**Son Preference and the Reproductive Behavior of Rural-Urban Migrant Women of Childbearing Age in China: Empirical Evidence from a Cross-Sectional Data**

### **Xiaojie Wang 1, Wenjie Nie <sup>1</sup> and Pengcheng Liu 2,\***


Received: 2 April 2020; Accepted: 1 May 2020; Published: 6 May 2020

**Abstract:** Son preference has been shown to influence the childbearing behavior of women, especially in China. Existing research has largely focused on this issue using cross-sectional data of urban or rural populations in China, while evidence from the rural-urban migrant women is relatively limited. Based on the data of China Migrants Dynamic Survey in 2015, we used logistic regression models to explore the relationship of son preference and reproductive behavior of rural-urban migrant women in China. The results show that the son preference of migrant women is still strong, which leads women with only daughters to have significantly higher possibility of having another child and results in a higher imbalance in the sex ratio with higher parity. Migrant women giving birth to a son is a protective factor against having a second child compared to women whose first child was a girl. Similarly, the effects of the gender of the previous child on women's progression from having two to three children showed the same result that is consistent with a preference for sons. These findings have implications for future public strategies to mitigate the son preference among migrant women and the imbalance in the sex ratio at birth.

**Keywords:** son preference; reproductive behavior; migrant women; subsequent parity

### **1. Introduction**

A strong preference for sons over daughters is common in East and Southeast Asia [1,2], notably in China. Son preference is often thought to be an important cause of imbalance in the sex ratio at birth [3–5]. According to the Global Gender Gap Report 2018, China ranked dead last among 149 countries in terms of "sex ratio at birth". This is the result of many factors, such as the Confucian cultural tradition, the socioeconomic system, and gender ideology. In the context of traditional patriarchal, patrilineal and patriarchal systems, sons are considered to have unique value, as they inherit the family name and property and represent an economic value premium to the family and parents [6–10].

Significantly different from that in other countries, the family planning policy in China is one of the limiting factors affecting fertility practices in the country [11–14]. Although the policy has been loosened by the shift from allowing one child to allowing two children for all couples, a clear limitation on the number of children still exists [15]. When there is a difference between the number of children allowed by the family planning policy and the number of children expected by the family, for individuals with a strong preference for boys, there is an incentive to conduct gender selection under the policy conditions or ignore the normal fertility limitation to have more children until a son is born. It is worth noting that the occurrence of the above two cases is strongly related to the individual's household registration status [16].

Another special policy in China, the household registration policy, also has a significant impact on women's reproductive behavior. The dual structure in urban and rural areas in China leads to significant differences in economic development, children-bearing concepts, and policy supervision [17–19]. The fertility behavior of urban residents is heavily regulated, and the opportunity cost of violating policies is extremely high, so the number of births remains low [20–22], and it is rare to maintain son preference through having more children. However, in addition to urban residents, there are large groups of rural-to-urban migrants in cities. Rural-to-urban migrants are people who have resided in an urban destination for at least 6 months and do not have local household registration [23]. The migrant population is too large to be ignored, accounting for approximately one-sixth of the total population of China [24]. According to the National Bureau of Statistics, the number of migrants reached 241 million in 2018, and half of them were migrant women. A considerable proportion of these women were of child-bearing age [25]. Although they make a living in the city, traditional concepts such as having a son to carry on the family name, raising children to provide for parents in old age and promoting the family status of mothers by giving birth to a boy are deeply entrenched in their minds, which motivates them to have more children and give birth to at least one boy [26]. Furthermore, the opportunity cost of violating policies is lower for migrant women than for urban women, so the fertility behavior of continuing to have children until a son is born is more likely to be observed in this group. However, scholars have pointed out that the well-bear and well-rear concept of urban residents may have a demonstratable effect on migrants and gradually change their fertility conception [19,27,28]. Studies have shown that migrants generally have lower fertility rates than rural residents and higher fertility rates than urban residents [27,29].

To a large extent, the birth conception and behavior of migrant women have a profound impact on the population changes in the places of domicile and migration, which will further affect the implementation of China's family planning policy, the overall fertility level and the changes in the gender structure [30]. Moreover, due to the high mobility of this type of population, the government has difficulties conducting cross-regional fertility regulations. Therefore, the constraints of the family planning policy on migrants are greatly decreased, so fertility intention is more likely to be implemented; that is, there may be a high degree of consistency between the migrating population's reproductive preference for sons and future fertility behaviors [31]. In summary, under the special policy background of China, choosing migrant women as the research object has certain practical significance for studying the relationship between the son preference and reproductive behavior of migrant women of childbearing age, predicting future population development trends and reversing the gender imbalance.

Studies on women's reproductive behavior in the existing literature focus on family reproductive intention, family planning, number of sons and contraceptive usage [8,12,14,32,33]. For example, prospective analysis is performed by directly asking questions such as "How many children do you expect to have?", "Would you like to have another child?" and other questions at the level of fertility intention to conduct a prospective analysis of fertility behavior. However, whether fertility intention can be transformed into actual fertility behavior is still limited by a variety of factors, such as national policies, family economic conditions and physical conditions of the couple, and the internal deviation between fertility intention and behavior cannot be effectively estimated. In particular, there is also a large deviation between son preference in terms of fertility intention and gender selection in actual reproductive behavior. Therefore, the direct investigation of fertility behavior can better reflect the issue of son preference.

The influence of son preference on the number of children and fertility behavior remains a controversial topic. Some studies suggest that a strong preference for sons will lead to the increases in fertility level desired by the state [34,35]. Others have shown that using sex-selective technologies will lead to a reduction in the number of children and thus a lower fertility level. In terms of gender composition, although evidence from India, South Korea, and Vietnam suggests that the preference for sons is closely related to family fertility behavior, couples with only girls are likely to have more children than families with only boys or gender-balanced families [2,8,36–39]. However, in the context of China's special policy, relatively few studies have focused on rural-urban migrant women to investigate the individual characteristics and social factors that may affect fertility behavior. The complex relationship between son preference and actual fertility behavior has not been effectively tested, and it remains to be explored whether the migrant experience can influence the son preference and fertility behavior of immigrant women. In addition, this article can help overcome the adverse bias between fertility intention and fertility behavior in previous studies, and the retrospective research method adopted largely makes the research conclusions more reliable and convincing.

Regarding the research on the son preferences of migrant women, this paper conducts a retrospective study based on the 2015 national monitoring survey data of the China Migrants Dynamic Survey (CMDS) and focuses on two aspects. The first is the pursuit of the subsequent parity progression. The second is the relationship between the sex composition of existing children and subsequent childbearing behavior. That is, in the context of China's special fertility policy, what are the factors that influence migrant women to have a higher number of children? If the gender composition of existing children fails to satisfy migrant women's son preference, will the women violate national policies and have more children? As the number of children increases, will the likelihood of having boys be increased through prenatal sex identification technology? Having more children can reduce the quality of childcare, while choosing to have boys following prenatal sex identification may cause severe gender imbalances, both of which pose challenges to improving the quality of the population and ensuring the gender balance in China.

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

### *2.1. Data Source*

The data in this paper are derived from the China Migrants Dynamic Survey in 2015 (CMDS2015). This is an annual large-scale national migrant population sampling survey initiated by the National Health Commission of the People's Republic of China and coordinated by the China Population and Development Research Center. CMDS2015 covers 31 provinces (autonomous regions and municipalities) in mainland China. The stratified, multistage and proportional PPS method is adopted for sampling. The survey covers the basic information of the migrant individuals and their family members, the mobility range of the migrant women, employment and social security, income and expenditure, residence, basic public health services, management of marriage and family planning services, children's mobility and education, etc. The survey questionnaire was presented through direct interviews by investigators who had been trained uniformly and reviewed by professional instructors. After the questionnaire was completed, it was properly kept in each city (district) and randomly checked by a panel of experts. Finally, the quality of the submitted questionnaire was monitored by logical verification and telephone return visits. The CMDS2015 was characterized by the full coverage and representativeness of the migrant population, and the questionnaire has good authenticity and reliability.

### *2.2. Study Design and Participants*

Cross-sectional data of rural-to-urban migrant women with one child were derived from CMDS2015 to explore the complex relationship of son preference and reproductive behavior. Given the research question of our study, we limited our sample to migrant women of childbearing age. Specifically, the sample selection criteria in this article are being married (first marriage or remarriage), having had at least one child, being aged 15–49 years old, having rural household registration, and being a non-local resident. Following the exclusion of non-qualifying individuals and invalid data samples resulting from refusal to answer, the number of valid samples was 36,182.

The dependent variable in this study is the fertility behavior of migrant women, that is, whether each respondent progressed from having one, two and three children to having two, three and four children, respectively. The independent variable is the gender of existing children. The control variables involved in this study are grouped into two categories. The first category elicited sociodemographic characteristics of the participants, including age, ethnic group, marriage duration, individual education level, spouse's education level, employment, mobility range and residence intention. The second category assessed the participants' fertility information based on the following items: the number of existing children and the birthplace of the first child. The proportion of boys being born should be around 51 percent without son preference, and if the proportion in our data is higher than this ratio, it means that the boy preference exists. Since the outcome variable of this study is a binary variable, we used logistics regression analysis to explore the effects of motivation and influencing factors, especially the sex composition of current children, on the subsequent childbearing behavior of migrant women.

### *2.3. Statistical Analysis*

A descriptive analysis was used to describe the characteristics of the participants and their reproductive behaviors. Variables that were significantly associated with fertility behavior by chi-square analysis were entered as independent variables in the binary logistic regression, including age, ethnic group, marriage duration, individual education level, spouse's education level, gender of the first child, mobility range and residence intention. The factors which were significantly related to reproductive behavior were then included in the logistic model analysis. A single-factor regression analysis was used to explore the effect of the gender of existing children on having another child. A multivariable logistic regression model was built to identify the determinants of reproductive behaviors. Through these analyses, crude ORs and 95% confidence intervals (CIs) were estimated for the gender preference and fertility behavior of migrant women. Statistical significance was defined as *p*-values < 0.05. Data analysis was conducted using the STATA 15.0. (Stata, College Station, TX, USA).

### **3. Results**

### *3.1. Sociodemographic Characteristics*

Table 1 presents the descriptive statistics of key variables, showing that in the overall sample of 36,182, most of the respondents were between the ages of 25 and 44 (81.88%), and the majority were ethnic Han, while less than 10% were minorities. Generally, the respondents were not highly educated, with only 6% of migrant women having a bachelor's degree or above and nearly 60% having a junior high school education (57.32%), while those who had no education or only primary education accounted for 2.25% and 16.4%, respectively. Most women had stable jobs (72.77%). Migrant women with one, two and three children accounted for 57.25%, 38.06%, and 4.18%, respectively, while less than 1% of the sample had more than three children. Most of the migrant women chose to have their first child in the hospital (89.36%), while 9.4% and 1.24% of migrant women chose to bear their first child at home or in a private clinic, respectively. Regarding the geographical range of migration, nearly half of them engaged in interprovincial migration (47.22%), 31.44% were intercity migrants, and 21.34% were intercounty migrants. A total of 85.03% of migrant women had the intention to live for a long time in their destinations.


**Table 1.** Sociodemographic characteristics of the study participants (*N* = 36,182).

### *3.2. Factors A*ff*ecting Childbearing Behavior*

Variables that were significantly associated with fertility behavior by chi-square analysis were entered as independent variables in the binary logistic regression. We took fertility behavior for the second child as an example and conducted an analysis. As shown in Table 2, the frequency of having a second child for women whose first child was a girl (53.53%) was significantly higher than those whose first child was a boy (33.18%). Migrant women who have been married for more than 5 years have a much higher proportion of progressing to second births (48.10%) than women who have been married for less than 5 years (6.32%), which shows the increased marriage duration improved the chance of having a second child. There is a significant positive correlation between age and having a second child. With increasing age, the proportion of second children gradually increases. In contrast, with the improvement of women's education level, the proportion of women choosing to have a second child gradually decreases from 73.13% to 13.17%. The education level of the husband shows a similar trend. Migrant women who have stable jobs and want to live in the destination cities have higher frequency of progressing to the next birth than women who do not work or want to stay.


**Table 2.** Distribution of second-child fertility by sociodemographic variables (*N* = 36,182).

### *3.3. Sex Composition and Subsequent Parity Progression*

Table 3 shows the distribution of the sex composition of existing children. Generally, the proportion of migrant women who progressed from having one, two, and three children to having two, three, and four children, respectively, is gradually decreasing. That is, among the children with a total sample size of 53,530, first and second children represented 36,182 and 15,468, accounting for 67.59% and 28.90%, respectively, while the number of third and fourth children sharply reduced to 1697 (3.17%) and 183 (0.34%), respectively. In terms of sex composition, the sex ratio significantly increases with the birth of more children. The sex ratio of the second children (1.42) is higher than that of the first children (1.13). Although the number of third and fourth children is relatively small, the sex ratio reaches a staggering 1.81 and 2.27, respectively. All the above are significantly higher than the natural sex ratio (1.06).


**Table 3.** Distribution of the sex composition of existing children.

Table 4 shows the effects of the sex composition of children at baseline on women's parity progression from having one, two, and three children to having two, three, and four children, respectively. Considering that younger women may not have completed their childbearing progression [40], we further limited the sample to migrant women of childbearing age over 35, and the sample size narrowed to 20,433. It can be seen that in the transition from the first to the second children, 68.60% of those whose first child was a girl chose to have a second child, while only 43.02% of those whose first child was a boy chose to have a second child. This suggests that the gender of the first child is significantly associated with the probability of progressing to the next parity. In the transition to having a third child, among women whose children were both daughters, the proportion of those who had a third child was 32.16%, significantly higher than the proportion among women whose children were both sons (less than 10%). Among women with three children, relative to women with three sons, those with three daughters had nearly four times the frequency of progressing to the next parity. In addition, women who had a son and a daughter were the most likely to stop having children. Similarly, among families with three children, women with two sons and one daughter were most likely to stop having children. Our findings show patterns of association between the sex composition of existing children and subsequent childbearing behaviors that are consistent with a preference for sons. (All the above model test results are statistically significant).

**Table 4.** Baseline sex composition on subsequent childbearing (*N* = 20,433).


Table 5 shows the results of the single-factor regression analysis of the gender of existing children on having another child. In the transition from the first to the second child, migrant women with one son is a protective factor against having a second child compared to women whose first child was a girl (OR: 0.43; 95% CI: 0.413–0.450). Among the migrant women with two children, those with one (OR: 0.19; 95% CI: 0.168–0.211) or two sons (OR: 0.21; 95% CI: 0.180–0.246) had a significantly reduced chance of progressing to the next parity compared to women with two daughters. Similarly, the effects of the gender of the previous child on women's progression from having three to four children show the same result that are consistent with a preference for sons. In addition, the son preference effect is significantly enhanced with the birth of more girls.


**Table 5.** Single-factor regression analysis of the gender of the previous child on having another child.

### *3.4. The E*ff*ect of Relevant Factors on Subsequent Childbearing*

Furthermore, we took fertility behavior for the second child as an example and conducted a multifactor analysis. The results are reported in Table 6. Binary logistic regression analysis was used to explore various factors influencing whether migrant women chose to have a second child. The results showed that ethnic group, education level, residence intention, mobility range and having a job had a significant influence on the fertility behavior of migrant women (*p* < 0.05). Women with one son is a protective factor against having a second child compared to women whose first child was a girl (OR = 0.35, *p* < 0.05). Compared with minority women, a smaller proportion of Han women chose to have a second child due to the restriction of family planning policy. Women over 25 were more likely to have a second child than women aged 24 or younger. Migrant women with age of more than 35 years old increased the chance of progressing to the next parity by more than three times compared to women less than 25 years old (OR = 3.75, *p* < 0.05) (OR = 3.53, *p* < 0.05). Migrant women with education is a protective factor; those with higher education levels reported lower possibility of having another child than those who had never gone to school. The marriage duration more than 5 years increased the chance of having a second child by more than seven times compared to a marriage duration less than 5 years (OR = 7.42, *p* < 0.05). Having a job and the education level of the spouse were negatively correlated with having another child. Women with residence intention were more likely than women without residence intention to have a second child (OR = 1.19, *p* < 0.05). In addition, those who chose to have their first child at home (OR = 1.89, *p* < 0.05) or in a private clinic (OR = 2.08, *p* < 0.05) increased the chance of having a second child than those who had their first child in the hospital.


**Table 6.** Binary logistic regression analysis of factors influencing reproductive behavior.

### **4. Discussion**

China's migrant population is large, and there is a trend of further growth in the future [23,24,41]. The fertility concepts of migrant women, to some extent, aggravate the complexity of the fertility level and gender structure of cities and even the whole of China, which will have a profound impact on future marriage, family structure, family planning policy and social development [19,27]. Based on the data of CMDS2015, this paper puts forward new views on the son preference and reproductive behavior of rural-urban migrant women of childbearing age in China.

The fertility concepts of migrant women tend to be markedly different from those of their rural or urban counterparts, which are influential in women's preference regarding childbearing. Although the emerging literature indicated that the modern outlook of fertility in urban areas might reduce the parental preference for sons [27,42], some scholars argued that the son preference remained strong among migrants [7]. Our analysis confirmed the latter view: migrants still have a strong son preference in the context of China's special policy, which makes having only daughters less desirable and results in a serious imbalance in the sex ratio.

Our findings showed the patterns of association between the sex composition of existing children and subsequent childbearing behaviors that are consistent with the preference for sons, which leads women with only daughters to have significantly higher intentions of having another child and results in a higher imbalance in the sex ratio with higher parity. To elaborate, our study found that migrant women whose first child was a girl were more likely to have another child compare with women whose first child was a boy. The same is true for the second and third children. It can be said that although the migrant women live in the city, the traditional concept of son preference had not changed significantly, which motivated them to have more children and give birth to at least one boy [7]. In addition, this study found that migrant women with both sons and daughters had the lowest odds of having another child. This also indicates that migrant women have a significant preference for sons. However, if the son preference is satisfied, the sex composition of children including both boys and girls is ideal for migrant women.

Socioeconomic characteristics can explain why some migrant women were more inclined to have more than one child. The results showed that with the increase in age, the likelihood of having another child gradually increased. A possible reason is that with the increase in age, individuals obtain a certain amount of savings and a stable work and residence environment that ease the pressure of raising another child, on the basis of which son preference can be achieved. Education level was found to have a significant effect on fertility. For better-educated migrants, the likelihood of having another child decreased greatly, and the evidence of a son preference declined. On the one hand, the increase in education improves migrant women's awareness of and compliance with national policies. On the other hand, highly educated women pay more attention to the quality of childrearing than to the quantity of children [27]. Therefore, encouraging women to receive higher education is an important measure to alleviate China's sex ratio imbalance.

In addition, the sex ratio of first children is slightly higher than the sex ratio under natural conditions (1.04–1.06) [7], indicating that the motivation for the sex selection of first children is not significant. The sex ratio of the second children (1.42) was significantly higher than that of the first children (1.13). Moreover, under family planning policy restrictions, the gender imbalance among third and fourth children worsened further, and the fourth children were more than twice as likely to be boys than girls. This result indicated an obvious tendency for gender selection and a motivation to ignore family planning restrictions and have more children in the case that the gender preference for sons was not satisfied. Therefore, migrant women who gave birth to the first child at home or in a private clinic had significantly higher odds of having a second child than those who gave birth to the first child in the hospital because the former can avoid inspection by family planning authorities.

Our study makes a possible contribution to the literature regarding the effect that son preference and the sex composition of existing children has on the reproductive behavior of women of childbearing age in the rural-urban migrant population. First, most studies of son preference focus on rural areas or locally representative samples, and there are relatively few studies on migrant women, a special group of women who differ from women in rural and urban areas. This paper expands the research scope of relevant fields. Second, most studies tend to explore fertility behaviors through fertility intentions, which are measured at a particular time point and could be unreliable because intentions are fluid. With data and retrospective birth histories, we can use logistic regression analysis to estimate the parity progression among migrant women conditional on sex composition.

The limitations of the study also should be highlighted. First, our study is based on a cross-sectional study that focuses on the number and gender of existing children, which makes it difficult to analyze the dynamics of female fertility. Women's reproductive behavior could be tracked using more effective methods. If large-scale data covering migrant women can be collected in future studies, through a longitudinal study, a more comprehensive understanding of the gender preference and reproductive behavior will become possible. Second, married women rarely migrate independently. In fact, they often migrant with their husbands [43]. Under the gender norms of a patriarchal society, immigrant women may have little influence on family decisions, such as those involving the optimal number of births. Future research could also examine gender preference of husbands in the rural-urban migrant population. Third, there are various differences in the inflow and outflow areas, and the lack of regional cultural surveys have limited the conduct of our research. This requires more detailed data, which are not typically covered by the China Migrants Dynamic Survey. Thus, future research and evaluation studies with a special focus on these issues will be critical to push research in such a direction.

### **5. Conclusions**

China's long-term gender imbalance deserves much attention. Gender imbalance is closely related to son preference. Therefore, the change in the fertility concept of migrant women plays an important role in reversing the gender imbalance. However, under the existing household registration system, migrant women have not been truly integrated into urban society, which makes it difficult for them to adapt to both the urban and the rural cultures. They have become a marginal group that is separated from rural and urban areas but also closely related to these areas, which complicates the transformation of their fertility concept. Therefore, how to carry out selective intervention through public policies, especially aiming to reverse the son preference of migrant women, is a problem worthy of attention in the future.

**Author Contributions:** X.W.—conceived and designed the study protocol, literature review, revised the manuscript, and approved the final manuscript. W.N. —data analysis, prepared the initial manuscript draft. P.L.—cleaned and analyzed the data, review and edit the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by National Social Science Foundation of China (19CRK016).

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

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

**Rugsapon Sanitya 1, Aniqa Islam Marshall 1,\*, Nithiwat Saengruang 1, Sataporn Julchoo 1, Pigunkaew Sinam 1, Rapeepong Suphanchaimat 1,2, Mathudara Phaiyarom 1, Viroj Tangcharoensathien 1, Nongluk Boonthai <sup>3</sup> and Kamheang Chaturachinda <sup>3</sup>**


Received: 7 April 2020; Accepted: 2 May 2020; Published: 4 May 2020

**Abstract:** Although physicians in Thailand can carry out abortions legally, unsafe abortion rates remain high and have serious consequences for women's health. Training programs for healthcare providers on the 'Care of unplanned and adolescent pregnancies for the prevention of unsafe abortions' have been implemented in Thailand with the aim of providing information and challenging negative attitudes about abortions. This study investigated the participants of the training courses in order to: (i) evaluate their knowledge and attitudes towards safe abortions; and (ii) investigate the factors that determine their knowledge and attitudes. A pre-post study design was applied. Descriptive statistics were calculated to provide an overview of the data. Bivariate analysis, a Wilcoxon signed rank test and a multivariable analysis using multiple linear regression were applied to determine the changes in attitudes and assess the likelihood of behaviour change towards adolescents and women experiencing unplanned pregnancy and abortions, according to demographic and professional characteristics. Having had the training, healthcare providers' change in attitudes towards adolescents and women experiencing unplanned pregnancies and abortions were found to be 0.67 points for the nine responses of attitudes and 0.79 points for the 14 responses on various abortion scenarios. Changes in attitude were significantly different among the varying health professional types, with non-doctors increasing by 0.53 points, non-obstetricians and non-gynaecologists increasing by 0.46 points and obstetricians and gynaecologists (OBGYN) increasing by 0.32 points. Positive attitudes towards unplanned pregnancies and unsafe abortions and attitudes towards abortion scenarios significantly increased. The career type of the health professional was a significant factor in improving attitudes. The training program was more effective among non-doctor healthcare providers. Therefore, non-doctors could be the target population for training in the future.

**Keywords:** abortion; training; health professionals; unplanned pregnancy; Thailand; pre-post evaluation

### **1. Introduction**

Annually worldwide, approximately 42 million women with unintended pregnancies choose to undergo abortion. Of these, approximately 20 million procedures are classified as unsafe abortions [**?** ]. An unsafe abortion is defined as the "termination of an unwanted pregnancy either by persons lacking the necessary skills or in an environment that does not conform to minimal medical standards or both" [**?** ]. This includes less-safe abortions, which are conducted using outdated methods by a trained provider or conducted using safe methods but not by a trained individual, as well as least safe abortions which are conducted using dangerous methods by untrained individuals [**?** ]. Safe abortion encompasses clinical practice on abortion, such as the use of manual vacuum aspiration (MVA) and many wider social determinants including abortion-related laws and regulations. Women often resort to unsafe abortions due to barriers in accessing safe abortions, such as restrictive laws, unavailability of services, high cost, stigma, unnecessary requirements to obtain services and conscientious objections of healthcare providers [**?** ]. Unsafe abortions can result in severe maternal consequences, including chronic health complications and disabilities, and are one of the leading causes of maternal death [**?** ]. The majority of unsafe abortions, 97% globally, occur in developing countries in Africa, Latin America and Asia [**?** ]. According to the World Health Organization, "As a preventable cause of maternal mortality and morbidity, unsafe abortions must be dealt with" [**?** ].

Maternal health in Thailand has continued to suffer from unsafe abortion-related complications. A study conducted by the Thai Department of Health, found that 28.5% of hospital admissions from 787 government hospitals were a result of induced abortions. One third of these cases developed serious complications, for which over half had undergone abortions by unqualified healthcare providers [**?** ]. Another study conducted in a Thai public hospital found that of all the women admitted for the treatment of complications from abortions, 36.8% underwent unsafe abortions [**?** ]. Similarly, a study in the South of Thailand reported that unsafe abortions accounted for 35.7% of all abortions and were significantly associated with maternal, financial or family problems [**?** ]. The fatality rate of abortions in Thailand is 300 per 100,000 abortions; this is a high rate compared with the fatality rate of 1 per 100,000 abortions in developed countries [**?** ]. Adolescents are particularly vulnerable, as they are more likely to seek unsafe abortions and experience severe complications [**?** ]. In Thailand, approximately 25.9% of all pregnancies are among adolescents, of which 14.4% result in abortions, making up 18% of the total abortions in the country [**?** ]. The adolescent pregnancy rate in Thailand is 39 per 1000 women and according to the Division of Strategy and Planning of the Ministry of Public Health (MOPH) approximately half of adolescent pregnancies are at a high risk of undergoing unsafe abortions [**?** ].

The Thai Penal Code 305 and the Thai Medical Council Regulation 2003 (BE 2548) permits abortions to be carried out by qualified physicians in certain conditions, including where there is risk to the health and life of the mother or child and pregnancies resulting from rape or incest. However, many women still face barriers to safe abortion services for various reasons, including lack of awareness of abortion services, societal stigma and taboo and the perception that abortion is absolutely illegal [**? ?** ]. Healthcare provider attitudes towards abortions are also one of the major barriers preventing women from accessing safe abortion services in Thailand and therefore healthcare providers play a decisive role in ensuring safe abortions. [**?** ]. Providers with negative attitudes towards abortion can lead women with unplanned pregnancies to risk their life and seek unsafe abortions conducted unlawfully by untrained health personnel [**? ?** ]. Therefore, to rectify this problem, efforts to change the notion of abortion among healthcare providers is necessary.

The Women's Health and Reproductive Rights Foundation of Thailand (WHRRF), a non-profit organization that advocates the health and reproductive rights of women in Thailand, was established in 1998. The organization has been working in collaboration with the Thai Health Foundation, the Royal Thai College of Obstetricians and Gynaecologists (RTCOG) and the Thai Medical Council (TMC) to advocate improved health and standards of quality health services for women in the country, including access to safe abortion services. The WHRRF has been conducting a training program for healthcare professionals on the 'Care of pregnant adolescents and women for the prevention of unsafe abortions', to provide information and shift negative attitudes. Over the period of August 2017 to April 2018, three cohorts of health professionals participated in the program, held at the Royal Thai College of Obstetricians and Gynaecologists in Bangkok. The trainings were organized by the president and

vice-president of the RTCOG as well as the president, vice-president and the secretary of the WHRRF and the advisor to the Thai Medical Council. The health professionals were invited to attend the training following a formal invitation letter sent to each Provincial Health Office by the WHRRF. Medical officers from the Department of Obstetricians and Gynaecologists of Ramathibodi Faculty of Medicines and Prince Songkla Faculty of Medicines were invited to conduct the training. The three-day training consisted of three modules. The first module explained the current global and national situation on abortion, including the debates on abortions, the effect of unsafe abortions on the socio-economic situation and health of women, the situations in which unsafe abortions occur, the risk of unsafe abortions and the benefits of access to safe abortions. The second module explained the methods used to provide safe abortions, with a focus on the use of manual vacuum aspiration (MVA) [**?** ]. The final module demonstrated the method by instructors for the correct application of MVA's and allowed participants to practice using MVAs on simulation models. Despite the active participation of the program, the impact of the program on the attendees was never evaluated. This study therefore aimed to: (i) evaluate the knowledge and attitudes towards safe abortion among the training courses participants and (ii) investigate the factors that determined the knowledge and attitudes among the training course participants.

### **2. Methods**

### *2.1. Study Design and Participants*

A pre-post study design was applied. A self-administered survey was filled out by each participant before and after the training. The time used for responding to the questionnaire was about 10–15 min for pre-test and for post-test. The course attendees were asked to return the filled questionnaire form to the course facilitator. Data from health professionals who attended the training program on the 'Care of pregnant adolescents and women for the prevention of unsafe abortions' between August 2017 and April 2018 were included.

### *2.2. Ethics Approval*

This study is part of the routine monitoring system by IHPP on progress and access to sexual and reproductive health. IHPP is a research institute of the Ministry of Public Health, and therefore it was not necessary to obtain ethics approval. Despite this, the researchers strictly followed ethical standards where all individual information was strictly kept confidential and not reported in the paper.

### *2.3. Data Collection and Measures*

Data from surveys conducted by the Women's Health and Reproductive Rights Foundation of Thailand were obtained. The survey was composed of four main parts: (1) demographic characteristics; (2) work experience with adolescents and women who have had unplanned pregnancies and undergone abortions; (3) perceptions towards adolescents and women with unplanned pregnancies and unsafe abortions (9 sub-questions); and (4) scenarios on abortions (14 sub-questions). Details of all the sub-sections are presented in the results section.

The demographic measures comprised sex (male, female), age (years), type of health profession (doctor, nurse, pharmacist and welfare workers) and the specialization of doctors (general practice, obstetrics and gynaecology, family medicine, preventative medicine, among others). The measures of experience of working with adolescents and women who had experienced unplanned pregnancies and undergone abortions included: (a) counselling experience with adolescents and women who had unplanned pregnancies (ever, never); (b) treating adolescents and women who had undergone abortions already (ever, never); (c) regulations and guidelines on the termination of pregnancies (know, do not know); (d) knowledge of manual vacuum aspiration (MVA) as a tool for safe abortion (know, do not know; ever seen, ever heard, never seen, never heard); and (f) experience in using manual vacuum aspiration (MVA) (yes, no). The measures of the attitudes of health professionals

were rated on a five-point Likert scale ranging from 1 representing 'strongly disagree' to 5 'strongly agree'. Nine items measured the general attitudes towards adolescent pregnancy and fourteen items measured the attitudes towards various scenarios for when abortions may be performed.

### *2.4. Analysis*

Pre- and post-surveys were matched using a participant code number that was provided to each participant when they registered to the training and recorded onto both the pre-survey and post-survey. Participants who did not complete both the pre- and post-surveys or whose information about demographic characteristics was incomplete were excluded from the analysis. All analyses were conducted using the Stata/IC version 14 (StataCorp LLC, Texas, TX, USA) and the statistical significance was assessed at alpha of 0.05. and the statistical significance was assessed at alpha of 0.05.

### *2.5. Descriptive Statistics*

Descriptive statistics were calculated for all the demographic characteristics and experiences of working with adolescents and women with unplanned pregnancies and having undergone abortions, including percentages for categorical variables, and mean and standard deviation (SD) for continuous variables. The mean, SD, median and interquartile range (IQR) were calculated for each item for participant attitudes towards adolescents and women with unplanned pregnancies and abortions and attitudes towards various abortion scenarios.

### *2.6. Bivariate Analysis*

A bivariate analysis was conducted to determine the significance of the change in attitudes towards adolescents and women with unplanned pregnancies and having undergone abortions as well as the attitudes towards various abortion scenarios. The Shapiro–Wilk test was used to examine distribution normality. The Wilcoxon signed rank test was applied to avoid assumptions of normal distribution needed for the paired sample t-tests. The Kruskal–Wallis rank test was applied to analyse the significance of the change in attitudes according to demographic characteristics and professional experiences, including sex, age, profession as well as knowledge and experience on abortion including aspects of regulations, treating, counselling and MVA.

### *2.7. Multivariable Analysis*

Multivariable analysis was applied to assess the likelihood of the change in attitudes towards adolescents and women with unplanned pregnancies and having undergone abortions as well as attitudes towards various abortion scenarios, using demographic and experience characteristics as predicator variables. Multiple linear regression was applied. The outcome variables were the change in the average score between the pre-test and post-test results on attitudes towards adolescents and women with unplanned pregnancies and having undergone unsafe abortions (Y1) and the response to abortion scenarios (Y2) (part 3 and part 4 of the questionnaire). The predictor variables were selected from the results of the bivariate analysis. Those with statistical significance at *p* < 0.2 were included in the multivariable analysis. The authors also conducted a kernel-based regularized least squares method, a special case of linear regression which allows the researchers to account for data with non-normal distribution, in order to assess whether the results would differ from conventional regression analysis.

### **3. Results**

A total of 325 participants attended the training programme: 99, 147 and 79 in the first, second and third batch, respectively. Of these, 250 completed both the pre- and post-test surveys and had a matching survey pair. The demographics and professional experiences section were incomplete for three participants, who were excluded (Figure **??**). The data from 247 participants were included and analysed: 43.3% from the second training session, 30.8% from the first and the remaining 25.9% from the third.

**Figure 1.** Flow diagram of the included participants.

### *3.1. Demographics Characteristics and Work Experience*

The participant age ranged from 23 to 58 years with an average of 35.9 years. The majority of participants were female (84.6) and were nurses (55.5%). Of the participants that were doctors (43.72%), 52.8% were general practitioners and 36.1% were obstetricians and gynaecologists. The survey found that 63.2% of all participants had prior experience of counselling for unplanned pregnancies and 80.6% had previously treated patients that had undergone abortions. 52.5% of participants had previous knowledge on the regulations of the Medical Council on Practices Regarding Termination of Medical Pregnancy. With regards to manual vacuum aspirations, 65.23% had knowledge of MVA, 64% had knowledge on the requirements of the use of MVA, 67.21% had previously seen an MVA and 50.21% had used an MVA. Details on the demographic characteristics and experiences of all the participants are summarized in Table **??**.

**Table 1.** Participant Characteristics.



**Table 1.** *Cont*.

*3.2. Comparison of Pre-post Results on Attitudes towards Adolescents and Women Experiencing Unplanned Pregnancies and Unsafe Abortions*

The median pre-test and post-test responses were found to be significantly different for each of the nine responses on attitudes towards unplanned pregnancies and unsafe abortions at *p* < 0.001 (Table **??**). The median of the combined average of all nine responses was found to have significantly increased by 0.67 points at *p* < 0.001 (Figure **??**).

**Table 2.** The comparison of the pre- and the post-test attitudes towards adolescents and women experiencing unplanned pregnancies and unsafe abortions (*n* = 247).


Note: All *p*-Values were ≤ 0.001 using the Wilcoxon signed rank test.

**Figure 2.** Box plot of the average pre-test and post-test responses. Note: Attitudes towards adolescents and women unplanned for pregnancy, and unsafe abortions. Scenario: Response to example of scenarios on abortions. \* *p*-Value ≤ 0.001

The difference in the average responses for attitudes towards adolescents and women experiencing unplanned pregnancies and having undergone unsafe abortions was found to be significantly different according to career at *p* = 0.013. Although the median pre-test scores were equal for all career types, with an average score of 4, the median post-test scores highly varied (Figure S1). The greatest increase in positive change was found in non-doctor participants with an average increase of 0.53 points, compared with doctors who were not specialized in obstetrics and gynaecology with an average increase of 0.46 points, and obstetricians and gynaecologists with an average increase of 0.32 points. The differences in the average response were also found to be significant among participants who had prior knowledge of the regulations of the Medical Council on Practices Regarding the Termination of Medical Pregnancy (*p* = 0.0039). Those with prior knowledge had an average increase of 0.55 points, while those without prior knowledge increased by an average of 0.42 points (Table **??**).


**Table 3.** Differences in the pre- and the post-test changes in the responses according to demographic and professional characteristics.

Note: Mean (SD), Median [IQR]; \* Statistically significant (Defined as *p* < 0.05); OBGYN: Obstetricians and gynaecologists.

The predictor variables incorporated into the regression model included healthcare professions, prior knowledge of the regulations of the Medical Council on Practices Regarding Termination of Medical Pregnancy and experience of treating and counselling teenagers and women with unplanned pregnancies. The analysis determined that non-medical doctor health professionals were most likely to benefit from the training. Although both the multiple linear regression (Table **??**) and the kernel-based regularized least squares method (Table S2) showed similar trends, the results were only significant when using multiple linear regression (*p* = 0.041).

**Table 4.** Least-squares regression analysis on the attitudes towards adolescents and women experiencing unplanned pregnancies and having undergone unsafe abortions.


### *3.3. Comparison of Pre-post Results on Attitudes towards Various Scenarios for Abortions*

The median pre-test and post-test responses were found to be significantly different for each of the 14 responses concerning attitudes towards abortion scenarios (Table **??**) at *p* < 0.001. The median of the combined average of all 14 responses was found to have significantly increased by 0.79 points at *p* < 0.001 (Figure **??**). The difference in the average responses to examples of scenarios on abortions was also found to be significantly different according to career at *p* ≤ 0.001. The most significant increase was found in non-doctor participants, with an average increase of 0.84 points, compared with doctors not specialized in obstetrics and gynaecology with an average increase of 0.59 points and then obstetricians and gynaecologists with an average increase of 0.54 points (Table **??**). However, it is important to note that the pre-test scores of non-doctors were lower with a median score of 3.57, compared to the median scores of obstetricians and gynaecologists at 3.79 and other doctors with a score of 3.64 (Figure S1).

The predictor variables incorporated into the regression model included healthcare profession, prior knowledge of the regulations of the Medical Council on Practices Regarding Termination of Medical Pregnancy, and experience of counselling teenagers and women with unplanned pregnancies. Similarly to the analysis of attitudes towards adolescents and women experiencing unplanned pregnancies and having undergone unsafe abortions, the regression analysis determined that non-medical doctor health professionals were the most likely to benefit from the training. The results were found to be significant using both multiple linear regression, at *p* = 0.003 (Table **??**) and kernel-based regularized least squares method, at *p* = 0.006 (Table S1 and Table S2).

**Table 5.** The comparison of the pre- and the post-test responses to the examples of abortion scenarios (*n* = 247).



**Table 5.** *Cont*.

Note: All *p*-values were ≤ 0.001 using the Wilcoxon signed rank test.



\* Statistically significant (Defined as *p* < 0.05)

### **4. Discussion**

The evidence indicated that following the implementation of the training programme, healthcare providers' positive attitudes towards unplanned pregnancies and unsafe abortions significantly increased. Overall, career types of healthcare providers significantly contributed to changes in attitudes towards unplanned pregnancies and unsafe abortions. Detailed analysis suggested that non-obstetric and non-gynaecology doctors and supporting staff (such as nurses) tended to benefit most from the training. This was evident in the significant increase in their assessment scores relative to the scores for obstetrics and gynaecology doctors (as shown in **????**).

Our findings were similar to findings from other studies. For instance, in Zimbabwe, a study found that supporting staff apart from doctors (such as a nurse, midwife, senior nurse or hospital administrator) play an important role in supporting women in accessing safe abortions [**?** ]. A study by Cooper et al. gave a positive view on nurses' and midwives' attitudes towards abortion [**?** ]. However, a systematic literature review on the perceptions and attitudes of healthcare providers in sub-Saharan Africa and Southeast Asia by Ulrilka et al. suggested that nurses and midwives disliked being involved in abortion services, and commonly reported hesitancy in providing these services. The nurses' resistance to providing abortion services was a powerful barrier against access to safe abortion services, with nurses' and midwives' strong opposition to abortion affecting rural women in particular [**?** ]. Such findings however contradicted the findings in this study, as the results did not present strong negative attitudes towards abortion among nurses and supporting staff. In contrast, the findings showed that nurses and supporting staff could be potential target groups for further trainings, as their scores were enhanced the most compared with other health professionals.

This study has some policy implications. Firstly, this kind of training on safe abortion is useful to wider health professionals, institutions and organizations, which in turn can play an important role in creating awareness of unsafe abortion and providing safe abortion services. Secondly, abortion is not only a matter for obstetrics and gynaecologists. The study found that non-obstetric doctors and support staff can play an important role. Finally, this study found that knowledge of regulations on abortions is quite low as less than 50% of the participants had adequate prior knowledge on these (Table **??**). Therefore, the Royal Thai College of Obstetricians and Gynaecologists and the medical council should work together to communicate these regulations to health professionals and the wider public.

Some limitations remain. Firstly, the participation in this training was voluntary and this therefore created a risk of selection bias. Those who opt out from this kind of training may not have similar favourable attitudes towards abortion. Secondly, the exclusions of participants undermined the statistical power as some observations were dropped (Figure **??**). This may be a reason why some factors did not show statistical significance. However, researchers checked the demographic characteristics of the participants that were excluded and found no significant differences compared to the study sample. Thirdly, analyses by different methods can yield different results [**?** ]. The questionnaire in section three and four applied a Likert scale to analyse the attitudes of participants. In this study we opted to use regression analysis; however, if researchers devised and used other analytical methods such as the chi squared or logistic analyses, they may yield different results. However, researchers found the results to be quite valid as the kernel-based regularized least squares method was applied and there was no marked difference in the results. Fourth, participants' locations of work, either urban or rural settings, were not collected, and as the trainings were conducted only in Bangkok, it is expected that most participants were from the Bangkok metropolitan area. Therefore, differences in attitudes between urban and rural areas could not be clearly identified. Further studies should be conducted to analyse the urban and rural area differences in the attitudes by the location of their workplaces. Finally, this assessment is subjective. The results showing the positive attitudes of the participants who attended the training programme do not guarantee good practice in a real-life situation and the study sample does not necessarily reflect the viewpoints of all the health professionals in the country. Further research should assess real-life practices and attitudes, including both qualitative and quantitative components.

### **5. Conclusions**

Following the training, the score for positive attitudes towards unplanned pregnancies and unsafe abortions as well as attitudes towards abortion scenarios significantly increased. The main determinant, which significantly contributed to positive attitudes towards unplanned pregnancies and unsafe abortions, was the career type of the healthcare provider. In particular, non-doctor health professionals were likely to benefit the most from this kind of training and could be the target population for training in the future. Further research using both qualitative and quantitative methods should be conducted to assess the attitudes and real-life practices of healthcare professionals concerning abortions in Thailand.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1660-4601/17/9/3198/s1, Figure S1: Box Plot of the Average Pre-test and Post-test Response by career type, Table S1 Regularized least squares analysis on towards adolescents and women unplanned for pregnancy, and unsafe abortions; Table S2: Kernel-based regularized least squares analysis

**Author Contributions:** All authors designed the study. R.S.(Rugsapon Sanitya), A.I.M., R.S.(Rapeepong Suphanchaimat) and V.T. were responsible for data analysis. R.S.(Rugsapon Sanitya), A.I.M., N.S., S.J., P.S., R.S.(Rapeepong Suphanchaimat) and M.P. crafted the first draft of the manuscript. V.T., K.C., N.B., and R.S.(Rapeepong Suphanchaimat) revised and finalized the manuscript. All authors contributed toward data analysis, drafting and critical revision.

**Funding:** The authors gratefully acknowledge Walaiporn Patcharanarumol for the funding support through IHPP from the Thailand Science Research and Innovation (TSRI) under the Senior Research Scholar on Health Policy and System Research [Contract No. RTA6280007].

**Acknowledgments:** The investigators are immensely grateful for the support from IHPP's and WHRRF's staff.

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

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

**Leticia Avila-Burgos 1, Julio César Montañez-Hernández 1,\*, Lucero Cahuana-Hurtado 2, Aremis Villalobos 3, Patricia Hernández-Peña <sup>4</sup> and Ileana Heredia-Pi <sup>1</sup>**


Received: 18 March 2020; Accepted: 27 April 2020; Published: 29 April 2020

**Abstract:** The purpose of this study was to assess whether government policies to expand the coverage of maternal health and family planning (MHFP) services were benefiting the adolescents in need. To this end, we estimated government MHFP expenditure for 10- to 19-year-old adolescents without social security (SS) coverage between 2003 and 2015. We evaluated its evolution and distribution nationally and sub-nationally by level of marginalization, as well as its relationship with demand indicators. Using Jointpoint regressions, we estimated the average annual percent change (AAPC) nationally and among states. Expenditure for adolescents without SS coverage registered 15% for AAPC for the period 2003–2011 and was stable for the remaining years, with 88% of spending allocated to maternal health. Growth in MHFP expenditure reduced the ratio of spending by 13% among groups of states with greater/lesser marginalization; nonetheless, the poorest states continued to show the lowest levels of expenditure. Although adolescents without SS coverage benefited from greater MHFP expenditure as a consequence of health policies directed at achieving universal health coverage, gaps persisted in its distribution among states, since those with similar demand indicators exhibited different levels of expenditure. Further actions are required to improve resource allocation to disadvantaged states and to reinforce the use of FP services by adolescents.

**Keywords:** government health expenditure; adolescents; maternal health; family planning

### **1. Introduction**

Mexico and other Latin American countries need to invest in the sexual and reproductive health of adolescents [1–4]. Ensuring the availability of healthcare and promoting healthy behaviors in this population group generate economic benefits that improve future labor productivity. Efforts on behalf of adolescents also contribute to reducing risks, such as complications during pregnancy in very young women, as well as preventing premature births and low-birthweight babies [3,5]. From a human rights perspective, adolescents are entitled to receive the information and health services they need to survive, grow, and realize their full potential as individuals [1,3,5].

As a result of Mexican population dynamics, adolescents now exert an unprecedented impact on national demographics. With nearly one-fourth of the population being aged 10–19 years in 2014 [6], the health needs of adolescents have acquired particular relevance. Their behavior has evolved towards

early initiation of sexual activity [7,8] with limited use of contraceptive methods (CMs), especially among the poorer groups [9,10]. With a fertility rate that grew from 64 births per 1000 adolescents (15–19 years old) in 2009 to 77 births in 2014 [10,11], Mexico ranks first in adolescent fertility among Organization for Economic Cooperation and Development (OECD) member states [12]. The healthcare system needs to strengthen its response such that the health expectations and requirements of this age group are addressed and their access to health services is expanded.

Over two decades ago, Mexico launched its System of Social Protection in Health (*Sistema de Protección Social en Salud*) with the aim of achieving universal health coverage. Its principal component, the *Seguro Popular* health insurance scheme, was conceived as a mechanism for enhancing service coverage and providing financial protection to the 45% non-salaried population in Mexico without access to social security (SS) [13,14]. Within this scheme, improving maternal health (MH) through greater coverage has been considered a core objective. An example of this commitment was the introduction of the Healthy Pregnancy Strategy (*Estrategia Embarazo Saludable*) in 2008. This program incorporated all pregnant women without SS coverage into the *Seguro Popular,* providing them with a package of MH services free of charge [15]. Although adolescents without SS coverage were not explicitly targeted by these initiatives, they benefited from the increased supply of maternal health (MH) and family planning (FP) services [16,17].

In 2009, Mexico stepped up its efforts to improve availability of sexual and reproductive health services for adolescents. One upshot was the Model of Comprehensive Care for the Sexual and Reproductive Health of Adolescents implemented in 2013. Two years later, having recognized adolescent pregnancy as a national problem, Mexico established the National Strategy for the Prevention of Adolescent Pregnancy. This initiative was designed to reduce the fertility rate among adolescents aged 15–19 years and to eradicate pregnancy among those aged 10–14 years [8]. One of its principal components was greater coverage for sexual and reproductive health services and modern contraception methods (CMs). To this end, the National Strategy engaged numerous public health institutions—those pertaining to the Social Security Institute, which covered 43% of the population, and those serving the population without SS coverage through the *Seguro Popular*, the State Health Services, and the *IMSS PROSPERA* Program, renamed *IMSS Bienestar* at the close of 2018 [18,19].

These initiatives expanded the coverage of antenatal care services [16] and improved the availability of CMs for the adolescent population lacking SS coverage [7,9], thereby mobilizing greater resources from the government. To facilitate the transfer of funds to State Health Services, the *System of Social Protection in Health* introduced a per capita payment scheme based on the number of affiliates. This was intended to overcome the historical inertia in the distribution of funds [20].

The Mexican *Reproductive Health Account* system offers information on public maternal health and family planning (MHFP) spending from 2003 to 2015 [21], which serves to monitor and analyze the resources and performance of these services during this period [17,19,22,23]. However, data are not disaggregated by age group, and the consequences of changes in MHFP spending for adolescents without SS coverage are therefore unknown. To assess whether policies for expanding the coverage of maternal health and family planning (MHFP) are benefiting the adolescents in need, we (1) estimated MHFP expenditure for the entire adolescent population without SS coverage and determined its growth and distribution among MH and FP services and providers for the period 2003–2015; (2) analyzed the expenditure gaps among states by level of marginalization; and (3) examined the relationship between MHFP spending and demand indicators. Given that changes in the health system have centered on the population without SS coverage, we confined our analyses to this population.

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

We conducted an ecological study in order to estimate government MHFP expenditure for adolescents without SS coverage. To achieve this, we drew data for 2003–2015 from the *Reproductive Health Accounts* [21], constructed according to the *OECD System of Health Accounts* [24] and the *World Health Organization's Guide to Producing Reproductive Health Subaccounts* [25].

### *2.1. MHFP Expenditure for Adolescents*

Analysis included government strategies aimed at the population without SS coverage, namely the *Seguro Popular* insurance and care providers, the State Health Services, and the *IMSS PROSPERA* Program [19,21]. Expenditure for adolescents was grouped according to care providers, specifically hospitals and ambulatory-care centers [21,24]. The selection of beneficiaries focused on young people between 10 and 19 years old, the age range for adolescence established by the Specific Action Program for the Sexual and Reproductive Health of Adolescents [26].

To estimate expenditure for the adolescent population, we carried out the following procedures. First, we identified the number of MH and FP consultations offered by ambulatory-care centers to 10 to 19-year-olds without SS coverage. Calculations were undertaken by type of service (MH or FP), year, and state. Subsequently, we estimated the proportions of these consultations with respect to the total number of MH and FP consultations offered to the total population without SS coverage. Likewise, we weighted MH and FP expenditures incurred by hospitals using the proportions of in-hospital days and consultations offered to adolescents without SS coverage. Based on OECD and WHO methodology [24], we used these proportions to weight the MH and FP expenditures sustained by ambulatory-care centers and hospitals for the total population lacking SS coverage. The underlying assumption was that spending was similar for comparable types of in-hospital days and consultations regardless of the ages of patients. Analyses covered 131.9 million in-hospital days as well as 1571 million general and specialized in-hospital consultations. We obtained the data from the General Directorate of Health Information under the Federal Ministry of Health (*DGIS*) [27], and grouped them according to the International Classification of Diseases, 10th Revision (ICD-10) [28].

MH services included care during pregnancy, childbirth (vaginal or cesarean), and the postpartum period, as well as abortion. FP services included counseling, consultations, the provision of CMs (hormonal or otherwise), and the performance of definitive surgical procedures (tubal ligation and vasectomy) for the entire adolescent population, 10 to 19 years old. Consultations were grouped by provider (hospital or ambulatory-care center).

We calculated MHFP expenditure in constant US dollars and converted the figures to 2015 international dollars (Purchasing Power Parity, PPP, 2015 US\$1 = MXN 8.541) [29]. To establish a comparison among states, we adjusted MHFP expenditure by adolescent women aged 10–19 according to *DGIS* data [27].

### *2.2. Demand Indicators*

Demand has been defined as the proportion of a population experiencing health needs and requiring health services to satisfy them. For the purposes of this study, we defined the demand indicator for MH services as the number of pregnant adolescents without SS coverage, and for FP services as the number of sexually active adolescent women without SS coverage. To estimate these indicators, we used data from the 2009 and 2014 National Surveys of Demographic Dynamics (*ENADID*s) [30]. Although the *ENADID* design identifies fertility and pregnancy at the population level, it captures information only on pregnancies among 15- to 54-year-old women. Thus, our demand indicator for MH services included only 15- to 19-year-old adolescents who were pregnant at the time of the surveys or the year before. Under the demand indicator for FP services, we included adolescent women between the ages of 15 and 19 who were sexually active during the month prior to the surveys and those aged 10 to 19 who became sexually active during the survey period.

To ensure comparability, we constructed the MHFP indicators for adolescents in accordance with international practices. For instance, we considered women as the basis of our CM demand indicators given the distinct impact that the use/non-use of CMs exerts on their reproductive health and risk of pregnancy. Moreover, female contraceptives offer a wider range of methods and costs compared to male contraceptives, limited mostly to condoms available at lower costs [31,32]. For these reasons, most resource-tracking methodologies, including the Health Accounts, base their indicators of FP spending on women as the common denominator [25]. Finally, international recommendations on

investment in adolescent FP activities also consider women as the basis, particularly in light of the high-priority problem of adolescent motherhood [33]. We obtained data on the size of the adolescent population lacking SS coverage for the period 2010–2018 from the *DGIS* databases [27]. As information was unavailable for the period prior to 2010, we calculated data from 2003 to 2009 based on the average annual growth of the adolescent population without SS coverage during 2010–2018. States were grouped according to the State Marginalization Index of the National Population Council [34].

### *2.3. Analytical Strategy*

To assess growth in MHFP spending during the period 2003–2015, we estimated the average annual percent change (AAPC) in MHFP expenditure through Jointpoint regression models [35]. Given the inertial allocation in the health budget [36], we adjusted the models using the logarithm of expenditure with autocorrelated errors. For 2009 and 2011, we inserted nodes on the introduction of the Healthy Pregnancy Strategy and its efforts to enroll the entire non-SS population in the *SP*, respectively. For each program, we estimated the ambulatory-care center/hospital ratio as an indicator of the relative growth in expenditure at ambulatory- and primary-care centers.

To analyze the alignment of expenditure with the population requiring MH and FP services among states, we calculated and assessed the concentration indices (CIs) for 2009 and 2014 based on their concentration curves. This involved the following procedures: (a) we calculated the proportions of public MH and FP expenditures spent in state j with respect to total MH and FP expenditures at the national level, respectively; (b) we calculated the proportion of the population requiring these services in state j with respect to the total population requiring these services at the national level; and (c) we generated concentration curves by arranging the states on the x-axis according to the proportions of populations needing these services, from the lowest to the highest, and connecting them with their respective proportions of expenditures on the y-axis. The CI is the area between the concentration curve and the diagonal, and its values range from −1 to 1, with zero denoting equality [37]. Finally, we analyzed the expenditure and population data for 2014 by creating scatter plots and estimating Spearman correlations to ascertain their relationships and statistical significance. We used STATA version 13.0 for the analyses [38].

### *2.4. Ethical Considerations*

To obtain data on service production (in-hospital days as well as consultations offered at hospitals and ambulatory-care centers), we used secondary public sources and the *ENADID* database, neither of which contained personally identifiable information. For expenditures, we used the *Reproductive Health Accounts*. This study was approved by the Research Ethics Committee of the National Institute of Public Health (No. 577-2016).

### **3. Results**

In 2015, MHFP expenditure for the adolescent population without SS coverage totaled US\$428 million, 88% of which was spent on MH. Meanwhile, the AAPC for the period 2003–2011 amounted to 15.4% (CI95%: 14.3−16.5) (Figure 1b and Table 1). The MH/FP ratios stood at 21.3 in 2003 and 7.5 in 2015. At the national level, expenditure per adolescent woman in the 10–19 age group rose from US\$17 in 2003 to \$64 in 2015 (Figure 1a,b and Table 1).

MH and FP services demonstrated different rates of growth for expenditures, which modified the distribution of funds among healthcare providers (Figure 2). From 2003 to 2011, spending on antenatal care in ambulatory-care centers showed an AAPC of 29.3% (CI95%: 23.5–35.4). Accordingly, while 61% of expenditure in ambulatory-care centers was used to finance antenatal care in 2003, this figure had grown to 85% by 2015. Spending on postpartum care grew at an annual rate of 9.3% from 2003 to 2009 (CI95%: 5.4–7.8) and 5.5% during the rest of the period analyzed (CI95%: 4.2–6.8) (Table 1 and Figure 2a).

Hospital spending on MH complications and childbirth registered an AAPC of 14.3% (CI95%: 13.2–15.5 and 11.9–16.8, respectively) from 2003 to 2009; the AAPC for the rest of the period analyzed was lower. Meanwhile, hospital spending on antenatal consultations, at a lower level, showed an annual growth rate of 8.1% (CI95%: 4.0–12.5) from 2003 to 2009, and 3.0% (CI95%: 1.5–4.5) from that year until 2015 (Table 1 and Figure 2b). The ambulatory-care center/hospital ratio of expenditure on MH services reflected a higher growth rate in ambulatory-care centers—in 2003, for each dollar spent by hospitals on MH, ambulatory-care centers spent \$0.3. By 2015, the ratio was 0.6.

In contrast with MH, FP expenditure was predominantly spent at ambulatory-care centers (Figure 3). Until 2008, the trends and levels of FP expenditure at hospitals and ambulatory-care centers were similar. However, from 2009 to 2011, spending by ambulatory-care centers experienced its period of greatest growth, with an AAPC of 130.6% (CI95%: 25.5–323.7). Thus, although hospital spending grew at an annual rate of 6.6% (CI95%: 3.7–9.5), by the end of the period analyzed, the ambulatory-care center/hospital spending ratio for FP services was 11 (Figure 3 and Table 1).

Table 2 shows that national expenditure on MHFP per adolescent woman had an AAPC of 11.8% (CI95%: 9.2–14.5) during the period 2003–2015. The increase in spending per adolescent woman reduced the gap between states with higher and lower levels of spending by 48% (from 5.96 in 2003 to 3.15 in 2015). Furthermore, the expenditure gap between states with higher and lower levels of social marginalization diminished by 13% (1.92 in 2003 and 1.68 in 2015). Although regions classified as having moderate, high, and very high levels of marginalization increased their spending per adolescent by 12% per year, no significant differences emerged in the annual rates among the five regions.

**Figure 1.** Government Maternal Health and Family Planning (MHFP) expenditure for adolescents aged 10–19 without Social Security coverage, Mexico, 2003–2015, US\$ (PPP 2015).



*IJERPH* **2020**, *17*, 3097

**Figure 2.** Government Maternal Health (MH) expenditure for adolescent women aged 10–19, without Social Security coverage, by type of healthcare provider, Mexico, 2003–2015, US\$ million (PPP 2015).

**Figure 3.** Government Family Planning (FP) expenditure for the entire adolescent population aged 10–19 without Social Security coverage, by type of healthcare provider, Mexico, 2003–2015, \$ million (PPP 2015).

Finally, Figure 4 shows the concentration curve for expenditures on MH and FP and the correlations between expenditures and their respective demand indicators. In 2009 and 2014, half of the states (n = 16) concentrated less than 30% of the expenditures (concentration index = 0.32 for both years) (Figure 4a). However, for FP expenditure, the levels of inequality increased between 2009 and 2014 (CI = 0.30 and CI = 0.38, respectively) (Figure 4b). On the other hand, although a positive and significant correlation emerged between expenditure on MH at the state level and its demand indicator (rho > 0.9476, *p* < 0.05), differences were observed in the distribution of spending. For example, states such as Michoacan (MICH) and Guerrero (GRO) exhibited different levels of spending (Figure 4c) despite having comparable proportions of pregnant adolescents. The situation was similar for expenditure on FP (rho > 0.9016, *p* < 0.05), where Guerrero (GRO) and Mexico City (CDMX) presented comparable proportions of sexually active adolescents (Figure 4d) but different expenditure levels. Likewise, Figure 4c indicates that the MH expenditure levels in states with a low marginalization status, such as Jalisco (JAL) were similar to those of highly marginalized states and smaller populations, such as Tabasco (TAB) and San Luis Potosí (SLP).



(SLP), Campeche (CAMP), Michoacan (MICH) and Yucatan (YUC); moderate

(MOR), Sinaloa (SIN), Zacatecas (ZAC), Durango (DGO) and Nayarit (NAY); Low

Chihuahua (CHIH), Tamaulipas (TAMP) and Baja California Sur (BCS); very low

marginalization:

 Queretaro (QRO), Guanajuato (GTO), Tlaxcala (TLAX), Quintana Roo (QROO), Morelos

marginalization:

marginalization:

 Baja California (BC), Coahuila (COAH) Nuevo Leon (NL), and Mexico City (CDMX).

 Jalisco (JAL), Mexico (MEX),

Aguascalientes

 (AGS), Sonora (SON), Colima (COL),

(**c**) Spearman's rho for MH 2014 = 0.9476. (**d**) Spearman's rho for FP 2014 = 0.9014.

### **4. Discussion**

The adolescent population is one of the groups where public policy could have the most dramatic impact given the repercussions for health and wellbeing and the economic benefits these policies would generate [1,4]. In a context of limited resources, it is crucial to identify the areas of opportunity in which investments in the healthy development of this population group would be most efficiently used. In this regard, the results of this study indicate that efforts by the Mexican government to expand healthcare coverage for the population without Social Security (SS) coverage benefited the non-SS adolescent population. Thus, government expenditure on maternal health and family planning (MHFP) showed an average annual percent increase of 15% for the period 2003–2011 but no significant change in the remaining years. During 2003–2015, MHFP expenditure for non-SS adolescents represented 25% of government MHFP expenditure for the total population lacking SS coverage [21]. Nonetheless, despite the growth in spending, inequalities in distribution persisted. The adolescent population living in the most marginalized states and suffering from the greatest level of economic inequality, as well as historical differences in resource allocation [16,17,39] continues to demonstrate below-average levels of per capita spending.

In terms of sexual and reproductive health, it has been documented that the Mexican adolescent population has low rates of contraceptive use and difficulty in planning their sexual lives [7,9,40]. For these reasons, the first contact this population has with health services is generally to receive antenatal care, and their use of CMs frequently begins after the first pregnancy [8]. These patterns result in a situation in which antenatal, childbirth, and postpartum care claim 88% of MHFP expenditure for adolescents. The results show a greater increase in expenditure for MH services in ambulatory-care centers than in hospitals. This difference is explained primarily by a surge in the volume of antenatal consultations offered to adolescents without SS coverage (which rose from 66,436 in 2003 to 1.56 million in 2015) [21]. Despite an improvement in antenatal coverage for pregnant adolescents rising from 61% to 71.8% between 2000 and 2012 [16], the figures continue to be lower, especially in marginalized communities, than those achieved for pregnant adults (20 years and older) [16,41,42]. Because pregnancy during adolescence increases the risk of obstetric complications [1], it is hardly surprising that a third of MH expenditure was spent on hospital care for complications arising during pregnancy, childbirth, and the postpartum period.

With regard to FP expenditure, the observed annual increase of 130% incurred by ambulatory-care centers from 2009 to 2011 was a reflection of various events: (a) an increase in the number of people enrolled in the *Seguro Popular* from 9.1 million in 2009 to 43.5 million in 2010, which enhanced the coverage of health services, FP included, for the population lacking SS coverage [9,13,14,43]; (b) the strengthening of the FP program specifically for adolescents, leading to an increase in post-obstetric-event contraception [42] and which boosted the percentages of sexually active adolescents using a CM in their first sexual intercourse from 43% to 66% in women and from 70% to 85% in men, where the latter was attributed to greater male condom use [44]; and (c) the centralization and increase in the purchase of CMs [8]. As a result of higher FP expenditure, spending per adolescent woman 10–19 years old rose from US\$0.8 in 2003 to \$7.5 in 2015, reaching the threshold of \$2.93 per woman per year since 2011, as recommended by the Guttmacher Institute [33]. In spite of these advances, however, deficiencies in CM coverage persist, and sexually active adolescents continue to be the group with the lowest rate of CM use in Mexico [9,40]. Evidence indicates that investment in FP programs generates significant monetary and social returns [34,45]. It has been estimated that each additional dollar invested in satisfying the demand for modern CMs generates savings of US\$2–6 in healthcare spending for pregnant women and newborns [34]. If we also include the long-term effects of reducing maternal and infant mortality and increased economic growth, savings rise to US\$60–120 per dollar spent on FP services [45]. This underlines the relevance of continuing to invest in FP for adolescents, as well as the need to make more efficient use of these resources.

Our results document the growing importance of ambulatory-care centers as health-service providers. These providers will no doubt continue to increase in importance, given that the current health reform in Mexico, which has replaced the *Seguro Popular* with the Health Institute for Wellbeing, is oriented towards strengthening the primary-care model for the population lacking SS coverage [46,47]. Nonetheless, the heterogeneity that characterizes ambulatory-care providers in terms of resources and the quality of services delivered [16,48,49] suggests that it will be necessary to improve their capacity to respond, particularly with regard to their supply of modern CMs [49].

Although government MHFP spending has increased across the board nationally, discrepancies persist in its distribution among states, since those with similar adolescent populations in need of MH and FP services exhibit different levels of expenditure. This could be the result of various factors, such as the persistence of inertial allocations: (a) in 2015, 33% of the *Seguro Popular* budget earmarked for payroll continued to be allocated to State Health Services on the basis of long-established, routine procedures [39]; (b) the concentration of infrastructure and personnel have traditionally privileged some states at the expense of others; and (c) managerial capacities have diverged and continue to vary widely among State Health Services [17,48,50].

Our study had the following limitations: (1) our analysis was restricted to government schemes providing coverage for the population without SS coverage, which prevents generalizing results to the entire public health system. However, these schemes provided coverage to 45% of the Mexican population [13,14], and their expenditures represented 46% of total public health spending in 2016 [39]; and (2) using production data to estimate the distribution of expenditure among health conditions and/or diseases, which could lead to estimation errors. Nonetheless, the OECD has evaluated this methodology in various countries, demonstrating its validity and consistency [51]. The WHO [52], as well as the OECD [53] encourage its use. (3) The age range considered for the demand indicators of MH services was also 15–19 years. After reviewing a variety of data sources, such as the Birth Information Subsystem, we decided to use data from the National Survey of Demographic Dynamics (*ENADID*). We arrived at this decision because the Birth Information database omits deceased children and abortions. Information from the *ENADID* thus provided the closest approximation to the adolescent pregnancy phenomenon under study. In addition, it has been documented that pregnant women under 15 register their children with the appropriate authorities later than other women [54]. (4) Finally, it should be noted that expenditure was analyzed at the state level, without considering the wide variability in the distribution of local spending [17,50]. Future studies need to explore in greater detail the relationship among MHFP resources, as well as their distribution and health outcomes at the municipal level.

### **5. Conclusions**

Governments around the world have recognized the need to invest in the sexual and reproductive health of adolescents. Sustainable Development Goal 3.7 [55] calls upon countries to ensure universal access to sexual and reproductive healthcare services, including for FP. This will require additional resources. Financial evidence on the levels of expenditure allocated to these services and on its distribution throughout the population is a key input for planning public investment. It serves as a basis for governments and health authorities to define how much more they must invest, what types of services should be prioritized, and which areas can be improved with regard to equity and efficiency.

The findings of this study fill an information gap on the levels of investment in sexual and reproductive health services in Mexico for a group traditionally lacking visibility—the adolescents. Our results demonstrate that the health policies implemented between 2003 and 2015 increased expenditure on the sexual and reproductive health of adolescents without Social Security coverage; in spite of this, however, problems persist in ensuring an equitable distribution of these resources. Looking ahead, the implementation of specific policies for the prevention of adolescent pregnancy will require special attention as the current health reform evolves. It will be necessary to monitor the financial implications of the ensuing changes and their consequences for adolescent health services. Subsequent analyses will also need to combine the allocation of expenditure with results indicators in this population in order to understand the extent to which investments are provided equitably and efficiently.

**Author Contributions:** Conceptualization, L.A.-B.; data curation, formal analysis, J.C.M.-H.; validation, methodology, L.A.-B. and J.C.M.-H.; critical review of the manuscript, writing-review and editing, L.A.-B. and J.C.M.-H., I.H.-P., L.C.-H., P.H.-P. and A.V.; funding acquisition, I.H.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the CONACYT Sectoral Fund for Research in Health and Social Security in Mexico (Grant No. 261230, year 2015). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

**Acknowledgments:** We thank the National Center for Gender Equity and Reproductive Health of the Ministry of Health for access to and use of data from the Reproductive Health Accounts. We also extend our thanks to Luz Maria Montes Romero for her support in the search for and recuperation of the bibliography related to this article, as well as to Patricia Solis Albarrán for her valuable contribution in the English translation and editing of this work.

**Conflicts of Interest:** The authors declare having 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* **Economic Crisis and Sexually Transmitted Infections: A Comparison Between Native and Immigrant Populations in a Specialised Centre in Granada, Spain**

**María Ángeles Pérez-Morente 1, Adelina Martín-Salvador 2,\*, María Gázquez-López 3, Pedro Femia-Marzo 4, María Dolores Pozo-Cano 4, César Hueso-Montoro 4,\* and Encarnación Martínez-García <sup>4</sup>**


Received: 13 March 2020; Accepted: 31 March 2020; Published: 5 April 2020

**Abstract:** This study aimed to analyse the influence of the economic crisis on the prevalence of sexually transmitted infections (STIs) in the immigrant population compared to the native population. A cross-sectional study was conducted by reviewing 441 clinical records (329 Spanish nationals and 112 non-Spanish nationals) of individuals who, between 2000 and 2014, visited an STI clinic in Granada and tested positive for an infection. Descriptive statistical analyses were performed, and infection rates, odds ratios, and 95% confidence intervals (CIs) were calculated. The mean age was 28.06 years (SD = 8.30; range = 16–70). During the period 2000–2014, the risk of being diagnosed with an STI was higher among non-Spanish nationals than among Spanish nationals (odds ratio (OR) = 5.33; 95% CI = 4.78–6.60). Differences between both populations were less marked during the crisis period (2008–2014: OR = 2.73; 95% CI = 2.32–3.73) than during the non-crisis period (2000–2007: OR = 12.02; 95% CI = 10.33–16.17). This may be due to underreporting of diagnoses in the immigrant population. Immigrants visiting the STI clinic in Granada are especially vulnerable to positive STI diagnoses compared to the native population.

**Keywords:** sexually transmitted infections; economic recession; transients and migrants

### **1. Introduction**

Sexually transmitted infections (STIs) are a global public health problem. According to the World Health Organization (WHO), more than one million people contract an STI every day. An estimated 357 million people become infected with chlamydia, gonorrhoea, syphilis, or trichomoniasis each year [1].

According to the latest epidemiological report from the European Centre for Disease Prevention and Control, the number of gonorrhoea infections had increased by 17% in 2017 compared to the previous year. The United Kingdom was the country with the highest proportion of confirmed cases (75 per 100,000 inhabitants), followed by Ireland (47), Denmark (33), Iceland (29), Norway (27), and Sweden (25). The countries with the lowest proportion of reported cases (<1 per 100,000) were Bulgaria, Croatia, Cyprus, Poland, and Romania. There were 33,189 (7.1 cases per 100,000 inhabitants) new cases of syphilis. The highest rate was found in Iceland (15.4 cases per 100,000 inhabitants), followed by Malta (13.5), the United Kingdom (11.8), and Spain (10.3). The lowest rates (<3 cases per 100,000 inhabitants) were observed in Croatia, Cyprus, Estonia, Italy, Portugal, and Slovenia. Furthermore, 25,353 people

were diagnosed with Human Immunodeficiency Virus (HIV) in the European Union (6.2 cases per 100,000 inhabitants) [2].

In 2016, in Spain, the STI with the highest incidence rate was chlamydia trachomatis (17.85/100,000), followed by gonorrhoea (13.8/100,000), and syphilis (7.22/100,000). In the case of syphilis, it has been stabilised since 2011. However, gonococcal infection rates have increased steadily from 2001 to 2016, from 2.02/100,000 to 13.89/100,000 [3].

Men who have sex with men (MSM), sex workers, transgender people, intravenous drug users, and immigrants have been identified as the key populations at highest risk of contracting an STI [4].

Increased migration has contributed to the spread of HIV, Hepatitis B Virus (HBV), and other STIs, with the vast majority of cases occurring in migrants from low- and middle-income countries who have moved to high-income countries. Most interventions in the United States of America, Australia, and Europe focus on individual behaviour rather than on broader sociocultural factors [5].

According to the Spanish National Statistics Institute (INE, by its Spanish acronym), the immigrant population in Spain in 2018 increased by 100,764 to 4,663,726, an increase 23% larger than in 2017 [6]. In 2017, 36.1% of new Human Immunodeficiency Virus (HIV) infections were diagnosed in non-Spanish individuals, being more frequent in the Latin American population (19%) [7]. In a study conducted at a hospital in Madrid, half of the 371 new HIV diagnoses were made in immigrants [8]. The immigrant population constitutes a very diverse and particularly vulnerable group due to the socio-cultural context, the language barrier (in some cases), their economic level, and their employment and legal status in Spain. In addition, the WHO points out that insufficient data on STIs at the local level compromises the global response to this problem [4].

Periods of economic crisis represent one of the factors that increase the proportion of socially and economically vulnerable citizens, including immigrants [9]. Different studies have analysed the effects of the economic crisis on the immigrant population in several European countries [10–13]. The economic situation in Spain in recent years has increased the impact of communicable diseases, especially on the most vulnerable populations [14–16]; thus, hindering healthcare delivery to immigrants [17].

The present research is in line with these studies and aims to analyse STI-related differences between the pre-crisis and crisis periods by comparing the native population with the immigrant population, of those who visited the Sexually Transmitted Disease and Sexual Health Clinic in Granada (Spain). Population rates have been taken into account in this comparison, which we consider to be a differentiating feature of this study, with respect to previous studies.

The objectives of this study were as follows: to describe the evolution of STIs in the non-Spanish population in comparison to the native population living in Granada (Spain); to explore, in the former group, the potentially higher risk of contracting some of these infections in comparison to the rest of the population, using the period of economic crisis as a variable of interest.

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

An observational study was conducted by analysing the cases of service users diagnosed with STIs who visited the Sexually Transmitted Disease and Sexual Health Clinic in Granada, Spain, between 2000 and 2014, inclusive. This specialised centre, attached to the Andalusian Health Service (SAS, by its Spanish acronym), is the referral service for the entire province of Granada and, according to INE data, during the years analysed, a yearly average total of 550,000 individuals aged between 15 and 64 years old have used their services [18].

These records were taken from a randomly generated database within a larger project from which this study derives. A sample size was calculated to detect differences in the basic variables of STI presence in patients with a new clinical record. This calculation was made in order to detect differences in a binary variable, seeking to detect differences of 20% in two years, with a power of 80% and applying an error of α = 5% to the test. The sample size required to detect this difference was 97 clinical records per year. The sample was obtained from a database of new records for each year, from which the first and the last record number of each year were taken using systematic random sampling without replacement.

This database contains the clinical records of 1437 adult users without cognitive impairment whose reason for visit was suspicion of, or confirmed presence of, an STI. The sample analysed in the present study, which is a sub-sample of the aforementioned sample, consists of subjects who had been diagnosed with an STI.

Data were collected from the clinical records using four categories: symptoms; control; contact follow-up; and HIV. The country of birth was the independent variable (Spanish nationals vs. non-Spanish nationals), which was determined by means of an official identification document (i.e., national identity card, residence permit, work permit, or passport). Time was the main explanatory covariate (i.e., the 2000–2014 period), in which 2008 was considered to be the onset of the economic crisis in Spain, as numerous studies indicate [9,10,19]. Other variables included were: (a) socio-demographic variables: sex (male/female), age in years (analysed as a continuous variable), occupation (sex worker/former sex worker/other), employment status (employed/unemployed), level of education (no education or primary education/secondary education/higher education), living with a partner (yes/no), and sexual orientation identity (heterosexual/bisexual/homosexual); (b) variables related to clinical care received: reason for visit (according to the reasons provided in the clinical history), previous treatment (yes/no), number of subsequent visits, number of new subsequent episodes; and (c) risk factors for contracting STIs: regular partner having symptoms (yes/no), period of time since last sexual contact without a condom, number of partners in the last month, number of partners in the last year, lifelong sexual history (number of sexual partners throughout lifetime), drug use (yes/no), previous STIs, and age of first sexual intercourse in years (analysed as a continuous variable). The following variables, registered in the clinical records as nominal variables were transformed into quantitative variables for ease of analysis: period of time since last sexual contact without a condom, number of partners in the last month, number of partners in the last year, and lifelong sexual history.

To analyse the effect of the financial crisis on STI diagnoses, the annual rates of STI diagnosis per 100,000 inhabitants were calculated for Spanish nationals and non-Spanish nationals using the direct method, taking as the denominator the number of individuals over 15 years old residing in the province of Granada for each group, according to the data published by the INE in the continuous annual census [18]. These rates were plotted to highlight trends. Odds ratios (ORs) of STI diagnoses between non-Spanish nationals and Spanish nationals and 95% confidence intervals (CIs) were calculated for each year of the pre-crisis (2000–2007) and crisis (2008–2014) periods, as well as for the total study period (2000–2014).

Statistical analyses were conducted using the Statistical Package for the Social Sciences 22.0 (SPSS; International Business Machines Corporation [IBM], 2016, Armonk, NY, USA).

The study protocol was approved by the Biomedical Research Ethics Committee of the province of Granada (research protocol approved on 12 February, 2012 and 1 April, 2015), as well as by the Management Directorate of the Granada-Metropolitano Health District, to which the centre where data were collected is attached.

### **3. Results**

The total number of service users diagnosed with an STI was 441, of whom 329 were Spanish (74.6%) and 112 (25.3%) were immigrants. The mean age was 28.06 years (SD = 8.30; range = 16–70). Table 1 shows the characteristics of the sample separated by nationality (Spanish vs non-Spanish).


**Table 1.** Sample description of people diagnosed with a sexually transmitted infection (STI) by nationality (Spanish or non-Spanish) (*n* = 441).

SD: Standard Deviation; HIV: Human Immunodeficiency Virus. Period of time since last sexual contact without a condom: 1 = never, 2 = less than one month, 3 = one to six months, 4 = six to 12 months, 5 = more than 12 months; Number of partners in the last month: 1 = 0–1, 2 = 2, 3 = 3–5, 4 = more than 5; Number of partners in the last year: 1 = 0–1, 2 = 2, 3 = 3–5, 4 = 6–10, 5 = 11–20, 6 = more than 20; Lifelong sexual history: 1 = 0–10, 2 = 10–20, 3 = more than 20.

Table 2 shows the main diagnoses identified by population group, the most frequent being Human Papilloma Virus (HPV) infection. Non-Spanish nationals were disproportionately diagnosed with Gardnerella (16.1%) and syphilis (5.4%), whereas Spanish nationals were disproportionately diagnosed with Molluscum contagiosum (10.6%) and gonococcal infection (6.7%). The proportions observed in the case of HIV infection are similar in both groups. However, there is a greater proportion of Herpes simplex virus infection in Spanish nationals and a greater proportion of HBV infection in non-Spanish nationals.

**Table 2.** Distribution of STIs in non-Spanish and Spanish populations (*n* = 378). Granada, Spain, 2000–2014.


HPV: Human Papilloma Virus. HIV: Human Immunodeficiency Virus. HBV: Hepatitis B Virus.

The rates of STI diagnoses among the non-Spanish population residing in the province of Granada were higher than among the Spanish population in all the years analysed, with the exception of 2014 (2.05 vs. 5.52) (Figure 1).

**Figure 1.** Annual distribution of STI rates for Spanish nationals and non-Spanish nationals aged between 15 and 64 years old. Granada, 2000–2014. STIs: sexually transmitted infections. CI: confidence interval.

In both the pre-crisis (2000–2007) and crisis (2008–2014) periods, as analysed separately, as well as for the entire time period analysed as one, there was a higher risk of being diagnosed with an STI among immigrants than among Spanish nationals. However, the difference observed between the two populations was less pronounced during the crisis period (Table 3).


**Table 3.** Annual odds ratios for STIs in non-native populations versus native populations aged between 15 and 64 years old. Granada, Spain, 2000–2014.

OR: odds ratio. STIs: sexually transmitted infections. CI: confidence interval.

### **4. Discussion**

Regarding the socio-demographic profile of the sample analysed, it is worth noting that, in the immigrant population, a greater proportion of patients were women compared to in the Spanish population. However, the mean age of the sample was very similar in both population groups, being between 27 and 28 years old. There was a higher proportion of individuals with higher education in the native population than in the immigrant population. The high percentage of sex workers or former sex workers in the immigrant population analysed stands out. In reference to the number of partners in the previous month, in the previous year, and throughout their sexually active lives, the mean values of all three of these variables are higher in immigrants than in Spanish nationals.

The most frequently diagnosed STI, in both immigrant and native populations, was HPV infection. This is consistent with a study on STIs conducted in the same region, Andalusia [20], in which HPV infection was the most frequent infection as early as 2009. With respect to other diagnoses, such as hepatitis B and syphilis, also reported in other studies, the results are consistent with another study [21] that reports a higher prevalence of hepatitis B in the immigrant population compared to the native population. In contrast, the pattern is different for syphilis, with a higher prevalence in the native population than in the immigrant population.

Based on the results obtained from the analysis of the crisis and pre-crisis periods, it can be observed that the risk of being infected with an STI is greater in the non-Spanish population throughout the entire period analysed. In addition, the proportion of patients who were not Spanish nationals experienced a linear upward trend until 2012. In this year, the trend seemed to reverse, leading to a progressive reduction of the proportion of patients who were not Spanish nationals, compared to what it had been prior. As a result, the described profile points to a reducing effect on the risk for non-Spanish nationals relative to Spanish nationals during the crisis years. This trend can also be observed in the chart comparing the changes in STI rates. In the Spanish population, the rate of STIs remains relatively stable throughout the period studied. However, the rate of STIs in the non-Spanish population is higher and more erratic in the pre-crisis period than in the crisis period, during which this rate seems to stabilise at a level closer to that of the Spanish population.

There may be several explanations for the apparent drop in risk of STI diagnosis during the crisis period compared to the pre-crisis period, which are consistent with a potential underestimation of STI diagnosis in the immigrant population during the crisis years. For instance, the deteriorating social and working conditions following the onset of the crisis, which fundamentally affected the poorest and most vulnerable areas of society, including immigrants [22,23], may have caused many immigrants in Spain to return to their countries of origin [24] and slowed down the arrival of immigrants in Spain [25]. A previous study concluded that, between 2006 and 2012, the health status of the immigrants who had arrived in Spain prior to 2006 was worse in comparison to that of the native population. A possible explanation for this may be the loss of the healthy immigrant effect during the most severe impact of the economic crisis on immigrants [26]. In addition, the passage in 2012 of the Royal Decree-Law 16/2012, on urgent measures to guarantee the sustainability of the Spanish National Health Service and improve the quality and safety of its services [27], restricted access to health care for illegal immigrants. The fear of being diagnosed with an STI and its potential ramifications (e.g., losing their job and/or residence permit) stands in the way of carrying out diagnostic tests in the immigrant population, especially when access to treatment and healthcare is being restricted.

The effect of austerity policies on the reduction of preventive strategies should also be taken into account. The decrease in use of contraceptives since 2007, the absence of protective measures against STIs in one fifth of occasional or sporadic relationships, as well as the increasing incidence of syphilis, gonorrhoea, and HIV in certain groups of individuals, demonstrate that there is a need to place more emphasis on preventive strategies and to strengthen the commitments made by institutions concerning the most vulnerable areas and groups of individuals. The economic crisis weakens the educational and healthcare systems to the same extent that it weakens prevention and promotion measures relating to sexual health [9].

Finally, the representation of sex workers among the study population must also be considered. A previous study conducted in female sex workers in Spain shows that the prevalence of self-reported STIs experienced a significant increase between 2005 and 2011 (from 14% to 20.6%), pointing to inconsistent condom use as a factor worth considering [28]. In our study, a higher proportion of the immigrant population are sex workers compared to the native population, which may also explain the higher prevalence of STIs. However, further studies are needed to corroborate this association.

### *Limitations*

Caution must be exercised when interpreting the data obtained and generalising them to the immigrant population as a whole due to certain limitations of this study. For example, it should be taken into consideration that this study has been carried out in a specific geographical area. As a result, this study has a low degree of external validity. However, the close relationship that exists between population, culture, healthcare systems, and the use of healthcare services (such as STI-related services) justifies the need to perform field analyses such as this one. Additionally, given that this study has been conducted in a specific healthcare district, it provides varied and accurate local data on the composition of the population attached to this area, data that are not usually available from demographical sources.

The non-Spanish population analysed in this study is over-represented in comparison with the non-Spanish population officially resident in the province of Granada, which may indicate the large number of illegal immigrants that characterises international migration in Southern Europe. In Spain, official sources have estimated that, only during the period analysed, 195,458 undocumented individuals had arrived in Spain [29], with Granada being one of the most affected provinces [30].

Furthermore, despite having analysed a long period of time, the cross-sectional nature of this study does not allow causal associations to be established. As a result, the findings shown in this respect must be regarded as hypotheses to be tested with other more complex designs that facilitate the establishment of stronger causal relationships. There is an inconclusive result suggesting the need to further investigate the results for more conclusive outcomes in future studies.

### **5. Conclusions**

During the period 2000–2014, the risk of being infected with an STI is greater in the non-Spanish population throughout the entire period analysed. There was also a gradual decrease in the rate of STI diagnoses in the immigrant population from 2009 to the lowest level of the time series in 2014, which led to a lower risk of being diagnosed with an STI during the same period. The difference in risk observed between the two populations is less marked during the crisis period (2008–2014) compared to the non-crisis period (2000–2007). This may be attributed to an underreporting of diagnoses in the immigrant population during the crisis period.

Drawing on the epidemiological and social context, the findings of this study show a population profile (of the non-Spanish population), which is more vulnerable to STIs. It should be noted that, for example, the lack of STI diagnosis and treatment in illegal immigrants hinders STI control. This, in turn, may increase STI transmission likelihood and could result in a deterioration in the health of the affected group and, potentially, of the general population [8]. Public health policies must improve the control and treatment of existing cases, allocate more resources for the detection of unreported cases, and put in place more effective preventative measures at a lower cost.

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

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

**Acknowledgments:** The authors would like to thank all members of the STI clinic team in the province of Granada, as well as Nurses María Teresa Sánchez Ocón, Esperanza Cano Romero and María Visitación Mingorance Ruiz, for their help in collecting the data.

**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* **Understanding the Experiences and Needs of Migrant Women A**ff**ected by Female Genital Mutilation Using Maternity Services in Australia**

**Sabera Turkmani 1,\*, Caroline S. E. Homer <sup>2</sup> and Angela J. Dawson <sup>1</sup>**


Received: 5 February 2020; Accepted: 18 February 2020; Published: 26 February 2020

**Abstract:** Female genital mutilation (FGM) is a cultural practice defined as the partial or total removal of the external female genitalia for non-therapeutic reasons. Changing patterns of migration in Australia and other high-income countries has meant that maternity care providers and health systems are caring for more pregnant women affected by this practice. The aim of the study was to identify strategies to inform culturally safe and quality woman-centred maternity care for women affected by FGM who have migrated to Australia. An Appreciative Inquiry approach was used to engage women with FGM. We conducted 23 semi-structured interviews and three focus group discussions. There were four themes identified: (1) appreciating the best in their experiences; (2) achieving their dreams; (3) planning together; and (4) acting, modifying, improving and sustaining. Women could articulate their health and cultural needs, but they were not engaged in all aspects of their maternity care or considered active partners. Partnering and involving women in the design and delivery of their maternity care would improve quality care. A conceptual model, underpinned by women's cultural values and physical, emotional needs, is presented as a framework to guide maternity services.

**Keywords:** female genital mutilation (FGM); women's health needs; equality; quality of maternity care; midwifery continuity of care

### **1. Introduction**

Female genital mutilation (FGM) is defined as the partial or total removal of external female genitalia for non-therapeutic reasons [1]. This practice is deeply rooted in culture, with social or religious obligation and marriageability considered to be the most important reason for its continuation (UNICEF 2013). FGM is also performed for fear of being excluded from opportunities as a young woman [2]. FGM is traditionally practiced in 30 African and Middle Eastern countries, and some parts of Asia and South America [3]. Changing patterns of migration have led to an increase in the prevalence of women with FGM in many high-income countries [4,5].

It is estimated that globally over 200 million women and girls have undergone FGM and another three million women and girls are at risk annually [6]. There is a lack of reliable and high-quality data in relation to the numbers of women affected by FGM in high-income countries (HICs) [7]. It is, therefore, challenging for countries to develop effective policies, allocate relevant resources and evaluate the results of interventions [8,9]. In Australia, a recent report [10] estimated that in 2017 there were 53,000 women and girls affected by FGM in Australia, which represents a prevalence of 4 per 1000 girls and women.

Women affected by FGM in HICs are usually migrants or refugees and may have complex needs in addition to their clinical care [11,12]. These women are more likely to face socio-economic and cultural challenges due to language barriers, low education levels, and financial difficulties, which can hinder access to health services [13,14]. Migrant and refugee women from low- and middle-income countries, especially those from African countries, are reported to have poorer perinatal outcomes due to a higher rate of complications during pregnancy and childbirth [15,16]. FGM poses an additional burden to affected women and there are potential adverse consequences during pregnancy and childbirth such as an increased risk of caesarean section, post-partum haemorrhage, instrumental birth and prolonged labour [6].

Research has found that health services in some HICs may not be adequately prepared to provide quality care to FGM-affected women [17]. For example, many health professionals lack clinical skills and knowledge about the law in relation to FGM [17,18]. Health professionals have also been found to have a poor understanding of the cultural background of women affected by FGM and find communication challenging [19,20]. These issues, combined with inadequate support services, such as interpreting and counselling services, mean that many women may face difficulties expressing their needs [21].

The World Health Organization (WHO) highlights the importance of improving the quality of maternity care for women with complex needs to minimise further complications and harm [6]. The WHO's quality care standards outline eight domains of quality of care that encompass the provision of care and a woman's experience of care [22]. Quality of health services would improve if women trust and are confident to utilise the services on the basis of their positive and satisfactory experiences [23]. While research has provided insight into what constitutes quality maternal care from a health system perspective, gaps remain concerning the views and needs of women with FGM and what they regard as quality care. This study aimed to identify approaches to achieve culturally safe and high-quality woman-centred care for migrant women who have been personally affected by FGM.

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

The study employed Appreciative Inquiry (AI), a qualitative methodology to gain a deep insight into women's experiences of midwifery care in an Australian setting [24]. Appreciative Inquiry is a well-accepted methodology in health research [25]. This methodology has been used to explore patient experience of clinical healthcare [26] and to address the complex needs of families and children in primary health care [27]. McAdam and Mirza [28] used AI to describe the experience of marginalised youth engaged in drug and alcohol misuse and the implications of positive stories on health and social well-being.

We applied the four-stepped processes of AI to elicit examples of positive care interactions and envisage what best quality maternity care might look like in the future [29]. AI is open-ended, allowing a flexible approach to be taken depending on the needs of the participants [30]. The collaborative nature of AI is helpful in the development of confidence and can motivate participants to become actively involved in change [31]. We designed this research not only to identify the changes that are needed to improve maternity care but encourage women to become involved in this change as users and beneficiaries of maternity services.

Ethical considerations for this project were of particular importance as the migrant women with FGM often feel vulnerable, stigmatised and marginalised [32]. Ethical approval (UTS HREC REF NO. ETH17-1525) was obtained from the Human Research Ethics Committee of UTS in August 2017 before the recruitment or data collection process.

This study was conducted in Western Sydney, New South Wales (NSW), the area that has the largest number of non-English speaking women in this state of Australia [33]. The participants were English-speaking migrant and refugee women who were personally affected by FGM and lived in Sydney. The women had given birth in Australia in the last ten years or were currently pregnant. A written information package was given to women to invite them to participate. The research

was designed and conducted in direct consultation with experts in this area from government and non-governmental organizations, and an independent activist and advocate who is a survivor of FGM. In addition, a member of the community was involved throughout the study process and guided the development of the research tools, assisted with recruitment of the study population and ensured that the project was conducted in an ethical and culturally appropriate manner. Chain referral sampling was employed to approach potential women [34]. This method of sampling is useful for recruiting participants in research where the topic is sensitive or in populations that are stigmatised and hard to reach [35]. Participants signed the informed consent form prior to the commencement of interviews or group discussions. Participant anonymity was assured by allocating a code to the woman's name.

Data were gathered through in-depth interviews and followed by focus group discussions over five months (October 2017–February 2018). Interviews lasted for one hour and group discussions were conducted with five to eight women for two hours. The interviews and discussions were guided by questions (Supplementary Materials) following concepts of AI that were flexible enough to enable an exploration of ideas and experiences that women raised [36]. Interviews were held at a time and place convenient to women and group discussions were held in community centres. We offered small gift cards as compensation for their time and travel to the interview location.

Braun and Clarke's [37] approach to qualitative data analysis was followed because it offered a way of analysing the data according to the 4Ds representing the discovery, dream, design and deploy phases of AI in the first instance followed by a closer analysis of women's experiences as maternity service users.

The interview transcripts were first transcribed verbatim [38] enabling the researchers to familiarise themselves with the data tomake sense of it and reflect on overall meanings and general ideas [39]. Data were exported into the NVivo qualitative data management software to enable coding of the text according to the four phases of AI. Woman's narratives were coded into themes representing 4Ds and appropriate sub-themes [40]. The final step of data analysis involved the interpretation of data to draw recommendations for future maternity care policy and practice [39].

### **3. Results**

In total, 23 individual interviews and three focus group discussions were conducted. The women were from Sudan (*n* = 9), Somalia (*n* = 6), Sierra Leone (*n* = 3), Egypt (*n* = 2), Indonesia (*n* = 2), and Ethiopia (*n* = 1). The majority of women (*n* = 21) had undergone FGM when they were 0-10 years old. All of the women came to Australia as refugees except for four who entered the country on spousal visas or employment visas. English was the second language for all women (Table 1).

The main findings are presented under four themes in line with the 4D cycle of Appreciative Inquiry: Appreciating the positives in their maternity care (Discovering); Desiring the best in maternity services (Dreaming); Planning together for improved maternity services (Designing); Improving and sustaining (Developing/Deploying). The four themes and their associated sub-themes are further elaborated in the sections below (Figure 1).


**Table 1.** Demographic information.

**Figure 1.** Thematic data analysis based on 4Ds cycle (stages) of Appreciative Inquiry.

221

### *3.1. Appreciating and Discovering the Positives in Maternity Care (Discovering)*

Appreciating the positives in maternity care concentrated on women's description of events during their maternity care in Australia, and the strategies or approaches that they perceived to be useful, or inappropriate. For the most part, women were appreciative of, and satisfied with, the maternity care they received. This included being provided with respectful care, a feeling of having a safe service, receiving the required information, having access to skilled health care providers, and being able to have advance care planning and family support. Women frequently reported that "Maternity services are really good in Australia compared to where we came from".

Women felt that the maternity services are safe and technologically advanced in Australia and they expected their maternity care providers to have an appropriate level of knowledge about FGM, possess effective communication skills, be sensitive to their cultural needs and involve women in their care. For example:

*The good thing was always feeling safe, knowing there are all the facilities, medicines and machines and skills you might need available within the hospital. I really felt relaxed in both my deliveries. Overall pregnancy was a happy experience for me and I knew they would help me straightaway compared to my country where nothing is available.* (W18)

Most of the women appreciated maternity care providers who were helpful, sensitive and responsive to their needs, especially when they had no family and relatives around to support them. Women were impressed by the way maternity care providers made them feel cared for, particularly when they followed up to make sure women did not miss their appointments.

A few women thought that while caring for women with FGM was not a common experience for Australian maternity care providers, women expected providers to know how to deal with FGM and how to communicate with women. This woman said:

... *it is not like that the doctors and midwives in Australia come across a circumcised woman every day, you know. And I don't blame them if they are surprised or ask you millions of questions.* (W13)

Several women were worried about the care they might receive because they did not think maternity care providers were adequately prepared to manage FGM as this woman explained:

*The medical sta*ff *need to understand this issue [FGM] and be knowledgeable about it and if they don't have hands-on experience and skills please do not touch us and make our situation worse. You need to feel safe knowing that they get training before coming to women with FGM.* (W23)

The women acknowledged that developing trust with a maternity care provider was directly associated with the provider's competence. Some felt anxious and lost confidence in the ability of maternity care providers to deliver good care when they saw that their care providers were surprised or shocked when they encountered FGM. One woman in a focus group said:

*If these midwives and doctors know where to cut (de-infibulation), how to cut and when to cut it will be so helpful for us and for them because we will not have a problem and they will be relaxed and confident in what they do. Now, as soon as they see us they are shaking* ... *Oh my God. They can get advice from doctors and midwives who worked in our country and have real experience of treatment of women with FGM.* (FGD3)

### *3.2. Desiring the Best in Maternity Services (Dreaming)*

Women expressed their vision for the best maternity care in the future, including how they would wish to be treated within the healthcare system. They described the need for equality and for FGM-affected women to be treated the same way as other women. This included a desire for personalised care to be delivered by a provider from a similar cultural background with services

tailored to the needs of the individual woman. In practical terms, women described how individualised care should mean the provision of support services for women following de-infibulation.

Women believed that each pregnancy is an individual experience and expressed a clear understanding of the need for services to facilitate informed choice and shared decision making in a way that involves women with FGM in their own care. They wanted maternity care providers to listen to each woman and adjust care to suit her individual needs, rather than following the same course for every woman. One woman said:

*They need to listen to women as they know their body better. Not everything is going to be according to the recipe in the book. They have to look at each individual pregnancy separately.*

Women wanted to be treated in the same way as other women without being labelled as different while accessing Australian maternity services. They also wanted to have access to appropriate mental health support that took into consideration their special circumstances due to their FGM, for example:

*If a woman has undergone FGM they need to look after her even after birth and even if there is not any visible harm there is always a change and she needs that emotional support.* (W20)

Many women struggled with their body image and the emotional impact of de-infibulation, and some wished to see their bodies the way they were used to seeing it since childhood (infibulated). This was exacerbated by the fact that legislation in the state did not provide the option of re-infibulation. They perceived a reluctance of health staff to consider any form of reconstruction of the vulval or perineal area and attributed this to laws prohibiting re-infibulation. Women wanted reconstructive surgery to be part of the services offered to them. Most believed that their de-infibulation had been done 'badly' and their body would be 'in better shape' if they were re-infibulated after birth. They desired varying degrees of re-infibulation and used the term 'closed-back' when describing reconstructive surgery. This comment captures such feelings:

*After they open you during delivery I wish there is someone who stitches it very very nicely so it doesn't look very open.* (FGD1)

Most women felt embarrassed and uncomfortable with their bodies and described their vulva as 'ugly', 'too open', 'not in good shape', 'hanging skin', and 'horrible'. Some women chose to undergo a caesarean section to avoid de-infibulation or they travelled back to their home country to be re-infibulated as this woman did:

... *I went overseas and closed it by a midwife in my country. You know last time I [got] closed myself in Sudan it was because it was so big and ugly they left me totally open at least they could have stitched me back to make me look like normal.* (W22)

Women recognised and valued their capacity to experience birth as a normal process without unnecessary intervention. Some wished that their FGM was not considered as a barrier to undergoing normal labour and birth and questioned interventions such as caesarean section. Several women stated, 'We had our baby normally and easier in our country; why not here'.

Some women felt they were vulnerable, disempowered and dominated by maternity care providers and these providers took control of the situation. Some agreed to let their family members make decisions on their behalf, while others expressed their strong desire to be involved in a collaborative way with maternity care providers. Culture, personal attitudes, and emergencies were also identified by most women as factors influencing the degree to which they could be involved in decision making. For example:

*My husband and mother in law made the decision for me. If it was up to me I would have chosen a caesar straightaway. I did not want all that pain and trauma, but midwife went with my husband and mother in law's decision without listening to me.* (FGD1)

Many women perceived that their lack of health literacy and knowledge about access to certain options or health services led to their exclusion from decision making. One woman explained:

*Sometimes you are in a position where you have to follow whatever they say. Maybe because our knowledge is limited and the language also is a big, big problem.* (W18)

### *3.3. Planning Together for Improved Maternity Services (Designing)*

The 'design' phase of AI invited women, individually or as part of a group, to develop a plan for what they need to achieve their dream for quality care.

In designing future maternity services, women discussed the need for education initiatives that enabled maternity care providers to provide emotional support, promote cultural safety and communicate in ways that are appropriate for supporting women with FGM. The need for training to involve women themselves to improve provider understanding of and familiarity with the cultural beliefs behind the practice of FGM was highlighted, as these women explained:

*If I am a midwife I make you feel good and I need to understand what you believe in so I can understand if you see FGM as a good thing or bad thing. Then I can talk to you and guide you accordingly* ... *first you need to get a sense of what women believe in, otherwise they may not disclose anything.* (W12)

Women also noted that, while maternity care providers need to be respectful and integrate the cultural aspects into their services, they also need to be mindful of harmful cultural practices that may place women at risk. For example, in some cultures, women do not use direct communication to explain their problems related to childbirth, maternity care or FGM. Many women mentioned that they avoided disclosing their FGM as they thought this was culturally inappropriate, as explained here:

*I was shy and hide my FGM until birth and I am sure many other would do that. In our culture women won't talk about it believe me or not. There is shame and stigma with those topics'.* (W23)

Midwifery continuity of care was one of the models of care or services most appreciated by women who received it. Women understood midwifery continuity of care as being cared for by a known midwife over the entire period of pregnancy and childbirth and after birth. Being with the same midwife and building a relationship based on mutual trust and understanding was perceived to improve women's sense of safety and confidence and increase their involvement in their care. Most women, however, did not have access to this model of care. There were a few women who received midwifery continuity of care during pregnancy, but during labour and birth, their known midwives were not present. They expressed feelings of anxiety and distress with being cared for during labour and birth by midwives they had not met before. Women suggested that maternity services should be designed to enable all women to have access to such a model of care, for example:

*It is very important for women because we want to trust someone and by changing midwives and doctors we will be lost. I will also develop my confidence in her competence and make sure she can manage my birth and I am in safe hands. That's a huge support for me knowing that I am safe and someone knows my issues and concerns.* (W17)

Women viewed high-quality maternity care in terms of the way that maternity care providers had behaved towards them. Considering the sensitivity of a topic such as FGM, the women believed positive and effective communication was a key component of maintaining a sense of connection, trust, and collaboration with health providers. For example, they wanted to be heard, touched and welcomed. Many women indicated that building trust happened over time as they got to know their maternity care providers through their direct interactions. As this woman explains, this was especially important in addressing the embarrassment that many women felt because of their FGM:

*You know little by little each time after I started to visit the doctors and midwives and they didn't make me feel embarrassed [because of FGM] and they asked me so many questions when I went to them. And the way they talked to me was so good. You know, you feel so good when someone listens to you. They were not in a rush to get to the next patient and kick me out of their o*ffi*ce. They spend time with you and do what they need to do while they kept privacy.* (W13)

Women wanted to receive emotional support to address their trauma including dealing with health issues related to FGM.

*Sometimes you just want someone to talk to and ask for nothing else, just someone to ask you what your feelings after birth are or how you are because it is a hard time.* ... *I want a midwife or nurse to provide care for me beyond giving medicines, I want them to talk to me and support me emotionally and mentally.* (W17)

Many women felt that there was no transparent, clear, and mutual communication between them and their maternity care providers. As a result, women were often suspicious of the maternity care services they received and were not always willing to accept advice from maternity providers as illustrated with the following quote:

*Sometimes they don't even talk about FGM with us and just write everything down and say all is good without giving us the details. I think it is mostly because they don't know anything about FGM and they just look at you and they have no idea.* (W17)

### *3.4. Improving and Sustaining Maternity Services (Developing*/*Deploying)*

The final theme reflected strategies that women regarded as useful to support their plans to improve maternity services. The women's suggestions represented three levels of action: mobilising and enabling communities, strengthening maternity care systems and increasing government support.

Women believed that communities need to be mobilised to create a supportive environment in which pregnant women and new mothers affected by FGM can feel safe and healthy. Advocacy and campaigns for policy, professional practice, and at a community level were considered critical in creating a supportive environment to improve health outcomes for women in the long term. Raising community awareness, through formal and informal education, campaigns in the community and schools, and involving women, men and young people, were considered essential to delivering positive change. One woman said:

*Still many people in the community believe it is a good thing to do on their daughters [FGM].* ... *I will not let my daughter to undergo FGM but we need to remove pressure of the community on families. If no one wants a girl without FGM then everybody forced to do it. We need to end that by educating community and change this culture.* (W22)

Women believed the practice of FGM was continuing in their communities, even in Australia, and emphasised the need for a reporting system at the community level. Women stated that the success of community-based interventions, such as education and media campaigns, depends upon the involvement of all members of the community including religious and community leaders in the planning and implementation processes of change. Women emphasised the central role of families in bringing a sustainable change to stop a culture such as FGM as explained here:

*Change is dependent on families. In my family, I have already talked to my kids about the stu*ff *like FGM and the even bigger impact of it on society. I think that's how we will spread the word and stop it, otherwise it is never going to be stopped. Now people believe in this society that talking about this issue is wrong or Haram [prohibited by religion]. I don't care; I will talk to my children because I don't want them to grow up blindly.* (W18)

Men were regarded as important actors of change, but women thought that they lacked knowledge about the physical and mental health consequence of FGM. Women felt that men believe that FGM is women's business and that their views on cultural obligation enabled the continuation of FGM. The women, therefore, perceived that men's involvement as a crucial part of the solution to end this practice but it might be very challenging as men are not interested in taking part in such a movement. An example such as this was given:

*At the moment most of the trainings are for women. We need men to talk to men so we can engage them otherwise you cannot force them to sit in a class. You need to train more men to open up and talk about this issue with other men in the community and engage them at the same level as women. Men are still looking at it as a good thing.* (W15)

Women described feeling empowered when they shared their stories and regarded these as an important resource for mutual support and to educate the community and challenge cultural beliefs about FGM. Women also mentioned that they feared being socially ostracised by their families and communities if they expressed dissenting views. This woman explained:

... *We need to create an environment where people talk about it. You know it is very hard to disclose such issues at community level, as it is a very private matter. I guess if we bring up stories and how women are su*ff*ering this would be e*ff*ective to change this culture in the future. Imagine you're living for someone else's pleasure and you're getting none.* (W13)

Women considered government support as a cross-cutting issue linked to all future actions and approaches. Women used the word government to mean all high-level decisions, policy and funding at local, state and territory and federal levels. They wanted resources for improving the health of affected women, introducing FGM as a topic in the school curriculum and making meaningful linkages with communities. They believed such strategies would ultimately lead to the improvement of the health of women with FGM and society as a whole.

*They [policy makers] need to identify women with FGM as a priority at policy level and provide them with things they want. We want services which all women deserve* ... *. We are in a developed country and we should have access to standard care from an experienced health provider.* (W21)

Women also spoke of the need for mental health support and counselling services, both at facility and community level, for example:

*Make sure they [women a*ff*ected with FGM] are OK, mentally and physically. Do the follow up afterwards. Education and individualised support not only for women who have undergone FGM but also to train sta*ff *and the community. It goes both ways.* (W16)

Women pointed to cultural taboos that make it challenging to have open discussions about FGM with male members of the family. Several women made suggestions like this:

*Facilitating and funding community training such as workshops for men and women we can raise the awareness. It is also helpful to open the discussion around this issue. At the moment it is not culturally appropriate to even talk about it even in the family.* (W20)

### **4. Discussion**

This research identified the maternity care experiences of women affected by FGM and their views concerning the care they wished to receive in the future and how this might be achieved. Women in this study acknowledged that the maternity care they received had not always been at the level of quality that they desired or had expected. Women reported that being meaningfully involved in their care design and delivery was a crucial strategy for building trust and improving and validating

the quality of maternity services. It has previously been shown that women who are well educated and have adequate information about FGM are more likely to have control over health care, access to shared decision making. Making an informed choice is key to respectful care for women with FGM [41] and they are less likely to perform FGM for their daughter [42]. While most women were motivated to be involved in their care, they struggled with poor communication and a lack of information tailored to their individual needs as reported elsewhere [43,44]. Women wanted to be cared for by skilled and culturally competent providers who treated them as 'special' but also as normal and equal to 'other women'. This has been described by other research where they ensured equality by including Aboriginal and Torres Strait Islander midwives, who can interact holistically and provide culturally sensitive services [45]. Finally, women described the importance of having access to evidence-based models of care such as midwifery continuity of care and available services including, reconstructive surgery, management of trauma, emotional support, psychotherapy services and cultural support.

A conceptual framework (Figure 2) was developed based on the findings of this research that highlights four priority approaches required to achieve quality care for women with FGM: co-production, woman-centred care, equity and equality and evidence-based models of care. These approaches are underpinned by four strategies that facilitate women's engagement and include involvement in developing health information to being an equal partner in decision making and the co-design of maternity services.


**Figure 2.** Conceptual model of quality improvement within maternity services for women with female genital mutilation (FGM).

### *4.1. Co-Design of Health Literacy Interventions*

Women regularly described the need for information that is tailored to their individual needs and noted that support services, such as counselling, were not always accessible due to language and cultural barriers. Some women stated that these services were not available or integrated into maternity care. This is similar to other studies in high-income countries that have found that women affected by FGM do not always receive or understand the information and resources they required or needed because of social isolation, stigma and a lack of health literacy [46]. When women have lower levels of health literacy, they are less prepared to engage and comply with their care regimes or protocols and as a result, do not receive optimum care [47]. Improving the health literacy of women with FGM may change the attitudes of women towards their own FGM and reduce the likelihood of their daughter's being circumcised [48].

Every woman should feel empowered to build her capacity and skills to use health information effectively and make an informed choice [49]. Many women in our study stated that they were not adequately engaged in their health care because of low levels of health literacy, inadequate information and unfamiliarity with their health rights. Again, these findings concur with other studies [50,51] and confirm that women's participation in the process of health information design leads to more satisfying and positive experiences with enhanced health outcomes [52]. Health literacy programs that involve women designing and delivering programs not only build the capacity of women to facilitate the sharing of stories and experiences but also empowers women to support others in their community [53]. Such approaches are likely to be useful for women affected by FGM.

### *4.2. Co-Design of Evidence-Based Models of Care*

Most women in our study reported different types of FGM-related trauma, which affected their overall quality of life. Women expected health care providers to be responsive to their psychological, emotional and socio-cultural needs as found in other studies [54]. The central philosophy that underpins high-quality maternity care does not only involve a focus on physical health but also emotional well-being and includes quality of life issues [55]. Despite the emotional and mental consequences of FGM, most studies are focused on the physical aspects and implications [56,57]. Laio et al. [58] indicated that women affected by FGM are often silent about their emotional problems due to the stigma associated with FGM and have difficulty communicating with health providers. It is difficult for care providers to recognise or determine the level of psychological trauma that may be caused by FGM, but our study highlights the importance of these considerations.

FGM related trauma is important to note because it can negatively impact on childbirth and sexual relationships highlighting the need for individualised trauma-informed interventions for such vulnerable women. FGM related mental health issues such as PTSD, negative body image and feelings of shame and stigma may also affect women's health-seeking behaviour [59]. A trauma-informed model of care may be an approach to providing safe supportive care to women who have been affected by violence to reduce the consequences of trauma in their life [60].

Women should also be involved in the design of such trauma-informed services so that individual needs, views and experiences can be addressed in a collaborative way [61]. Efforts in the area of trauma-informed care currently focus on strengthening health provider's knowledge and skills based on their interactions with consumers, rather than understanding a women's experiences and needs [62,63]. Implementing participatory interventions, however, requires both the health system and community change [64]. Women need to be supported to become empowered to recognise their potential and utilise their capacity in the design and delivery of services [65]. Creating an environment of collaboration and mutual trust by engaging women and acknowledging their values and lived experiences may ensure that women's needs are understood and their views and culture are taken into account in service design, thereby, improving the quality of culturally safe care.

Correa-Velez and Ryan [66] emphasise the need for specific models of maternity care for marginalised and high-risk women, such as women with FGM, that encompass continuity of care plus educational interventions and the delivery of mental health support. Our study indicates that continuity of care can lead to improved interpersonal communication and can boost women's confidence, the collaboration between a woman and her provider and help facilitate women engagement in the process of care design and delivery [67]. Midwifery continuity of care enables health providers to consider the socio-cultural and emotional needs of marginalised women and, therefore, empowers women to achieve positive outcomes [68]. Such care models ensure the continuous assessment and evaluation of women's experiences, opinions and views that can improve the quality of care for marginalised groups [69].

### *4.3. Co-Design Approaches to Shared Decision Making*

The health system must offer women adequate support to enable them to be empowered to communicate, to ask for help and to question their care [60,70]. Patient participation in the process of service design and delivery is often missing as patients are perceived not to have adequate medical and clinical knowledge [71]. A review of the literature found that consumer involvement in the training of health providers ensures that the health system reflects their needs and desires in the design and delivery of services [72]. Collaborative partnerships have been found to have a positive impact on nursing practice by improving communication and shared decision making [73]. Another example from the field of mental health demonstrates the benefits of sharing the experiences and insights of patients through story-telling and using different aspects of personal experience in the development of a mental health assessment tool [74]. There is limited evidence in maternal health research and further research is needed to determine the best approach to engage women and evaluate the impact of their involvement in the co-design of education and training material, guidelines and health service processes.

### *4.4. Co-Design of Health Professional Education and Training*

Women in our study described the need for health providers to receive special training on the cultural aspects of care for women from diverse backgrounds. This would help to address their need for a model of maternity care that integrates a woman's cultural and individual values with excellent communication and referral paths to promote their well-being and safety as described in other research [75]. The involvement of women in teaching health professionals may be a useful strategy to increase the knowledge of clinicians. One study that investigated the outcomes of learning where consumers delivered classes found that nursing students improved their cultural knowledge and understanding of empathic care [76]. The involvement of mental health consumers in the education of nurses also showed improvements in nurses' communication skills and decreased cultural barriers for consumers as well as reduced discrimination [77]. The integration of cultural safety in practice is challenging as it requires the involvement of service users in the co-design of such services and involving a vulnerable population requires a paradigm shift in power differences between service users and health professionals [51]. Future health services need to be co-produced with women to disrupt the inherent power imbalances.

This study is one of the first of its kind in Australia to analyse this group of women's views and experiences of their maternity care. The use of AI as the methodology was unique and enabled women to focus on their positive experiences and come up with solutions for future action and changes within the health system. The study has highlighted the voices of women providing important knowledge to improve the quality of maternity care for marginalised women.

This study included only women who lived in Sydney, which is generally well resourced in terms of services for migrant populations. Therefore, the results may not be generalisable to the other states across Australia and suggested solutions and recommendations might be specific to the local context.

Sampling bias is a possible limitation. Potential women were recruited through chain referral sampling. Therefore, those who decided to participate in this study might be those who had more interest in this subject area, and this might have led the discussion either more positively or negatively.

### **5. Conclusions**

The engagement of individuals and communities is critical to the process of improving the quality of maternity health services and to address the socio-cultural needs of women affected by FGM. Empowering women and raising their awareness of their health care rights can help to engage women as active partners in the design and delivery of health information, models of care approach to shared decision making and health professional education and training which is based on their needs and context.

Further research is needed to explore the replicability of the suggested framework at policy and practice levels. Research is required to establish the feasibility of the co-production of maternity services and how this improves the quality of care and equitable health outcomes for women affected by FGM.

### **Supplementary Materials:** The following are available online at http://www.mdpi.com/1660-4601/17/5/1491/s1.

**Author Contributions:** Conceptualization, S.T., A.J.D. and C.S.E.H.; methodology, S.T. and A.J.D.; software, S.T.; validation, S.T., A.J.D. and C.S.E.H.; formal analysis, S.T.; investigation, S.T.; resources, A.J.D. and C.S.E.H.; data curation, S.T., A.J.D. and C.S.E.H.; writing—original draft preparation, S.T.; writing—review and editing, A.J.D. and C.S.E.H.; supervision, A.J.D. and C.S.E.H.; project administration, S.T. All authors have read and agreed to the published version of the manuscript.

**Acknowledgments:** This research was supported by the Australian Government Research Training Program (RTP). The authors are thankful for the support provided by the Faculty of Health, University of Technology Sydney (UTS). We also thank women who generously shared their perspectives and ideas and this study would not have been possible without them.

**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*

### *Review* **Prevalence of Intimate Partner Violence in Pregnancy: An Umbrella Review**

**Rosario M. Román-Gálvez 1,2, Sandra Martín-Peláez 3,4,\*, Juan Miguel Martínez-Galiano 5,6, Khalid Saeed Khan 3,6 and Aurora Bueno-Cavanillas 3,4,6**


Received: 1 December 2020; Accepted: 13 January 2021; Published: 15 January 2021

**Abstract:** Background: Intimate partner violence (IPV) is a public health concern, especially during pregnancy, and needs to be urgently addressed. In order to establish effective actions for the prevention of IPV during pregnancy, authorities must be aware of the real burden of IPV. This review aimed to summarize the existing evidence about IPV prevalence during pregnancy worldwide. Methods: A review of reviews was carried out. All published systematic reviews and meta-analyses published until October 2020 were identified through PubMed, Scopus, and Web of Science. The main outcome was the IPV prevalence during pregnancy. Results: A total of 12 systematic reviews were included in the review, 5 of them including meta-analysis. The quality of the reviews was variable. Physical IPV during pregnancy showed a wide range (1.6–78%), as did psychological IPV (1.8–67.4%). Conclusions: Available data about IPV prevalence during pregnancy were of low quality and showed high figures for physical and psychological IPV. The existing evidence syntheses do not capture the totality of the worldwide disease burden of IPV in pregnancy.

**Keywords:** intimate partner violence; pregnancy; prevalence; umbrella review

### **1. Introduction**

Intimate partner violence (IPV), defined as physical violence, sexual violence, harassment, and psychological assault (including coercive tactics) by a current or former intimate partner [1], is a public health concern that urgently needs to be addressed. During pregnancy, the woman experiences a situation of special dependence, both physical and emotional. In this period, exposure to violence affects not only the mother but also the fetus, which is at greater risk than in other stages of life [2]. In fact, IPV has been associated with adverse pregnancy outcomes including increased risk of human immunodeficiency virus infection [3], perinatal depression [4], insufficient weight gain during pregnancy [5], uterine rupture, hemorrhage, maternal death [6], prematurity, low birth weight, newborns small for gestational age [7], stillbirth [8], and reduced levels of breastfeeding [9]. At the same time, routine contacts with the health system offer

an excellent detection window to identify it and establish protective measures. Despite this, IPV during pregnancy is a neglected condition, even though it is more common than many maternal health conditions like preeclampsia and gestational diabetes [10].

IPV during pregnancy should be an avoidable global public health problem. However, in order to establish effective actions for the prevention of IPV during pregnancy, such as the performance of systematic screening and diagnosis of IPV in the antenatal visits, authorities must be aware of the real worldwide burden of IPV. However, information about IPV prevalence is not consistent. Whereas some studies indicate higher prevalence of IPV during pregnancy than before [11] or after [12,13] the pregnancy, other studies report a smaller prevalence [14,15]. Furthermore, the prevalence of IPV during pregnancy is reported to vary depending on the definition used [1], the screening strategy [16,17], and the development status of the population studied [10,18]. These factors make comparison between individually reported rates difficult.

This review aims to summarize the existing evidence about IPV prevalence during pregnancy worldwide through a synthesis of systematic reviews and meta-analysis. Prevalence studies provide a snapshot of a situation in a specific context, so it is important to bring together different existing studies for a global understanding. This work analyzes the existing reviews, identifying their strengths and limitations and laying the foundations for future reviews that clarify the situation of IPV during pregnancy in a complete and realistic way.

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

This umbrella review was written according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for systematic reviews and meta-analyses [19] and the Aromataris' guidelines for performing umbrella reviews [20].

### *2.1. Inclusion and Exclusion Criteria*

Systematic reviews and meta-analyses including observational studies reporting IVP prevalence suffered by women during pregnancy were considered in this review. The types of IPV were classified as physical, sexual, psychological and any type of IPV. Studies not following a systematic review approach, narrative reviews, and primary studies were excluded. No language restrictions were applied in this review.

### *2.2. Literature Search and Selection of Studies*

Relevant systematic reviews and meta-analyses according to inclusion criteria were identified through systematic searches of the following electronic databases: PubMed, Web of Science and Scopus, CYNAHL, PsycINFO, Social Science Database, and Sociological Abstracts. The full search strings can be found in Table 1. When the search engine used only allowed selecting systematic reviews or meta-analyses, the terms "systematic reviews" OR "meta-analyses" were not included in the search string; otherwise, those terms were included. Studies published from inception until October 2020 were included. Reference lists of identified studies were checked.


**Table 1.** Search strings.

### *2.3. Data Collection and Analysis*

Eligible studies were selected through a multistep approach (elimination of duplicates, title reading, abstract and full-text assessment). Two researchers (S.M.-P. and R.M.R.-G.) independently examined titles and abstracts, evaluating afterwards full texts according to the inclusion criteria described above. Any disagreement between the reviewers was resolved by means of a consensus session with a third reviewer (A.B.-C.). In case of ambiguity in reporting or lack of data, primary authors were contacted for clarification.

### *2.4. Data Extraction and Management*

Data were independently extracted by two researchers (S.M.-P. and R.M.R.-G.), and the following information was considered for each article: (1) first author and year of publication; (2) interval of time covered by the review; (3) countries (of studies included in the systematic review or meta-analysis); (4) number of studies included; (5) study design (of studies included in the systematic review or meta-analysis); (6) sample characteristics; (7) IPV as main outcome; (8) type of IPV investigated; (9) meta-analysis performance; (10) IPV outcome.

### *2.5. Quality Assessment Tools*

The updated AMSTAR 2 version for systematic reviews and meta-analyses was used to evaluate the methodological quality and risk of bias of studies included in the systematic review [21]. The overall final rating of each systematic review was judged as high, moderate, low, or critically low. In case of disagreements, a consensus session with the third reviewer (A.B.-C.) was held.

### **3. Results**

The electronic search initially resulted in 199 citations. A total of 80 studies were excluded after elimination of duplicates. From the 119 remaining, 61 were excluded after title and abstract screening and 58 full-text articles were selected and read. From those, a total of 12 systematic reviews were included in this umbrella review, of which 5 were meta-analyses. The reasons for exclusion were the lack of data about IPV prevalence during pregnancy (20), the use of violence other than IPV or the indistinct report of IPV or domestic violence (10), investigations on populations with a specific risk, supposedly different from the general population (12), and not being a systematic review or meta-analysis (4). The list of the excluded articles is presented in Supplementary Materials Table S1. Figure 1 shows the PRISMA flowchart and the study selection process.

Characteristics of included systematic reviews and meta-analyses are shown in Table 2. Only two reviews included global data [22,23], most of which were limited to a country or a group of countries, mainly from Asia [4,23–28] and Africa [4,26,27,29,30], followed by America [31,32], Europe [32], and Australia [32]. The number of studies included in the reviews giving information about IPV prevalence during pregnancy ranged from 2 [24] to 73 [23].

Most of the studies included in the selected reviews were cross-over studies, in which there was only a single evaluation of the women sometime during pregnancy. Some of the reviews also included cohort studies [4,24,25,30–32]. Reviews included studies giving information about IPV only at pregnancy [22–24, 27,29,30], both during pregnancy or at postpartum [4,30–32], during pregnancy, or having a child 2 years old or younger [25] or at current pregnancy or any pregnancy [26].

Wide differences were also observed regarding the type of IPV violence investigated. Of the selected reviews, nine investigated prevalence of physical violence [4,22,24–28,30,31], nine psychological violence [4, 22,24–28,30,31], ten sexual violence [4,22–28,30,31], and three any type of violence [26,29,32]. From the selected studies, four did not report IPV pregnancy during pregnancy as main outcome [24–26,32]. Five studies showed a summarized estimate of IPV during pregnancy [23,24,28–30]. In most of the reviews ranges of IPV prevalence are given [4,22,24,26,27,31,32].

**Figure 1.** PRISMA flowchart of the study selection process.

From the reviews included in the study, many showed data about any IPV prevalence worldwide. The data about prevalence of any kind of IPV during pregnancy were obtained from different countries [26,29,32].

The highest range of any IPV prevalence was obtained in Portugal, USA, and Australia [32] (15.4–40%), followed by Ethiopia [29] (26.1% (95% CI: 20–32.3)) and countries from the Arab League [26] (40.9–44.1%).

Regarding physical IPV during pregnancy, China [28] and Vietnam [25] showed the lowest ranges; (3.6% (95% CI: 1.6–6.2%)) and (3–8.5%) respectively. Higher ranges were found in countries from Latin America [31] (2.5–38.7%), Africa and Asia [4] (2–35%) followed by low- and middle-income countries [27] (5–52.8%), countries from the Arab League [26] (10.4–34.6%) and African countries [30] (22.5–40%), being the widest range the one found in Saudi Arabia [24] (21–78%). James and colleagues [22] showed a prevalence of 13.8% in all over the world.


**2.**CharacteristicsofthestudiesselectedfortheumbrellareviewofreviewsofworldwideprevalenceofIPVin




**Table 2.** *Cont*.

242

NR, not reported; Ph, Physical violence; Ps, Psychological violence; S, Sexual violence. \* Number of studies included in the review related to IPV during pregnancy, when in the review

reports IPV during pregnancy and others.

#### *IJERPH* **2021** , *18*, 707

As for any and physical IPV, China [28] showed the lowest and smallest ranges of psychological IPV prevalence during pregnancy (4.2% (95% CI: 1.8–7.5%)). Higher ranges were found in Vietnam [25] (6–32.5%) and countries from the Arab League [26] (23.4–32.6%), Latin America [31] (13–44%), and African countries [30] (24.8–49%). The widest ranges were found in low- and middle-income countries [27] (17–67.4%) and countries from Asia and Africa [4] (22–65%). James and colleagues [22] showed a prevalence of 28.4% throughout the world.

In general, sexual violence showed lower ranges of prevalence than the other types of IPV violence during pregnancy, being the lowest in China [28] (1.3% (95% CI: 0.6–2.5%)), followed by Vietnam [25] (3.4–10%), countries from the Arab League [26] (5.7–15.0%), low- and middle-income countries [27] (2.8–21%), Africa [30] (2.7–26.5%) and Latin America [31] (3–34.4%). The highest and widest ranges of sexual IPV prevalence during pregnancy were found in countries from Asia and Africa in the study of Halim and colleagues [4] (9–40%). Worldwide, prevalence of sexual IPV during pregnancy remained lower than 18% [22,23].

Quality assessment is reported in Table 3. Although all of the studies used a comprehensive literature search strategy, only two of the selected reviews did not include the components of PICO in their research questions and inclusion criteria [22,24], the reviews described the included studies in adequate detail, with the exception of three studies [22,26,32], and in only three of the reviews [22,25,27], authors did not report any statement about potential sources of conflict of interest. For some other aspects of the AMSTAR2 checklist, the quality remained low. Thus, none of the reviews included an explicit statement that the review methods were established prior to the conduct of the review nor the sources of funding for the studies included. Only one review provided a list of excluded studies and justified the exclusions [28]. Only three of the reviews explained their selection of the study design for inclusion in the review [24–26]. Half of the reviews performed study selection in duplicate [4,23–26,28], whereas only three did not perform data extraction in duplicate [4,22,31]. Half of the studies included did not use a satisfactory technique for assessing the risk of bias in individuals included in the review [22,23,27,29,31,32], and more than half did not account for risk of bias in individual studies when interpreting/discussing the results of the review [22–25,31–33]. Only two of the reviews provided a satisfactory explanation for, and discussion of, any heterogeneity observed in the results of the review [22,26]. In the reviews where meta-analysis was performed, the authors used appropriate methods for statistical combination of results [23,25,28–30], but in only three of them, the review authors assessed the potential impact of risk of bias in individual studies [25,28,30] and only two [28,29] carried out an adequate investigation of publication bias.


**Table 3.** Evaluation of selected IPV during pregnancy reviews based on AMSTAR 2 guidelines.

\* Each number corresponds with an AMSTAR 2 checklist item as follows: 1. Did the research questions and inclusion criteria for the review include the components of PICO? 2. Did the report of the review contain an explicit statement that the review methods were established prior to the conduct of the review and did the report justify any significant deviations from the protocol? 3. Did the review authors explain their selection of the study designs for inclusion in the review? 4. Did the review authors use a comprehensive literature search strategy? 5. Did the review authors perform study selection in duplicate? 6. Did the review authors perform data extraction in duplicate? 7. Did the review authors provide a list of excluded studies and justify the exclusions? 8. Did the review authors describe the included studies in adequate detail? 9. Did the review authors use a satisfactory technique for assessing the risk of bias (RoB) in individual studies that were included in the review? 9. Did the review authors report on the sources of funding for the studies included in the review? 10. If meta-analysis was performed did the review authors use appropriate methods for statistical combination of results? 11. If meta-analysis was performed, did the review authors assess the potential impact of RoB in individual studies on the results of the meta-analysis or other evidence synthesis? 12. Did the review authors account for RoB in individual studies when interpreting/discussing the results of the review? 13. Did the review authors provide a satisfactory explanation for, and discussion of, any heterogeneity observed in the results of the review? 14. If they performed quantitative synthesis did the review authors carry out an adequate investigation of publication bias (small study bias) and discuss its likely impact on the results of the review? 15. Did the review authors report any potential sources of conflict of interest, including any funding they received for conducting the review? NA, Not applicable.

### **4. Discussion**

The aim of this umbrella review was to provide a summary of the evidence currently available on global IPV prevalence in women during pregnancy. Despite the fact that the selected reviews were recent, they are of low quality as assessed against most of the AMSTAR2 recommendations. There were only two reviews giving worldwide IPV prevalence during pregnancy [22,23], both of them complying with less than half of the AMSTAR2 criteria.

### *4.1. Limitations*

We selected systematic reviews for prevalence of IPV in pregnancy, yet we obtained very diverse data. In the reviews, sometimes the concepts of IPV and domestic violence were mixed together. Most of the included studies were cross-sectional self-report surveys, which may have been associated with inaccurate recall [24,28,31]. They did not always specify the gestational time point of IPV evaluation. It was common to find a mix among studies assessing IPV at any time of pregnancy or even after pregnancy. This is important since data can vary depending on the gestational age when IPV is measured, antenatally or after delivery. One review [4] also included studies that assessed violence for a period that is inclusive of, but not exclusive to, pregnancy.

Some of the included reviews, in spite of being systematic reviews, showed possible bias in studies included for evidence synthesis. They generally failed to adequately address the heterogeneity of results [22,28,32]. Others had a very narrow geographical coverage [24–26,29]. In addition, sample sizes of the included studies were generally small [29–31], and the use of standardized and validated IPV instruments was low. Geographical coverage of the reviews selected was mainly focused on low-income countries, a fact that invites readers to infer that IPV is a problem exclusive of those countries, which is far from the reality [34].

The main strength is that we have conducted an umbrella review following up the PRISMA and Aromataris' guidelines. Our search has been exhaustive, collecting all kind of IPV.

### *4.2. Implications*

Whereas the consequences of IPV during pregnancy on the mother and on the newborn are widely known [3–9], the frequency and types of IPV in that period are not fully characterized. WHO recommendations on antenatal care for a positive pregnancy experience advise considering clinical inquiry about the possibility of IPV at antenatal care visits when assessing conditions that may be caused or complicated by IPV [35]. Other prenatal care guidelines affirm that clinical practitioners should be aware of the possibility of IPV, but do not include any specific recommendation related to the screening [36]. It is well known that IPV is associated with adverse mental health and obstetrical health consequences for the mother, fetus, and child, but women are reluctant to speak about this topic without a previous inquiry [37]. The American College of Obstetricians and Gynecologists guidelines recommend screening for IPV at the first prenatal visit, at least once per trimester, and at the postpartum checkup [38]. However, the overall rate of screening asymptomatic women is distressing [39]. Due to the high prevalence of this serious problem, estimated violence during pregnancy ranges from 15 to 40.5% for any type of violence, figures higher than those previously reported by Perttu et al. [40]; it is vital to have a correct estimation of its magnitude. These evaluations are necessary to underscore the importance of systematic screening: only when health staff are aware of the right prevalence and repercussion of IPV will they be able to cope with the screening barriers and to identify the most vulnerable populations by introducing screening programs in antenatal care.

Isolated prevalence studies may underestimate the true IPV prevalence due to barriers to open disclosure. These barriers could vary in different cultures and religions; e.g., widespread social norms in some regions support husbands' right to physically discipline wives. Abused women often face high social, economic, and legal barriers to divorce, a situation that is made worse by unresponsive law enforcement and health care institutions. In this social context, women are often reluctant to report violence to authorities and may hesitate to disclose violence to survey interviewers. In many societies, women are also reluctant to report violence because they are ashamed of living under this kind of relationship.

Summarizing the figures of IPV prevalence in pregnancy is needed to highlight the public health importance of this problem, with rates in some studies reported to be over 50%. These figures point towards the need of systematic screening in pregnancy. However, the analysis of published IPV reviews showed weaknesses in the research available on this topic. This umbrella review allows us to identify some methodological aspects that should be addressed in future reviews, related to geographic scope, study selection, and bias assessment.

### **5. Conclusions**

Available data about IPV prevalence during pregnancy are of low quality. The existing evidence syntheses do not capture the totality of the disease burden in IPV in pregnancy. Despite there being wide variability in existing prevalence figures, it is worth noting that no less than 1 out of 50, and as many as 1 out of 2 women, could be suffering physical IPV in pregnancy. Psychological IPV violence is reported to be even more frequent in the published reviews. The existing evidence syntheses do not capture the totality of the worldwide disease burden of IPV in pregnancy.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1660-4601/18/2/707/s1, Table S1: List of excluded articles.

**Author Contributions:** R.M.R.-G. and S.M.-P. were responsible for data collection and analysis, data extraction and management, and quality assessment. A.B.-C. resolved any disagreement between R.M.R.-G. and S.M.-P. R.M.R.-G., S.M.-P., K.S.K., A.B.-C., and J.M.M.-G. participated in study design, analysis, and data interpretation. All authors have read and agreed to the published version of the manuscript.

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

**Acknowledgments:** K.S.K. is a Distinguished Investigator funded by the Beatriz Galindo (senior modality) Program grant given to the University of Granada by the Ministry of Science, Innovation, and Universities of the Spanish Government.

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

### **References**




© 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*

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

### **Nuria Infante-Torres 1, Milagros Molina-Alarcón 2, Angel Arias-Arias 1, Julián Rodríguez-Almagro 3,\* and Antonio Hernández-Martínez <sup>3</sup>**


Received: 2 September 2020; Accepted: 21 October 2020; Published: 23 October 2020

**Abstract:** To evaluate the association between prolonged second stage of labor and the risk of adverse neonatal outcomes with a systematic review and meta-analysis. PubMed, Scopus and EMBASE were searched using the search strategy "Labor Stage, Second" AND (length OR duration OR prolonged OR abnormal OR excessive). Observational studies that examine the relationship between prolonged second stage of labor and neonatal outcomes were selected. Prolonged second stage of labor was defined as 4 h or more in nulliparous women and 3 h or more in multiparous women. The main neonatal outcomes were 5 min Apgar score <7, admission to the Neonatal Intensive Care Unit, neonatal sepsis and neonatal death. Data collection and quality assessment were carried out independently by the three reviewers. Twelve studies were selected including 266,479 women. In nulliparous women, a second stage duration greater than 4 h increased the risk of 5 min Apgar score <7, admission to the Neonatal Intensive Care Unit and neonatal sepsis and intubation. In multiparous women, a second stage of labor greater than 3 h was related to 5 min Apgar score <7, admission to the Neonatal Intensive Care Unit, meconium staining and composite neonatal morbidity. Prolonged second stage of labor increased the risk of 5 min Apgar score <7 and admission to the Neonatal Intensive Care Unit in nulliparous and multiparous women, without increasing the risk of neonatal death. This review demonstrates that prolonged second stage of labor increases the risk of neonatal complications in nulliparous and multiparous women.

**Keywords:** Apgar score; meta-analysis; Neonatal Intensive Care Unit; neonatal morbidity; newborn care; labor stage; second; systematic review

### **1. Introduction**

The second stage of labor is the period of time between full cervical dilatation and birth of the baby, during which the woman has an involuntary urge to bear down, as a result of expulsive uterine contractions [1].

The description of the onset of the second stage of labor in clinical practice is often not precisely known. If complete dilatation is found on vaginal examination, it remains uncertain how long this cervical status has been present [2].

Multiple observational studies [2–4] have observed an increase in maternal complications associated with a prolonged second stage of labor, such as operative vaginal delivery, third-/fourth-degree perineal lacerations, caesarean delivery, urinary retention, postpartum hemorrhage and chorioamnionitis, as well as an increase in neonatal complications like seizures, hypoxic-ischemic encephalopathy, sepsis and increased mortality. However, the criteria these studies used to define the second stage of labor are heterogenous.

Thus, diagnosis and management of prolonged second stage of labor and its complications are difficult and often pose a dilemma to the obstetrician regarding timing and type of intervention [5]. Additionally, evidence on the duration of the second stage of labor is of very low certainty [1] and it is unclear whether there is a point of time from which the risk of perinatal complications increases and at which health professionals should intervene to prevent adverse events [3,6].

Nevertheless, there are professionals involved in childbirth care that try to reduce the duration of the second stage by obstetric interventionism in order to avoid neonatal complications. Paradoxically, these interventions, such as immediate pushing (initiated as soon as complete dilation is identified) [7], instrumental birth [8] or fundal pressure [9], may themselves increase the risk of neonatal morbidity.

In the past, a prolonged second stage of labor had been defined as a period of time that lasted beyond 2 h with epidural analgesia or 1 h without epidural analgesia for multiparous women. For nulliparous women, a prolonged second stage is defined as a period of time that lasted beyond 3 h with epidural analgesia or 2 h without epidural analgesia [10]. Recently, though, the American College of Obstetricians and Gynecologists (ACOG) [11] and the National Institute for Health and Care Excellence (NICE) [12] have allowed longer durations in specific cases. In spite of this, the correct management of the second stage of labor should be individualized according to birth progress, fetal malposition or the use of epidural analgesia [11,12]. For example, the Eunice Kennedy Shriver National Institute of Child Health and Human Development document suggested allowing one additional hour for the use of epidural analgesia. Thus, at least 3 h in multiparous women and 4 h in nulliparous women would be considered to diagnose a prolonged second stage of labor [11].

Thus, our objective was to evaluate the evidence on the association between prolonged second stage of labor (defined as 4 h in nulliparous women and 3 h in multiparous women) and the risk of adverse neonatal outcomes.

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

This systematic review with a meta-analysis was done according to PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) [13,14].

### *2.1. Data Sources and Searches*

The adopted search strategy was: "Labor Stage, Second" (Mesh) AND (length OR duration OR prolonged OR abnormal OR excessive). Studies were identified in three main databases: PubMed [15], Scopus [16] and EMBASE [17], from 1 January 1990 to 1 November 2019. As well as published studies, we included non-published studies which had been included in the conference proceedings of the main scientific associations and indexed in the databases consulted. All languages were included. The search results for each database are provided in detail in Table A1.

All members of the research team had prior training in the methodology of systematic reviews, literature reviews and critical reading. AAA and AHM are also experts in meta-analysis.

Studies were included according to four criteria: (I) duration of second stage of labor greater than 4 h in nulliparous women; (II) duration of second stage of labor greater than 3 h in multiparous women; (III) studies reporting neonatal outcomes in relation to duration of second stage of labor; (IV) studies that stratified results by parity. Reference lists from the selected studies were also examined to locate further studies not identified using the search strategy. Two authors (NIT and AAA) independently performed the literature search and excluded any articles that did not meet the established inclusion criteria. A third author (MMA) was consulted to resolve any disagreements or uncertainty regarding inclusion.

### *2.2. Main Outcomes*

The primary outcomes were 5 min Apgar score < 7, admission to the Neonatal Intensive Care Unit, neonatal sepsis and neonatal death. All neonatal outcomes examined by the available studies were included in this review. The definitions of some of the variables included in our study are shown in Table 1.






NR: not reported.


### *2.3. Data Extraction and Quality Assessment*

Data collection and quality assessment were carried out independently by the three reviewers (NIT, AHM and JRA). We tried to contact the authors of several studies to provide us with data that did not appear in their manuscripts.

We used the Joanna Briggs Institute Critical Appraisal tools for use in JBI Systematic Reviews to assess the risk of bias in each included study [24]. Eleven domains were assessed to appraise the methodological quality of a study and to determine the extent to which a study had addressed the possibility of bias in its design, conduct and analysis.

### *2.4. Data Synthesis*

For the categorical results, the odds ratio (OR) was used along with its 95% confidence intervals (95% CI). To calculate the OR, either the Mantel–Haenszel fixed-effects or Der Simonian–Laird random-effects models were used, depending on whether there was heterogeneity between the studies. Heterogeneity was assessed using the I<sup>2</sup> and the statistical Cochran's Q tests. I2 values of < 25%, 25–50 and >50% normally correspond to small, medium and large heterogeneity, respectively [14,25,26]. Publication bias was also evaluated using the Egger asymmetry test and funnel plots [14,27]. Statistical significance was defined at the ≤0.05 level.

All calculations were done with the StatsDirect statistical software, version 2.7.9. (Stats Direct Ltd., Cheshire, England) [14].

### **3. Results**

### *3.1. Study Selection*

A total of 1868 studies were selected from the literature search. After removing any duplicated articles, 267 were selected by title and abstract. After applying the inclusion/exclusion criteria, twelve articles were selected for the qualitative and quantitative analyses (meta-analysis) (Figure 1).

**Figure 1.** PRISMA flow diagram of the literature reviewing process.

### *3.2. Study Characteristics*

The description of the studies included in this systematic review are shown in Table 2. The sample included 268,624 women. The selected studies were conducted in Canada [4,18,28], United States [19–21, 29,30], China [31], Sweden [22,32] and Spain [23]. The sample size of these studies ranged from 307 [31] to 121,490 [4]. All studies were restricted to singleton infants with cephalic presentation. Eight of these articles studied nulliparous women [18,20–22,28,30–32], two studied multiparous women [19,23] and two studied both (nulliparous and multiparous women) [4,29]


**2.**Characteristics of the studies analyzed.

**Table**


**Table 2.** *Cont.*


**Table 2.** *Cont.* NC: not calculated, NR: not reported, EA: epidural analgesia, BW: birthweight, WG: weeks gestation, \*: no data on

nulliparous/multiparous.

### *3.3. Study and Data Quality*

The included studies had a low risk of bias, except for three studies that did not identify confounding factors [18,28,31] and four studies that did not include strategies to deal with confounding factors [18,28,30,31] (Table A2).

With regard to the selection of subjects, all studies except one [31] specified inclusion and exclusion criteria, selecting all women (nulliparas and/or multiparas) with singleton cephalic presentation that reached second stage of labor within a specific period of time.

Seven of the studies included in the meta-analysis [4,18,20,21,23,28,32] correctly defined prolonged second stage of labor (in this case, second stage of labor longer than 4 h in nulliparas and longer than 3 h in multiparas). Conversely, only three of them [18,21,28] established the maneuver used once prolonged second stage of labor was diagnosed (instrumental birth, continuing maternal pushing, caesarean, etc.).

As for data and information collection, five studies [4,22,23,30,32] included missing or incomplete data as exclusion criteria, so they were not included in the analysis.

### *3.4. Main Outcomes and Meta-Analysis*

### 3.4.1. Nulliparous Women

### 3.4.1.1. min Apgar score <7

To determine the relation between prolonged second stage of labor in nulliparous women (Table A3) and risk of low 5 min Apgar score (<7), six studies were included (*n* = 116,624) [4,18,28,30–32]. A significant increase in low 5 min Apgar score was observed when the second stage of labor lasted more than 4 h with respect to when the second stage of labor was ≤ 4 h. (OR = 1.65; 95% CI: 1.20–2.27). For this analysis, a random-effects model was used since heterogeneity was observed (Cochran's Q *p*-value = 0.0041; I2 = 71.0) (Figure 2a; Table 3).







### Admission to Neonatal Intensive Care Unit

To assess the risk of admission to the Neonatal Intensive Care Unit, eight studies were employed (n = 156,650) [4,18,20–22,28–30].

The risk significantly increased when the second stage of labor lasted more than 4 h with respect to when the second stage of labor was ≤ 4 h (OR, 1.63; 95%CI 1.44–1.84). For this analysis, a random-effects model was used since medium heterogeneity was observed (Cochran's Q *p*-value = 0.057; I2 = 48.8) (Figure 2B; Table 3).

### Neonatal Sepsis

By combining three studies (*n* = 82,053) [4,20,21], we found that the risk of neonatal sepsis increased when the duration of the second stage of labor was longer than 4 h with respect to when the second stage of labor was ≤ 4 h (OR, 1.57; 95% CI 1.07–2.29). For this analysis, a fixed-effects model was used since no heterogeneity was observed (Cochran's Q *p*-value = 0.7962; I2 = 0.0) (Table 3).

### Neonatal Death

Two studies (*n* = 28,032) [18,21] were employed to determine the relationship between prolonged second stage of labor and risk of neonatal death, and no differences were found (OR, 7.21; 95% CI 0.37–139.71) (Table 3).

### Other Neonatal Outcomes

No significant associations were reported between prolonged second stage in nulliparous women and 1 min Apgar score < 1, 5 min Apgar score ≤ 3, umbilical artery pH < 7, acidosis, meconium-stained amniotic fluid, meconium aspiration, birth depression, minor or major trauma, birth trauma, shoulder dystocia, brachial plexus injury, Erb's palsy, resuscitation at birth, heart compressions, hypoxic ischemic encephalopathy, hypothermia treatment or composite neonatal morbidity. When the results of two studies were combined [20,22], only an increased risk of neonatal intubation in women with a second stage of labor > 4 h was observed (OR, 2.19; 95% CI 1.23–3.90) (Table 3).

### 3.4.2. Multiparous Women

### 3.4.2.1. min Apgar Score < 7

To determine the relation between prolonged second stage of labor in multiparous women (Table A4) and risk of low 5 min Apgar score (< 7), three studies were included (*n* = 72,857) [4,19,23]. A significant increase in low 5 min Apgar score was observed when the second stage of labor lasted more than 3 h with respect to when the second stage of labor was ≤ 3 h (OR, 3.67; 95% CI 2.49–5.43). For this analysis, a fixed-effects model was used since no heterogeneity was observed (Cochran's Q *p*-value = 0.987; I2 = 0.0) (Figure 2C; Table 3).

### Admission to the Neonatal Intensive Care Unit

To assess the risk of admission to the Neonatal Intensive Care Unit, three studies were employed (*n* = 76,692) [4,19,29]. The risk significantly increased when the second stage of labor lasted more than 3 h with respect to when the second stage of labor was ≤ 3 h (OR, 2.41; 95% CI 2.02–2.88). For this analysis, a fixed-effects model was used since no heterogeneity was observed (Cochran's Q *p*-value = 0.417; I2 = 0.0) (Figure 2D; Table 3).

### Neonatal Sepsis

None of the studies that analyzed multiparous women considered this variable when assessing neonatal morbidity in relation to the duration of the second stage of childbirth (Table 3).

### Neonatal Death

None of the studies that analyzed multiparous women considered this variable when assessing neonatal morbidity in relation to the duration of the second stage of childbirth (Table 3).

### Other Neonatal Outcomes

No significant associations were reported between prolonged second stage in multiparous women and umbilical artery pH < 7.0, umbilical artery pH < 7.10, umbilical artery base excess ≥12, meconium aspiration, shoulder dystocia, prolonged neonatal stay, advanced neonatal resuscitation, birth depression, minor or major trauma or any perinatal morbidity. After combining two studies [19,29], only an increase in the risk of meconium staining was observed (OR, 1.29; 95%CI, 1.01–1.66), and an increase in composite neonatal morbidity (OR,1.97; 95% CI, 1.39–2.80) was observed after another two studies were combined [19,23] (Table 3).

### 3.4.3. Publication Bias

We did not observe publication bias for the study in any of the variables studied (Tables A3 and A4).

We can observe a summary of results obtained following meta-analysis of all variables studied in nulliparous and multiparous women in Table 3.

### **4. Discussion**

### *4.1. Main Findings*

Our meta-analysis results suggested that duration of second stage of labor of more than 4 h in nulliparous women increased the risk of low 5 min Apgar score < 7, admission to the Neonatal Intensive Care Unit, neonatal sepsis and neonatal intubation. In multiparous women, when the second stage of labor was longer than 3 h, the risk of 5 min Apgar score < 7, admission to Neonatal Intensive Care Unit, meconium staining and composite neonatal morbidity increased.

However, a prolonged second stage of labor did not increase the risk of any of the other variables studied, such as umbilical artery pH < 7, birth depression, neonatal death meconium aspiration or shoulder dystocia.

### *4.2. Comparison with Existing Literature*

The literature has very limited data on neonatal outcomes of women with duration of second stage of labor of more than 4 h in nulliparas and of more than 3 h in multiparas. We were only able to locate 12 articles with these durations for this review.

An example of this is a recent systematic review by Gimovksy et al., which evaluated the maternal and fetal morbidities associated with prolonged second stage of labor in nulliparous women with epidurals, in which the authors defined prolonged second stage as greater than three hours [33]. Only two papers were included in this systematic review, and very discordant neonatal outcomes were analyzed, which did not allow the results to be combined in order to establish conclusions that would be useful for decision-making in clinical practice.

Another systematic review studied the influence of prolonged second stage of labor on the risk of adverse maternal and neonatal outcomes from 1980 until 2005 [34]. It did not report associations between prolonged second stage and adverse neonatal outcomes, but most of the studies analyzed in this review defined the prolongation of the second stage as more than 2 h, without differentiating according to parity. In addition, it did not conform to the new recommendations of allowing longer durations.

Only one randomized controlled trial [35] specifically addressed the effect of this change in obstetric practice on maternal and neonatal outcomes. In that trial, a policy of extending the second stage of labor for at least 1 h in nulliparous women with epidural anesthesia with respect to "usual labor" (3 h) decreased the incidence of caesarean birth by more than half compared with the common practice (19.5%, 8 of 41, vs. 43.2%, 16 of 37; RR, 0.45; 95% CI, 0.22–0.93). Maternal or neonatal morbidity were not statistically different between the groups. Unfortunately, the trial was underpowered to detect significant differences in the frequency of adverse maternal or neonatal outcomes between groups because the sample studied was very small (only 78 nulliparous women) (35).

However, Zipori et al. [36] recently published another study comparing maternal and neonatal outcomes over two distinct time periods. In period I, the duration of the second stage of labor was considered prolonged according to ACOG limits, and it was called a "classic labor curve" (10). The "new labor curve" of period II allowed nulliparous and multiparous women to continue the second stage of labor for an additional 1 h before diagnosing second-stage arrest. Primary caesarean deliveries decreased with the new policy of labor management, with a small rise in instrumental deliveries, but it also increased other immediate maternal and neonatal complications, such as higher rate of lower umbilical artery cord pH.

### *4.3. Strengths and Limitations*

One of the strengths of this study is that it is the first systematic review to define prolonged second stage of labor according the most recent recommendations (11), that is, 4 h for nulliparous women and 3 h for multiparous women. Most of the studies had large sample sizes with sufficient numbers of participants in each group to lend power to the findings, and the majority of them used methods to control for potential confounding factors.

Among the limitations of our systematic review is that neonatal outcome measures were discordant in the included studies, meaning it was difficult to combine data to summarize important clinical findings, and that the definition of two variables (admission to NICU and composite neonatal morbidity) differed among included studies. None of the studies considered the pushing duration or pushing techniques employed (delayed pushing or immediate pushing). Finally, since they were observational studies, there is a risk of confounding bias even though many of the studies included techniques to control confounding.

### **5. Conclusions**

In nulliparous women, a prolonged second stage of labor is not related with an increased risk of neonatal death. However, it is related with an increased risk of 5 min Apgar score < 7, admission to the Neonatal Intensive Care Unit, neonatal sepsis or intubation. In multiparous women, a prolonged second stage of labor is related with an increased risk of 5 min Apgar score < 7, admission to the Neonatal Intensive Care Unit, meconium staining and composite neonatal morbidity.

These potential risks associated with a prolonged second stage of labor in both nulliparous and multiparous women should serve as an incentive for professionals involved in childbirth care to increase supervision of mothers who exceed these durations.

More studies are needed, especially clinical studies, to guarantee the safety of newborns when the second stage of labor exceeds 4 h in nulliparous women and 3 h in multiparous women.

**Author Contributions:** Conceptualization, N.I.-T. and J.R.-A.; Methodology, M.M.-A. and A.A.-A.; Formal Analysis, A.H.-M. and J.R.-A.; Writing—Original Draft Preparation, N.I.-T. and M.M.-A.; Writing—Review & Editing, N.I.-T. and A.A.-A.; Supervision, J.R.-A. 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**
