Table 3). These data indicate that AEG::SOE2 may be a target for a rapid, accurate, and cost-effective POC test. Such a test would allow for screening of individuals with active *T. vaginalis* infection or permit identification of those

previously exposed to the organism. Another reason for moving toward a serum-based diagnostic is the demonstration of positive IgG seroconversion in relation to *T. vaginalis* and PCa development and progression [8–11]. The next step now appears to be development of a platform incorporating AEG::SOE2 in order to demonstrate a POC diagnostic for broad application of *T. vaginalis* surveillance.

As discussed recently [25], little is known of the temporal nature and duration of the serum IgG antibody responses among patients after infection with *T. vaginalis* and after diagnosis and cure. Such a POC test would also permit the medical community to understand the specific IgG response to the parasite in relation to active and/or past infections. My laboratory showed the short-lived nature of both serum and vaginal IgG to trichomonad cysteine proteinases after treatment of patients [28,29]. My laboratory also reported that IgG to a 230-kDa trichomonad protein was still evident in vaginal washes of patients at 4-weeks post cure [30]. I believe that these earlier findings support the view that a serodiagnostic test is necessary in order to understand the antibody responses of infected individuals. The availability of proteins unique to *T. vaginalis,* like α-actinin [24] and AEG::SOE2, provides the opportunity to elucidate the extent and nature of the antibody response issues in the future. The data (Figure 4 and Table 2) presented here show that AEG::SOE2 is equivalent to ACTP2 in serum IgG immunoreactivity. The presence of serum IgG antibody to epitopes unique to *T. vaginalis* proteins further reinforces the legitimacy of the approach taken here for identifying a specific and novel target for *T. vaginalis,* and although speculative, it may be possible to develop tests for other STI microbial pathogens and additional infectious diseases using this approach.

These novel, chimeric SOE proteins are comprised of immunogenic epitopes unique to the microbial pathogen of interest and, in this case, *T. vaginalis*. It is intriguing to consider that these SOE proteins may have efficacy as vaccines. At present, there is no evidence of immune protection against *T. vaginalis* despite the presence of both serum and vaginal antibody responses among patients with trichomonosis [28–30]. The many trichomonad proteases that degrade immunoglobulins may be a reason for immune evasion [29]. Reports have proposed that whole *T. vaginalis* organisms or lysates may be used as vaccines for *T. vaginalis* [32,33]. I argue against using *T. vaginalis* organisms or lysate as vaccines. One reason is that serum IgG antibody is made to epitopes of these enzymes, which in fact have amino acid sequence identities with human enzymes (referred to as trichomonad non-unique epitopes) [21]. In other words, as mentioned for Table 1, 10 of 12 epitopes of A, 9 of 18 epitopes of E, and 12 of 19 epitopes of G had sequence identity to human, bacterial, parasite, and fungal enzyme proteins. This is an important finding in and of itself that should be considered when studying host antibody responses to microbial pathogens. Whether antibodies to these trichomonad, non-unique epitopes common to human proteins mediate auto-immune reactions and, therefore, possible tissue damage is presently unknown. I believe that this issue must be considered within the framework of pathogenesis of trichomonosis and also for other infectious diseases. Indeed, it has been shown that human serum antibody to α–enolase of group A streptococcus cross-reacts with host tissues [34]. Thus, the approach described here may circumvent potential immune-crossreactive problems posed by using whole cell or lysate vaccines.

Another reason against whole organisms and lysates as vaccines is that we now know that *T. vaginalis* acquires onto its surface numerous host serum proteins [22,23,35–38]. The coating of the parasite surface with host proteins [36,38] may represent yet another mechanism for parasite evasion of immune-antibody responses. Further, the *T. vaginalis* surface-associated E and G metabolic enzymes are ligands that bind host proteins, such as plasminogen, fibronectin, collagen, and laminin [22,23]. As these host proteins may play a role in pathogenesis, the proteins are referred to as host-pathogenicity factors [22,23,35–38]. It is possible that host proteins on *T. vaginalis* may have altered structures exposing epitopes to host antibody responses, creating possible auto-antibodies and adverse reactions with tissues. This possibility deserves more attention in host–parasite interactions.

Of interest is that trichomonad lysates were recently compared with α-actinin by ELISA for IgG reactivity [39], and both were found to be equivalent in serum IgG detection. Here, too, I argue that lysates are inappropriate diagnostic targets for the same reasons mentioned above. Seropositive reactions may be due, in part, to IgG antibody responses to metabolic enzyme epitopes common to other pathogens. In this scenario, infections by other bacterial, parasite, and fungal pathogens may induce IgG not only to the epitopes of the enzymes A, E, and G used here but also to epitopes of other proteins that are immuno-crossreactive with *T. vaginalis* proteins. This conclusion has merit based on the findings presented here. This, then, would lead to false-positive reactions for this STI. Therefore, future serodiagnostic targets for this and other infectious diseases must have specific epitopes that are unique to the pathogen causing the disease. These concerns are relevant to the development of effective infectious disease diagnostics and vaccines and are also important considerations for surveillance and interventions of STIs and infectious diseases [40].

Finally, I have shown that the chimeric AEG::SOE2 protein is immunogenic, as evidenced by IgG antibody made by immunized mice (Table 4). Importantly, that anti-AEG::SOE2 serum detects whole organisms reaffirms the surface location of these proteins on *T. vaginalis* [22,23]. More importantly, multi-epitope constructs like AEG::SOE2 may be tested as a vaccine candidate for *T. vaginalis*, and this may be verified as was recently shown by others for α-actinin [41]. I believe that the approach described here may lead to future specific diagnostic targets and that such targets, comprised of immunogenic epitopes unique to the pathogen of interest, can be possible effective subunit vaccine candidates.

### **5. Conclusions**

A stepwise approach is presented that may have applicability for infectious diseases by producing a unique and specific serodiagnostic target to an infectious agent, and such SOE proteins can be tested for specificity as a diagnostic target. *Trichomonas vaginalis* was used as a model to test the viability of the approach. This approach includes i) identification of immunogenic surface proteins; ii) epitope mapping; iii) selection of epitopes with amino acid sequences unique to the trichomonad proteins; and iv) construction of a hybrid String-Of-Epitopes AEG::SOE2 protein comprised of A, E, and G epitopes. Finally, because of the highly immunogenic nature of the epitopes, as evidenced by patients' serum IgG reactivities (Figure 3; Figure 4 and Table 3), the fact that AEG::SOE2 itself elicited IgG antibodies via immunization (Table 4), and the fact that the epitopes have no identity to other known proteins in databanks, this novel SOE protein of *T. vaginalis* and SOE proteins of infectious agents in general may have potential vaccine applicability.

### **6. Patent**

J.F. Alderete. Strings of Epitopes Useful in Diagnosing and Eliciting Immune Responses to Sexually Transmitted Infections. No. 9910042, 5 March, 2018.

J.F. Alderete. Strings of Epitopes Useful in Diagnosing and Eliciting Immune Responses to Sexually Transmitted Infections. No. 10386369, 20 August, 2019.

**Funding:** This research received no external funding.

**Acknowledgments:** I thank a past collaborator at Washington University at St. Louis for providing sera that was highly seropositive for ACT-P2. I also thank past collaborators of the School of Medicine at The University of Texas Health Science Center at San Antonio for obtaining women and men patient sera used in prior studies by my laboratory there and used here at Washington State University. I want to acknowledge Grace Alderete for her voluntary assistance and laboratory maintenance throughout this work.

**Conflicts of Interest:** The author declares no conflict of interest. I alone designed the study and was responsible in the collection, analyses, or interpretation of data; in the writing of the manuscript; and in the decision to publish the results.

### **References**

1. Hobbs, M.M.; Sena, A.C.; Swygard, H.; Schwebke, J.R. Trichomonas Vaginalis and Trichomoniasis. In *Sexually Transmitted Diseases*; Holmes, K.K., Sparling, P.F., Stamm, W.E., Piot, P., Wasserheit, J.N., Corey, L., Cohen, M.S., Watts, D.H., Eds.; McGraw-Hill MedicalHolmes: New York, NY, USA, 2008.


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Validation of the Women's Views of Birth Labor Satisfaction Questionnaire (WOMBLSQ4) in the Spanish Population**

**María Dolores Pozo-Cano 1, Adelina Martín-Salvador 2,\*, María Ángeles Pérez-Morente 3,\*, Encarnación Martínez-García 1, Juan de Dios Luna del Castillo 4, María Gázquez-López 5, Rafael Fernández-Castillo <sup>1</sup> and Inmaculada García-García <sup>1</sup>**


Received: 4 June 2020; Accepted: 31 July 2020; Published: 2 August 2020

**Abstract:** The satisfaction of women with the birth experience has implications for the health and wellness of the women themselves and also of their newborn baby. The objectives of this study were to determine the factor structure of the Women's Views of Birth Labor Satisfaction Questionnaire (WOMBLSQ4) questionnaire on satisfaction with the attention received during birth delivery in Spanish women and to compare the level of satisfaction of pregnant women during the birth process with that in other studies that validated this instrument. A cross-sectional study using a self-completed questionnaire of 385 Spanish-speaking puerperal women who gave birth in the Public University Hospitals of Granada (Spain) was conducted. An exploratory factor analysis of the WOMBLSQ4 questionnaire was performed to identify the best fit model. Those items that showed commonalities higher than 0.50 were kept in the questionnaire. Using the principal components method, nine factors with eigenvalues greater than one were extracted after merging pain-related factors into a single item. These factors explain 90% of the global variance, indicating the high internal consistency of the full scale. In the model resulting from the WOMBLSQ4 questionnaire, its nine dimensions measure the levels of satisfaction of puerperal women with childbirth care. Average scores somewhat higher than those of the original questionnaire and close to those achieved in the study carried out in Madrid (Spain) were obtained. In clinical practice, this scale may be relevant for measuring the levels of satisfaction during childbirth of Spanish-speaking women.

**Keywords:** validation study; satisfaction questionnaire; birth attention; patient satisfaction

### **1. Introduction**

The birth of a child is one of the most significant events in the lives of women and their families. Knowing the level of satisfaction regarding the care received during the birth and postpartum periods is of special interest as it may help to improve the quality of health systems [1]. In Western countries, these experiences are becoming less frequent due to the drop in birth rates observed in many of them, especially in southern European countries, and it is hoped that the birth experience can become as rewarding as possible, despite not being free of serious consequences for the health of women and their newborns [1–3].

Many authors have explained the importance of women's satisfaction with the birth process, because it influences such important aspects as the maintenance of breastfeeding [4], which is crucial for the health of mothers and newborns [5,6].

When women experience unsatisfactory or traumatic births, their memories will be of pain, anger, fear, or sadness, and they may even suffer from post-traumatic stress disorders or may not remember anything about the delivery process [7–9]. Furthermore, a bad experience in a previous delivery increases the anxiety and fear in subsequent deliveries [10,11]. The proximity of childbirth activates memories of previous traumatic experiences and abuse as well as psychiatric disorders in women that can trigger a fear of vaginal childbirth and increase the demand for caesarean births, thus increasing the risks to maternal and perinatal health [12].

This is why, at present, the perceived satisfaction regarding care received during the birth is considered an essential indicator to measure quality of care [13]. Hodnett describes the personal expectations of pregnant women, the support and quality of the relationship with health professionals, especially midwives, and the participation of women in decision making as the most influential elements [14].

There are various instruments that measure the satisfaction of women with childbirth [2,15–22]. The Women's Views of Birth Labor Satisfaction Questionnaire (WOMBLSQ4) [23] has been used extensively in the recent literature and identifies women's satisfaction with their birth labor and delivery experiences, as well as the pain relief received during and after. It was developed in the United Kingdom by Smith and has been translated into French and validated to be applied to French-speaking women in University Hospitals in Geneva (Switzerland), and into Spanish, where Marín-Morales et al. did the same with women who gave birth in hospitals in Madrid (Spain) [23–25].

In the Autonomous Community of Andalusia (Spain), the Public Health System is committed to achieving excellence in healthcare. This is understood as a comprehensive concept involving multiple variables, among which citizen satisfaction is an inalienable element [26]. Birth care is focused on women, providing them with personalized care and promoting their autonomy and their role in decision-making [27].

The version translated into Spanish also presents discrepancies in the number of factors with respect to the original version and its translation into French due to significant convergence problems that make it necessary to eliminate the "control" factor, thus leaving the scale in nine dimensions. In order to assess the satisfaction of Andalusian mothers in the process of birth labor and to check the structure of the instrument, the WOMBLSQ4 scale was translated into Spanish and validated.

The objectives of this study are to determine the factor structure of the WOMBLSQ4 questionnaire on satisfaction with the care received during birth in Spanish women and to compare the level of satisfaction of pregnant women during the birth process with other studies that validated this instrument.

### **2. Materials and Methods**

### *2.1. Sample and Data Collection*

A cross-sectional study was carried out between January and March 2019 in puerperal women who had given birth in the Public University Hospitals of the city of Granada (Spain). In the year prior, an average of 5000 deliveries had taken place at both hospitals. Through intentional sampling, 385 Spanish-speaking puerperal women aged 18 years old or older were selected by collaborating with midwives in the studio. The included women voluntarily agreed to participate and signed an informed consent self-completed questionnaire that was delivered in a sealed envelope and later collected by the principal investigator. Those who did not understand Spanish and had elective caesarean births were excluded.

Postpartum surveys were administered to 450 women, of whom 15 refused to complete them, 40 did not deliver babies, and 10 did not provide informed consent. The questionnaires that were complete for all items were considered valid. The final sample consisted of 385 women, which constitutes a response rate of 85.5%.

### *2.2. Materials*

To evaluate women's satisfaction with care received during delivery, the final version of the WOMBLSQ4 scale was used, which consists of 32 questions with Likert-type responses and 10 dimensions: professional support during the birth (5 questions), expectations of delivery (4 questions), assessment at home at the beginning of birth labor (3 questions), first contact with the newborn (3 questions), support of the husband/partner during labor (3 questions), pain relief during labor (3 questions), pain relief immediately after delivery (3 questions), continuity (2 questions), environment during delivery (2 questions), and control (2 questions). The measure of general satisfaction involved two questions [23]. The factorial validity of the scale was confirmed, as well as an adequate global reliability (Cronbach's alpha 0.89), and the validity of the subscales was also shown (Cronbach's alpha values ranged between 0.62 and 0.91). The score for each dimension was obtained by adding the values obtained in each question (some of them with an inverse score), and later on, the result was transformed so that the minimum possible score was 0 and the maximum possible one was 100 (total satisfaction in the dimension) [23–25]. Higher scores indicated greater satisfaction on the part of the women.

For this research, the scale was translated into Spanish by two English language translators and its final content was agreed upon by three midwives with extensive experience in childbirth assistance. The translated questionnaire was piloted to 50 women, and it was demonstrated that the instrument presented an excellent level of comprehension and an adequate completion time since, when collecting it, the participants were asked if they had difficulty completing it, if they understood all the questions, and if it seemed too long.

In addition, the following sociodemographic variables were incorporated: age, marital status, educational level, and employment situation.

### *2.3. Data Analysis*

A descriptive analysis was performed in which means and standard deviations were calculated for the quantitative variables and frequencies and percentages for the qualitative ones. The factorial structure of the scale was explored by extraction of the main components followed by a Varimax rotation. In the first analysis, the Kaiser–Meyer–Olkin (KMO) sample adequacy measure was calculated, accepting values greater than 0.70 as optimal measures. Subsequently, the Bartlett sphericity test was applied to show significant differences between the items in the correlation and the unit matrix.

Next, the communality of each of the items on the scale was studied, and those that showed values less than 0.30 were eliminated, as they were poorly represented in the factorial set obtained. Those factors with eigenvalues greater than 1 were considered, and the percentage of variance explained with the said factors was determined to assess the weight of each one. After the rotation and analysis of the item saturation table, these were assigned to the dimension in which their saturation was highest. Once the items were eliminated, the previous steps were repeated in order to obtain the final factor structure. The internal consistency of each of the subscales was measured using Cronbach's alpha. Data analysis was performed with the SPSS v. Statistical package. 26.0 (International Busines Machines Corporation (IBM), Armonk, NY, USA) for Windows.

### *2.4. Ethical Considerations*

The study complies with the standards of good clinical practice, explicit in the European Directive 2001/20/EC and Law 14/2007 (of 3 July) on biomedical research. The treatment of personal data in health research is governed by the provisions of the Organic Law 3/2018, 5 December, Protection of Personal Data and Guarantee of Digital Rights in Spain. The research protocol obtained a favorable resolution from the Ethics and Research Committee of Health Institutions.

### **3. Results**

The sociodemographic characteristics of the analyzed sample are reflected in Table 1. The mean age of the participants was 31.62 years (SD 5.32), with a range of 18 to 46 years old. Regarding the level of education, almost half (175, 46.2%) had a university-level education. In relation to marital status, the majority were married or had a partner (359, 94.0%).


**Table 1.** Sociodemographic characteristics of the sample.

Regarding the labor situation, 184 (47.9%) were employed workers.

### *3.1. Exploratory Facial Analysis*

To carry out the factor analysis, firstly, all items on the scale were considered, and a mean KMO sample adequacy of 0.80 was obtained, with the result of the Bartlett sphericity test being statistically significant (*p* < 0.001).

Of the 32 items in the original questionnaire, only three showed communalities below 0.50: 25 (I am satisfied with just one or two things about the labor care that I received: 0.441), 31 (I didn't need a lot of pain relief after the birth: 0.475) and 12 (The way my labor care was provided could not have been improved: 0.476). However, they have not yet been removed from the questionnaire.

Using the main components method, nine factors that showed self-values greater than one, explaining 68.0% of the global variance, were extracted. A new dimension (3) was designed pain during and after delivery—after merging dimensions six (pain during delivery) and seven (pain after delivery) from Smith's original questionnaire [23].

Table 2 shows the Cronbach's alpha and variance explained by each factor, as well as the saturations of each item, once the Varimax rotation had been performed.




Item 12 (the way my labor care was provided could not have been improved) showed a saturation of 0.55 and the generalization of its statement could be confusing. Item 25 did not saturate well with respect to the other two (0.46), and due to its statement (I am satisfied with just one or two things about labor care that I received), it did not seem to correspond to the being analyzed. Finally, both items were removed from the questionnaire.

Later on, a second analysis was performed with the remaining items, obtaining a sample adequacy of KMO of 0.86 and maintaining statistical significance in the Bartlett sphericity test (*p* < 0.001). This time, only items 3 and 31 showed communalities of less than 0.50, (0.44 and 0.47, respectively), although we decided to keep them in the model. The number of factors extracted by the principal component method with eigenvalues greater than 1 was also nine, which explained 70.0% of the global variance.

Table 3 represents the saturation level in the rotated components, the corresponding Cronbach's alphas, and the variance explained by each factor.

In Table 3, the dimension of pain again appears to be merged. Item 31 (I didn't need a lot of pain relief after the birth) has a saturation level close to 0.50 and continues to remain on the scale, although it is poorly associated with the other items, because its contents belong to this dimension.

Table 4 shows the Cronbach's alpha values from the validation carried out in this study as well as those from the English version, the French adaptation, and the puerperal period in Madrid (Spain). It can be seen that the Cronbach's alpha values of this study are in the range of previous studies or, in some cases, even higher.

### *3.2. Level of Satisfaction in the Di*ff*erent Versions*

Table 5 shows the mean scores in each of the dimensions for the different versions. It can be seen that the three best valued dimensions in the four versions were professional support, support of the husband, and first contact with the newborn.






**Table 4.** Cronbach's alpha values from the different studies analyzed.

**Table 5.** Average scores in the different versions.


### **4. Discussion**

The response rate was 85.0%, similar to that obtained in the study by Floris et al. [24] in Geneva, Switzerland, and somewhat higher than that obtained by Marín-Morales et al. [25] in Madrid (Spain).

The scale designed in this study showed a high validity and some good psychometric characteristics for measuring childbirth satisfaction in women based on their sociocultural environment.

It can be used in primiparous and/or multiparous women; pregnant women of low, medium, and high risk; puerperal women who have had vaginal births, whether spontaneous or instrumental; and even those who have had unscheduled caesarean sections. However, it is not applicable for women who have had scheduled (elective) caesarean births since, in most cases, these women would not be able to complete some items. In these circumstances, other dimensions not considered in the original scale should be considered.

The obtained percentage of women's satisfaction with the care received during their births by factor analysis was somewhat lower than that shown by the original scale [23]; however, we understand that it is adequate for identifying those aspects that can be improved.

In relation to the first dimension, "Professional support", the psychometric characteristics of our study showed slightly lower values than those of Smith [23], but higher than those achieved by Floris et al. [24] and Marín-Morales et al. [25]. This is the first factor identified in all of these studies and the one that best explains women's satisfaction. It is made up of five items, all of them stated in a positive way, and results similar to those of previous studies were obtained [28–31]. All of them indicate that the kind and correct treatment of professionals and good communication favor the satisfaction of women, especially highlighting the role of the midwife as the professional who provides the most support during the delivery process, describing her as "competent", "inspiring confidence", or "wonderful" [32].

In the second dimension, "Expectation", made up of four questions expressed in a positive way, the parameters obtained are similar to those obtained in the previous dimension. In their research, Melender et al. postulated that if the expectations of the pregnant women are in accordance with their lived experiences during childbirth, their evaluation of childbirth will be satisfactory [33]. Many women look for information on the sensations that they may experience during the labor process and idealize how it should go. In preparation for childbirth sessions the expectations created, information from other mothers, previous experiences, and the signals of their own bodies influence the elaboration of a mental image of delivery [34]. In other cases, despite experiences of severe pain or complications during a previous delivery that are different from their expectations, women feel motivated and encouraged to have another child due to having received good support from the midwife during the process [32].

The third dimension, "Pain during and after childbirth", integrates two dimensions of the original version, which also appears in the French version. Our results coincide with a study carried out on Spanish women from Madrid (Spain). In both cases, the items were related to pain. In this new dimension, there are 2 items stated positively and 4 negatively, which coincides with the original and French versions. The reliability of the original version is superior to that of the other studies, while the results of this investigation are superior to the version carried out with women from the center of Madrid (Spain) and partially superior to the French version.

Item 31 "I didn't need a lot of pain relief after the birth" is the only one of these new dimensions that has an adjusted value in our study. This is probably explained by the fact that most of the women in our sample received epidural analgesia for childbirth, and the effects of this analgesia remain in the immediate postpartum period [35–37]. This fact means that puerperal women do not need many analgesics in this period.

In the early puerperium period (from 3 h after birth to 10–15 days later), women may experience pain due to uterine involution, the presence of hemorrhoids, and even breast pain [38,39]. There may also be perineal pain following or without an episiotomy [39,40], but in most cases, pain relief is necessary. On the other hand, this item is not relevant for those women who wanted a natural childbirth and therefore would consider it unnecessary. Therefore, it could be a potentially upgradeable question.

The fourth dimension, "Home assessment", presents a very similar Cronbach's alpha value in the four versions, the highest of which corresponds to the original version. In all versions, with the exception of the version from Madrid (Spain), this dimension consists of three items written in negative form. The coefficients obtained in our study were slightly lower than those obtained by Smith [23], but higher than those found by Marín-Morales et al. [25].

In Spain, as in most countries of the Organization for Economic Cooperation and Development (OCEDE) [41], except in the Netherlands and the United Kingdom, there is no culture of maternal care at home [42,43]. In the Spanish Health System, both public and private, most pregnant women go to hospitals when they have their first contractions to be evaluated at the beginning of labor and births are completed in them. Anecdotally, births at home are attended by professionals from the private sector.

In relation to the fifth dimension, "Support from husband", the psychometric parameters obtained are similar to those in the original version. Item 29 "I would have preferred to have more support from my partner/husband" is negative, and its value is the lowest on our scale. This factor identified in the analysis is a component of satisfaction that, in our environment, is favored by legislation [44]. Along the line of humanization of perinatal care, in 1995, in the Autonomous Community of Andalusia, the right of pregnant women to be accompanied by a person they trust during the prepartum, delivery,

and postpartum periods was legislated [45]. This legal support for the figure of the companion has been highly valued in numerous studies [30,32,46].

The sixth dimension, "Holding baby", consists of two negative and one positive item, and the psychometric characteristics found in our study have somewhat higher values than those obtained in Madrid (Spain) [25]. Early contact with the newborn, in addition to being an indicator of women's satisfaction, favors the establishment of an emotional bond between mother and child [47]. The search for greater prominence, that is granted to women in our Public Health System through the implementation of the Childbirth and Birth Plan of the Autonomous Community of Andalusia, provides mothers with the possibility of expressing their preferences during birth as the right to have their son or daughter by their side during the hospital stay [46,48]. Whenever the state of the newborn and the mother allow it, skin-to-skin contact between the two should be promoted, since it provides benefits to both the mother and the newborn: maintenance of a good body temperature, an increase in blood glucose levels, and helps to maintain breastfeeding and weight [49–52].

The seventh dimension, "Knowledge of women by professionals during birth assistance", showed an adequate Cronbach's alpha value. The title of this dimension is formulated in the same sense as the French version and both differently from the original version. The content of the questions is focused on the continuity raised by Australian authors [53,54] and adapted to the title formulated in our research.

The values of the parameters obtained in the eighth dimension, "Environment", are adequate and most of them are superior to those collected the study of the Center of Spain [25]. This dimension consists of two questions and one of them is negative. An environment that facilitates intimacy, silence, environmental warmth, and the absence of medicalized furniture contributes to the satisfaction of women [55,56]. However, in the qualitative study carried out by Jenkins et al. in the state of New South Wales (Australia), most women did not highlight the environment as one of the three most important aspects in their care [53].

"Control" was the ninth dimension and showed a great relationship with the satisfaction of women and their experience in childbirth. Various authors have pointed out that perceived control over the situation increases satisfaction; this is a dimension that has been widely incorporated in Anglo-Saxon research [53,55,57–60]. However, in our study, similarly to that of Smith and Floris et al., this dimension was the last to be shown as a factor and also the one that explained the smallest percentage of satisfaction [23,24]. Probably for this reason, this dimension was excluded in the version carried out by Marín-Morales et al. with women from Centre of Spain [25]. This result shows that, despite the fact that, in recent years, it has been an incentive by the State and Health Institutions for women and families to take control and responsibility over their health through the inclusion of the Autonomy Law of the patient (2002), the change in the healthcare model based on paternalism has prevailed for so many years in our healthcare system that it is not yet something that the population considers to be of special importance [61].

The two questions about General Satisfaction, number 12 (the care during the delivery process could not have been better) and number 25, (I am satisfied with only one or two things about the care I received during the delivery process) were not analyzed as in the original study. Both questions, especially the last one, are so general that they suggest a certain ambiguity, since it is difficult to assess satisfaction with one or two of these aspects. However, the resulting mean scores are mostly somewhat higher than those obtained by Smith [23] in the original version, although they are similar to those obtained by Marín-Morales et al. [25] in their study carried out in Madrid (Spain).

We understand that, at present, childbirth satisfaction questionnaires should incorporate the control dimension, since, in the current context, the empowerment of women and decision-making during childbirth is a priority in healthcare [62].

### *4.1. Limitations*

Given the good understanding of the scale items and the good adaptation of the scale, it would have been interesting to have expanded the sample to other Spanish-speaking areas.

### *4.2. Recommendations for Future Research*

It would be advisable for future research to merge dimensions 6 "pain relief during childbirth" and 7 "pain relief immediately after childbirth" into one dimension, as shown in the original version. Both dimensions are related to pain relief.

In this dimension, there is item 31, which asks about the need to relieve pain immediately after delivery. We believe that this question should be changed, as many analgesics are not necessary immediately after delivery, only if in the postpartum period.

### **5. Conclusions**

In the resulting model from the WOMBLSQ4 questionnaire, the nine dimensions measure the level of satisfaction with childbirth care in puerperal women. The resulting average scores are mostly somewhat higher than those obtained in the original version, and close to those achieved in the study carried out in Madrid (Spain). In clinical practice, this scale is relevant for measuring satisfaction levels in Spanish-speaking women.

**Author Contributions:** Conceptualization, M.D.P.-C. and A.M.-S.; methodology, M.D.P.-C., A.M.-S., M.Á.P.-M., J.d.D.L.d.C. and I.G.-G.; software, J.d.D.L.d.C. and I.G.-G.; investigation, M.D.P.-C.; data curation, M.D.P.-C., E.M.-G., J.d.D.L.d.C. and I.G.-G.; writing—original draft preparation, M.D.P.-C., A.M.-S., M.Á.P.-M., M.G.-L., R.F.-C., I.G.-G.; writing—review and editing, M.D.P.-C., A.M.-S., M.ÁP.-M., E.M.-G., M.G.-L., R.F.-C. and I.G.-G; supervision, M.D.P.-C and I.G.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We gratefully acknowledge the women who participated in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Contribution of Chronic Fatigue to Psychosocial Status and Quality of Life in Spanish Women Diagnosed with Endometriosis**

**Antonio Mundo-López 1,2, Olga Ocón-Hernández 3,4, Ainhoa P. San-Sebastián 1, Noelia Galiano-Castillo 3,5,6, Olga Rodríguez-Pérez 1, María S. Arroyo-Luque 1, Manuel Arroyo-Morales 3,5,6, Irene Cantarero-Villanueva 3,5,6, Carolina Fernández-Lao 3,5,6,\* and Francisco Artacho-Cordón 1,3,6,7,\***


Received: 15 April 2020; Accepted: 23 May 2020; Published: 28 May 2020

**Abstract:** Aim: To analyze the levels of chronic fatigue in Spanish women with endometriosis and its relationship with their psychosocial status and quality of life (QoL). Methods: A total of 230 Spanish women with a clinical diagnosis of endometriosis were recruited. Chronic fatigue (Piper Fatigue Scale) and pelvic pain (Numeric Rating Scale) were evaluated. An on-line battery of validated scales was used to assess psychosocial status [Hospital Anxiety and Depression Scale, Scale for Mood Assessment, Pain Catastrophizing Scale, Pittsburgh Sleep Quality Index, Gastrointestinal Quality of Life Index, Female Sexual Function Index and Medical Outcomes Study-Social Support Survey] and QoL [Endometriosis-Health Profile questionnaire-30]. Associations between fatigue and both psychosocial and QoL outcomes were explored through multivariate regression models. Results: One-third and one-half of women showed moderate and severe fatigue, respectively. Fatigue was associated with higher anxiety and depression, poorer sleep quality, poorer sexual functioning, worse gastrointestinal health, higher catastrophizing thoughts, higher anger/hostility scores and lower QoL (*p*-values < 0.050). Moreover, fatigue and catastrophizing thoughts showed a mediating effect on the association between pelvic pain and QoL. Conclusion: This work reveals the important role of fatigue in the association between pain, psychosocial status, and QoL of Spanish women with endometriosis.

**Keywords:** chronic fatigue; endometriosis; psychosocial status; quality of life

### **1. Introduction**

Endometriosis, characterized by the ectopic development of endometrial-like tissue, is among the most commonly diagnosed benign diseases in women of reproductive age [1]. Diagnostic delay and the fact that diagnosis is often overlooked by primary care doctors make the prevalence of the disease difficult to establish. Nevertheless, prevalence estimates range from 1–2% when considering populations at low risk to 10% when high-risk populations are considered [2]. However, despite the

benign nature of this disease, huge direct and indirect costs (raising up to more than \$12,000 – \$15,000 in some countries) have been evidenced to be associated with endometriosis [3].

Pain in the pelvic region is acknowledged to be the most characteristic symptom of women with endometriosis, which is intensified during the menstruation period (dysmenorrhea) and during the performance of daily activities such as defecation (dyschezia), urination (dysuria) or sexual relationships (dyspareunia) [4,5]. The contribution of pelvic pain (PP) to the psychosocial status and the quality of life (QoL) of women with endometriosis has been extensively addressed [6–10]. Additionally, chronic fatigue, i.e., the perception of physical tiredness and lack of energy distinct from sadness or weakness, is another endometriosis-related symptom, as recently identified in women with endometriosis [11]. However, the role of chronic fatigue on patients' lives has been poorly addressed, although a few qualitative studies have indicated that affected women ascribed social and work impairments to fatigue [12,13]. Contrary, there is consistent evidence of the relevant role of fatigue in different subsets of patients experiencing chronic pain, suggesting that fatigue hinders the completion of routines and significant activities, and therefore, severely reduces QoL [14,15].

Moreover, the relevant contribution of chronic fatigue to the presence of psychosocial impairments in patients with autoimmune diseases [16] or neurological problems [17] has been reported. However, contrary to the well-established relationship between chronic pain, psychosocial problems and QoL in women with endometriosis, there is a scarcity of published studies addressing the contribution of chronic fatigue to the symptomatic burden in women with endometriosis under medical treatment. Thus, the aim of this study was to explore the presence of chronic fatigue in Spanish women diagnosed with endometriosis and its contribution to the psychosocial status and QoL.

### **2. Material and Methods**

### *2.1. Study Population*

A total of 230 women with a clinical diagnosis of endometriosis, from different regions of Spain, were enrolled in this cross-sectional study from January to July 2019. Recruitment of women was carried out in combination with both gynecologists and Spanish associations of endometriosis patients, which advertised the study in their social networks. The inclusion criteria were: to have attended a gynecological visit with any participating gynecologist or to belong to any of the Spanish associations of endometriosis patients; to be diagnosed with endometriosis (either by laparoscopy, magnetic resonance or ultrasound imaging, or based on symptoms); and to have the ability and availability to use an electronic device with internet connection (computer, tablet or mobile phone). The exclusion criteria were: to live in another country or to be a non-Spanish speaker. The survey was designed to consider researchers', gynecologists' and patients' opinion about the most relevant aspects that should be addressed. Interested women received a link for the completion of an on-line questionnaire. Prior to this, women were informed about the nature and objectives of the study, and they were requested to read and sign the informed consent. No personal information was asked in the questionnaire, and data was extracted to create an anonymized database. This study was carried out following the principles of the Declaration of Helsinki and Biomedical Research Law 14/2007 and was approved by the Research Ethics Committee of Granada (1733-N-18).

### *2.2. Assessment of Self-Reported Intensity of Chronic Fatigue and Pain*

Chronic fatigue was assessed through the Spanish version of the Piper Fatigue Scale (PFS) [18,19]. PFS is a validated 22-item tool for self-reported chronic fatigue in breast cancer survivors, but also in patients with gynecological disorders [20] or coronary diseases [21]. It includes four dimensions of subjective fatigue: "behavioral/severity", "affective meaning", "sensory" and "cognitive/mood". Scores range from 0 to 10, with higher scores indicating greater fatigue. It has demonstrated high reliability and validity (Cronbach's alpha 0.86). Participants were divided into three groups according

to the clinically significant fatigue criteria: mild (≤4.0), moderate (4–7) or fatigued (≥7), according to the value obtained for the PFS total score [18,22].

Intensity of self-reported PP during the last week was assessed through a numeric rating scale (NRS). This 11-point Likert scale (0 = no pain; 10 = unbearable pain) is one of the best single-item methods available to estimate the intensity of pain [23–25]. Pain severity was categorized as mild (0–3), moderate (4–7) and severe (8–10), as reported elsewhere [26].

### *2.3. Psychosocial Assessment*

Participants were asked to complete the Hospital Anxiety and Depression Scale (HADS), the Scale for Mood Assessment (EVEA), the Pain Catastrophizing Scale (PCS), the Pittsburgh Sleep Quality Index (PSQI), the Female Sexual Function Index (FSFI), the Gastrointestinal Quality of Life Index (GIQLI) and the Medical Outcomes Study-Social Support Survey (MOS-SSS).

Anxiety and depression were assessed through the Spanish version of the HADS [27], a self-assessment mood scale validated for its use in non-psychiatric hospital outpatients to determine their levels of anxiety and depression [28]. It has two subscales (anxiety and depression), each ranging from 0 to 21, showing adequate reliability (Cronbach's alpha 0.86) [29]. Higher scores on the subscale indicate higher degrees of anxiety and depression [30]. For both subscales, available cut-off scores allowed the identification of non-cases (≤7) mild (8–10), moderate (11–14) and severe cases (15–21) [28,31]. Additionally, the Spanish version of the EVEA scale was partially used to evaluate "anger/hostility" and "happiness" through the corresponding subscales. Item scores are evaluated with Likert scales ranging from 0 to 10, and the values per category are obtained from mean scores. EVEA subscales have shown good reliability (Cronbach's alpha range between 0.88 and 0.93) [32].

Catastrophic thoughts about pain were assessed through the Spanish version of the PCS, a 13-item, validated, self-report instrument with adequate reliability (Cronbach's alpha 0.79) [33]. This measure has a 5-point Likert-style response scale and the scoring range is 0–52, with higher scores indicating higher levels of catastrophic thoughts. Previous studies have shown a cut-off of more than 30 points to be associated with clinical relevance [34].

Sleep quality was assessed using the Spanish validated version of the PSQI [35]. The PSQI is a 19-item, validated, self-report scale used to measure quality and patterns of sleep, with adequate reliability (Cronbach's alpha 0.87). Scores range from 0 to 21, with higher scores representing poorer sleep quality [36]. It has been proposed that a total score ≤5 indicates good sleep quality while a total score > 5 indicates poor sleep quality [36].

Sexual function was assessed through the Spanish version of the FSFI [37]. This is a 19-item questionnaire, validated, multidimensional self-report instrument for assessing the major aspects of female sexual dysfunction and sexual satisfaction [37,38]. The FSFI score ranges from 0 to 36. Higher scores represent better sexual function, considering that patients with a FSFI total score below 26 are categorized as sexual dysfunctional, whereas those scoring at or above this cut-off score are categorized as sexually functional [39]. Adequate reliability has been reported (Cronbach's alpha >0.70 for all domains).

Digestive complaints were assessed through the Spanish version of the GIQLI [40], a self-administered 36-item questionnaire concerning digestive symptoms, physical status, emotions, social dysfunction and effects of medical treatment. Each item scores from 0 to 4 with the total score ranging from 0 to 144, higher scores representing better quality of life. GIQLI also measures physical well-being, mental well-being, digestion and defecation [41].

The Spanish version of the MOS-SSS scale was used to assess the extent to which the person has the support of others to face stressful situations [42]. It is comprised of 19 items with a 5-point Likert-style response, with higher scores representing better social support. This measure has shown good psychometric quality in different studies using diverse populations and clinical scenarios (Cronbach's alpha 0.94) [43].

### *2.4. Quality of Life*

The Spanish version of the validated Endometriosis Health Profile-30 (EHP-30) questionnaire was used for the assessment of QoL in participating women [44]. This 30-item scale has five subscale scores: pain, control and powerlessness, social support, emotional well-being and self-image. Each subscale is standardized on a scale of 0–100, where 0 indicates the best health status and 100 the worst health status. Scale scores for each scale are calculated from the total of the raw scores of each item in the scale divided by the maximum possible raw score of all the items in the scale, multiplied by 100. This instrument has shown good internal consistency reliability, with Cronbach's alpha >0.88 for all subscales.

### *2.5. Statistical Analysis*

The sociodemographic and gynecological characteristics of participants and scores for PP and chronic fatigue were expressed as geometric means (GMs) with geometric standard deviation (GSD), or as percentages, depending on the continuous or categorical nature of the variable. Scores for QoL, i.e., psychosocial outcomes, were expressed as GM with GSD, as minimum and maximum values, and as percentiles (25, 50, and 75). When clinical cut offs were available, variables were categorized and expressed as percentages.

To improve normality of the data, psychosocial outcomes were log-transformed and, therefore, β coefficients are also presented as exp(β). Associations between fatigue severity (mild/moderate/severe), psychosocial outcomes and QoL were assessed by using linear regression models adjusted for sociodemographic and gynecological characteristics, including age, schooling, civil status, severity of premenstrual syndrome (none, mild, moderate or severe), type of diagnosis, time since diagnosis and number of surgeries. Additional models adjusted for severity of last week PP are also presented. Moreover, the mediation effect of fatigue and pain catastrophizing thoughts on the relationship between last week PP intensity and QoL was assessed through the macro PROCESS for Statistical Package for the Social Sciences (SPSS) [45], and mediating effects were considered significant when zero was not located within the confidence intervals.

The statistical significance level was set at *p* = 0.05. Analyses were performed using SPSS v23.0 statistical software (IBM, Chicago, IL, USA), while figures were designed with Graphad Prism 5.0 software (San Diego, CA, USA). The post-hoc analysis to estimate the power (1-β) of the statistical analysis was conducted using G\*Power 3.1.9.7 statistical software (Düsseldorf University, Düsseldorf, Germany). For the main analysis between chronic fatigue and QoL, it revealed that, for an R2 of 0.28 assuming an α-error of 0.05, the power was >0.99.

### **3. Results**

A total of 241 women were interested in the study. However, 11 (4.6%) women did not meet inclusion/exclusion criteria. Finally, 230 women agreed to participate. Baseline characteristics of the study population are summarized in Table 1. The mean (±SD) age of the study population was 36.7 ± 5.2 years, the majority of them hold a university degree (53.9%), are currently working (64.3%) and declared the presence of premenstrual syndrome at any level of severity (56.6%). A total of 155 out of 230 women had a laparoscopic confirmation of the presence of endometriosis lesions at the time of this survey, while in 62 (27.0%) the diagnosis was based on magnetic resonance imaging (MRI) and/or ultrasound (US) imaging techniques. Only 13 (5.7%) were diagnosed based on symptoms but not confirmed by MRI and/or US imaging. Finally, the mean time since diagnosis was 5.0 ± 5.3 years, and 68 (29.6%) had undergone at least two endometriosis surgeries.

### *3.1. Intensity of Chronic Fatigue and Pain in Spanish Women Diagnosed with Endometriosis*

Self-reported severity of chronic fatigue and last week PP are summarized in Table 2. GM (±GSD) intensity of chronic fatigue was 5.9 ± 1.7 points, with almost half of the participating women reporting

severe fatigue. Concerning last week PP intensity, GM (±GSD) was 5.0 ± 1.9. A total of 46.3% and 27.1% of the entire study population showed moderate and severe PP during the last week. Using multivariate linear regression modelling, a positive association was found between intensities of both chronic fatigue and last week PP scores after adjustment for potential confounders (Supplementary Table S1).


**Table 1.** Baseline characteristics of the study population (N = 230).

\* Geometric mean ± geometric standard deviation; PMS: premenstrual symptoms; MRI: magnetic resonance imaging; US: ultrasound.

**Table 2.** Intensity of chronic fatigue and pelvic pain in women with endometriosis.


\* Geometric mean ± geometric standard deviation.

### *3.2. Psychosocial Impairments and Quality of Life in Spanish Women Diagnosed with Endometriosis*

Table 3 summarizes the results from the descriptive analysis of each analyzed psychosocial dimension. Considering available cut-off scores (not shown in tables), anxiety was present in 169 out of 230 (73.4%) of the women, with 58 (25.2%) and 48 (20.9%) showing moderate and severe anxiety, respectively. Similarly, depression was detected in 111 (48.3%) of the women, with 45 (19.6%) and 19 (8.3%) showing moderate and severe depression, respectively. Additionally, pain catastrophizing thoughts were found in 108 (47.0%) of the participants, while poor sleep quality and sexual dysfunction were found in 187 (81.3%) and 174 (75.7%) of the participating women, respectively. Moreover, anger/hostility and happiness dimensions, assessed through the EVEA subscales, showed a GM (±GSD) of 12.2 ± 2.6 and 11.1 ± 2.3 points, respectively. GM (±GSD) MOS-SSS score was 73.7 ± 18.7 points, while for gastrointestinal problems, GM (±GSD) GIQLI score was 65.2 ± 1.4 points. Regarding QoL, GM (±GSD) EHP-30 score was 55.0 ± 1.7 points.


**Table 3.** Psychosocial status and quality of life in women with endometriosis.

GM: geometric mean; GSD: geometric standard deviation.

### *3.3. Contribution of Fatigue Intensity to Psychosocial Impairment in Spanish Women*

Results from the multivariate analyses assessing associations between self-perceived fatigue severity and psychosocial impairments are depicted in Figure 1. Results from the bivariate and multivariate analyses are summarized in Supplementary Table S2. After adjustment for potential confounders (sociodemographic and gynecological characteristics and intensity of PP during the last week), moderate and severe fatigue was found to be related to anxiety and depression, poorer sleep quality, poorer sexual functioning and less gastrointestinal quality of life in an intensity-dependent manner, while higher PCS and EVEA-anger/hostility scores were associated with severe fatigue. Moreover, multivariate logistic regression analyses that were run in parallel when cut-off points were available showed similar results (data not shown in tables). Sensitivity analyses stratified by endometriosis diagnosis yielded similar results.

**Figure 1.** Influence of chronic fatigue on psychosocial status in Spanish women with endometriosis. Results from multivariate linear regression analyses adjusted for age (years), schooling, civil status, number of surgeries, type of diagnosis, time since diagnosis, number of children, premenstrual symptom severity and last week pelvic pain intensity. # *p*-value <sup>≤</sup> 0.05 between mild and moderate groups; \* *p*-value ≤ 0.05 between mild and severe groups.

### *3.4. Contribution of Pain, Fatigue and Psychosocial Impairment to Quality of Life in Spanish Women*

Results from the multivariate linear regression analyses are summarized in Table 4. Severity of chronic fatigue and last week PP were associated with poorer QoL in an intensity-dependent manner. Moreover, anxiety, depression, anger/hostility and catastrophizing thoughts were associated with poorer QoL. Similarly, poorer gastrointestinal health, sexual function and sleep quality were also related to poorer QoL, although the latter showed a close to statistically significant association with QoL when models were further adjusted for intensity of PP during the last week (*p*-value 0.059).

Mediation effects of fatigue and psychosocial impairments on QoL were also accomplished (Figure 2). All chronic fatigue, gastrointestinal complaints, sexual function, anxiety, depression, anger/hostility, sleep quality and catastrophizing thoughts showed a mediating effect on the association between last week PP and QoL when assessed on an individual level (data not shown). However, when the combined mediating effect was evaluated, only chronic fatigue and catastrophizing thoughts revealed a statistically significant mediating effect on the association between last week PP and QoL (1.128 and 0.863, respectively; *p*-value < 0.05).

### **4. Discussion**

To our knowledge, this study constitutes the first attempt to objectively evaluate levels of endometriosis-related fatigue in a comprehensive population, and to address its relevant contribution to the psychosocial status and QoL in Spanish women diagnosed with endometriosis. Moreover, our results suggest that endometriosis-related fatigue and catastrophizing thoughts also exert a mediating effect on the association between intensity of PP and poorer QoL in affected women, evidencing that these factors also need to be addressed within appropriate treatment approaches in women with endometriosis.



(yrs), schooling, civil status, number of surgeries, type of diagnosis, time since diagnosis, number of children and PMS severity; \*\* Additionally adjusted for intensity of pelvic pain during last week.

**Figure 2.** Mediating effect of chronic fatigue and catastrophizing thoughts on the association between pelvic pain and quality of life (QoL) in Spanish women with endometriosis. \* Mediating effects were considered significant when zero was not located within the confidence intervals. Analyses were conducted with the macro PROCESS for SPSS.

Previous studies have stated that women with endometriosis frequently report the presence of chronic fatigue [46], with authors defending the effect of endometriosis on its generation, independently from other symptoms of the disease [11]. Our study shows that 85.3% of the patients enrolled have moderate to severe fatigue. Our findings are in accordance with previous studies where affected women were asked if they felt fatigue. Hence, a total of 50.7% and 27.1% of women reported frequent and occasional fatigue, respectively [11]. Similarly, Surrey et al. [47] recently reported that 54–74% of affected women with moderate to severe pain reported experiencing fatigue. Although the underlying mechanisms are not still fully elucidated, it has been reported that the endometriotic lesions usually develop a complex and dynamic environment dominated by inflammatory, angiogenic, and endocrine signals [48]. Similarly, Suryawanshi et al. [49] reported that endometriotic lesions generate a specific immune microenvironment similar to a tumor-like inflammatory profile. Thus, in accordance with the positive correlation between inflammatory cytokines and fatigue shown in cancer patients [50], elevated cytokine levels found in endometriosis might be hypothesized to play a role in the development of fatigue symptomology in these affected women.

Regarding the relationship identified in this study between levels of fatigue experienced by women with endometriosis and severity of PP, it was not unexpected, as this association has been previously stated in different populations suffering different chronic conditions such as rheumatic diseases [51] or cancer [52]. In fact, both symptoms have been found to be related to an inflammatory microenvironment. Hence, studies from basic sciences evidenced that changes in immune surveillance and central sensitization were related to the pathophysiology of endometriosis [53]. Interestingly, a misbalance in estrogen levels, as widely reported in patients with endometriosis, may be the first responsible for the generation of an inflammatory microenvironment [48] that ultimately can lead to the development of not only PP [54] but also endometriosis-related fatigue. Moreover, the dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis has been reported to contribute to the development of fatigue in chronic illness [55].

Our sample of patients show high levels of psychosocial impairments such as anxiety, depression, poor sleep quality or sexual dysfunction. In this respect, several studies have stated the association between endometriosis and psychological impairments, with depression and anxiety as the most common disorders related to endometriosis, deeply affecting the QoL of these women [6,56–59]. Interestingly, our results suggest an association of chronic fatigue and psychosocial factors, as reported in another multicenter study by Ramin-Wright et al. [11] that comprised 554 women with endometriosis, where fatigue was associated with insomnia and depression among other factors. In this respect, a previous review stated the influence of chronic fatigue on different inflammatory conditions and the possible association between inflammation, pain and depression [51,60]. Moreover, in addition to poorer sleep quality and depression, our study suggests for the first time that the presence of chronic fatigue is associated with higher levels of anxiety and anger/hostility, as well as poorer sexual function. In accordance with our results, it has been reported that fatigue was associated with poorer sexual functioning in women with chronic conditions such as breast cancer [61] or multiple sclerosis [62]. In this regard, fitness level, crucially related to the presence of chronic fatigue, has been recently identified as a strong predictor of sexual function in middle-aged adult women [63]. Although the molecular links between chronic fatigue and psychosocial impairments remain unclear, it has been suggested that the HPA axis might be behind this cluster of symptoms typically observed in cancer patients [64]. Interestingly, an aberrant HPA response has been reported in women with endometriosis [65]. Moreover, social factors might also contribute to the development of fatigue. Hence, in a different subset of patients, it has been reported that social support, through promoting self-confidence and rational thoughts, may have an impact on the reinforcement of the immunity system, and in turn, on the reduction of fatigue levels [66]. Care practitioners and clinicians' perception of women's experiences of endometriosis [67] and low-value healthcare [68] might also contribute to the endometriosis-related burden of symptoms. Additionally, diagnosis delay, infertility and worries related to low work productivity or job loss, in addition to depressive symptoms or disturbed sleep, might also negatively impact on energy and vitality [69–71], revealing the complex inter-relationships between physiological and psychosocial factors in women with endometriosis.

Regarding the interrelationship between intensity of PP, chronic fatigue and psychosocial impairments, our findings indicate that chronic fatigue and catastrophizing thoughts may mediate the association between last week PP intensity and QoL. A similar relationship was described in a previous work showing an association between pain and psychological stress with a worsening of the QoL in women living with endometriosis [72]. In this study, we have added for the first time the contribution of chronic fatigue, in addition to psychosocial distress, on QoL impairment in women with endometriosis. Taken together, these studies would support the idea that pain is associated with chronic fatigue and negative emotions [73,74], that in turn could affect QoL in women with endometriosis. Therefore, our data supports the necessity of multimodal treatments that address fatigue and psychosocial distress in addition to PP intensity in order to improve QoL in women with endometriosis, in line with previous suggestions [11,72]. Thus, besides medical therapy [47], physical and psychological interventions might be beneficial in endometriosis treatment, as evidenced for a variety of chronic illnesses in women [75–78]. More attention should be paid to non-pharmacological approaches to manage the symptom burden of this silenced female disease.

Regarding limitations, study population might not be fully representative of Spanish women with endometriosis. In this regard, although we have included affected women from all Spanish regions, the presence of a selection bias is plausible, given that participants might have a different symptom burden than non-participants. In this regard, it has been reported that outcomes related to QoL are influenced by recruitment strategy [79]. Secondly, this study has a cross-sectional design that does not allow the assessment of the causal relation between studied variables. The absence of a control group also limited understanding of the differential impact of this symptom on lives of women with and without endometriosis. Nevertheless, in a case-control study comprised of 25 women with endometriosis and 25 healthy controls, we have recently reported that mean fatigue score was 2.9 ± 2.0 among controls and

5.3 ± 2.3 among women with endometriosis [80]. In fact, a large majority of healthy women had mild fatigue and none of them had severe fatigue. Contrarily, the majority of women with endometriosis (72.0%) had moderate-severe fatigue. Moreover, the information retrieved in the present study was obtained from self-administered questionnaires and, therefore, a risk of information bias could also exist. However, the use of validated scales for this assessment may counteract this information bias. Finally, we have no information about medication taken by the women during the study, although all participants reported to be adhering to the prescribed medical treatment. However, it is possible that fatigue in endometriosis could be partially attributed to side effects from medication [46]. In addition, contraceptive hormonal therapy is usually prescribed to many women with endometriosis, and its use has been previously associated with different depression symptoms [81].

### **5. Conclusions**

This work provides preliminary evidence of the relevance of chronic fatigue for the psychosocial status and the QoL of women living with endometriosis. We consider that it has important implications for the evaluation and treatment of this population, as the main goal of their management is usually to ameliorate symptoms and to improve general QoL. The habitual treatment of the disease is focused on classic symptoms such as pain or infertility [1,82], but our findings also support the importance of addressing fatigue when treating patients with endometriosis, highlighting the necessity for an interdisciplinary management of the disease. Thus, our results warrant future studies that assess the effectiveness of multidisciplinary approaches (i.e., physical and psychological rehabilitation interventions, in addition to medical therapy) for symptom management.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1660-4601/17/11/3831/s1, Table S1. Relationship between intensity of chronic pelvic pain and chronic fatigue; Table S2. Relationship between intensity of chronic fatigue, psychosocial status and quality of life in women with endometriosis.

**Author Contributions:** F.A.-C., C.F.-L., A.M.-L. and M.S.A.-L. conceived, designed, and implemented the study. O.O.-H. helped with the design and the implementation of the study and coordinated the patient recruitment. I.C.-V., N.G.-C., C.F.-L. and F.A.-C. analyzed and interpreted the data and A.M.-L. and A.P.S.-S. drafted the paper. M.A.-M., I.C.-V. and O.R.-P. contributed to the interpretation of the data and critically revised the article for important intellectual content. All authors provided a critical revision of the manuscript and helped with data interpretation and manuscript preparation. F.A.-C. and C.F.-L. contributed equally to this work. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Health Institute Carlos III (ISCIII)-FEDER (grant number PI17/01743) and donations from particular women with endometriosis that believed in this project from the first time. It was also partly supported by the PAIDI group CTS-206 (Oncología Básica y Clínica) funds. This study takes place thanks to the additional funding from the University of Granada, Plan Propio de Investigación 2016, Excellence actions: Units of Excellence; Unit of Excellence on Exercise and Health (UCEES), and by the Junta de Andalucía, Consejería de Conocimiento, Investigación y Universidades and European Regional Development Fund (ERDF), ref. SOMM17/6107/UGR.

**Acknowledgments:** The authors are indebted to all participants, without whom this work would not have been possible. We are grateful to Ana Yara Postigo-Fuentes for her assistance with the English language. This paper is part of the thesis developed by A. Mundo-López in the Official Doctoral Program in Clinical Medicine and Public Health of the University of Granada.

**Conflicts of Interest:** The authors declare no conflicts of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Laws Restricting Access to Abortion Services and Infant Mortality Risk in the United States**

**Roman Pabayo 1,\*, Amy Ehntholt 2, Daniel M. Cook 2, Megan Reynolds 3, Peter Muennig <sup>4</sup> and Sze Y. Liu <sup>5</sup>**


Received: 28 April 2020; Accepted: 20 May 2020; Published: 26 May 2020

**Abstract:** *Objectives:* Since the US Supreme Court s 1973 *Roe v. Wade* decision legalizing abortion, states have enacted laws restricting access to abortion services. Previous studies suggest that restricting access to abortion is a risk factor for adverse maternal and infant health. The objective of this investigation is to study the relationship between the type and the number of state-level restrictive abortion laws and infant mortality risk. *Methods:* We used data on 11,972,629 infants and mothers from the US Cohort Linked Birth/Infant Death Data Files 2008–2010. State-level abortion laws included Medicaid funding restrictions, mandatory parental involvement, mandatory counseling, mandatory waiting period, and two-visit laws. Multilevel logistic regression was used to determine whether type or number of state-level restrictive abortion laws during year of birth were associated with odds of infant mortality. *Results:* Compared to infants living in states with no restrictive laws, infants living in states with one or two restrictive laws (adjusted odds ratio (AOR) = 1.08; 95% confidence interval [CI] = 0.99–1.18) and those living in states with 3 to 5 restrictive laws (AOR = 1.10; 95% CI = 1.01–1.20) were more likely to die. Separate analyses examining the relationship between parental involvement laws and infant mortality risk, stratified by maternal age, indicated that significant associations were observed among mothers aged ≤19 years (AOR = 1.09, 95% CI = 1.00–1.19), and 20 to 25 years (AOR = 1.10, 95% CI = 1.03–1.17). No significant association was observed among infants born to older mothers. *Conclusion:* Restricting access to abortion services may increase the risk for infant mortality.

**Keywords:** US state laws; abortion; infant mortality

### **1. Introduction**

The infant mortality rate (IMR), the number per 1000 live births of infant deaths before the age of 1, is one of the best predictors of a nation s life expectancy and widely used as an indicator of population health [1]. In 2011, around 24,000 infants died in the United States (US), resulting in an IMR of 6.1, nearly twice the Organization for Economic Co-operation and Development (OECD) average of 3.4/1000 [2]. Known risk factors for infant mortality include individual-level risk factors (e.g., socio-economic status, such as mother's education or household income) and larger contextual factors (e.g., neighborhood environment or state-level policies that affect access to health care) [3].

Within the US, IMRs vary greatly across socio-demographic groups. Infants born to Black mothers, single mothers, and low-income mothers have higher IMRs compared to infants born to White mothers, married mothers, and moderate or high-income mothers [4]. For example, in 2016, the IMR among

non-Hispanic Blacks was 11.4 per 1000, whereas the IMR for non-Hispanic Whites was 4.9 [5]. These socio-demographic groups may have limited access to resources, such as psychosocial supports, material resources, and vital health and reproductive services.

Abortion, a medical procedure terminating a pregnancy, is recognized as a key component of reproductive health services [6], which the United Nations recognizes as an important predictor of national well-being and population health, generally, and maternal and infant mortality, specifically [7–9]. Beyond the abortion laws examined in this study, states continue to create policy restricting access to safe abortions. Currently, the Supreme Court is hearing one such case affecting a provider's rights to provide abortion services. Beyond the immediate constitutional value of any restrictive abortion law, courts should consider the health impact of such laws on women and their future unborn children.

Restrictive abortion policies may have a detrimental effect on both maternal and infant health via several mechanisms. First, restrictive abortion policies may jeopardize patient health by undermining providers medical counsel. For example, some US states require counseling that provides inaccurate information about negative mental health consequences of abortion or a link between abortion and increased risk of breast cancer [10]. Second, restrictions may increase psychological distress, which is a known risk factor for poor birth outcomes. Both human and animal studies show that psychological stress produces a cascade of neuroendocrine changes that increase the risk of serious birth complications (e.g., pre-eclampsia) and thereby increase infant mortality [11–14].

Third, these policies may increase the risk for postpartum depression by inducing dramatic, but unwanted life changes. Postpartum depression can contribute to infant death in cases where the mother is unable to properly care for her infant [15,16]. Previous research indicated that women in states that prohibit the Medicaid funding of abortions have significantly higher rates of postpartum depression than those in states that fund Medicaid abortions [17]. Mental illness among pregnant women has been associated with increased risk for infant mortality [18]. Postnatal depression has also been observed as a risk factor for sudden infant death (SIDS) [19].

Fourth, restrictive policies may encourage the delivery of infants whom the mother is unable to economically or emotionally support. For example, women from low socioeconomic backgrounds who have limited resources and income may be less able to pay to terminate their pregnancies out of pocket and more likely to carry their infant to full term [20]. Moreover, women from lower socioeconomic backgrounds may also experience limited access to prenatal care and health care, and sufficient living expenses for optimal health and growth, all of which may have an adverse effect on maternal and infant health [21]. Findings from a previous study suggests denying women abortions may be associated with greater poverty than enabling women to postpone childbearing [22].

As shown in ecological studies conducted in the US, increased state funding for family planning and abortion services is associated with lower infant mortality rates, especially for low-income women of color [7]. Conversely, restrictions on abortion services have been associated with increased infant and maternal mortality risk [23,24]. However, because these previous studies are based on aggregated data, they limit our ability to make inferences at the individual level. Moreover, few of these studies have examined interactions between restrictive abortion policy and individual-level characteristics. Therefore, the overall objective of this study is to identify the association between the number and type of state-level restrictions to abortion services during year of birth and infant mortality risk while controlling for both individual and state-level confounders. Additionally, we examine whether observed relationships are heterogeneous across socio-demographic groups, such as mother s age, race, and educational attainment. By studying these heterogeneous associations, we aim to determine who is at greatest risk of infant mortality when such state-level abortion policies are enacted.

### **2. Methods**

We obtained data from the US Cohort Linked Birth/Infant Death (LBID) Data Files on infants born 2008–2010, which is provided by the National Center for Health Statistics (NCHS). Infants were followed until their first birthdays. US state laws require birth certificates to be completed for all births. Of all deaths that occurred, around 98% were linked to the corresponding birth certificate. Instructed by US state laws, each birth requires a birth certificate to be completed, which contains information on maternal and socio-demographic characteristics and place of birth. Federal law mandates the national collection and publication of births, deaths, and other vital statistics. These data are then compiled by the National Vital Statistics System. Those with missing socio-demographic data, foreign residents, and those with records that were a mismatch between state of birth and mother s state of residence were excluded from this analysis. Ethical approval was obtained from the University of Nevada, Reno, Institutional Review Board (code 791378-1).

### **3. Measures**

### *3.1. Outcome*

Infant mortality (death within 365 days of birth) was our outcome of interest.

### *3.2. Main Exposure*

Our main measure of exposure was the category of restrictive state abortion law. Since the US Supreme Court s 1973 *Roe v. Wade* decision legalizing abortion nationwide, individual states have enacted laws regulating the access and availability of abortion services in the hopes of moderating the effects of federal decision. Data on restrictive abortion laws for each US state were obtained in 2008, 2009, and 2010 from the Guttmacher Institute [25–27]. From 2008–2010, there were five types of state-level restrictive abortion laws directly impacting patients that the Supreme Court regarded as constitutionally permissible. These types of restrictive laws consist of: (1) Medicaid funding restrictions—prohibitions against use of state public funds to pay for abortions for indigent women [25–27]; (2) parental involvement laws—requirements that a parent be notified or give consent for an unmarried teen minor to obtain an abortion; (3) mandatory counselling—requirements that an abortion provider give or offer their patients information about abortion (usually written in a way to dissuade women from completing the abortion) by providing information on fetal development, fetal pain, fetal age, possible future health risks [e.g., substance abuse, breast cancer, suicide, or infertility], adoption options, and available public assistance to the birth mother); (4) mandatory waiting period—-requirements that a specified time period (usually 24 h) elapse before the procedure can be performed; (5) two-visit laws—requirements that women make two separate trips to the abortion provider prior to the procedure, elevating the monetary and time burdens for prospective patients.

Table 1 shows a complete list of the status of the five restrictive abortion laws for all 50 states and the District of Columbia in 2008. Over the next two years, several states enacted a few more abortion restrictive laws. For example, in 2008, the only restrictive law in Arizona was mandatory parental involvement [25–27]. By 2010, Arizona had added mandatory counselling and mandatory wait times. We looked at the relationship between each type of restriction and the number of restrictions during the year of birth of each infants and risk for infant mortality.


**Table 1.** Abortion restrictions by State, 2008.

Source. Guttmacher Institute (2008) [25].

For this study, we included several additional state-level and individual-level covariates that could act as confounders of the relationship between restrictions to abortion services and infant mortality risk. State-level covariates include median income, proportion of population that is African-American, population size, and US census division. Individual-level maternal covariates include mother s age, race/ethnicity, education, marital status, and nativity (US vs. foreign-born).

### **4. Statistical Analysis**

Infants with missing data on their mother s education were excluded from the analyses. We used multilevel logistic modeling (mothers and their infant nested within states) to determine the association between abortion-restriction laws during the year of birth and infant mortality risk. We tested the relationship between each of the five types of restrictive abortion laws separately and looked at the total number of restrictions (no restrictions, one or two restrictions, and three or more restrictions). The internal consistency of restrictive laws was high (Cronbach s alpha = 0.82).

To investigate the potential effect of restrictive laws and risk of infant mortality, we adopted a step-up approach, where we systematically add variables to the models. The null model was first estimated to compute the overall predicted probability, which indicates the average probability of infant mortality across all states. Additionally, the 95% plausible value range, which is an indication of the degree of variability of the likelihood of infant mortality. For example, the plausible value range allows us to compute the range of plausible proportion (i.e., the maximum and minimum values) experiencing infant mortality across the US states. Second, we identified the crude relationship between each type of restrictive law at the year of birth (2008, 2009, or 2010) and infant mortality risk. Third, we fit our logistic regression model including state- and individual-level characteristics. Finally, we stratified the analyses by age (≤19 years, 20–25 years, 25–30 years, 30–34 years, and ≥35 years of age), race (Black vs. White), and education (less than high school vs. high school or more) to determine if relationships were heterogeneous across socio-demographic groups (results not shown).

### **5. Results**

### *Characteristics*

Socio-demographic characteristics of infants born 2008 to 2010 are presented in Table 2. Over half of the mothers were White (53.5%), around a quarter were Hispanic (24.4%), and around 15% were Black. Just over a half of the mothers (52.5%) had some post-secondary education, almost two-thirds were married (59.0%) and roughly three-quarters were US-born (76.4%). Characteristics of the fifty states and District of Columbia also appear in Table 1. The median state-level income in 2010 was \$49,973.65 (SD = 8130.60) and the average proportion of African-Americans in a state's population was 12% (SD = 11.8).

By 2010, the most common state-level restrictive abortion law was mandatory parental involvement (*n* = 35) and the least common restrictive abortion law was the two-visit law (*n* = 7). Nine states (17.7%) had no restrictive laws, while seven (13.7%) had all five laws enacted.

From 2008 to 2010, there were 71,528 infant deaths corresponding to an infant mortality rate of 6.0 deaths/1000 births. The overall predicted probability by year was 0.61%, 0.60%, and 0.57% in 2008, 2009, and 2010, respectively. The plausible value range indicates that there is considerable variability in the cumulative incidence of infant death across US states. The plausible value range for infant mortality was 0.44–0.86%, 0.43–0.84%, and 0.40–0.80%, in 2008, 2009, and 2010, respectively.

When we tested the crude relationship between each of the five types of abortion restrictive laws, each was significantly associated with an increased odds for infant mortality (Table 3): the presence of Medicaid restrictions (OR = 1.22, 95% CI = 1.10–1.36); parental involvement (OR = 1.26, 95% CI = 1.13–1.40); mandatory counselling (OR = 1.08, 95%CI = 1.00–1.17); mandatory wait period (OR = 1.09, 95% CI = 1.01–1.18); and two-visit laws (OR = 1.18, 95% CI = 1.01–1.38). However, after adjusting for individual- and state-level characteristics, infants born in states with parental involvement laws (OR = 1.10, 95% CI = 1.02–1.19) had an increased risk for infant mortality compared to those who lived in states without such laws. No other laws remained significant when tested separately.

When we assessed the crude relationship between the total number of laws, compared to no restrictions, one or two restrictive laws (OR = 1.16, 95%CI = 1.01–1.34), and three or more restrictions (OR = 1.26, 95% CI = 1.11–1.45) were associated with an increased odds for infant mortality (Table 3). Adjusted analysis of the total number of restrictive laws resulted similar trends but resulted in decreased OR estimates. For example, infants born in states with one or two (OR = 1.08, 95%CI = 0.99–1.18) or

three or more restrictive laws (OR = 1.10, 95%CI = 1.01–1.20) had greater risk for infant mortality than those born in states with no restrictive laws (Table 3).


**Table 2.** Characteristics of mothers and US infants born 2008–2010.

In analyses stratified by age, in comparison to those infants born in states with no parental involvement laws, those who were born in states with this restrictive law were at greater risk for infant mortality among mothers aged ≤19 years of age (OR = 1.09, 95% CI = 1.00–1.19), and 20–25 years (OR = 1.10, 95% CI = 1.03–1.17), but not among mothers in older age categories (Table 4). No other findings were significant. When stratified by race, White infants born in states with one or two restrictions were more likely to die than those born in states with no restrictions (OR = 1.15, 95% CI = 1.04–1.27). Infants born to White mothers who were living in states with mandatory wait periods were significantly less likely to die (OR = 0.93, 95% CI = 0.88–0.99). Among Blacks, infants born in states with Medicaid Restrictions (OR = 1.08, 95% CI = 1.00–1.16) and Parental Involvement (OR = 1.08, 95% CI = 0.99–1.17) were more likely to die. Blacks were more likely to experience infant mortality in states with one or two restrictions (OR = 1.07, 95% CI = 0.97–1.19) and in states with 3 or more restrictions (OR = 1.09, 95% CI = 0.99–1.19) in comparison to those infants born in states with no restrictions. When analyses were stratified by education, infants whose mothers had less than high school education born in states with Medicaid Restrictions (OR = 1.08, 95% CI = 1.01–1.17) and one or two restrictions (OR = 1.13, 95% CI = 1.04–1.12) were more likely to die compared to those born in states with no such restrictions. Infants born to mothers with a high school education in states with Parental Involvement laws (OR = 1.11, 95% CI = 1.03–1.21) or in states with 3 or more restrictions were more likely to die (OR = 1.11, 95 %CI = 1.01–1.21) than were those born in states with no restrictions.







### **6. Discussion**

Since the federal legalization of abortion in 1973, many US states have successfully enacted laws restricting access to abortion, the majority of which started in the past two decades [25–27]. As a result, a woman s access to abortion services varies greatly across US states. We exploited variation in state-level restrictive abortion laws and the individual risk for infant mortality, with special attention to differential effects by maternal characteristics. We observed a significant relationship between the number of restrictive abortion laws and infant mortality risk, indicating a potential additive effect. More specifically, infants born in states with three or more restrictive laws were significantly more likely to die before their first birthday than were those born in states with no restrictions.

We posit that the 10% increase that we observed is meaningful from a population health standpoint. There were 22,000 infant deaths in the United States in 2017, a disproportionate number of which occurred in states with restrictive abortion laws [28]. A 10% reduction in infant mortality in these states could eliminate hundreds of excess infant deaths per year.

These findings are consistent with ecological studies that have identified a significant relationship between the legalization of abortion within the US and state funding for family planning and abortion services and infant mortality rates. For example, Krieger et al. observed US infant death rates declined most quickly between 1970 and 1973 in states that legalized abortion in 1970 [29]. They also found that, since 2000, the rate ratio for infant death comparing states in the top funding quartile to states with no funding for abortion services revealed an average 15% reduction in risk among the top funders [7].

The Sexual and Reproductive Framework (SRJ) defines reproductive rights as human rights and recognizes the multiple forms of oppression that impact individuals decisions about their sexual and reproductive health [30]. By interpreting our results using a SRJ framework, we further emphasize the need for legislation to acknowledge how women's sexual and reproductive health decisions are shaped by social and contextual factors as well as by individual-level resources [30]. Our results support SRJ assertions by suggesting that unintended, and differentially strong adverse effects associated with abortion policies that restrict women's reproductive decision-making. Governments should promote sexual health policies that respect women's reproductive decisions as part of their efforts to enhance population health [31,32]. Furthermore, it could be argued that preventing unwanted pregnancies itself and the consequent need for abortions is crucial for reproductive, maternal, and infant health. Targeting upstream factors—e.g., improved and more expansive reproductive services, especially access to contraception—could decrease the number of unwanted pregnancies, thus eliminating the need for abortions.

One reason for the observed findings could be that restrictions on abortion, through their limiting of the right to make a medical choice, could themselves have detrimental effects on the health of the mother, and consequently on the infant's health as well. Abortion restrictions impede a woman s ability to make health decisions and to exercise autonomy over her reproductive life. This autonomy has been identified as a fundamental human right and an important determinant of women s health [33]. Thus, when women are compelled to carry a pregnancy to term, giving birth may have detrimental effects on the woman's own health and, therefore, on their infant's well-being. Additionally, social stigma attached to abortion may also have an effect on mental health of women seeking to terminate their pregnancy. For example, researchers observed that women who were denied an abortion and, therefore, carried their pregnancies to term experienced psychological distress a year later [34], which can detrimentally affect the health of their infants. Researchers may want to explore casual methods to understand whether abortion laws cause infant mortality. However, given how such laws are implemented the common assumptions of causal methods may be violated. Furthermore, future research is needed to determine whether the mechanisms linking abortion restriction laws leads to adverse infant health independently or through its effects on the mother.

In addition to directly impacting health and wellness, the state laws analyzed here may be markers for an array of factors comprising a state's socio-political environment that fosters inequality and harms population health. In other words, states who enact restrictive abortion laws are more likely to support and fund reproductive and women s health. This can have detrimental consequences on women s health and therefore on, maternal and infant health [35].

We also observed significant heterogeneous associations across socio-demographic groups. When stratified by mother s age (21 and younger and 22 and older), mandatory parental involvement, was significantly associated with an increased risk for mortality among infants of mothers from both age groups. One explanation may be that states with parental involvement laws are associated with an increase in the price of an abortion by 14% [36]. This increase in price may further act as a barrier for young mothers in obtaining an abortion. Among infants born to Black mothers, those born in states with Medicaid restrictions and parental involvement laws were more likely to die than were those born in states without these laws. Black mothers are disproportionately more likely to be from lower socioeconomic status backgrounds and are therefore less apt to have access to health services. Additionally, Black girls are more than twice as likely as White girls to become pregnant [37]. Low socioeconomic status and teenage motherhood are each risk factors for infant mortality.

One unexpected finding was that among infants born to White mothers, those in states with mandatory wait periods were significantly less likely to die in comparison to those states without this law. Mandatory wait periods may be effective at delaying an abortion, which in theory could encourage women to use unsafe practices [38]. The protective finding of mandatory wait periods against infant mortality among White mothers requires further study, including, potentially a qualitative analysis.

Infant mortality rates have been on the decline in recent years. The drop from 6.1/1000 live births in 2009 to 5.8/1000 in 2015 may be explained by decreasing birth rates among young teenage girls [39]. Infants of adolescent mothers are at greater risk for mortality than are those born to adult mothers [39]. Younger mothers may not have sufficient access to resources and health care that are essential for optimal health for their newborns. As births to teen mothers decrease, infant mortality rates could drop as a result.

### *Strengths and Limitations*

This study s results should be interpreted in light of several limitations. Data analyzed for this study was collected from infants born 2008–2010, which was the most recent data available at the time of this investigation. However, one reason for using not using the most recent data is that the CDC does not release individual-level data immediately since it takes years for the data to be prepared, cleaned, and de-identified. Once this has been completed, the data is made available for public use. A delay in our investigation is further caused since obtaining state residence information requires additional paperwork and approval. Thus, linked state-individual data from 2008 to 2010, is the most temporally appropriate data to be utilized for this investigation. Since the study was not a randomized controlled trial (i.e., we could not randomize mothers into states with and without restrictions on abortion) we cannot assess whether observed associations are causal. Furthermore, because we only had data from 2008–2010 and were limited by the number of states changing or adding restrictive abortion laws, we did not utilize a quasi-experimental approach. Furthermore, endogeneity could be an issue due to residual confounding: potential confounders such as individual household income and other socioeconomic conditions were not available. Therefore, the inability to use a quasi-experimental study design and endogeneity due to residual confounding limits our ability to draw causal inferences. Lastly, although we were able to examine the moderating effects of education, we could not test whether household income acts as an effect modifier of the relationship between abortion restrictions and infant mortality.

### **7. Conclusions**

Although we cannot conclude a causal relationship, results from this investigation indicate that the number of state-level restrictions on abortion may be a significant risk factor for infant mortality. In particular, mandatory parental laws that require either parental permission or notification for a minor to have an abortion are significantly associated with infant mortality. Future studies should identify the extent to which this relationship is causal. For example, researchers can take advantage of the passage of new laws against abortion, such as banning abortions once a fetal heartbeat is detected, or restricting providers with unnecessary requirements. These changes present a unique opportunity to better understand how and why these restrictions may cause adverse health outcomes for mothers and affect infant mortality risk.

**Author Contributions:** R.P. conceptualized and designed the study, conducted the analyses, drafted the initial manuscript, and reviewed and revised the manuscript. S.Y.L. helped to conceptualize and design the study, interpret the findings, wrote subsequent drafts of the manuscript, and critically reviewed the manuscript for important intellectual content. A.E., D.M.C., M.R., and P.M., helped to interpret the results, reviewed and revised the manuscript, and critically reviewed the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. All authors have read and agreed to the published version of the manuscript.

**Funding:** National Institutes of Health Research, National Institute on Minority Health and Health Disparities 1R15MD010223-01. Roman Pabayo is a Tier II Canada Research Chair in social and health inequities throughout the lifespan. The authors have no financial relationships relevant to this article to disclose.

**Acknowledgments:** We thank Elizabeth Nash, Senior State Issues Manager at the Guttmacher Institute for the data on State-level Abortion Restrictive Laws.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Abbreviations**

OR—Odds ratio. AOR—Adjusted odds ratio. OECD—Organization for Economic Co-operation and Development. SD—Standard deviation. LBID—Linked birth and infant death records. NCHS—National Center for Health Statistics. Using the 2008–2010 U.S. Cohort linked birth/infant death data files, this study identifies the relationship between state-level restrictive laws and infant mortality risk.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*

### *Article* **Psychometric Properties of the Condom Use Self-E**ffi**cacy Scale among Young Colombians**

### **Vanessa Sanchez-Mendoza 1,2,\*, Encarnacion Soriano-Ayala <sup>3</sup> and Pablo Vallejo-Medina <sup>4</sup>**


Received: 24 March 2020; Accepted: 21 May 2020; Published: 26 May 2020

**Abstract:** (1) Background: This study evaluated the psychometric properties of the Condom Use Self-Efficacy Scale among Colombian youth. (2) Method: A total of 2873 men and women between 18 and 26 years old (*M* = 21.45, *SD* = 2.26) took part in this study. All participants answered a socio-demographic survey, the Condom Use Self-Efficacy Scale, the UCLA Multidimensional Condom Attitudes Scale, The Condom Use Errors and Problems Scale, and the Sexual Assertiveness Scale. Sampling was web-based, and the survey was distributed via Facebook. (3) Results: The Condom Use Self-Efficacy Scale demonstrated adequate reliability (ordinal α ranged = 0.76 to 0.92). Exploratory and confirmatory factor analysis suggested a four-factor structure with an explained variance of 69%. This dimensionality was also invariant across gender. Moreover, positive attitudes toward condom use were significantly associated with appropriation and assertiveness. Two dimensions (appropriation and partner disapproval) showed significant gender differences. (4) Conclusions: The Spanish–Colombian version of the Condom Use Self-Efficacy Scale is a psychometrically adequate instrument to measure perceived condom use self-efficacy. This scale can be used in both research and professional settings to measure self-efficacy at using condoms in young people.

**Keywords:** HIV prevention; unintended pregnancies; condom use; sexual risk; Latins; psychometric; validity; sexual behavior; STI prevention

### **1. Introduction**

Sexual health is a significant interest area for public health in Colombia. The Colombian government's 2030 Agenda for Sustainable Development has established goals and indicators concerning health and well-being. Some of those goals are related to sexual and reproductive health: (1) to reduce the HIV/AIDS deaths tolls from 4.9 per 1000 people to 2.4 per 1000 people, and (2) to increase the percentage of women between the ages of 15 and 49 years old who use contraceptive methods from 68.1% to 81.4%. Those goals will be achieve over the next ten years [1]. Additionally, the government created the Ten-Year Public Health Plan 2012–2022 and The National Sexual and Reproductive Rights Policy. Both documents are written understanding health as a human right, gender differences in access to healthcare, taking a life course approach, and attending to social and behavioral factors [2].

In 2018, 36.2 million people lived with HIV/AIDS around the world, 1.7 million individuals became infected in 2018. Although the Caribbean, Central Africa, Europe, and North America reported significant decreases in new infections between 2010 and 2018, Latin America saw a 7% increase in cases [3]. In Colombia, HIV increased by even more than Latin America as a whole. Between 2015 and 2019, new HIV infections increased by 17.3%. The distribution of new infections by population group (among people aged 19–45 years) in Latin America shows that 40% of new infections occur amongst men who have sex with other men, 25% occur amongst sex workers, illegal drug users, and transgender people, and 35% occur in other segments of the population. In 2019, Colombia had 17,502 new cases. Males comprised 80.8%, and sexual intercourse was the cause of infection in most cases (17,219) [4].

Other sexually transmitted infections (STIs) are also a major problem. Global data show that more than a million people are infected daily, and an estimated 376 million new infections occur each year [5]. More than 500 million people are estimated to suffer from the herpes simplex virus, and more than 290 million women have one kind of human papillomavirus infection [6]. In Colombia, 94,000 people sought medical attention due to an STI between 2009 and 2011; according to individual service provision records, the age of highest prevalence was between 20 and 29 years of age, and the female population was the most affected [7]. The impacts of these infections vary across populations, depending on the biological, social, behavioral, and economic factors affecting also the reactions to and physiological results of STIs. Some of these are infertility [8], social inequality [5], depression and associated mental health disorders [9], and increases in healthcare expenditures [10].

Finally, another problem related to sexual health is unintended pregnancies. Latin America is one of the regions with the highest rate of unintended pregnancies; 122 out of 1000 women aged 15–44 are pregnant, but 55% were unintended. Latin America is the only region in the world where the percentage of unplanned pregnancies exceeds planned ones [11]. In Colombia, poverty and social factors are the major reasons to explain the high rates of teen pregnancy, one in every five women between 15 and 19 years of age are or have been pregnant and just 33.6% are planned [12]. Unintended pregnancies in women cause problems such as maternal death (830 women die during childbirth every day), it perpetuates the cycle of poverty in developing countries where 99% of mortality occurs and 7% of women drop out of school due to pregnancy, and finally it increases the risk of stillbirth due to preeclampsia, placental abruption, or maternal malnutrition [6].

Condoms are inexpensive and highly effective means at preventing HIV, STIs, and unintended pregnancies. In Colombia, consistent condom use is uncommon (22%) [13], its use oftentimes is doubtful and incorrect [14]. The greatest predictor of regular, consistent condom use is a person's attitudes towards condoms (i.e., personal positive evaluation that condoms actually work) [15,16], subjective norms or beliefs (i.e., perceived normative support for condom use) [17–19] and self-efficacy (i.e., their belief that they can actually use condoms) [16,20]. Self-efficacy is supported by the social cognitive theory, according to which people's behavior depends on how they evaluate their own abilities [21,22]. Thus, one of the reasons why young people take high risks when they engage in sexual activity may be associated with low perceived self-efficacy about their competence to negotiate condom use. Unlike the attitudes and subjective norms, there are no validated or translated standardized scales in Colombia to evaluate this key construct of sexual health.

Previous studies described above underscore the importance of socio-cognitive variables, such as self-efficacy, to increase condom use frequency and contribute to the prevention of STIs [23,24]. One of the scales used to assess self-efficacy at using condoms is the Condom Use Self-Efficacy Scale (CUSES) [25], which assesses a person's feelings of confidence about buying condoms, putting them on and taking them off, and also the act of negotiating the use of it with a sexual partner. This scale has been used with college students from different cultural backgrounds [26] and other young people's samples [23,24,27]. Validation studies have reported on the adequate psychometric properties of the CUSES as a valid and reliable instrument to measure condom use self-efficacy in different populations. Although developing a scale using the intended respondents' native language remains the best option to measure self-efficacy adequately in a given context, validating and using an existing scale is not only economical, it also provides backgrounds for comparison in other settings [28].

The present study evaluated the psychometric properties of the CUSES in Colombia and established relationships between this scale and results from other validated instruments administered to Colombian participants.

### **2. Method**

### *2.1. Sample*

Participants were recruited through Facebook. Initially, 3649 young people clicked on the scale; 281 of these acknowledged informed consent but failed to respond, seven failed to indicate their sex, 15 were over 26 years of age, 24 were not Colombian, and 449 did not answer the CUSES fully. A total of 2873 young people completed the sociodemographic questionnaire and the CUSES. Inclusion criteria were being between 18 and 26 years of age and being Colombian. Exclusion criteria were failure to provide informed consent or not completing the questionnaires. The mean age was 21.45 years (*SD* = 2.26). A total of 33.8% (*n* = 971) were male and 66.2% (*n* = 1902) were female. The characteristics of the sample are described in Table 1.


**Table 1.** Sample description.

**Males (***n* = **971) Females (***n* = **1902) Total (***n* = **2873)** *N* **%** *N* **%** *N* **%** *Do you have a partner that you have been seeing for more than 6 months?* Yes 491 (50.7) 1179 (62.1) 1670 (58.1) No 477 (49.3) 719 (37.9) 1196 (41.6) *How often do you use condoms during sexual intercourse?* Every time 271 (28) 334 (17.6) 605 (21.1) Usually 248 (22.6) 345 (18.2) 593 (20.7) Frequently 105 (10.8) 184 (9.7) 289 (10.1) Sometimes 55 (5.7) 125 (6.6) 180 (6.3) Occasionally 92 (9.5) 253 (13.3) 345 (12) Rarely 105 (10.8) 321 (16.9) 426 (14.9) Never 93 (9.6) 336 (17.7) 429 (15)

**Table 1.** *Cont.*

Three subsamples were randomly obtained from the total sample; data from 873 participants were subjected to exploratory factor analysis (EFA), data from 1000 participants were subjected to confirmatory factor analysis (CFA), and data from 1000 participants were subjected to invariance analysis. Remaining analyses used the full sample.

### *2.2. Instruments*

**Sociodemographic variables.** An ad hoc questionnaire was created in order to register sex orientation, age, nationality, and city of residence, schooling, gender, sexual orientation, and marital status of the participants. This questionnaire has previously been used in the Colombian context [13].

**Condom Use Self-E**ffi**cacy Scale (CUSES)** [25]. The scale evaluates a person's feelings of confidence about buying condoms, putting them on and taking them off, and negotiating their use with a sexual partner. It is composed of 15 statements with five Likert-type response options from Strongly disagree = 0 to Strongly agree = 4; seven items are reverse-coded. Its original version includes four factors: (a) appropriation, with four items related to confidence in performing the technique necessary to use a condom in a sexual intercourse; (b) partner's disapproval, including five items related to confidence in dealing with possible rejection from a sexual partner due to request to use a condom; (c) assertiveness, including three items related to an individual's ability to ask a partner to use a condom; and (d) intoxication, with three items referred to the respondent's confidence in their ability to use a condom while under the influence of alcohol or other drugs or when they are over-excited. The present study applied for first time the Colombian validated scale for Colombian youth [28] 29. Once reversed items (items 5 to 9) are re-coded, higher scores mean higher self-efficacy.

**The UCLA Multidimensional Condom Attitudes Scale (MCAS)** [29]. The MCAS scale evaluates people's attitudes toward condom use. It is composed of five dimensions consisting of five items each (reliability, pleasure, stigma, negotiation, and shame). It is answered using Likert-type response options from 1 = totally disagree to 7 = totally agree. The present study applied the Colombian validated scale [30]. Higher scores indicate better attitudes toward the use of condoms. Obtained data in this study showed alphas ranging from 0.71 to 0.87.

**Condom Use Error**/**problems Survey (CUES)** [31]. The scale measures errors and problems associated with the use of condoms, considering the last three times a condom was used during the past three months as the recall period. Male (putting on a condom themselves) and female (putting the condom on their partner) versions are available. The scale is composed of 16 items with four Likert-type response options ranging from failure to perform an action to always performing it. High scores indicate a high frequency of errors/problems during any of the last three occasions or during the last three months. The present study applied the Colombian validated scale. [14].

**Sexual Assertiveness Scale (SAS)** [32]. The scale measures assertiveness with respect to initiation, refusal, and sexually transmitted disease/pregnancy prevention (STD-P P) with a regular partner. It is composed of 18 Likert-type items (Never = 0 to Always = 4). The brief (9 items) version was used, which has been validated to Colombia [33]. A high score represents high sexual assertiveness. The present study used only the initiation and STD-P P items. Initiation (assertiveness) is related to a person's feelings of confidence about initiating sexual intercourse and making suggestions about physical intimacy. Alphas in the present study for those scales were 0.74 (initiation) and 0.85 (STD-P P).

### *2.3. Procedure*

The questionnaire was published and distributed using Facebook. Payment of USD 150 was made to the virtual platform to promote the scale from 2 October to 17 October 2019. Responses were collected using a secure third-party survey provider (Survey Monkey, https://www.surveymonkey.com). The targeted population was young people aged 18–26. Online participants confirmed their willingness to participate through an online consent procedure, and questionnaire completion was also online. All internet protocol (IP) addresses were logged to discourage multiple responses from a single individual. By limited responses to one per IP address, the quality of the online data were higher.

**Data analysis.** Results were processed using R [34] (Version 3.6.0) and the R Studio interface [35] (Version 1.1.463). A polychoric matrix was for reliability and factorial analysis calculation; thus, the α presented is not Cronbach's but ordinal. The number of dimensions to be extracted was calculated with the following methods: optimal coordinates, acceleration factor, parallel analysis, Eigenvalues (Kaiser criterion), Velicer Minimum Average Partial MAP, Bayesian Information Criterion BIC, sample size adjusted BIC, Very Simple Structure VSS complexity, and VSS complexity 2. The mode and the quality of the indicators indicated the *n* of factors. EFA was computed through an maximum likelihood robust ML-R) method using varimax rotation on the polychoric matrix of sub-sample 1. CFA was performed using a Weighted Least Square Mean and Variance Adjusted-Robust (WLSMV-R) estimator on a polychoric matrix based on sub-sample 2. Five different models were tested. The fit indexes used were root mean square error approximation (RMSEA) [36] and its 90% confidence interval (90% CI), the comparative fit index (CFI) [37], and Tucker Lewis Index (TLI) [38]. Values up to 0.08 for RMSEA are usually considered as acceptable, but it is desirable not to exceed a 0.06 threshold; while a value above 0.90 is acceptable, one higher than 0.95 is desirable for CFI and TLI [39]. For the invariance across gender, as before, data were derived from a polychoric matrix, in this case of sub-sample 3. WLSMV-R was also used. Invariance indicators were: a −0.01 change in CFI, paired with changes in RMSEA of +0.015 concerning the least restrictive model [40] and a non-significant increase of the χ<sup>2</sup> using the Scaled Chi-Squared Difference Test [41]. Progressive invariance was tested for four models (configural, metric, scalar, and strict).

The following packages were also used: ggplot2 for data visualization [42] (Version 3.1.1), psych (Version 1.8.12) psychometric (Version 2.2) and psycho (Version 0.4.9.1) [43–45] were used for estimating some psychometric properties. While lavaan [46] (Version 0.6–5), semPlot [47] (Version 1.1.2), and semTools [48] were used for calculating and plotting the Structural Equation Model.

### **3. Results**

The number of dimensions to isolate was evaluated using sub-sample 1. The methods (optimal coordinates, parallel analysis, Kaiser, and VSS complexity 2) suggested a three-dimensional model. Some of these methods are popular (Kaiser) and often recommended (optimal coordinates and parallel analysis). Nevertheless, we decided to compare the original four-dimensional structure of the scale with the three-dimensional structure (Table 2). The three-dimension (3-D) proposal explains 62% of the variance and has RSMEA = 0.127 and TLI = 0.85 as exploratory estimates. The four-dimension (4-D) model explains 69% of the variance, and the exploratory indicators RMSEA and TLI were 0.086 and 0.93, respectively. Item complexity was similar in both models; therefore, CFA results were needed to decide on the model.


**Table 2.** Exploratory factor analysis based on the polychoric matrix using maximum likelihood and varimax rotation.

\* *D* = Number of dimensions' model; *h2* = communality of the item; *u2*= uniqueness of the item; *com*= Hoffmann's item complexity. Weights lower than 0.30 are hidden; boldface represent correct item-factor weight.

CFA was performed using the second sub-sample. Five different models were tested: (1) a one-dimension model (1-D); (2) the three-dimension model previously explored by independent factors (3D I); (3) the three-dimension model previously explored with related factors (3D R); (4) four independent factors explored previously and proposed by previous theory (4D I); and finally, (5) four related factors explored previously and proposed by previous theory [25] (4D R). The main results are shown in Table 3. Model 4D R seems to have the best fit; it is the only one in which all fit indexes are acceptable. Standardized weights for the chosen model are shown in Figure 1.


**Table 3.** Fit indexes for the model tested.

\* χ*<sup>2</sup>* = Chi-Square statistic; *Df* = degrees of freedom; *CFI* = Comparative Fit Index; *TLI* = Tucker Lewis Index; *RMSEA* = root mean square error of approximation.

Once the four-dimension model was confirmed to be the best, an analysis of gender invariance was performed. For that purpose, we used sub-sample 3. As shown in Table 4, a strict level of invariance was achieved. Based on the three indicators taken into account for the four levels, just a significant increase of the χ<sup>2</sup> was observed for the weak invariance. However, the other indicators (ΔCFI and ΔRMSEA) seem to behave appropriately to suggest gender invariance.


**Table 4.** Fit indexes for the invariance 4-D R model.

\* χ*<sup>2</sup>* = Chi-square statistic; *Df* = degrees of freedom; *Conf* = configural invariance; *Metric* = metric invariance; *Scalar* = strong invariance; *Strict* = strict invariance.

The psychometric properties of the items began to be obtained. Once strict invariance across gender was tested and no differences were found, descriptive psychometric properties were assessed. In Table 5, ordinal alphas above 0.75 for all dimensions are shown. A couple of items (3 and 15) improve α if eliminated, but no further actions were considered necessary. Overall, corrected correlations between total and items are higher than 0.50. Item distributions cannot be considered normal, especially when considering kurtosis.


**Table 5.** Some psychometric item properties.

\* *M* = mean; *SD* = standard deviation; *CI-TC* = corrected item-total correlation; *App* = appropriation; *Pds* = partner's disapproval; *Ass* = assertiveness; *Scn* = self-control.

Criterion validity was also explored. To this aim, a correlation matrix between the CUSES subscales and other theoretically related instruments was created. Positive and significant correlation were observed and confirmed. Significantly low and moderate correlations were observed (see Table 6). Finally, the CUSES sub-scales were compared for gender. Figure 2 shows significant differences in appropriation and partner's disapproval (see Figure 2A). Men (*M* = 13.86; *SD* = 2.68) have more appropriation of condom use than women (*M* = 11.85; *SD* = 3.45), and women's score (*M* = 17.68; *SD* = 3.07) is higher in partner's disapproval (women are better at handling disapproval) than men's score (*M* = 16.9; *SD* = 3.45) (see Figure 2B). Neither assertiveness (men (*M* = 10.89; *SD* = 1.70); women (*M* = 10.82; *SD* = 1.77)) nor self-control (men (*M* = 8.77; *SD* = 2.47); women (*M* = 8.80; *SD* = 2.40)) show significant differences between men and women (see Figure 2C,D).



**Table**

157

condoms UCLA; *Plea* = sexual pleasure associated with condom use UCLA; *Sham* =

condoms UCLA; *CUEP* = condom use

errors/problems;

 *Init* = initiation Sexual Assertiveness

embarrassment

 Scale (SAS); *STI\_P\_P* = STI and pregnancy prevention SAS.

 about the purchase of condoms UCLA; *Stig* = stigma attached to persons who use

**Figure 2.** Densiogram distribution across gender. (**A**) gender differences in appropriation (App); (**B**) gender differences in partner's disapproval (Pds); (**C**) gender differences in assertiveness (Ass); (**D**) gender differences in self-control (Scn).

### **4. Discussion**

This study assessed construct and criterion validity of the CUSES [25] adaptation into Colombian Spanish language, as well its reliability and certain psychometric properties. From the socio-cognitive perspective, studies on the reduction of sexual risk behaviors among young people are widespread and cover previously mentioned factors, such as knowledge about condom use, stigma related to its acquisition, and others. The use of CUSES in the present study contributes to the recognition of perceived self-efficacy as a factor mediated by aspects such as self-control, partner´s disapproval, and condom use appropriation.

Exploratory and confirmatory factor analyses suggested a four-factor model (appropriation, partner´s disapproval, assertiveness, and self-control). This four-factor structure was found to be invariant across gender. Reliability indexes and factors related with other similar variables were adequate. Gender-based differences were also observed. These results show that the current version of the CUSES is valid and reliable for its use in Colombia.

Our first factorial analysis suggested a three-factor scale. Shaweno and Tekletsadik [24] also observed a three-dimensional structure in a brief 9-item version. However, the majority of previous studies [25,26,28] observed four dimensions. Therefore, we also considered a four-dimension structure with even better indicators. This model explained 69% of the variance. Other studies have shown a range of variance ranging from 48.2% [23] to 73.72% [28]. Furthermore, the confirmatory analysis presented better-fit indexes for this four-dimension model that was finally settled down with a strict invariance across gender. To the best of our knowledge, this is the first CUSES study including a strict invariance test. This level of invariance allowed for the comparison of the mean of the factors with a minimal bias across gender due to variance and covariance errors [49]. In other words, items were measured with the same precision in each group and no gender biases were found, this factor is key to better understand sexual behaviors [50]. Our fourth factor was termed self-control, which is consistent with the "intoxicants" [26], "pleasure and intoxicants", [28] or "intoxicant control" [24] factors. The different label used in our study was due to the fact that, in the cultural adaptation, we used an adjustment of cultural words and context [51]. Thus, "heat of passion" is not related to the consumption of alcohol or psychoactive substances in the Colombian cultural context, although, taken together, all these factors are associated with self-control.

All items show adequate psychometric properties and total item correlations above 0.50, as recommended by the literature [52]. In addition, slight improvements can be observed when removing item 3 of the appropriation factor ("I feel confident I could gracefully remove and dispose of a condom when we have intercourse") or item 15 of the self-control factor ("I feel confident I could stop to put a condom on myself or my partner even in the heat of passion"). However, its level of adjustment is good and the improvements if the items were eliminated are considered negligible. Another indicator is item improvements. Item distribution is not normal and a remarkably high kurtosis is found; however, standard deviations are close to 1, indicating adequate score variability [53]. Alphas are observed to be similar to those reported in other studies, with values ranging from 0.76 to 0.92. These values are considered adequate for both research and professional practice.

Consistent with prior research, we find that the CUSES is related with attitudes toward condom use [27]. Concerning relationships between CUSES and other measures (i.e., criterion validity), they are found to be low or moderate in most cases. We find associations with the UCLA Multidimensional Condom Attitudes Scale (MCAS) factors, in which higher scores represent better attitudes toward condom use. Ritchwood, Penn, Peasant, Albritton, and Corbie-Smith [54] observed that greater condom use self-efficacy was predicted by favorable attitudes toward condom use, which is similar to the association between partner´s disapproval, and condom use negotiation observed in the present study. Likewise, the present study finds an association between partner´s disapproval and attitudes toward condom use stigma, which has been related with reduced safer sex practices [55]. These findings highlight the necessity of interventions or training programs to improve young people's abilities to cope with factors that could result in not using a condom. Although the Colombian Ministry of Health has designed advertising campaigns and prevention programs aimed to young people population, it is important to design strategies considering academic curricula. Additionally, a low and inverse association between self-efficacy factors and Condom Use Errors and Problems (CUEP) factors is found: scores on condom use self-efficacy increase as those associated with errors and problems with condom use decrease, which has previously been reported [56]. This observation is interesting because actual problems and errors seem to differ from one's perception of the correct use of condoms, or self-efficacy. Significant relationship with variables such as initiation assertiveness (Init) and sexually transmitted disease/unintended pregnancy prevention assertiveness (STD-P P) were also verified according to assumptions. Thus, condom use self-efficacy certainly has an impact in delaying young people's sex initiation and reducing sexually transmitted infections and unplanned pregnancies [57].

Gender comparisons show that men scored higher in appropriation and lower in partner´s disapproval than women. This may suggest a need for sexual health intervention programs aimed at increasing condom use self-efficacy differentially for men and women. For men, programs should emphasize how to deal with rejection by the sexual partner, and for women, training should be oriented toward appropriation. Differentiated interventions should also be designed to prevent STIs and unintended pregnancies. In this regard, the instrument's subscales provide information on the effectiveness of negotiating condom use that could be associated with cultural factors in the Colombian context [58]. Highly typified gender roles, benevolent sexism, and prejudices associated with female sexuality and women's sexual autonomy [59,60] have also been described by Peasant, Parra, and Okwumabua [61] as factors affecting condom use self-efficacy among women, including their ability to successfully negotiate condom use without compromising their relational goals.

A relevant strength of the present study is the sample size and its heterogeneity across the cities, that means that the CUSES can be used to obtain a reliable and valid measurement of condom use

self-efficacy among Colombian young people, and it can be used as a baseline characterization scale for other Latin American Spanish-speaking populations to help in the early identification of youth condom use self-efficacy. Finally, this is the first successful adaptation in the continent.

Limitations of the present study is the non-randomized sampling approach used, the sample was skewed female and heterosexual an, the study excluded people without internet access because of the web-based survey method used. Future studies can further explore the differences in scores obtained by Spanish speaking groups at high risk of acquiring HIV/STI (an example, homosexual males, men who have sex with men and sex workers).

### **5. Conclusions**

The CUSES is the first instrument used in Colombia to measure condom use self-efficacy in Spanish language. The reliability of the CUSES has been established in different studies since 1994 [23–27]. The present study allows CUSES use in Colombia with enough psychometric guarantees for its use for both, social/health and research interest. Some gender differences in condom self-efficacy and uncommon condom use were observed, suggesting the need to implement sexual health programs in Colombia considering these differences.

**Author Contributions:** V.S.-M. and P.V.-M.: design of the study; acquisition, analysis, and data interpretation; formal analysis of data; drafting—original draft preparation; drafting—final version; have approved the submitted version. E.S.-A.: design of the study; drafting—review and editing; has approved the submitted version. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

International Journal of *Environmental Research and Public Health*
