**(Don't) Look at Me! How the Assumed Consensual or Non-Consensual Distribution A**ff**ects Perception and Evaluation of Sexting Images**

**Arne Dekker 1,\*,**†**, Frederike Wenzla**ff **1,**†**, Anne Daubmann 2, Hans O. Pinnschmidt <sup>2</sup> and Peer Briken <sup>1</sup>**


Received: 8 April 2019; Accepted: 13 May 2019; Published: 17 May 2019

**Abstract:** The non-consensual sharing of an intimate image is a serious breach of a person's right to privacy and can lead to severe psychosocial consequences. However, little research has been conducted on the reasons for consuming intimate pictures that have been shared non-consensually. This study aims to investigate how the supposed consensual or non-consensual distribution of sexting images affects the perception and evaluation of these images. Participants were randomly assigned to one of two groups. The same intimate images were shown to all participants. However, one group assumed that the photos were shared voluntarily, whereas the other group were told that the photos were distributed non-consensually. While the participants completed several tasks such as rating the sexual attractiveness of the depicted person, their eye-movements were being tracked. The results from this study show that viewing behavior and the evaluation of sexting images are influenced by the supposed way of distribution. In line with objectification theory men who assumed that the pictures were distributed non-consensually spent more time looking at the body of the depicted person. This so-called 'objectifying gaze' was also more pronounced in participants with higher tendencies to accept myths about sexual aggression or general tendencies to objectify others. In conclusion, these results suggest that prevention campaigns promoting 'sexting abstinence' and thus attributing responsibility for non-consensual distribution of such images to the depicted persons are insufficient. Rather, it is necessary to emphasize the illegitimacy of the non-consensual distribution of sexting images, especially among male consumers of the material.

**Keywords:** eye tracking; non-consensual image sharing; intimate images; objectification; objectifying gaze; rape myth acceptance; sexting

#### **1. Introduction**

Sexting, the sending of intimate or explicit personal pictures, videos, or texts [1], has become common practice within different age groups [2–5]. Definitions vary, and the confusion of consensual and non-consensual sexting proves to be a central conceptual problem. [6,7]. While consensual sexting refers to the purposeful, active, and often pleasurable sending of one's own images, the non-consensual sharing of sexting images happens against the will or without the knowledge of the person depicted [8]. This non-consensual sharing is one of the most frequently discussed risks in the context of sexting [9–18]. If sexting images are forwarded against the will of the person depicted (e.g., in their circle of friends) or published on the internet, this poses a serious risk to mental health. Situations in which victims are

exposed to public humiliation and online bullying can lead to grave psychosocial consequences, in some cases even suicide [3,7].

Not only in the public debate but also in 'sexting abstinence' campaigns [19], sexting, in general, is deemed dangerous [20]. Not differentiating between consensual and non-consensual sexting can lead to victim blaming if the depicted producers of the images are held responsible for the unintended dissemination [7]. This mechanism has been criticized in the theoretical context of 'rape culture' [21–23] and linked to the broader concepts of 'sexual objectification' [24–27] and 'rape myth acceptance' [26,28,29]. Objectification theory postulates that in western societies women are sexually objectified, treated as objects and are only considered worthy to the extent that their bodies give pleasure to others [29] (for reviews [28,30]). Sexual objectification can be seen as a continuum ranging from acts of violence to subtler acts such as objectifying gazes [30,31]. These gazes, conceptualized as visually inspecting (sexual) body parts, have been empirically demonstrated using eye-tracking technology [32]. Additionally, people who sexually objectify others have been shown to be more likely to accept rape myths [24,25], which serve to normalize sexual violence, e.g., through victim blaming (for reviews [27,33]). These subtle myths have been conceptualized as cognitive schemes [34] and demonstrated to influence eye movements [35,36].

Although research has evolved around non-consensual sexting and its correlates [7,9,20], little effort has been conducted to investigate reasons for consuming such images. The question arises why people consume non-consensual sexting material when mere comparisons with consensual material do not reveal apparent differences in image content. Is there a specific attraction in the non-consensuality itself, at least for some of the consumers? Against this background, we experimentally investigate the question of how the supposed way of distribution (consensual vs. non-consensual) influences the perception of sexting images. Thus, the study promises important findings for future prevention efforts.

In accordance with the objectification theory we expect differences in evaluation and perception of sexting images depending on their supposedly consensual or non-consensual forwarding. In line with previous research, we argue that increased objectification is associated with higher attractiveness ratings of the objectified person [37] and a more pronounced objectifying gaze [32]. We further hypothesize that supposedly non-consensually forwarded images are considered as more intimate and their further distribution as more unpleasant. Overall tendencies for other objectification and higher rape-myth acceptance are also expected to increase objectification.

A large part of the scientific literature on sexting focuses on the behavior of adolescents. This may reflect widespread societal fears, but, in fact, sexting experience is significantly higher among adults than among adolescents. In a current systematic review [3] the prevalence estimate of studies of adolescents sending messages containing sexually suggestive texts or photos was found to be 10.2% (95% CI (1.77–18.63)), while the estimated mean prevalence of studies of adults was 53.31% (95% CI (49.57–57.07)). Against this background, and also because the present experimental study does not focus on a representative image of the user population, we have decided to examine a sample of adults. We assume that the mechanisms shown are comparable in adolescents, but this must be demonstrated by future research.

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

#### *2.1. Participants*

A total of 76 participants (57% female, Mage = 31.99, SDage = 10.28) were recruited via university newsletters. They were informed about the tasks and the stimulus content but were kept naïve to the full purpose of the experiment. Participants provided written consent to study participation. No compensation was given. The ethics committee of the state chamber of psychotherapists of Hamburg (Psychotherapeutenkammer Hamburg) approved the study protocol of the present study (03/2015-PTK-HH).

#### *2.2. Stimuli and Apparatus*

Volunteers personally known to the authors but unknown to the study participants provided 14 semi-nude sexting images [38]. One additional image per gender was obtained from freely available internet sources for public presentation purposes, resulting in a set of 16 pictures (50% female).

Stimulus presentation and data collection were conducted on a 22-inch widescreen monitor (1680 × 1050 pixels) using SensoMotoric Instruments (SMI GmbH, Teltow, Germany) software ExperimentCenterTM. A remote eye tracker (SMI, RED system) recorded eye movements at 120 Hz from 50 cm viewing distance using a head-chin rest.

#### *2.3. Questionnaires*

Individuals' objectification of others was assessed using a German translation of the modified version of the Self-Objectification Questionnaire [39] for other objectification (Other Objectification Scale, OOS [40]). The scale consists of 10 body attributes, five competence-based (i.e., strength) and five are appearance-based (i.e., physical attractiveness). Participants were asked to rank how important they perceive each attribute (10 = "most important"; 1 = "least important") separately for men and women. Possible scores range from −25 to 25 with higher scores indicating higher levels of objectification.

Participants further completed an 11-item short version of the German Acceptance of Modern Myths About Sexual Aggression Scale (AMMSA) [41] which had been used successfully in other eye tracking studies already [35,36]. Each item was rated on a 7-point scale (1 = "completely disagree"; 7 = "completely agree").

#### *2.4. Procedure*

Participants read an introductory text stating that the study aimed to understand more about the evaluation of sexting images. Depending on the condition, picture distribution was either described as voluntary (consensual condition) or as unwanted, against the will of the depicted person (non-consensual condition). The manipulation was strengthened by asking participants to state three feelings the image distribution could have evoked in the depicted persons. Following, participants saw the images three times with different tasks. Pictures were randomized within blocks, starting with the male images. Pictures were presented individually on full screen for 5 seconds, preceded by a black fixation cross on the left side shown for 1 second. The first task was to freely view the pictures. Second, participants rated the sexual attractiveness of the depicted person. For the third task, participants were asked to evaluate how intimate they considered the image content and how unpleasant further picture distribution would be for the depicted person (ranging from 1 = "not at all ... "; 7 = "very ... "). After completion of the sociodemographic information, and the questionnaires, participants were thanked and debriefed.

#### *2.5. Data Reduction and Data Analysis*

To account for repeated measures made on the same subject, a mixed model approach was employed. We examined the fixed effects of the independent variables condition (consensual vs. non-consensual distribution), gender (women vs. men), image gender (female vs. male images), of their three and two-way interactions and of the OOS score and AMMSA score on the ratings of (1) sexual attractiveness, (2) intimacy of image content, and (3) perceived unpleasantness of picture distribution. Random intercepts were assumed for participants. We report the marginal means and their 95%-confidence intervals. We report the results of the final models after a backward elimination of the non-significant effects according to Kleinbaum et al. [42]. All statistical tests were two-tailed (α = 0.05).

The eye tracking data were analyzed using the same model as described above with the objectifying gaze as the dependent variable. The objectifying gaze was operationalized as the relative time spent looking at the body compared to the time spent looking at faces [32]. We created two areas of interest (AOI) on each image, one containing the head and the other containing all the rest of the body. The total dwell time for both AOIs, i.e., the overall time viewing the person depicted, was set to 100%. For the following analysis, we focus on the percentage of that time directed at the body. Accordingly, an increase in viewing time on the body always results in a decrease of dwell time on the face, since both values always add up to 100%. So a stronger objectifying gaze refers to relatively longer viewing time on the body and shorter viewing time on the face.

Computations were done using the GENLINMIXED (Generalized linear mixed model) routine of SPSS version 22 (IBM Corporation, Armonk, NY, USA) and eye tracking data reduction was realized using the standard settings of BeGazeTM (SMI, Teltow, Germany), providing gaze information such as duration (dwell time).

#### **3. Results**

#### *3.1. Participants*

Prior to data analysis participants were excluded due to poor recordings (*n* = 5), non-heterosexual orientation (*n* = 3), or due to inadequate responses to the manipulation check (*n* = 10) as rated by four independent raters. A total of 58 participants (57% female, Mage = 31.45, SDage = 10.18) remained for data analysis (see Table 1). Table 1 also shows the means of participants' AMMSA and OOS scores. In this context, it is particularly important that the mean values of the two study groups do not differ.



The means do not significantly differ between conditions (*p* > 0.08). OOS score = Scores on the Other Objectification Scale (Strelan and Hargreaves, 2005) separately for the objectification of women and of men; the possible range is from −25 (low objectification) to 25 (high objectification). AMMSA score = Scores on the 11-item short version of the Acceptance of Modern Myths About Sexual Aggression scale (Gerger et al., 2007); the possible range is from 1 (low acceptance) to 7 (high acceptance). <sup>a</sup> *n* = 25. <sup>b</sup> *n* = 33. <sup>c</sup> *n* = 24.

#### *3.2. Ratings*

Separate models were conducted for each of the three explicit ratings, namely sexual attractiveness of the person depicted, perceived intimacy of the image content, and unpleasantness of further distribution. Only the significant effects of the final models are reported here.

For attractiveness ratings, we did not find that condition (consensual vs. non-consensual distribution; see Table 2) had any effect. We did, however, find that gender had an effect as well as an interaction effect between participant gender and image gender. Overall, men rated the images of men as more attractive (M = 4.17, SE = 0.32) than women did (M = 3.02, SE = 0.31; *t*(924) = 3.25, *p* < 0.001). Women also rated the images of men as less attractive than images of women (M = 4.46, SE = 0.32, *t*(924) = 9.36, *p* < 0.001). No other effects reached significance.

Concerning the intimacy ratings, we found an interaction effect between condition and gender (*p* = 0.008, see Table 2). Pairwise contrasts revealed that women who assumed non-consensual distribution regarded the images as more intimate (M = 4.86, SE = 0.25) than women who assumed consensual distribution (M = 4.56, SE = 0.26; *t*(924) = 2.58, *p* = 0.01).

Analyzing influences on how unpleasant further distribution was considered for the depicted person, we found that condition (consensual vs. non-consensual distribution; *p* < 0.001) had a highly significant effect (see Table 2). Pairwise contrasts revealed that participants assuming non-consensual sharing considered further distribution as more unpleasant (M = 4.63, SE = 0.28) than participants who assumed consensual sharing (M = 4.26, SE = 0.28; *t*(924) = 3.74, *p* <.001). We also found an interaction effect between gender and image gender. Women rated the unpleasantness of further distribution lower for images of men (M = 4.08, SE = 0.40) than male participants did (M = 4.41, SE = 0.40; *t*(924) = 2.50, *p* = 0.013). Furthermore, the AMMSA score reached significance (coefficient = −0.13, *p* = 0.002), indicating that the higher participants scored on the AMMSA-scale, the less unpleasant they considered picture distribution for the depicted person.


**Table 2.** Final models of the influences on ratings of sexual attractiveness, intimacy, and presumed unpleasantness of further distribution.

Fixed effects (*df* 1 = 1, *df* 2 = 924). AMMSA score = Scores on the 11-item short version of the Acceptance of Modern Myths About Sexual Aggression scale (Gerger et al., 2007). <sup>a</sup> The coefficient value indicates the increase of the rating per score-increase of 1 (e.g., unpleasantness rating decrease of −0.13 per AMMSA score increase of 1).

#### *3.3. Eye Tracking Analysis*

Regarding eye movements, we were interested in the objectifying gaze, operationalized as the relative time viewing the body. We found a significant interaction of condition and gender (*F*(1,834) = 8.36, *p* < 0.001). Men in the non-consensual condition demonstrated a stronger objectifying gaze as they looked significantly longer at bodies *(*M = 54.37, SE = 8.99) than men in the consensual condition (M = 46.52, SE = 9.01; *t*(834) = 4.25, *p* < 0.001) (see Figure 1). Within the non-consensual condition, men also demonstrated the objectifying gaze more than women did, spending more time looking at bodies than women did (M = 49.53, SE = 8.97; *t*(834) = 3.07, *p* = 0.002). Notably, there was no such gender difference within the consensual condition (*p* > 0.05).

**Figure 1.** Estimates of the mean proportion (and standard error) of dwell time spent on the body by condition and gender. \*\*\* *p* < 0.001; \*\* *p* < 0.01.

The effects of the OOS score and the AMMSA score were significant (*p* < 0.001), indicating that relative dwell time on the body increases for higher scores. In other words, this reveals a more pronounced objectifying gaze for higher tendencies to objectify and accept myths about sexual aggression (see Table 3).


**Table 3.** Influences on the proportion of dwell time spent looking at the body.

Fixed effects (*df* 1 = 1, *df* 2 = 834). OOS score = Scores on the Other-Objectification questionnaire (Strelan and Hargreaves, 2005). AMMSA score = Scores on the 11-item short version of the Acceptance of Modern Myths About Sexual Aggression scale (Gerger et al., 2007). <sup>a</sup> The coefficient value indicates the increase of dwell time on the body per score-increase of 1 (e.g., dwell time increase of 2.96 per AMMSA score increase of 1).

#### **4. Discussion**

We demonstrate that not only explicit ratings but also the implicit viewing behavior are influenced by the assumed consensual or non-consensual distribution of sexting images.

#### *4.1. Image Evaluations*

Participants who assumed the non-consensual distribution of a sexting image, namely the sharing against the will of the person depicted, rated the further distribution of the images as more unpleasant. This finding demonstrates that not only the picture content itself or personal feelings about sexting but also the surrounding information is considered when estimating the unpleasantness of further picture distribution. Interestingly, women rated the unpleasantness of distribution lower for images of men than male participants did. Seeing images of other men, the risk saliency of becoming a victim and having one's images shared non-consensually might have increased for men, leading to higher ratings of unpleasantness. Due to the common stories of non-consensual sexting involving women, female participants might be aware of personal risks at any time independent of the condition. As the potential consequences of forwarding are more severe for women [43,44], female participants might consider further forwarding as less unpleasant because of the less severe consequences for men. However, it is important to note that the images of men and women should not be compared directly with each other in this study as picture compositions varied. Men were usually posing less sexually than women, which is due to the naturalistic creation of the images, but likely influences the ratings of unpleasantness.

Overall, higher general rape myth acceptance led to lower ratings of perceived unpleasantness of further distribution in both conditions. Higher endorsement of rape myths is indicative of a higher likelihood of victim blaming, which is in line with the common risk discourses on sexting [7,12,22,45]. Accordingly, considering non-consensual sharing a risk inherent in sexting allows minimizing the expected level of unpleasantness of further distribution. The depicted person is deemed responsible for having taken the image to begin with and hence either stupid or reckless. In other words, the estimated unpleasantness decreases when victim blaming increases. This is crucial as this pattern is not only typical for cases of revenge pornography [46] but also for other forms of sexual harassment [26,47] and has even found its way into 'sexting abstinence' campaigns [20]. Concerning the perceived intimacy of the images, women assuming non-consensual distribution rated the images as more intimate for both genders than women assuming consensual sharing. Men, however, did not differ between the consensual or non-consensual distribution of images of either men or women. This could be attributed to the fact that women are more likely to be victims of non-consensual sexting [3] and to be victimized in general in most forms of online gender-based violence [19,48]. Being aware of the potential personal risk might make women more sensitive to the intentions of the depicted person and violations of privacy.

Unlike expected, the assumed way of distribution did not affect how participants rated the sexual attractiveness. Previous research linking objectification and attractiveness ratings presented women in casual wear and the same women in bikinis [46]. Such a strong manipulation allows for large differences between conditions. Using the same semi-nude images in both conditions as done in our study might not have been a strong enough manipulation to affect explicit attractiveness ratings. The exhibited interaction effect between gender and image gender, more precisely higher ratings of male images by men, is likely due to factors inherent in the images and not the context. Therefore, we do not consider them as relevant for this study.

#### *4.2. The Objectifying Gaze*

The objectifying gaze, defined as the relative amount of time looking at the body, was influenced by condition and participant gender. Men assuming non-consensual distribution displayed the objectifying gaze more than men assuming voluntary sharing and more than women assuming either manner of distribution. Hence, we were able to demonstrate for the first time that the supposed way of distribution influences how participants look at images and how strongly they display the objectifying gaze. Previous research suggests that especially women are sexually objectified in the media [26,49,50] and during interpersonal interactions [51,52]. The objectifying gaze has been linked to negative social perceptions, dehumanization, and self-objectification [53–55]. While an appearance-focus in women has been linked to negative social perceptions [54,55] and severe mental health problems [55], no comparable research on men exists.

Although mostly discussed for men, women are thought to have internalized the objectifying gaze so much that they demonstrate it toward other women as well [56]. However, in our study, only men assuming non-consensual distribution differed from the other participant groups, albeit unaffected by the gender of the depicted person. Unlike other studies [57–60], we did not find systematic influences of image gender on viewing behavior. We suggest that our manipulation might have evoked other task demands that resulted in viewing patterns different from free viewing conditions, possibly covering influences of image gender [61]. In line with previous research, higher general tendencies to objectify others, as well as higher acceptance of rape myths, were related to a more pronounced objectifying gaze [35]. Numerous gender-specific functions and consequences have been reported for rape myths acceptance (for a review see [62]). Still, due to cultural changes, rape myths and sexist beliefs have become increasingly subtle as taken into consideration and measured by the acceptance of modern myth about sexual aggression scale applied here [63]. This study is the first to consider the influences of both biases on eye movements and suggests that subtle attitudes indeed affect viewing behavior. These influences and their implications should be further investigated in the context of sexual aggression.

#### *4.3. Limitations and Future Research*

Our study was conducted in the laboratory with well educated, heterosexual participants viewing images of young, attractive adults who were semi-nude, unlike in most severe cases of non-consensual image sharing [64]. Accordingly, the generalizability of our results needs further investigation. Future research has to take intersectional influences (e.g., skin color or age) into account, as these factors are relevant in the context of objectification [50]. Concerning participants, intersectionality is also important, as cultural influences regarding eye-movements [65], sexual objectification [66], and sexual harassment [67,68] have been found. Other reasons for fixating more on bodies (e.g., social comparison) or avoiding faces (e.g., shame) should be explored as well.

As mentioned above, in this study we have focused on adult participants for two main reasons: First, the prevalence of sexting among adults is actually higher than among adolescents. Secondly, we were not interested in a representative image of the user population, but in an experimental comparison of two equivalent groups. Nevertheless, it is possible that the correlations shown do not exist among adolescent users. For this reason, a replication of the present study with adolescent participants would be desirable.

Although we demonstrated that the supposed manner of distribution affects the perception of sexting images, qualitative research asking consumers of non-consensual sexting for their motives seems like an important step to further identify the beliefs behind such behavior, (e.g., the enjoyment of power) [69]. Another aspect is the perceived agency of the depicted person that might be decreased by non-consensual forwarding, which in turn could facilitate objectification. This idea needs further investigation.

Since everyday sexual objectification is common [70], it is crucial to examine and develop theories regarding possible outcomes and further explore the similarities between sexual assault and non-consensual pornography, or technology-facilitated violence in general.

As rapid changes of the technological landscape routinely link new types of specific behavior (e.g., non-consensual sexting) to existing theory (e.g., on sexual objectification) they can inform the creation of prevention programs [46,71]. The well-researched theory of 'sexual double standards' suggests that women's sexuality is often perceived as pure and damageable through active desire, holding women responsible for protecting themselves from aggressive male sexuality [72,73]. This leads to the paradoxical position for women of experiencing social and cultural pressure to be sexy while simultaneously risking negative social consequences when portraying themselves in such manner online [74,75]. Considering the sexual double standard allows us to understand nonconsensual sexting as reaffirming stereotypical gender roles that place women under the control of men [53,55]. As girls are more likely to engage in sexualized self-presentations on social network sites and more attention is paid to their physical appearance than that of boys [76], gendered aspects need to be considered [17,77]. While arguments have been made to consider sexting as an empowering (social) media production [78,79] and to frame sexy appearance as a feminist act to counter the negative effects of objectification [80], this positive reframing carries the potential negative effect of normalizing unwanted sexual attention, which may outweigh the possible benefits of individual self-preservation [71].

#### **5. Conclusions**

In conclusion, we demonstrated that viewing behavior and evaluation of sexting images are influenced by their supposed consensual or non-consensual distribution. In line with objectification theory, an 'objectifying gaze' was more pronounced in men who assumed non-consensual picture distribution, meaning they spent a relatively longer time looking at the body of a depicted person. This 'objectifying gaze' was also more pronounced for participants with higher tendencies to accept myths about sexual aggression or general tendencies to objectify others. The results suggest that prevention campaigns that focus on a general message of sexting abstinence and thus attribute responsibility for non-consensual distribution of such images to the persons depicted are insufficient. Rather, it is necessary to emphasize the illegitimacy of the non-consensual distribution of sexting images, especially among male consumers of the material. This can be done, for example, in the context of school educational events, but there is also at least one example of an appropriate public prevention campaign: http://notyourstoshare.scot/. Only with these or comparable measures can the serious psychological consequences of public humiliation and online bullying be prevented in the long term.

**Author Contributions:** Conceptualization, A.D. (Arne Dekker), F.W., and P.B.; methodology, A.D. (Arne Dekker), F.W.; software, not applicable; formal analysis, F.W., A.D. (Anne Daubmann), H.O.P.; investigation, F.W.; resources, A.D. (Arne Dekker), P.B.; data curation, F.W.; writing—original draft preparation, A.D. (Arne Dekker), F.W.; writing—review and editing, A.D. (Arne Dekker), F.W., P.B.; visualization, F.W.; supervision, P.B.; project administration, A.D. (Arne Dekker); funding acquisition, P.B.

**Funding:** This research was funded by the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF, 01SR1602).

**Acknowledgments:** We would like to thank all volunteers for providing their images.

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

#### **References**


© 2019 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/).

### *Article* **Sexual Satisfaction and Mental Health in Prison Inmates**

**Rodrigo J. Carcedo 1, Daniel Perlman 2, Noelia Fernández-Rouco 3,\*, Fernando Pérez <sup>1</sup> and Diego Hervalejo <sup>1</sup>**


Received: 30 April 2019; Accepted: 14 May 2019; Published: 17 May 2019

**Abstract:** The main goal of this study was to investigate the association between sexual satisfaction and mental health, and the combined effect of two previously found, statistically significant moderators: partner status and sexual abstinence. In-person interviews were conducted with 223 participants (49.327% males and 50.673% females). The effect of sexual satisfaction on mental health and the interactions of sexual satisfaction × partner status, sexual satisfaction × sexual abstinence, and sexual satisfaction × partner status × sexual abstinence were examined using simple moderation and moderated moderation tests after controlling for a set of sociodemographic, penitentiary, and interpersonal variables. Results revealed a direct relationship between sexual satisfaction and mental health only for the sexually abstinent group. Partner status was not significant as a moderator. It seems that the lack of sexual relationships is more powerful as a moderator than the lack of a romantic relationship. Additionally, the sexually abstinent group showed lower levels of sexual satisfaction in those with a partner outside or inside prison, and lower mental health independently of the current romantic status, than sexually active inmates. These findings point to the importance of sexual satisfaction to mental health in sexual situations of extreme disadvantage.

**Keywords:** sexual satisfaction; sexual abstinence; partner status; mental health; prison inmates

#### **1. Introduction**

More than 10 million people are living in jails and prisons worldwide [1], and considerably larger numbers of ex-prisoners are living in society [2]. A high prevalence of mental health problems is present in prison populations [3]. There is also increasing epidemiological evidence that prisoners are more likely to suffer from mental health problems than the average population [4–7].

In the most representative Spanish study that included 28.8% of the inmate populations in five different prisons, the lifetime prevalence rate of mental disorders was 84.4%. The prevalence of any mental disorder in the last month before the time of interview was 41.2% [8]. These results were confirmed more recently by a study with a smaller sample size (*n* = 184), obtained from three prototypical Spanish prisons [9]. A total percentage of 90.2% inmates had suffered a mental disorder during their lives. Also, 55.2% were suffering a mental disorder at the time. Finally, in this study, the inmate population was 5.3 times more likely to have a mental health problem than the general population.

These mental health problems are risk factors for a range of adverse outcomes in prison and on release including self-harm [10], suicide [11–16], and violence inside prison [17], and reoffending in released prisoners [2,18,19].

In sum, most prevalence studies have been conducted in developed countries and consistently show that a very high proportion of prisoners suffer from poor mental health [3,20]. Despite the high level of need, these disorders are frequently underdiagnosed and poorly treated [20]. In addition, a growing literature documents the detrimental consequences of incarceration for mental health [21–24]. For example, early scholars believed being imprisoned is associated with having higher rates of mental health disorders than inmates would have had if they had remained in the community [25]. Massoglia found evidence of persisting elevated mental health issues in previously incarcerated individuals [26]. Furthermore, incarceration is negatively associated with finances [27], family ties [28], and physical health [29] as well as a greater risk for sexual victimization [20].

All this makes the mental health status of current and former prison inmates an important public health issue [3]. Following the World Health Organization's definition, this study will consider mental health as not merely the absence of illness but "a state of well-being in which every individual realizes his or her own potential, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to her or his community" [30]. Thus, this concept includes mental illness but also understands mental health as a positive dimension of well-being [31].

One of the possible causes of prison inmates being an at-risk population for poor mental health is that they encounter difficulties in having a satisfactory sex life [32–35]. Linville found that approximately 75 percent of a sample of 100 male inmates in a minimum-security prison reported emotional problems due to sexual deprivation [36]. As a result of the sexual deprivation inmates experience, they may seek relief in alternative, less satisfactory and/or riskier ways [37]. Different studies have demonstrated a high rate of masturbation [38–40], and the presence of consensual homosexual behavior as alternative forms of sexual behaviors [41,42]. Such behaviors are sometimes coercive [43–45], and can lead to the transmission of sexual diseases such as HIV [46]. Conjugal visitations have been suggested as one possible solution. Consistent with this view, states that permit conjugal visits have lower instances of reported rape and other sexual offenses in their prisons [47]. Nonetheless, the low frequency of visits, the lack of good conditions [48], and their being restricted to married or committed partners limits the efficacy of conjugal visits.

All these experiences are evaluated by prison inmates determining their level of sexual satisfaction. Sexual satisfaction has been defined as "an affective response arising from one's subjective evaluation of the positive and negative dimensions associated with one's sexual relationship" [49] (p. 258). It is regarded as a fundamental dimension of the quality of sexual activity. Research on sexual satisfaction in prison inmates has generally shown very low levels of sexual satisfaction except for those with a romantic partner inside the same prison and those who did not remain abstinent [48,50,51]. Taken altogether, sexual needs are not well satisfied in prison.

Arguably, sexual satisfaction can be considered an essential component of general well-being and mental health. Empirically, higher sexual satisfaction is associated higher mental health and lower depression [52–54]. The recognition of the need to be loved, appreciated and cared for, and of the desire for intimate relationships that provide emotional sustenance and empathy, have been considered important aspects for maintaining mental health in prisons [30].

#### *1.1. The Sexual Satisfaction and Mental Health Relationship Moderated by Partner Status and Sexual Abstinence*

Research on the relationship between sexual satisfaction and mental health in prison inmates is in a fledgling state. Researchers have largely overlooked the part sexual satisfaction can play in inmates' mental health and well-being. Research involving these variables conducted with other populations is more extensive.

#### 1.1.1. Research Conducted Outside Prisons

The consequences of a satisfying sex life are important areas of research that are gaining increasing attention in the psychological and medical literature, suggesting that sexuality maintains its importance even in the context of serious health concerns [55]. In this way, higher sexual satisfaction is associated with low levels of sexual anxiety [56,57], low psychopathological symptoms [57,58], and good mental health [59,60].

Furthermore, fostering patients' quality of life and mental health are key aims of health care in which subjective factors are commonly seen as central [61]. One subjective factor that has received very little attention is patients' sexual satisfaction, although Mallis et al.'s results showed that sexual satisfaction and quality of life are "strongly connected" (p. 447) [62]. Other research has found sexual dissatisfaction is higher in patients with depression than in those without depressive symptoms [63].

Turning to relationship status and its role in the association between sexual satisfaction and mental health, in both non-clinical and clinical samples, partnered compared to single individuals have tended to report higher sexual satisfaction and sexual activity [52,64,65]. Having a partner does not necessarily mean that couples live together or that they have an active sex life, but it increases the likelihood that partners do have consistent sexual contact. Furthermore, tight-knit social structures such as being in a close relationship often, but not always, lead to better mental health outcomes [66]. Consistent with the beneficial view of tight structures, Holt-Lunstad, Birmingham, and Jones [67] and others (e.g., [68]) have found that being married is associated with better mental and physical health. Analyses such as this typically lump everyone together and do not examine other predictors of mental within subgroups.

Although the moderating effect of partner status between sexual satisfaction and mental health was not specifically investigated in the aforementioned studies of partner status, there is important evidence that the negative aspects of romantic life (e.g., loneliness and dissatisfaction, two aspects related to the fact of not having a partner or not having a satisfactory relationship for meeting one's emotional needs) predict personal well-being more strongly than the positive aspects (e.g., marital satisfaction) [69]. Complementing the negative is stronger than the positive, other non-prison studies have found a strong relationship between sexual satisfaction and general well-being including mental health for those who had been sexually deprived due to the presence of sexual dysfunctions [70,71], physical disabilities [72], amputations [73], and having had germ-cell tumor therapy [74].

Other interpersonal variables, a category in which sexuality belongs [75], have a differential effect on mental health depending on partner status. For example, friendship quality only correlated significantly with depression among a group of college students without a romantic partner whereas no association was found in the group in a current romantic relationship [76].

Furthermore, Taleporos and McCabe compared the strength of this relationship for a group of people with and without sexual difficulty (physical disability vs. no physical disability) [72]. In this case, for both genders, the relationships between sexual satisfaction and indicators of mental health such as depression and self-esteem were stronger for people with physical disabilities than for able-bodied people. In other words, sexual satisfaction was a stronger predictor for the mental health of the group in a less favorable and more restrained condition. This situation might be comparable with the situation of sexually abstinent prison inmates who have shown much lower levels of sexual satisfaction than sexually active ones. In this comparison it is the sexually abstinent inmates who are in a more restrained and difficult situation.

Complementing Taleporos and McCabe's results, Laumann et al. found that in the cluster of countries where average levels of sexual satisfaction were low (male-centered regimes; in a worse situation with a less freedom of choice) there was a stronger relationship between sexual well-being and happiness, which may be considered as an indirect indicator of positive mental health status, than in the cluster of countries where average levels of sexual satisfaction were higher (gender-equal sexual regime; in a better and free situation) [77]. If results in this vein generalize, one would then expect the association between sexual satisfaction and mental health to be stronger among sexually abstinent inmates. Also, based on prisoners' previously mentioned negative feelings toward abstinence and the available data, we would expect the sexually abstinent inmates to have low sexual satisfaction.

#### 1.1.2. Research Conducted in Prison Contexts

Sexual satisfaction, mental health and other well-being-related measures have been found to be significantly correlated in studies conducted in prison settings [48,50,51,78]. The findings revealed that higher levels of sexual satisfaction were associated with higher levels of mental health and other well-being related measures.

Typically, in these studies the association between sexual satisfaction and mental health has been examined without considering the participants' relationship status. The meaning of sexual experiences may vary depending on individuals' romantic situation, especially among prison inmates who have stringent restrictions imposed on their sexual activities. In fact, research has shown that prison inmates without a partner or with a partner outside the prison had lower levels of sexual satisfaction and mental health than those inmates with a partner inside the same prison [48]. In a later study, a moderating effect of partner status on the relationship between sexual satisfaction and mental health was found. Lower sexual satisfaction was associated with lower mental health only for those without a partner [50]. These latter findings illustrate a pattern suggested in non-prison studies that the association between sexual satisfaction and mental health is intensified for those in a less desirable romantic status.

In arguing that a lack of sexual satisfaction can negatively impact prison inmates' mental health, most authors [33–35,79] were referring mainly to inmates who had not had heterosexual relationships during their incarceration. Thus, these investigators were defacto ignoring inmates who were engaging in sanctioned sexual activities with their partners. Sexual satisfaction reflects a self-evaluation of one's current sexual life; sexual abstinence refers to a complete lack of sexual relationships during a period of time. In reporting their sexual satisfaction, abstinent inmates were reporting on their satisfaction with not having sanctioned partnered sex whereas partnered inmates were reporting on the partnered sexual activities they were permitted to have. As has been found, an inmate may have been sexually abstinent during the last 6 months, yet show reasonable high sexual satisfaction [32]. By contrast, an individual may have been sexually active and show low sexual satisfaction. Thus in a noteworthy way, the referent for their judgments of sexual satisfaction is different for abstinent inmates than it is for partnered inmates.

This opens the possibility that the relationship between inmates' sexual satisfaction and mental health may be different for sexual abstainers than for sexually active individuals. An earlier prison study found such a moderating effect [51]: sexual satisfaction was significantly associated with psychological health only for the group of inmates who had not had sexual relationships during the last 6 months, in other words, sexual abstainers.

In sum, previous research findings showed lower levels of sexual satisfaction and mental health in sexually abstinent inmates [32,51]. More importantly, an association between low sexual satisfaction and low mental health was only found for those who did not have a partner in the same prison (versus without a partner) [50] and those who remained sexual abstainers (versus non-abstainers) [32,51]. However, these two interaction effects have not been tested together to study (a) whether both are significant, (b) whether the proportion of variance for which they account is similar or different, and (c) whether there is a higher order interaction formed by sexual satisfaction, partner status, and sexual abstinence.

This study will focus on the new knowledge gained by including both abstinence and partner status. This current investigation also refines a previous study [50] because it includes three different partner statuses (no partner, partner outside of prison, and partner inside the same prison) instead of two (partner vs. no partner). Clearly there is a need of differentiating inmates with a partner inside or outside because these situations delineate different experiences.

In addition, this study benefits from a larger sample size and the addition of a set of control variables that have previously been demonstrated to have significant effects on mental health. Namely, poorer mental health has been exhibited by inmates who are younger, Caucasian [80], and married [80,81]; who have longer sentences and a longer expected time prior to their release [82]; who report poor general health [83]; who show higher levels of social and emotional loneliness [32,50,51,78]; and who, based on non-prison studies [84–86], masturbate more frequently. All these variables will be entered in the models as covariates.

#### 1.1.3. Research Questions

Flowing from the summary of the aforementioned evidence found, two research questions emerge, a first and central question and an ancillary second one: (a) Will partner status and sexual activity level play a moderator role in the relationship between sexual satisfaction and mental health, after controlling for sociodemographic (sex, age, and nationality), penitentiary (total time in prison and estimated time to parole), and personal, social, and sexual well-being aspects (self-rated health, social, family, and romantic loneliness, and frequency of masturbation)? (see Figure 1) and (b) Will partner status and sexual abstinence be associated with inmates' sexual satisfaction and mental health, after controlling for sociodemographic (sex, age, and nationality), penitentiary (total time in prison and estimated time to parole), and personal, social, and sexual well-being aspects (self-rated health; social, family, and romantic loneliness; and frequency of masturbation)?

#### **2. Experimental Section**

#### *2.1. Participants*

Participants for this study were entirely inmates from the medium-security Topas penitentiary, located in Salamanca (Spain). This prison houses men and women in the same prison but in different modules. The prison administration decided from which men's and women's modules the investigators could recruit participants. After stratifying by gender, 80% of the participants were randomly selected, whereas 20% were selected under a "snowball" sampling scheme [36]. Participants were excluded from this study if they (a) had been in prison for less than 6 months, the time considered necessary to become adapted to prison life and develop new relationships inside the facility; (b) did not speak Spanish or English; (c) had been diagnosed with a serious mental disorder; or (d) were not in an optimal condition to be interviewed (e.g., under the influence of drugs or expressing high levels of anxiety or distrust toward the interviewer). Only twelve potential participants declined being interviewed. All of the participants found the interview to be a positive experience.

Due to the difficulties collecting information from this specific population, we retained for analyses in the present report participants in two of Carcedo et al.'s previous studies [50,51] that had 119 and 173 participants, respectively. For this study, a sample of 223 inmates from 20 to 62 years old (M = 35.172, SD = 7.823) was used. We selected the increase in sample size to ensure reasonable power for testing the interaction effects of interest in the current analyses. This increase resulted in successfully having at least 10 participants per subgroup formed by crossing partner status (inside, outside, no partner) and sexual abstinence categories (abstinent vs. non-abstinent). Although males and females in prison are not equal in number, we selected a roughly equal number of male (*n* = 110) and female (*n* = 113) participants in order to explore the possible effect of sex on the results and, consequently, the results' interpretation and discussion. Nationality was encoded in two levels: Spanish nationality (*n* = 103) and foreign, unspecified origin country (*n* = 120). Regarding the two moderators in this study, 76 inmates had no partner (34.080%), 61 had a partner outside the prison (27.354%), and 86 had a partner inside the prison (38.565%); also, 122 inmates reported having had sex in the last six months (54.709%) and 101 kept sexually abstinent (45.291%).

In comparison with inmates with a current romantic partner outside the prison, those in a relationship inside the same prison presented a higher frequency of in-person contact (*t* (145) = −12.413, *p* > 0.001; outside: M = 3.311, SD = 1.679; inside: M = 5.698, SD = 0.510; variable coding: 1 "never", 2 "more than 6 months", 3 "3–6 months", 4 "each 1–2 months", 5 "each 7–15 days", and 6 "every day or almost every day", and satisfaction with the current relationship (*t* (145) = −2.746, *p* < 0.001; outside: M = 3.510, SD = 1.678; inside: M = 4.358, SD = 1.326; variable coding: 1 "totally unsatisfied" and 5 "totally satisfied") and lower duration of the union in months (*t* (145) = 3.588, *p* < 0.001; outside: M = 102.459, SD = 100.593; inside: M = 49.831, SD = 77.172).

All sexually active inmates reported they had engaged in heterosexual behavior at least once in the last six months. Regarding frequency, 63.934% of sexually active inmates had had sexual relationships at least once every 15 days, 24.590% every 2 months, and 11.475% every 6 months. Most of these sexual relationships occurred in conjugal visit rooms (76.471%), but also in other locations inside the prison (shared areas such as the sociocultural module, prison laundry, kitchen, gym, etc. (19.328%) and family visit rooms (1.681%)) and during furloughs outside the prison (2.521%). Inmates reported that their sexual relationships had included vaginal coitus at least once in the last 6 months. It is also important to mention that three sexually active participants also reported to have had some homosexual contact in prison. Finally, no sexually active inmate was convicted of sex crimes.

Preliminary analyses did not find any significant effect of sex, in the presence of partner status and sexual activity level, on sexual satisfaction and mental health nor a moderating effect between these two variables. Therefore, both sexes were analyzed together and sex was only included as a control variable in all the analyses.

#### *2.2. Design and Procedure*

This study used a short-term longitudinal design. Two interview sessions were carried out with a difference of a week between them. The main associated variable and control variables were extracted from the first interview and the outcome (mental health) was taken from the second one. Each participant was interviewed in a private room located in his or her prison module, separated from the rest of the inmates. The interviews were kept short (approximately 30 min without counting the time dedicated to create a good relationship) to ensure that participants did not get tired and to avoid "interrogation effects".

All the interviews were conducted by the same interviewer to foster consistency. Before starting the interview, the interviewer spent a significant amount of time building a trustful relationship with every inmate (usually about 20–30 min, but depending on the speed of establishing rapport, in some cases it took up to 2 h). Afterwards, participants were invited to participate and were informed about the possibility of leaving the study whenever they wished to do so. Participants were informed about the confidentiality and anonymity of the study and all the participants signed consent forms. We consider that respecting all of these conditions is extremely important in collecting good-quality data from this population. Ignoring these conditions can easily increase distrust among the prison inmates. Finally, it is important to state that this study respected the norms of the Declaration of Helsinki's ethical principles for medical research involving human subjects.

#### *2.3. Measures*

#### 2.3.1. Sexual Satisfaction

The sexual satisfaction subscale of the Multidimensional Sexual Self-Concept Questionnaire (MSSCQ) [87] was used to measure the main variable of this study. A total of five items were scored on a five-point Likert-type scale that ranged from 1 (not at all characteristic of me) to 5 (very characteristic of me). Cronbach's alpha for this scale was 0.960.

#### 2.3.2. Moderating Variables: Sexual Activity Level and Partner Status

This variable was recorded as 0 for the inmates who had experienced sexual relationships in the past 6 months (non-abstinent), and 1 for the inmates who had not (abstinent). Sexual relationships were understood as any sexual behavior with another person including vaginal or anal intercourse, oral sex, and mutual masturbation and genital caresses, excluding kisses, hugs, and non-genital caress. Partner

status was coded to have three categorical levels: no partner (0), partner outside (1), and partner inside the prison (2). Partner status was defined as a relationship deemed, in the inmate's mind, as one that both partners considered serious.

#### 2.3.3. Outcome Variable: Mental Health

This construct was measured with the short Spanish version of the Psychological health subscale included in the World Health Quality of Life scale (WHOQOL-BREF) [88]. Six items were scored on a five-point Likert-type scale that ranged, with different labels, from 1 (not at all; very dissatisfied; never) to 5 (extremely-completely; very satisfied; always). Cronbach's alpha was 0.709. Sample items include "To what extent do you feel your life to be meaningful?" and "How often do you have negative feelings such as blue mood, despair, anxiety, depression?" This scale was selected for multiple reasons: It is brief; it conceptualizes mental health not only as the absence of illness but also the presence of positive aspects of mental health; and its concurrent validity as indicated by its high correlation (*r* = 0.70) with the widely used SF-36 (36-Item Short-Form Health Survey) mental health subscale [89].

#### 2.3.4. Control Variables: Sociodemographic, Penitentiary, and Personal, Social, and Sexual Well-Being Variables

Considering sociodemographic variables, sex was codified as 0 for male and 1 for female inmates, age was asked directly to each inmate and confirmed against inmate penitentiary records for accuracy, and nationality was dichotomized into Spaniards (0) versus foreigners (1). Regarding penitentiary variables, total time in prison refers to the total time spent in prison for previous and current offenses. This information was collected by reviewing inmates' penitentiary records, and it was recorded in months. Estimated time to parole was captured by asking the inmates how much time they expected to be in prison from that moment, based on the information they possessed. This variable was also computed in months.

With respect to personal, social, and sexual well-being variables, self-rated health was measured by asking the participants "in general, would you say your health is: excellent (4); very good (3); good (2); fair (1); or poor (0)?" [90]. The short version of the Social and Emotional Loneliness Scale for Adults (SELSA-S) [91] was used to measure both types of loneliness. SELSA-S consists of three subscales labeled (a) social loneliness, (b) family-emotional loneliness, and (c) romantic-emotional loneliness. Participants rated 15 items (five per scale) on a seven-point Likert-type scale that ranged from 1 (strongly disagree) to 7 (strongly agree). Cronbach's alphas were 0.829, 0.898, and 0.840 for social, family-emotional, and romantic-emotional loneliness, respectively. Finally, masturbation frequency was codified into six levels based on the frequency inmates reported having masturbated during the last 6 months: (1) never, (2) less than once a month, (3) once or twice a month, (4) once or twice a week, (5) once a day, (6) twice a day or more.

Each scale or subscale score was obtained by adding the item scores and dividing them by the number of items answered. Higher scores represented higher levels in that dimension for all the variables included in this study.

#### *2.4. Statistical Analysis*

A 3 × 2, partner status (no partner, partner outside the prison, and partner inside) by sexual activity (abstinent vs. non-abstinent inmates) ANCOVA was used to first analyze the differences in sexual satisfaction and then performed again with mental health as the outcome variable. Each analysis controlled for sociodemographic, penitentiary, personal, social, and sexual well-being variables. If the partner status by sexual activity interaction between factors was statistically significant, Bonferroni post-hoc tests for multiple comparisons were conducted. Statistical significance was defined as *p* < 0.05.

The Breuch–Pagan test was conducted to test heteroscedasticity between sexual satisfaction and mental health. The macro heteroscedasticity test for SPSS [92] was utilized for this purpose. To study the relationships of sexual satisfaction with mental health and the moderating effects of partner status and sexual activity level, the PROCESS 3.2. macro for SPSS [93] was utilized. PROCESS's models number one and two for two-way interactions (also called simple moderation), and three for the three-way interaction (also named moderated moderation) were used. Additionally, 95% confidence intervals were calculated based on 5000 bootstrap samples. The HC3 heteroscedasticity-consistent standard error estimator was applied [94] due to the violation of homoscedasticity. All the statistical analyses were conducted using the IBM SPSS 23 package (IBM Corp., Armonk, NY, USA).

#### **3. Results**

Descriptive information for the variables considered in this study are included in Table 1. With sexual satisfaction as the outcome variable, the 3 × 2 partner status by sexual activity level ANCOVA yielded significant effects for sexual activity level (F (1, 207) = 47.115, *p* < 0.001, η<sup>2</sup> <sup>p</sup> = 0.185) and the partner status <sup>×</sup> sexual activity level interaction (F (2, 207) <sup>=</sup> 14.638, *<sup>p</sup>* <sup>&</sup>lt; 0.001, <sup>η</sup><sup>2</sup> p = 0.124). Bonferroni post-hoc comparisons revealed lower levels of sexual satisfaction in sexually abstinent inmates in comparison with non-abstinent for those who had a partner outside (*p* < 0.001; abstinent: M = 0.838, SE = 0.237; non-abstinent: M = 2.979, SE = 0.178) or inside the same prison (*p* < 0.001; abstinent: M = 1.094, SE = 0.302; non-abstinent: M = 3.006, SE = 0.155). However, no differences in sexual satisfaction between abstinent and non-abstinent inmates were found for those who were not involved in a romantic relationship (*p* > 0.05; abstinent: M = 2.462, SE = 0.197; non-abstinent: M = 2.312, SE = 0.354).

In the 3 × 2 ANCOVA with mental health as the outcome measure, sexual activity level yielded a significant effect (F (1, 207) = 10.182, *p* < 0.01, η<sup>2</sup> <sup>p</sup> = 0.047). Those who were sexually abstinent (M = 3.260, SE = 0.081) presented lower levels of mental health in comparison with non-abstinent inmates (M = 3.633, SE = 0.080). The effect due to partner status was non-significant.

Regarding associations with mental health, the Breuch–Pagan test yielded a significant result for heteroscedasticity (LM = 3.883, *p* < 0.05). Thus the HC3 heteroscedasticity-consistent standard error estimator was used [42] to run the regression model. The three-way interaction of sexual satisfaction <sup>×</sup> partner status <sup>×</sup> sexual activity level was not significant (Δ*R*<sup>2</sup> <sup>=</sup> 0.001, F (2, 201) <sup>=</sup> 0.174, *p* > 0.05). By contrast, the two-way sexual satisfaction × sexual activity level interaction was statistically significant (Δ*R*<sup>2</sup> = 0.016, F (2, 205) = 8.298, *p* < 0.01), whereas the sexual satisfaction <sup>×</sup> partner status interaction was not (Δ*R*<sup>2</sup> = 0.007, F (1, 205) = 1.590, *p* > 0.05). In the former case, the conditional effects of sexual satisfaction at the values of the moderators showed lower levels of mental health only for those who were abstinent during the last six months. This result was found significant across the three levels of partner status (see Table 2 and Figure 1) and for the whole sample (sexual abstinent group: B = 0.176, SE = 0.089, *t* = 1.990, *p* < 0.05, 95% CI = (−0.346, 0.129)). No significant effect was observed for sexually active individuals.


**Table 1.** Correlations, means, and standard deviations of the variables considered in this study.

2.

3.

4.

5.

9.

*p*< 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001. *%*(1): Percentage of group with label "1" (females, foreigner, partner status, and abstinent) for dichotomous

 variables.

#### *J. Clin. Med.* **2019**, *8*, 705

\*


**Table 2.** Multiple regression analysis on mental health and conditional effects of sexual satisfaction at values of the moderators (partner status and sexual activity level).

\* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001. B, Unstandardized coefficient; SE, Standard Error.

As clearly can be seen in Figure 1, the interaction effect of sexual satisfaction × sexual activity presents a similar pattern for the groups of inmates without a partner, and in a current relationship outside or inside the prison. It is important to highlight that overall a decrease in sexual satisfaction of the sexually abstinent group is associated with a reduction of mental health levels, and the contrary, an increase in sexual satisfaction is related to an improvement in mental health.

**Figure 1.** Sexual satisfaction × sexual activity level interaction associated with mental health for three different partner status groups.

#### **4. Discussion**

A direct relationship between sexual satisfaction and mental health was only found for the sexually abstinent group in this study. Partner status did not appear as a significant moderator. However, among those with a partner outside or inside prison, the sexually abstinent group showed lower levels of sexual satisfaction and mental health than sexually active inmates.

Again, sexual satisfaction was found to be significantly associated with mental health, as in other prison studies [36,50,51,78] and non-prison studies [52–54,56–60,63]. In this study, however, the sexual satisfaction, mental health association was only obtained for those who had remained sexually abstinent for at least the last six months. Previous research testing just one moderator has found that higher levels of sexual satisfaction were associated with higher levels of mental health only for prison inmates without a partner [50] and inmates who were sexually abstinent [51]. The current study examined the impact of both moderators, partner status and sexual activity level, together on the sexual satisfaction, mental health association. The results of this analysis showed that only the sexual satisfaction × sexual activity interaction was statistically significant. Neither the sexual satisfaction × partner status interaction nor the three-way interaction was significant. Thus a key implication of this study is that the lack of sexual relationships is more powerful as a moderator than the lack of a romantic relationship.

An important question here is why the lack of sexual relationships emerged in the regression analysis as significantly associated with mental health, whereas partner status did not. We speculate that sexual needs may be more important or basic than the emotional needs associated with romantic relationships. Sexuality, and more specifically sexual desire, comprises cognitive, emotional, and physiological processes and is consubstantial to the fact of being humans. Sexual desire may be a stimulus that sparks the inmates' sensitivity to their lack of sexual satisfaction. Lack of sexual contact in prison has even been named by inmates as "sexual torture" [32]. By contrast, wishing to be in a romantic relationship in a prison where the pool of eligible partners may not be especially attractive, may produce either lower levels of reactance and/or lower levels of dissatisfaction with not having a partner. In addition, intimacy and emotional needs can be solved by other ties, like close friends [75]. This suggests that future research could profitably focus on the role of sexual desire levels as a means of dealing effectively with inmates' sexual deprivation and/or the role that non-romantic, personal relationships contribute to improving their mental health. Hence, the damaging impact of being exposed to circumstances perceived as negative could be lessened by promoting positive interpersonal experiences and healthy interactions within inmates' daily experiences.

Prison settings are an unconventional, yet potentially diagnostic, context in which to study the sexual satisfaction, mental health association. Similar contexts would be worth considering in other public health studies. The meaning of sexual satisfaction may be completely different for those who are sexually inactive or suffering from serious restrictions vis-a-vis sexual activities than for those who are sexually active. The current results have possible implications for other populations whose freedom to choose has been reduced or eliminated due to constraining situations or who are involved in more negative or difficult circumstances. As noted previously, strong associations between sexual satisfaction and mental health or other well-being related measures have been found in other populations afflicted by different medical conditions [70–74] or living in a more sexually restrained culture [77].

This study also found significantly lower levels of sexual satisfaction in the abstinent group. This result is consistent with previous research in prison [32,51]. Furthermore, there were parallel significant differences in sexual satisfaction between abstinent and sexually active inmates in both the groups with a partner outside and inside the prison (not all inmates with a partner were sexually active), but not for those without a partner. Having a partner and not having access to sexual relationships can generate even more reactance and/or create a worse position than not having a partner and sexual relationships. Additionally, sexually abstinent inmates showed low mental health. Presumably the abstinent inmates were in a worse situation and experiencing greater reactance to the loss of freedom with respect to their sexual lives than the sexually active inmates. All these results are consistent with previous research developed within prison contexts, highlighting inmates' difficulties in meeting their sexual needs [32–35,51] and, as a consequence of this, presenting mental and emotional health problems [36].

Findings stemming from the two research questions of this study point to the crucial role sexual abstinence can have for mental health in some circumstances. Low sexual satisfaction (only for inmates with a partner outside or inside the prison) correlated with poorer mental health and a significant relationship between sexual satisfaction and mental health was observed in sexual abstainers. The abstinent group may be increasing their desire for sexual relationships due to their sexual deprivation [95]. Individuals wish to operate with a freedom to choose behaviors to satisfy their needs and if their freedom is reduced, threatened, or eliminated, individuals will become "motivationally aroused" to regain this freedom (see reactance theory [96,97]). Also, as seen in our results, this group is afflicted by sexual dissatisfaction, possibly the result of a large gap between their desires and their reality. Negative information and events (e.g., being abandoned by partners, losing friends, etc.) per se have been shown to have more impact on individuals' judgments and well-being than positive ones (e.g., gaining friends, partners, etc.) (see "the bad is stronger than good approach" [98]), especially in stigmatizing contexts [99]. This association has also been found in romantic relationships in non-prison studies [69].

The reactance and the bad is stronger than good explanations complement one another but do differ. The reactance interpretation sees motivation as a triggering factor in the linkage between sexual abstinence and mental health. The "bad is better than good" interpretation places primary emphasis on evaluation per se as crucial in the sexual abstinence-mental health association. Future research might profitably examine whether the processes implied by one of these explanations is more applicable than the processes implied by the other and test this current study's findings in other populations where individuals are afflicted by sexual deprivation or restriction due to different medical conditions or social factors.

In sum, this current investigation has found (a) lower levels of both sexual satisfaction and mental health in the sexually abstinent group, and (b) a stronger sexual satisfaction and mental health association in that group. Our perspective is that sexual satisfaction has been strongly correlated with mental health for the abstinent inmates likely because they are in a sexually worse or more deprived situation, and a similar, strong sexual satisfaction-mental health correlation should be observable in other comparably compromised situations or populations.

Our findings have important implications. First, inmates, especially those who are not sexually active, may benefit from prison policies that ease access to romantic and, especially, sexual relationships. We would note that inmates scoring higher on mental health have lower levels of misconduct [100] and lower recidivism rate after release [101]. Promoting positive mental health in prison inmates during incarceration and therefore increasing the likelihood of a successful reentry into society is a central concern with important consequences for public health, security, and the economy. According to this, clinical interventions to increase sexual access could be introduced to enhance inmates' sexual satisfaction. This in turn should be associated with an increase in their mental health. Such changes, however, should take into account the risk profile of inmates because it may be an important variable influencing the choice of interventions.

Assuming inmates will not be able to engage in sexual activities with a partner, other policies and interventions may also be helpful. A shift in cognitions and/or attitudes might influence inmates' evaluation of their sexual satisfaction. Cognitive restructuring techniques might be useful in this regard. Also, helping inmates to focus on other activities, especially ones that they pursue passionately, may relieve part of the distress associated with abstinence. In his dual theory of passion, Vallerand has shown that what he calls harmonious engagement in activities leads to psychological well-being [102]. Finally, increasing privacy in prison cells could facilitate masturbation as another way to obtain some sexual pleasure. Future research should address possible differences in sexual satisfaction between inmates who do, or do not, share their cells with other inmates. Also, it would be worthwhile to compare inmates living alone in a cell but in different prisons where inmates have more or less privacy (e.g., cameras in the rooms, prison officers entering in the cell without asking in advance, etc.).

We also believe that clinicians working with other populations who see their sexual freedom threatened (e.g., physical disabilities, older adults in nursing homes, etc.) can benefit from considering the implications of this study. Populations at risk of mental health problems should also be questioned about the presence or absence of sexual activity in their lives as a means of improving diagnosis and a more accurate intervention plan. Including sexual satisfaction in any diagnosis of mental health and its subsequent intervention seems sensible to consider, especially for those who have difficulties in meeting their sexual needs. Working on external impediments or barriers to having access to sexual relationships should be addressed too. Finally, clinical strategies aimed at reducing patients' reactance and negative evaluations of their sexual deprivation coupled with helping patients discover and perform new highly motivating activities may help patients overcome part of the distress associated with their actual sexual situation.

Apropos of the limitations of this work, this study is correlational so causation is difficult to infer although we used a short-term longitudinal design. Also, a few participants affirmed engaging in homosexual behavior. Despite our stressing the confidentiality and anonymity of the study, homosexual contacts might have been underreported by the inmates. The Spanish context is conservative in character, where heteronormativity (the cultural assumption that heterosexuality is the only valid social norm) is tied deeply to culture [103]. These values are definitely prone to be found in prison inmates too [104]. In this context it is not easy to acknowledge engaging in homosexual behaviors. However, all the participants pointed out they felt very comfortable during the interview and disclosed information that they considered sensitive and important.

#### **5. Conclusions**

In sum, correctional systems often adopt deprivation as a solution to inmates' sexual desires during incarceration. This study offers evidence regarding the importance of sexual satisfaction for their mental health, especially for abstinent inmates. A clear implication of this work is to urge prison administrators to find different solutions for inmates' sexuality that helps them to deal with their sexual desires. But not only that, this study adds new evidence to highlight the importance of considering sexual satisfaction as a predictor of mental health especially in those populations whose freedom to engage in partnered sexual activity has been threatened. From a public health perspective, the association between sexual satisfaction and mental health can vary depending on an individual's sexual activity level, as has been found in this study. Clinicians and health professional should take into consideration this possibility as part of their patients' evaluation and intervention.

**Author Contributions:** Conceptualization, R.J.C., D.P. and N.F.-R.; Methodology, R.C., D.P., F.P., and N.F.-R.; Software, R.J.C. and F.P.; Validation, R.J.C., F.P. and N.F.-R.; Formal analysis, R.J.C., F.P., and N.F.-R.; Investigation, R.J.C. and D.H.; Resources, R.J.C. and D.H.; Data curation, R.J.C. F.P., N.F.-R., and D.H.; Writing—original draft preparation, R.J.C., D.P., F.P., N.F.-R., and D.H.; Writing—review and editing, R.J.C., D.P., F.P., N.F.-R., and D.H.; Visualization, R.J.C., F.P., N.F.-R., and D.H.; Supervision, R.J.C.; Project administration, R.J.C.; Funding acquisition, R.J.C.

**Funding:** This research was funded by the regional education authority of Castile and Leon (Junta de Castilla y León, ref. SA007B08).

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

#### **References**


© 2019 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/).

### *Article* **Sexual Distress in Patients with Hidradenitis Suppurativa: A Cross-Sectional Study**

#### **Carlos Cuenca-Barrales 1, Ricardo Ruiz-Villaverde <sup>1</sup> and Alejandro Molina-Leyva 2,3,4,\***


Received: 7 April 2019; Accepted: 16 April 2019; Published: 18 April 2019

**Abstract:** Hidradenitis suppurativa (HS) is a chronic auto-inflammatory skin disease with a great impact in quality of life. However, there is little research about the impact of HS on sex life. The aims of this study are to describe the frequency of sexual distress (SD) in patients with HS and to explore potentially associated epidemiological and clinical factors. We conducted a cross-sectional study by means of a crowd-sourced online questionnaire hosted by the Spanish hidradenitis suppurativa patients' association (ASENDHI). Sexual distress (SD) was evaluated with a Numeric Rating Scale (NRS) for HS impact on sex life. A total of 393 participants answered the questionnaire. The mean NRS for HS impact on sex life was 7.24 (2.77) in women and 6.39 (3.44) in men (*p* < 0.05). Variables significantly associated (*p* < 0.05) with SD in the multiple linear regression model were sex, with a higher risk in females, the presence of active lesions in the groin and genitals and NRS for pain and unpleasant odor; being in a stable relationship was an important protector factor. Regarding these results, it seems that SD in HS patients is due, at least in part, to disease symptoms and active lesions in specific locations, emphasizing the importance of disease control with a proper treatment according to management guidelines. Women and single patients are more likely to suffer from sexual distress.

**Keywords:** sexuality; mental health; mental disorder; sexual dysfunction; hidradenitis suppurativa

#### **1. Introduction**

Hidradenitis suppurativa (HS) is a chronic auto-inflammatory skin disease characterized by recurrent nodules, abscesses and fistulae and which involves hair follicles, predominantly in intertriginous areas [1]. These lesions cause pain, unpleasant odor, itching and suppuration. When the disease progresses to advanced stages, there may be a permanent negative effect on body image due to scarring.

According to recent studies, the reduction in HS patients' quality of life is one of the most significant among dermatological patients [2,3] and similar to other non-dermatological illnesses such as chronic obstructive pulmonary disease, diabetes mellitus, cardiovascular disease and cancer [4]. Some research indicates that pain or pruritus may negatively affect quality of life [5,6].

Sexuality is a basic need and one which cannot be separated from other aspects of human life, being extremely important for maintaining good mental health [7]. Several studies show a direct relationship between sexual function and quality of life [8,9]. Sexual functionality can be impaired by chronic diseases because of factors related to the disease itself, its treatments, or alterations in body image [10]. Due to the chronic relapsing course of HS and the disease's characteristics, HS may affect patients' sexuality. Numerous publications have associated HS with depression, anxiety, low self-esteem, loneliness, stigmatization, suicide risk, or impact on working life [2,3,11–16]. However, there is little research about the impact of HS on sex life.

The aims of this study are to describe the frequency of sexual distress (SD) in patients with HS and to explore potentially associated epidemiological and clinical factors.

#### **2. Experimental Section**

#### *2.1. Patients and Design*

We conducted a cross-sectional study by means of a crowd-sourced online questionnaire. Participants were recruited from 1 March to 1 April 2018. The Spanish hidradenitis suppurativa patients' association (ASENDHI) hosted the questionnaire and invited people with HS to participate in the study [17].

The selection criterion was self-referred diagnosis of HS. Participants were aware of the questionnaire's anonymity and the use of their data for research purposes. The study was approved on May 2017 by the ethics committee of Hospital Universitario San Cecilio and is in accordance with the World Health Organization Declaration of Helsinki.

#### *2.2. Questionnaire*

The questionnaire was developed with Google® Forms suite. Socio-demographic data, biometric parameters, use of medication for other comorbidities and several characteristics of the disease, such as age of onset, time under medical attention and affected areas were collected. Disease severity was assessed by patients' self-reported Hurley stage, since patients with HS are capable of self-assessing their Hurley stage with a good correlation with physician assessment [18].

Disease activity was assessed by Patients' Global Assessment (PtGA), including five categories (inactive, very low, low, mild and severe) [19], and intensity of symptoms by Numeric Rating Scales (NRS) [20]. These scales show the subjective impact of the disease on patients, with equal or greater importance than objective scales.

SD was evaluated with a NRS for HS impact on sex life, in which participants were asked to measure from 0 to 10 how much the disease affects their sex life. This scale reflects the subjective suffering and distress caused by the disease to patients' sex lives. Its concordance with the Female Sexual Function Index-6 (FSFI-6) and the International Index of Erectile Function-5 (IIEF-5), two validated questionnaires that explore female sexual dysfunction and erectile dysfunction respectively, was also assessed.

#### *2.3. Statistical Analysis*

Statistical analyses were performed using JMP version 9.0.1 (SAS institute, Inc., Cary, NC, USA). When there were missing data in any of the variables of interest, patients were excluded from the study. When missing data were found in other variables, they were imputed. To explore the characteristics of the sample, descriptive statistics were used. Continuous variables were expressed as means and standard deviations. Qualitative variables were expressed as absolute and relative frequencies.

The main outcome of interest was SD, measured by the NRS for HS impact on sex life. To explore possibly associated factors, simple linear regression was used for continuous variables, Student's t-test for dichotomous variables, and one-way analysis of variance for nominal variables with two or more categories (Levene's test was used to assess the equality of variances, standardized residual plots to check independence and Normality was assumed because of the sample size). Significantly associated variables (*p* < 0.05) or those showing trends towards statistical significance (*p* < 0.20) were included in a multiple linear regression model to assess the factors associated with SD. Statistical significance was considered if *p* values were less than 0.05.

The correlation of NRS for HS impact on sex life with FSFI-6 and IIEF-5 was checked with simple linear regression. Student's *t*-test was used to assess differences between NRS for HS impact in sex life means in participants with and without sexual or erectile dysfunction according to the FSFI-6 or IIEF-5 scores, respectively. The cut-off point for sexual dysfunction using the NRS for HS impact on sex life was assessed by ROC curve analysis.

#### **3. Results**

#### *3.1. Baseline*

Three hundred and ninety three participants answered the questionnaire. Seven of them filled out the questionnaire incompletely, so the final sample consisted of 386 participants (319 (82.6%) from Spain, 57 (14.8%) from abroad, and 10 (2.6%) did not provide their country of residence). The ratio of women to men was 3.8:1 (306 (79.27%) women and 80 (20.73%) men). Their socio-demographic characteristics and comorbidities are shown in Table 1; current smoking was higher among men, body mass index was 1.5 greater in women, and the prevalence of diabetes mellitus type II and antidepressant consumption was higher among women, but these differences did not reach statistical significance. HS baseline characteristics are shown in Table 2. Age of onset was earlier in women (19.09 ± 7.1 vs. 23.57 ± 9.45, *p* < 0.0001), with a medium diagnosis delay of 11.23 ± 9.55 in women and 8.86 ± 9.13 in men. The groin was the location most affected in women, either by active lesions (65.7%) or scars (57.2%). In men, groin was the location more frequently affected by active lesions (53.8%), and axilla by scars (47.5%). Genitals were affected by active lesions in 111 (36.3%) of women and in 31 (38.8%) of men, and by scars in 82 (26.8%) of women and in 28 (35%) of men. The presence of active lesions in the perianal region (35 (43.8%) vs. 50 (16.3%), *p* < 0.0001) and on the buttocks (35 (43.8%) vs. 95 (31%), *p* <0.05) were higher among men, while the breast region was more frequently affected in women (90 (29.4%) vs. 2 (2.5%), *p* < 0.0001).


**Table 1.** Socio-demographic characteristics and comorbidities.

Continuous variables are expressed as means ± standard deviation and qualitative variables as absolute (relative) frequencies. BMI: body mass index. HBP: high blood pressure. DM2: diabetes mellitus type 2.




**Table 2.** *Cont.*

Continuous variables are expressed as means ± standard deviation and qualitative variables as absolute (relative) frequencies. PtGA: Patient's Global Assessment; values range from 1 (inactive disease) to 5 (severe disease). NRS: Numeric Rating Scale; values range from 0 (no symptoms) to 10 (maximum intensity of symptoms).

#### *3.2. Sexual Distress and Related Factors in Patients with Hidradenitis Suppurativa*

The mean NRS for HS impact on sex life was 7.24 (2.77) in women and 6.39 (3.44) in men (*p* < 0.05). Results from univariate analysis of factors possibly related to NRS for HS impact on sex life are shown in Table 3.

**Table 3.** Univariate analysis of factors associated with sexual distress in patients with HS.



**Table 3.** *Cont.*

*p*-values of variables significantly associated are marked with \* PtGA: Patient's Global Assessment; values range from 1 (inactive disease) to 5 (severe disease). NRS: Numeric Rating Scale; values range from 0 (no symptoms) to 10 (maximum intensity of symptoms). 81

No x = 7.62 (0.3)

Variables that were significantly associated or showed trends towards statistical significance (*p* < 0.20) were included in the multiple linear regression model, whose results are shown in Table 4. Variables significantly associated with SD were sex, with a higher risk in females, the presence of active lesions in the groin and genitals and NRS for pain and unpleasant odor; being in a stable relationship was an important protector factor for SD. Current smoking, PtGA, time under medical attention and treatment with adalimumab showed trends toward statistical significance.


**Table 4.** Multivariate analysis of factors associated with sexual distress in patients with HS.

*p* values of variables significantly associated are marked with \* PtGA: Patient's Global Assessment; values range from 1 (inactive disease) to 5 (severe disease). NRS: Numeric Rating Scale; values range from 0 (no symptoms) to 10 (maximum intensity of symptoms).

#### *3.3. Correlation between NRS for HS Impact on Sex Life and FSFI-6*/*IIEF-5 Scores*

Scores from NRS for HS impact on sex life and FSFI-6 showed a negative correlation (<sup>β</sup> <sup>=</sup> <sup>−</sup>0.15 <sup>±</sup> 0.02, *r*<sup>2</sup> = 0.16, *p* < 0.0001), indicating a good concordance between both questionnaires. Scores from NRS for HS impact on sex life and IIEF-5 also showed a negative correlation (<sup>β</sup> <sup>=</sup> <sup>−</sup>0.21 <sup>±</sup> 0.05, *r*<sup>2</sup> = 0.15, *p* < 0.001). The mean score on the NRS for HS impact on sex life was 8.27 ± 0.21 in women with sexual dysfunction, and 6.16 ± 0.21 in women without sexual dysfunction (*p* < 0.0001). In men, the mean score on the NRS for HS impact on sex life was 7.31 ± 0.47 in those with erectile dysfunction, and 5 ± 0.58 in those without erectile dysfunction (*p* < 0.01).

In women, a score of 8 or more on the NRS for HS impact on sex life was indicative of sexual dysfunction according to FSFI-6 scores, with a sensitivity of 73% and a specificity of 64% (Figure 1). In men, a score of 9 or more on the NRS for HS impact on sex life was indicative of erectile dysfunction according to IIEF-5 scores, with a sensitivity of 52% and a specificity of 81% (Figure 2).

**Figure 1.** ROC curve analysis for comparison between scores of NRS of HS impact on sex life and FSFI-6.

**Figure 2.** ROC curve analysis for comparison between scores of NRS of HS impact on sex life and IIEF-5.

#### **4. Discussion**

To our best knowledge, this is the largest cross-sectional study about the impact of HS on sexuality. Socio-demographic and disease characteristics did not differ from those previously reported in the literature, and were representative of the general HS population [21–28].

The mean NRS score for HS impact on sex life was significantly higher in women, which tallies with previous research that indicates higher sexual distress in women than in men with HS [29] or psoriasis [30]. These differences have been associated with cultural aspects and differences in emotional and neuroendocrine responses to disfigurement, and with the earlier onset of HS in women (4.5 years earlier in our sample) [29]. A higher prevalence of lesions at the lower abdomen has also been posed as a reason for this higher distress in women [29], but in our sample we only observed more involvement below the abdomen in the groin.

Although in psoriasis the involvement of the anogenital area has been related to sexual dysfunction [31–33], in HS anogenital involvement has been related to a reduction in quality of life [2,34], but there are no locations related to sexual dysfunction or to sexual distress [29,34]. In our investigation, we found an association between active lesions in the groin and genitals and SD, so a properly medical/surgical intervention at this level could turn into a better sexual life. In previous research about sexual health in patients with HS, samples were taken from hospital departments [29,34] and from a patient's association [34], and there were no important differences in patients' baseline characteristics, with the exception of a more prevalent Hurley III stage in our sample. Therefore, these findings were probably made possible due to the larger size of our sample.

Moreover, subjective symptoms caused sexual distress. The intensity of pain and unpleasant odor were related with higher scores on NRS for HS impact on sex life. This association may be due to factors directly related to the nature of the sexual act and/or to psychological factors that could be related to disease activity [2], highlighting the importance of symptom management to improve sexual health in patients with HS. Other factors such as antidepressant or benzodiazepine use were not statistically associated with SD, suggesting that SD is directly related to organic symptoms.

The absence of a stable relationship was not associated with sexual dysfunction in previous research [29,34]. Nevertheless, we observed that the presence of a stable partner was importantly related to lower SD. Since having a partner is associated with less self-consciousness and less orgasm difficulty in both men and women [35], probably feelings of shame, distrust, shyness and rejection influence SD, which could be lessened by the trust built in a relationship.

There were other factors that showed trends toward statistical significance in the multiple linear regression model: (1) PtGA, pointing to the importance of disease activity in sexual distress and the need to control the inflammatory load; (2) current smoking, because it is related to greater disease activity, since it favors follicular occlusion, a proinflammatory state with activation of neutrophils and Th17 lymphocytes, induces biofilm formation and suppresses notch signaling, among other effects [36]; moreover, smoking cessation is associated with clinical improvement [37]; (3) time under medical care, since it reflects time of disease evolution, with cumulative life course impairment [38]; and (4) Treatment with adalimumab, probably because in our sample it is a better predictor of severity than Hurley, since the patients treated with adalimumab are the most severe.

Finally, we found a good correlation between the scores on NRS for HS impact on sex life and those of FSFI-6 and IIEF-5, which indicates an association between the subjective and objective involvement of the sexual sphere in participants. However, despite this concordance, the ROC curve analyses revealed that the NRS for HS impact on sex life was not a good tool to assess sexual dysfunction in women or erectile dysfunction in men, because the cut-off points reach neither an acceptable sensitivity nor specificity. It is important to the clinician to distinguish between sexual distress and sexual dysfunction, because the first reveals the suffering of the subject, whereas the second may mean a worse sexual experience for both members of the stable/sporadic relationship. Therefore, clinicians should assess both aspects when patients' sexuality is addressed.

There are some methodological weaknesses in our study: (1) A possible selection bias, since it only represents patients in contact with support groups and Internet access. The elderly, who may use the Internet less frequently, or those with low sociocultural status or fear of new technologies, could be under-represented [39]. Moreover, people already concerned about sexual problems may have been more likely to answer the questionnaire. Nonetheless, the baseline characteristics of our sample did not differ from those previously reported in the literature, either in hospital-based or population-based studies. Given the scarcity of information about HS and sexuality we consider that this study is a good introduction to the problem, and could lay the foundation for future research. (2) A possible classification bias, because it was an online questionnaire and HS diagnosis could not be confirmed; HS characteristics were also self-referred. Nevertheless, an informed population can properly identify HS, because of its apparent and distinctive clinical manifestations. Since a patients' association hosted the questionnaire, it is expected that the participants did suffer from the disease.

#### **5. Conclusions**

This is the largest cross-sectional study about HS and sexuality. We have observed important sexual distress in patients with HS. Factors related to SD were female sex, the presence of active lesions in the groin and genital areas, and the intensity of pain and unpleasant odor. Being in a stable relationship has been an important protector factor against SD. Regarding these results, it seems that SD in HS patients is due, at least in part, to disease symptoms and active lesions in specific locations, emphasizing the importance of proper control of the disease based on management guidelines to improve their sexual health. Women and single patients are more likely to suffer from sexual distress, so special medical care should be given to them.

**Author Contributions:** Conceptualization, C.C.-B. and A.M.-L.; methodology, C.C.-B. and A.M.-L.; software, A.M.-L.; validation, C.C.-B., R.R.-V. and A.M.-L.; formal analysis, C.C.-B.; investigation, C.C.-B. and A.M.-L.; resources, A.M.-L.; data curation, C.C.-B.; writing—original draft preparation, C.C.-B.; writing—review and editing, R.R.-V. and A.M.-L.; visualization, C.C.-B., R.R.-V. and A.M.-L.; supervision, R.R.-V. and A.M.-L.; project administration, A.M.-L.

**Acknowledgments:** We would like to thank José Juan Jiménez Moleón for his contribution to the development of this study; Charlotte Bower, for improving the English of this manuscript; the Spanish hidradenitis suppurativa patients' association (ASENDHI) for their help and valuable collaboration to develop the study, as well as for hosting the questionnaire; and all the patients who have participated in this survey. The results of this study are part of Carlos Cuenca-Barrales' PhD.

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

#### **References**


© 2019 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/).

### *Article* **Predictors of Sexual Dysfunction in Veterans with Post-Traumatic Stress Disorder**

**Marina Letica-Crepulja 1,2,\*, Aleksandra Stevanovi´c 1,2,3, Marina Protuđer 4, Božidar Popovi´c 5, Darija Salopek-Žiha <sup>5</sup> and Snježana Vondraˇcek <sup>5</sup>**


Received: 18 February 2019; Accepted: 27 March 2019; Published: 29 March 2019

**Abstract:** Background: The problems in sexual functioning among patients with post-traumatic stress disorder (PTSD) are often overlooked, although scientific research confirms high rates of sexual dysfunctions (SD) particularly among veterans with PTSD. The main objective of this study was to systematically identify predictors of SD among veterans with PTSD. Methods: Three hundred veterans with PTSD were included in the cross-sectional study. The subjects were assessed by the Mini-International Neuropsychiatric Interview (MINI) and self-report questionnaires: PCL-5, i.e., PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) with Criterion A, International Index of Erectile Function (IIEF), Premature Ejaculation Diagnostic Tool (PEDT), and Relationship Assessment Scale (RAS). Several hierarchical multiple regressions were performed to test for the best prediction models for outcome variables of different types of SD. Results: 65% of participants received a provisional diagnosis of SD. All tested prediction models showed a good model fit. The significant individual predictors were cluster D (Trauma-Related Negative Alterations in Cognition and Mood) symptoms (for all types of SD) and in a relationship status/relationship satisfaction (all, except for premature ejaculation (PE)). Conclusions: The most salient implication of this study is the importance of sexual health assessment in veterans with PTSD. Therapeutic interventions should be focused on D symptoms and intended to improve relationship functioning with the aim to lessen the rates of SD. Psychotropic treatment with fewer adverse sexual effects is of utmost importance if pharmacotherapy is applied. Appropriate prevention, screening, and treatment of medical conditions could improve sexual functioning in veterans with PTSD.

**Keywords:** post-traumatic stress disorder; sexual dysfunction; veterans; predictors

#### **1. Introduction**

The problems in sexual functioning among patients with post-traumatic stress disorder (PTSD) are often overlooked clinically and receive little attention in research. However, an increasing body of scientific research regarding sexual dysfunctions (SD) among veterans who were exposed to military trauma confirms much higher rates of problems in sexual functioning among veterans with PTSD than in those without PTSD or in adults without exposure to military trauma [1–5]. The rates of SD differ across the studies, mainly because of methodological differences. Systematic reviews reported a prevalence of SD between 8.4% and 88.6% among male veterans with PTSD [3,5]. Persons with PTSD, compared with similarly exposed survivors without it, have an increased risk of SD implying that PTSD, rather than trauma exposure per se, is the more proximal antecedent to sexual problems [3,6–10]. Studies revealed correlation of PTSD with a variety of impairments in the specific domains of sexuality (desire, arousal, orgasm, resolution) [1–7]. On the other hand, the specific PTSD symptoms or PTSD symptom clusters may influence the prevalence of SD unevenly. The emotional numbing and avoidance cluster, for example, appeared to be intimately tied to impairment in sexual functioning and higher level of sexual anxiety [2,11,12].

#### *1.1. Predictors of Sexual Dysfunction in Veterans with PTSD*

Only a few studies and systematic reviews have addressed the possible predictors that have an impact on sexual functioning in the population of veterans with or without PTSD. Considering the relationship between overall PTSD symptom severity and SD, studies revealed conflicting results [5]. Particular PTSD clusters and symptoms have been studied, and it was hypothesized that autonomic arousal, anger/hostility [13], emotional numbing/avoidance symptoms [2,11,12], and chronic autonomic arousal and intrusive symptoms [3,14,15] were mostly associated with sexual problems among veterans with PTSD. Recent studies indicate that emotional numbing may impede intimacy and attachment, thus serving as a potential mechanism through which symptoms of PTSD may drive problems and predict SD in these patients. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [16], numbing symptoms (low positive emotions and negative emotional state) were included in the new D symptom cluster (Trauma-Related Negative Alterations in Cognition and Mood). These and other symptoms from this cluster, such as diminished interest or participation in significant activities, a feeling of detachment or estrangement from others, and guilt and shame, may impede sexual functioning in veterans with PTSD. SD is more common among veterans who are male, older, separated, divorced, or widowed, have lower annual income, mental health diagnoses—particularly PTSD—hypertension, and are prescribed psychiatric medications [1,4,17]. Returning combat veterans with SD have a reduced quality of life, decreased sexual intimacy, and increased health-care utilization [18]. PTSD is associated with impairments in romantic relationship satisfaction [19,20]. Recent research revealed that marital dissatisfaction is the factor that mediates the relationship between the number of PTSD symptoms and sexual dissatisfaction [21]. Considering the specific types of SD, age appeared to be the only significant predictor of erectile dysfunction; age, race, depression, and social support predicted self-reported sexual arousal problems; and race, combat exposure, social support, and avoidance/numbing symptoms of PTSD predicted self-reported sexual desire problems in male combat veterans seeking outpatient treatment for PTSD [2].

#### *1.2. Predictors of Sexual Dysfunction in the General Population*

Generally speaking, the predictors, risk, or etiological factors of SD can be separated in two groups: "organic" (such as diabetes, peripheral vascular disease or venous leaks, injury of the spinal cord, etc.) and "non-organic" (such as anxiety, depression, cultural taboos, ignorance, relationship problems, poor communication skills, etc.). However, there is substantial evidence indicating a multifactorial etiology of sexual function and dysfunction, meaning that the sexual response can be described as a complex interaction of psychological, interpersonal, social, cultural, physiological, and gender-influenced processes [22,23]. SD is strongly associated with certain health conditions and diseases, psychiatric disorders, medication or substance use, lack of knowledge, psychological or behavioral factors, relationship and cultural factors processes [23].

#### *1.3. Study Background*

More than 20 years after the Homeland War in Croatia (1991–1995), veterans still suffer from numerous health problems. Patients and/or health professionals may be reluctant to mention and discuss sexual symptoms [24], and a huge proportion of SD remains undiagnosed. Despite that, clinical observations and rising awareness have encouraged the recognition and assessment of SD in this patient group, and case reports [25] and research articles [26,27] regarding SD in veterans with PTSD in Croatia have been published.

The main objective of this research is to systematically identify predictors of SD among veterans with PTSD. The main hypothesis of the study is that SD are predicted by overall PTSD symptom severity and by severity of D symptom cluster (Trauma-Related Negative Alterations in Cognition and Mood).

#### **2. Experimental Section**

#### *2.1. Participants and Procedure*

Participants were male war veterans (*N* = 300) recruited from a pool of patients referred to the Regional Center for Psychotrauma (RCP) and Department of Psychiatry within the Clinical Hospital Center (CHC) Rijeka, the Referral Center for PTSD of the Ministry for Health of the Republic of Croatia (*N* = 250), and the Daily Hospital for PTSD and Department of Psychiatry within the General Hospital (GH) Našice for treatment. Most of the veterans participated in operations on different and almost all battlefields. Thirteen of those whom we approached refused to participate, while two patients did not complete the questionnaires.

Eligibility was determined by meeting diagnostic criteria for war-related PTSD as defined in DSM-5 [16]. Three patients were not eligible for the study as they did not meet the criteria for PTSD diagnosis. We continued recruiting patients until the number of 300 participants was reached. There were no differences in sociodemographic characteristics between those who refused to participate, those who did not complete the questionnaires, and those who were not eligible for the study.

The inclusion criteria for the study were: participation in the Homeland War as a soldier, experiencing at least one war-related traumatic event defined in the DSM-5 criteria for PTSD (personal experience of combat or exposure to a war zone), male gender, and age below 65. The exclusion criteria for the study were: active psychosis, moderate or high suicide risk measured by the Mini-International Neuropsychiatric Interview (MINI) for DSM-IV [28], and deformities, injury, or mutilation of the genital organs. None of the participants met the exclusion criteria.

Research consisted of two parts, i.e., a clinical interview and self-report questionnaires. The interviews were conducted by five psychiatrists and two psychologists from the two study sites. Sociodemographic data were collected through a questionnaire created for study purposes. The interviews and filling in of the questionnaires were usually completed in one or two sessions. The study was approved by the Ethics Committees of the Faculty of Medicine, University of Rijeka, CHC Rijeka, and GH Našice. Written informed consent was obtained from all participants after detailed information about the study was provided to them.

The study sample included a total of 300 male veterans. At the time of participation in the study, the majority of participants were in ambulatory treatment (66.8%), while other participants were involved in day-hospital treatment (19.3%) or club for PTSD (7.5%), or were hospitalized (6.4%). Table 1 provides further information on sample demographics.


**Table 1.** Sociodemographic characteristic and differences according to the presence of sexual dysfunction.

<sup>1</sup> *n* = 256, range 7–35; <sup>2</sup> *n* = 148; <sup>3</sup> *n* = 98.

#### *2.2. Measures*

#### 2.2.1. PTSD Checklist for DSM-5 (PCL-5) with Criterion A

The PCL-5 with Criterion A [29] is a self-report measure was revised to match the adapted DSM-5 criteria for PTSD. The interpretation of the PCL-5 should be made by a clinician. A PTSD diagnosis can be made provisionally considering items rated 2 = moderately or higher as a symptom endorsed according to the DSM-5 diagnostic rule (at least one B, one C, two D, and two E symptoms present). DSM-5 symptom cluster severity scores can be obtained by summing the scores for the items within a given cluster, i.e., cluster B (items 1–5), cluster C (items 6–7), cluster D (items 8–14), and cluster E (items 15–20). A total symptom severity score (range 0–80) can be obtained by summing the scores for each of the 20 items. Preliminary validation work was sufficient to make a cut-point score of 33, which was chosen for the purpose of this study [29]. Previous validation studies showed good psychometric properties for evaluating PTSD [30–33]. Cronbach's alpha in our study for clusters of symptoms ranged from 0.67 to 0.85, and to 0.89 for total PCL-5. The Criterion A measure was included in the assesment according the criteria of DSM-5 [16].

#### 2.2.2. The International Index of Erectile Function (IIEF)

IIEF [34] is a widely used, multi-dimensional self-report instrument for the evaluation of male sexual function over the last four weeks [34]. It consists of 15 questions grouped into five domains that assess erectile function (Q1,2,3,4,5,15), intercourse satisfaction (Q6,7,8), orgasmic function (Q9,10), sexual desire (Q11,12), and overall satisfaction (Q13,14). Each item is rated from 1 (very low; almost never or never; extremely difficult) to 5 (very high; almost always or always; not difficult). Scores for domains are calculated as the sum of the answers, with lower scores indicating worse functioning. The score for erectile function can be calculated and used to classify the severity of dysfunction as severe, moderate, mild, or no dysfunction. For other domains, a higher score indicates better function. The IIEF meets psychometric criteria for test reliability and validity, has a high degree of sensitivity and specificity, and correlates well with other measures of treatment outcome [34–37]. Cronbach's alpha was 0.96 for erectile function, 0.91 for orgasmic function, 0.89 for sexual desire, and 0.91 for intercourse satisfaction and overall satisfaction.

#### 2.2.3. Premature Ejaculation Diagnostic Tool (PEDT)

PEDT [38,39] is a self-report instrument for the evaluation of the presence and severity of premature ejaculation. Each PEDT item is rated from 0 (not difficult at all; almost never or never; not at all) to 4 (extremely difficult; almost always or always, extremely), with a higher score indicating more difficulties with premature ejaculation. Previous validation studies have shown satisfactory feasibility, reliability, and validity of the PEDT [38,39]. Cronbach's alpha for PEDT scale in our study was 0.87.

#### 2.2.4. Male Sexual Dysfunction Criteria

The DSM-5 [16] classification recognizes four male sexual dysfunctions: delayed ejaculation (DE), erectile disorder (ED), male hypoactive sexual desire (HSD), and premature (early) ejaculation. To be diagnosed with SD, the symptoms must be present for at least six months, cause significant distress, and not be caused exclusively by a non-sexual mental disorder, significant relationship distress, medical illness, or medication. Also, these diagnoses are applicable to men who engage in non-vaginal sexual activity, but unfortunately, the specific duration criteria remain unknown [16]. For the purposes of this study, the following criteria were applied for a provisional diagnosis:


#### 2.2.5. Relationship Assessment Scale (RAS)

The RAS [40,41] is a seven-item measure of global relationship satisfaction. Responses are on a five-point Likert scale, and either the total or the average score can be used in the interpretation. Average scores range from 1 to 5; total scores range from 7 to 35 (used in this study). Higher scores indicate greater relationship satisfaction. The reliability and validity of the English RAS have been established [41]. Cronbach's alpha in our study was 0.87.

#### 2.2.6. Mini-International Neuropsychiatric Interview (MINI)

Comorbid psychiatric disorders were diagnosed using the Croatian version of MINI for DSM-IV [28]. It is a brief and valid structured clinical interview meeting the need for a short but accurate structured psychiatric interview for multicenter clinical trials and epidemiology studies, to be used as a first step in outcome tracking in nonresearch clinical settings. This interview enables researchers to assess the 17 most common psychiatric disorders in DSM-IV.

#### 2.2.7. Anatomical Therapeutic Chemical (ATC) Classification System

Self-reported data about drug consumption are classified in accordance with the ATC classification [42]. In brief, the ATC system classifies therapeutic drugs. The purpose of the system is to serve as a tool for drug utilization research in order to improve the quality of drug use. In the ATC classification system, the drugs are divided into different groups according to the organ or system on which they act and their chemical, pharmacological, and therapeutic properties. Drugs are classified into five different groups.

#### *2.3. Data Analysis*

#### 2.3.1. Data Analysis Plan

The aim of the study was to assess the predictive models of several sexual dysfunctions in male veterans with PTSD. Average score of erectile function, orgasmic function, sexual desire, intercourse satisfaction, overall satisfaction (all measured by IIEF), and premature ejaculation (measured by PEDT) were the outcome variables. Prediction variables were characteristics identified as relevant for sexual dysfunction in previous studies. Two sets of hierarchical regression analyses were executed for each of the sexual functions (one without and one with relationship satisfaction) in order to assess the best models for the overall sample of veterans with PTSD and for the subset of veterans in relationship. In order to control for covariances, predictor variables were entered in the following steps/models: (1) sociodemographic variables, (2) comorbid disorders (psychiatric and others), (3) medication used (psychotropic and other drugs), (4) variables related to PTSD (deployment duration and PTSD symptoms), and (5) relationship satisfaction (subset sample of veterans in relationship). The exclusion criterion for dichotomous predictors was set to 10 or less events per variable [43]. The inclusion criterion for a prediction variable was a significant association with the outcome variable.

#### 2.3.2. Statistical Analysis

Statistical analysis was performed with Statistica software, version 12 (Dell Inc. Inc., Tulsa, OK, USA). Data are presented as *N* (%) or M (sd). Chi-square tests for categorical variables and independent sample *t*-tests for continuous variables were used to compare veterans with or without provisional diagnosis of sexual dysfunction. Pearson and Spearman correlation coefficients were calculated between sexual functions and the variables of interest. Several hierarchical multiple regressions were performed to test for the best prediction models for the outcome variables of erectile function, orgasmic function, sexual desire, intercourse satisfaction, overall satisfaction, and premature ejaculation. All models were controlled for basic assumptions. Two issues with multicollinearity were encountered, i.e., between cluster D and cluster E symptoms with overall PTSD symptoms, and between in-a-relationship status and relationship satisfaction in the subset sample. Overall PTSD symptoms were excluded from both sets of samples, and the in-a-relationship status variable from the subset sample. Missing values were controlled for listwise. Probability significance was set to *p* ≤ 0.05.

#### **3. Results**

#### *3.1. Sociodemographic Data*

Sociodemographic data for the overall sample are presented in Table 1.

#### *3.2. Trauma Exposure and PTSD*

The average duration of active participation in the Homeland war was 30 (19.516) months, ranging from 1 month to 70 months.

Twenty-three percent of participants had sought psychiatric help in the period from 1991 to 1995, while the war was ongoing. The average intensity for overall PTSD symptoms was 57.5 (10.92) within the range of 33 to 80. The average intensity of B symptoms was 15 (3.25), of C symptoms 6.2 (1.47), of D symptoms 18.7 (5.15), and of E symptoms 17.7 (3.90).

#### *3.3. Prevalence of SD and Association with Sociodemographic Data and PTSD*

The average score for erectile function was 16 (9.71), which relates to moderate dysfunction. The average score for orgasmic function was 5.8 (3.31) (theoretical maximum = 8), for sexual desire 5.8 (2.47) (theoretical maximum = 8), for intercourse satisfaction 6.51 (4.71) (theoretical maximum = 12), and for overall satisfaction 6.3 (2.44) (theoretical maximum = 8). The average score for PEDT was 7.43 (5.14) within the range of 0 to 20.

According to provisional criteria for male sexual dysfunction (described in methodology), the following rates were found: DE 124 (44%, *n* = 282), ED 134 (46.2%, *n* = 290), HSD 128 (44.6%, *n* = 287), and PE 59 (21.3%, *n* = 277). Overall, on the basis of self-reported data, 98 (35.1%) of veterans with PTSD did not meet, while 181 (64.9%) participants met provisional criteria for at least one male SD in the last month. Out of possible four SD, one SD had 49 (17.6%) participants, two SDs had 36 participants (12.9%) participants, three SD had 80 participants (28.7%), and four SD had 16 participants (5.7%).

As presented in Table 1, participants in a relationship and participants with medium income were less likely to have a provisional diagnosis of SD. Participants who met the provisional diagnosis of SD were significantly less satisfied with their relationship compared to participants without SD. Veterans with SD had significantly greater severity of cluster D, cluster E, and overall symptoms of PTSD. They did not differ for duration of deployment or for cluster B and cluster C symptoms.

Prevalence of comorbid disorders and drug use and association with SD are presented in Supplementary Materials: Material S1.

#### *3.4. Prediction Models of Sexual Dysfunctions among War Veterans with PTSD*

Predictor variables for each model (i.e., sexual function) were selected on the basis of the following criteria: variables with events greater than 10 and significant correlation with the outcome variable (Table S2). However, some variables were included regardless, such as age and all clusters of PTSD symptoms. Also, analysis showed great correlation coefficients between overall PTSD symptoms intensity and cluster D and E symptom intensity (variance inflation factor (VIF) > 8). Because of the multicollinearity issues, overall PTSD symptoms were not included in the models. The variable "in a relationship" had high multicollinearity with relationship satisfaction (VIF > 8), and, therefore, only relationship satisfaction was included in the models for the subset of veterans in a relationship. The final steps for all tested models are presented in Supplementary material (Tables S3 and S4). An overview of individual significant predictors for each sexual function is given in Tables 2 and 3.



\* *<sup>p</sup>* ≤ 0.05; \*\* *<sup>p</sup>* ≤ 0.01; <sup>1</sup> Alcohol use disorders; <sup>2</sup> Hypertension, essential; significant values are in bold.


**Table 3.** Overview of individual significant predictors in the final step of hierarchical regression analysis for the subset sample of veterans in a relationship.

\* *<sup>p</sup>* ≤ 0.05; \*\* *<sup>p</sup>* ≤ 0.01; <sup>1</sup> Hypertension, essential; <sup>2</sup> Hyperplasia of prostate; <sup>3</sup> Relationship satisfaction; significant values are in bold.

#### 3.4.1. Erectile Function

The initial model tested for erectile function included age, low income, medium income, not married, married, "in a relationship" status (Model 1: *R*<sup>2</sup> = 0.134, *F* = 7.060, *p* < 0.001). In the second step ongoing major depressive episode (MDE), panic disorder lifetime, essential hypertension, and hyperplasia of prostate (Model 2: *R*<sup>2</sup> = 0.181, *F* = 5.816, *p* < 0.001) were included; in the third step, use of antidepressants, hypnotics, and sedatives (Model 3: *R*<sup>2</sup> = 0.184, *F* = 5.514, *p* < 0.001) was added; in the fourth step, war deployment in months and cluster B, C, D, and E symptoms were added (model 4: *R*<sup>2</sup> = 0.257, *F* = 5.651, *p* < 0.001). Significant predictors did not change through the models. The final model explained 25.7% of the variance of erectile function. Variables with significant independent contribution were being in a relationship, having essential hypertension, and severity of D cluster symptoms (Table 2).

In the subset of participants in a relationship, Model 1, containing age, low income, medium income, not married, and married, was not significant, since the variable relationship status was removed. Model 2 accounted for 9.3% (*F* = 2.200, *p* = 0.019), Model 3 for 9.98% (*F* = 2.097, *p* = 0.22), and Model 4 for 20.1% (*F* = 3.273, *p* < 0.001) of the variance of erectile function. The final model with relationship satisfaction added explained 27.9% of the variance of erectile function (*F* = 5.457, *p* < 0.001). Significant individual predictors were having essential hypertension, severity of cluster D symptoms, and relationship satisfaction. (Table 3).

#### 3.4.2. Orgasmic Function

The initial model tested for orgasmic function included age, higher education, low income, medium income, married, and "in a relationship" status (Model 1: *R*<sup>2</sup> = 0.100, *F* = 4.619, *p* < 0.001). In the second step, ongoing MDE, panic disorder lifetime, essential hypertension, hyperplasia of prostate, and disorders of lipoprotein metabolism were added (Model 2: *R*<sup>2</sup> = 0.165, *F* = 4.397, *p* < 0.001); in the third step, use of antidepressants, hypnotics, and sedatives (Model 3: *R*<sup>2</sup> = 0.189, *F* = 4.751, *p* < 0.001) was included; in the fourth step, cluster B, C, D, and E symptoms (Model 4: *R*<sup>2</sup> = 0.248, *F* = 4.628, *p* < 0.001) were added. Higher education level was a significant individual contributor until psychotropic medication was introduced in the third step. The final model explained 24.8% of the variance of orgasmic function. Significant independent predictors were being in a relationship, use of antidepressants, having hypertension, and severity of cluster D symptoms (Table 2).

In the subset of participants who were in a relationship, Model 1 was not significant and accounted for 5.4% variance of orgasmic function. Model 2 (*R*<sup>2</sup> = 0.121, *F* = 2.601, *p* = 0.004), Model 3 (*R*<sup>2</sup> = 0.152, *F* = 3.098, *p* < 0.001), and Model 4 (*R*<sup>2</sup> = 0.244, *F* = 3.839, *p* < 0.001) were all significant. The final model explained 29.5% of the variance of orgasmic function (*F* = 4.679, *p* < 0.001). Significant individual predictors were: use of antidepressants, presence of essential hypertension, severity of cluster D symptoms, and relationship satisfaction (Table 3). There was no significant change in the significance of individual predictors through the models.

#### 3.4.3. Sexual Desire

The initial model for sexual desire included age, low and medium income, and "in a relationship" status (Model 1: *R*<sup>2</sup> = 0.055, *F* = 4.131, *p* = 0.003). In the second model, alcohol use disorder (AUD) was added (Model 2: *R*<sup>2</sup> = 0.074, *F* = 4.521, *p* = 0.001); in the third model, antidepressant use was included (Model 3: *R*<sup>2</sup> = 0.095, *F* = 4.901, *p* < 0.001); in the fourth model, cluster B, C, D, and E symptoms (Model 4: *R*<sup>2</sup> = 0.166, *F* = 5.482, *p* < 0.001) were added. All the models were significant, and there was no change in the significance of individual predictors. The final model explained 16.6% of the variance of sexual desire in the entire sample. Predictors with independent contribution were being in a relationship, presence of an AUD, use of antidepressant, and severity of cluster D symptoms (Table 2).

In the subset sample of veterans in a relationship, the sociodemographic variables entered did not significantly contribute to the variance of sexual desire (Model 1: *R*<sup>2</sup> = 0.035, *F* = 2.219, *p* = 0.068) Addition of AUD in step two (Model 2: *R*<sup>2</sup> = 0.054, *F* = 2.789, *p* = 0.018), antidepressant in step three (Model 3: *R*<sup>2</sup> = 0.081, *F* = 3.536, *p* = 0.002), and clusters of PTSD symptoms in step four (Model 4: *R*<sup>2</sup> = 0.172, *F* = 4.930, *p* < 0.001) significantly increased the variance of sexual desire. The final model which included relationship satisfaction explained 19.6% of sexual desire in veterans in a relationship (*F* = 5.227, *p* < 0.001). As in the total sample, predictors with significant independent contribution were use of an antidepressant and severity of cluster D symptoms, but not AUD. A significant contributor was also relationship satisfaction (Table 3). The significant predictors did not change through the models.

#### 3.4.4. Intercourse Satisfaction (IS)

The initial model for the intercourse satisfaction consisted of age, low income, medium income, not married, divorced, married and relationship status (Model 1: *R*<sup>2</sup> = 0.139, *F* = 6.748, *p* < 0.001). In the second model, ongoing MDE, other anxiety disorders, essential hypertension, and hyperplasia of prostate were added (Model 2: *R*<sup>2</sup> = 0.168, *F* = 5.287, *p* < 0.001); in the third (final) model, war deployment in months, cluster B, C, D, and E symptoms (Model 3: *R*<sup>2</sup> = 0.251, *F* = 5.714, *p* < 0.001) were included. There was no change in individual predictors through the models, and the final model explained 25.1% of intercourse satisfaction in veterans with PTSD. The identified significant predictors were being in a relationship, presence of essential hypertension, and severity of cluster D symptoms (Table 2).

In the subset sample of veterans who were in a relationship, the final model accounted for 31.6% of intercourse satisfaction (*F* = 7.382, *p* < 0.001). All tested models were significant (Model 1: *R*<sup>2</sup> = 0.067, *F* = 2.982, *p* = 0.008; Model 2: *R*<sup>2</sup> = 0.102, *F* = 2.788, *p* = 0.003; Model 3: *R*<sup>2</sup> = 0.188, *F* = 3.987, *p* < 0.001). The significance of predictors did not change through the models. In contrast to the overall sample, essential hypertension was not a significant predictor of IS among veterans in a relationship. Severity of D cluster symptoms and relationship satisfaction were independent significant contributors (Table 3).

#### 3.4.5. Overall Satisfaction

In the first model for overall satisfaction, the following variables were entered: age, low income, medium income, and "in a relationship" status (Model 1: *R*<sup>2</sup> = 0.061, *F* = 4.474, *p* = 0.002). In the next step, comorbid diseases, ongoing MDE, panic disorder lifetime, other anxiety disorders, AUD, essential hypertension, and hyperplasia of prostate were entered (Model 2: *R*<sup>2</sup> = 0.140, *F* = 4.371 *p* < 0.001); in the third step, use of antidepressants (Model 3: *R*<sup>2</sup> = 0.145, *F* = 4.125, *p* < 0.001) was included; in the fourth step, clusters B, C, D, and E symptoms (Model 4: *R*<sup>2</sup> = 0.210, *F* = 4.652, *p* < 0.001) were added. Recurrent panic disorder was a significant predictor until PTSD symptoms were entered in the last step. The final model explained 21% of the variance of overall satisfaction in veterans with PTSD. Significant individual predictors of overall satisfaction were being in a relationship, presence of an AUD, presence of essential hypertension, and severity of cluster D symptoms (Table 2).

All the models tested for overall satisfaction among veterans with PTSD who were in a relationship were significant (Model 1: *R*<sup>2</sup> = 0.038, *F* = 3.188, *p* = 0.024; Model 2: *R*<sup>2</sup> = 0.117, *F* = 3.457, *p* < 0.001; Model 3: *R*<sup>2</sup> = 0.136, *F* = 3.717, *p* < 0.001; Model 4: *R*<sup>2</sup> = 0.215, *F* = 4.570, *p* < 0.001). The "other anxiety disorders" variable was a significant predictor until PTSD symptoms were entered in the fourth step. The final model in the subset sample explained 38.4% of the variance of overall satisfaction (*F* = 9.651, *p* < 0.001). Significant individual predictors were: presence of hyperplasia of prostate, use of an antidepressants, severity of cluster D symptoms, and relationship satisfaction (Table 3). It is important to note that relationship satisfaction by itself (β = 0.435) explained most of the variance of overall sexual satisfaction.

#### 3.4.6. Premature Ejaculation

In the first model of premature ejaculation in the overall sample, the following sociodemographic variables were entered: age, not married, married, and "in a relationship" status (Model 1: *R*<sup>2</sup> = 0.054, *F* = 3.779, *p* < 0.01). In Model 2, diabetes mellitus (DM) was added (*R*<sup>2</sup> = 0.067, *F* = 3.795, *p* < 0.01), and cluster B, C, D, and E symptoms were added in Model 3. Significant individual predictors were DM and severity of cluster D symptoms (Table 2).

Similar findings were reported in the subset sample of veterans in a relationship, as the final model contributed to 10.7% of the variance of premature ejaculation (*F* = 3.019, *p* < 0.001) with the independent significant contributors DM and cluster D symptoms (Table 3). Model 1, containing sociodemographic variables (*R*<sup>2</sup> = 0.016, *F* = 1.257, *p* = 0.290), and Model 2 (*R*<sup>2</sup> = 0.035, *F* = 2.079, *p* = 0.054), containing comorbid diseases, did not contribute significantly to the variance of premature ejaculation. Model 3, which included clusters of PTDS symptoms, was significant (*R*<sup>2</sup> = 0.104, *F* = 3.277, *p* < 0.001). Relationship satisfaction added in the final model did not alter significantly the variance explained.

#### **4. Discussion**

To the best of our knowledge, the present study is the first to suggest patterns of association of PTSD with different types of SD and to determine the predictors of this relationship. The results of the study support the main hypothesis that SD in veterans with PTSD are predicted by the severity of the D cluster of PTSD symptoms. The second part of the hypothesis that states SD are predicted by overall PTSD symptom severity is partially supported. We found that veterans with SD had significantly higher PTSD symptom scores than veterans without SD. Furthermore, overall PTSD symptom severity was significantly correlated with all types of SD (DE, ED, HSD, and PE) as well as intercourse satisfaction (IS) and overall satisfaction (OS). Analysis revealed significant multicollinearity of this predictor with D symptoms of PTSD, which implies that the association of PTSD symptom severity with SD is mediated and mostly depends on the quantity and severity of trauma-related negative alterations in cognition and mood. Previous studies found high rates of SD among male veterans with PTSD [1–5]. The results of our study are consistent with the scarce but increasing body of research that indicates that the severity of PTSD measured by overall scores on PTSD scales is not a significant predictor of SD in veterans with PTSD [2,5,11,12].

Beside the prevalence and correlation of SD with PTSD, it is important to understand the background of this relationship. A high score of D symptoms (Trauma-Related Negative Alterations in Cognition and Mood) appears to be the most prevalent predictor of SD among veterans with PTSD, emerging as a significant predictor of all types of SD (DE, ED, HSD, PE) as well as of IS and OS. D cluster includes three new symptoms according to the DSM-5 classification: negative expectations of self, others, or the world (replacing the sense of foreshortened future), persistent distorted blame of self or other for trauma, and pervasive negative emotional state. The presence of these symptoms and/or other symptoms from the D cluster, such as diminished interest or participation in significant activities, a feeling of detachment or estrangement from others, or a persistent inability to experience positive emotions, precludes a person's capacity to engage adequately in sexual behavior(s). As a result, D symptoms predict lower levels of satisfaction in sexual life. The current DSM-5 classification

embraces the four-factor model, as it provides a better representation of PTSD's latent structure than the tripartite model of DSM-IV [43–46], which has received extensive criticism [47]. Our findings in veterans are consistent with prior research demonstrating that avoidance/numbing symptoms of PTSD are strongly linked to self-reported problems in sexual functioning. Nunnink and colleagues found that self-reported symptoms of emotional numbing predicted a greater likelihood of endorsing sexual problems [11]. The results of another study that investigated predictors of ED and self-reported sexual problems among 150 male combat veterans seeking outpatient treatment for PTSD revealed, beside various demographic, physical, and psychosocial risk factors, a significant zero-order correlation between avoidance/numbing symptoms and SD [2].

Partner relationship is the next prominent predictor of SD in veterans with PTSD. Results in the overall sample revealed that being in a partner relationship reduces the risk of DE, ED, HSD, IS, and OS. Being in a relationship has no predictive value for PE. Analysis in the sample of participants who were in a partner relationship indicated that a low level of relationship satisfaction was a significant predictor of DE, ED, HSD, and IS and OS. Relationship satisfaction was not a significant predictor of PE. The association of PTSD with impairments in romantic relationship satisfaction has been previously reported [11,19,20]. A recent meta-analysis of 23 studies found an association between the emotional numbing and avoidance symptom cluster and parent, child, family, and marital/partner functioning problems [48]. Sexual functioning and relationship satisfaction are also robustly, positively correlated in many different samples across a variety of adult populations, including those who are dating [49,50], in long-term relationships [51], and married [52,53]. A lower level of relationship satisfaction in our study sample was an independent predictor of SD and was not mediated by the severity of any PTSD cluster. Sexual functioning is one of the essential domains of relationship functioning. Association between SD and quality of relationship is bidirectional and reciprocal. Relationship problems caused by family stressors, economic reasons, lifestyle, etc. inevitably affect sexual functioning. Problems in sexual functioning may have an impact on all other domains of a relationship. In the context of PTSD, the quality of a relationship also depends on the accommodation capacities of the partner for mutual acceptance, which is important for healthy sexual functioning. Additionally, PTSD may affect relationship and sexual functioning indirectly through changes of behavioral patterns. For example, insomnia and nightmares are less likely to have a direct impact on sexual functioning than numbing symptoms. On the other hand, these symptoms may lead to sleeping in separate beds, allowing or encouraging the rituals and avoidant behavior that lessen the quality of a relationship and sexual functioning. This finding implies that therapeutic efforts directed to promoting relationship satisfaction in veterans with PTSD could have a positive effect on sexual functioning in most of its domains. Interestingly, being in a relationship and relationship satisfaction are not significant predictors of PE. This finding could be explained by considering PE symptoms as more of an individual than a relational problem, which in turn is not worsened or maintained by disturbances in a partner relationship.

Antidepressant use is a significant predictor of the impairment of orgasmic functioning and sexual desire, i.e., veterans with PTSD that use antidepressants have increased risk for DE and HSD. Surprisingly, antidepressant utilization did not show predictive values for ED and OS. Adverse sexual effects are frequent with commonly prescribed psychotropic drugs and are usually underestimated [24,54]. The recent clinical guidelines highlight antidepressants as first-line pharmacotherapeutic agents in the management of PTSD [55,56]. In spite of increasing rates of drug utilization (80%) among veterans with PTSD [57], some studies revealed a marked inconsistency with the current guidelines for treatment of PTSD, particularly in the post-conflict settings [58]. In that context, our finding of antidepressant use as a significant predictor of DE and HSD is important, bearing in mind that 41.5% of our participants have DE and 45.4% have HSD. The findings are consistent with a meta-analysis which revealed increased rates of SD among patients in treatment with antidepressants [54]. Furthermore, higher rates of total and specific-treatment emergent SD and specific phases of dysfunction were found for drugs with a predominantly serotonergic action, including selective serotonin reuptake inhibitors (SSRIs) and serotonin and norepinephrine reuptake

inhibitors (SNRIs) [55,59]. Ejaculation-delaying effect of antidepressants on orgasmic function is, on the other hand, the basis for the use of either tricyclic antidepressant or SSRIs in treatment of PE. Among other medications from this pharmacological group, paroxetine has the most prominent ejaculation-delaying effect [60] caused by its impact on serotonergic receptors, cholinergic receptor blockade, and inhibition of nitric oxide synthase [61–64]. It is also supported by the results of this study, as antidepressants are not significant predictors of PE.

Arterial hypertension was a significant predictor of ED, DE, IS, and OS in the overall sample. It was a significant predictor of ED and DE in the sample of veterans in a relationship. These findings are consistent with those of numerous studies that emphasize high blood pressure as a risk factor for SD [65–67]. Actually, vasculogenic ED is considered part of a systemic vasculopathy and has a known relationship with cardiovascular risk factors such as hypertension, diabetes, dyslipidemia, and smoking [68]. A research that included 1255 male participants revealed that lower systolic and diastolic blood pressure were associated with better sexual functioning [67].

The significant predictor of PE in the overall sample and among participants in a relationship was DM. Patients with DM have higher rates of various SD directly related to the deleterious complications of their disease. [69–71]. DM is also indirectly related to SD through anxiety and depression that are often experienced by men with DM [72]. Of these, ED was most commonly reported [69–71]. Some studies reported higher rates of PE in patients with DM, indicating duration, severity, and poor metabolic control as the main risk factors for PE in diabetic patients. On the other hand, a close relationship between ED and PE exists. Some authors suggest that the longer the erectile problem, the worse the anxiety, and the more marked the PE [73]. Because of performance anxiety regarding their erectile reliability, patients could rush through an intercourse, with PE as a deleterious consequence [74].

AUD was a significant predictor of SD and OS in the overall sample and in those who patients were in a relationship. The results are consistent with the findings of a previous research and meta-analysis [65]. This finding is important in the context of populations of veterans with PTSD, as repeated heavy drinking is one of the common strategies to alleviate trauma symptoms that may lead to the development of AUD. The prevalence of AUD in PTSD is also high. For example, in the US, 42% of PTSD subjects met criteria for AUD diagnosis [75]. The prevalence of alcohol-induced sexual dysfunction is unclear, probably because of underreporting. Sexual disorders ranging from 8% to 95.2% have been reported in men with chronic alcohol use [76–80]. The common dysfunctions reported were lack of sexual desire [79,80], premature ejaculation [81,82], and erectile dysfunction [76,82–84].

Although the prevalence of the SD was not the main focus of this research, it is indicative that none of the participants reported being diagnosed with SD. Only one veteran with PTSD reported the utilization of a medication prescribed to treat ED (sildenafil). This finding is completely inconsistent with data from previous studies suggesting that SD is strongly related to PTSD, particularly war-related PTSD [1–5]. A backup check of medical records confirmed only one diagnosis of SD recorded in the study sample. Two widely used instruments for the assessment of the presence and severity of the different types of SD were applied with restrictive criteria for severity of SD symptoms, consistent with DSM-5 for diagnosis of SD (i.e., present in at least 75% of sexual activity occasions) in order to avoid over-diagnosing minor and potentially transient problems in sexual functioning. According to that criteria, the following rates were found: SD in 64.9% of patients, DE in 44%, ED in 46.2%, HSD in 44.6%, PE in 21.3%. The rates of SD differ across studies [6–10], mainly because of different methodological approaches. Predominantly, two methods for identifying SD have been used in research. In some studies, the estimation of SD diagnosis was based on reported patients' symptoms and problems in sexual functioning, with wide criteria for SD applied. In another study, the presence of SD was considered if SD diagnosis was recorded or medication for SD was used, which may be a more conservative approach. Both methods for identifying SD may be problematic. If we chose the second approach, we could conclude that veterans with PTSD in our sample had superior sexual functioning. Therefore, we chose the first approach, bearing in mind that self-reported symptoms in questionnaires

can be used only for an estimation and provisional diagnosis of SD. Clinical interviews are irreplaceable and necessary to sufficiently diagnose SD if they are conducted by well-trained personnel, who are also trained about social stigmatization. Conversely, they may contribute to underreporting biases arising from personal concerns about social stigmatization and lack of privacy, particularly in older or less educated participants [85]. The rates of SD in this study confirm that the complete absence of SD diagnosis in our clinical setting could not be a consequence of non-clinically significant problems among veterans. A dramatically higher self-reported prevalence of SD suggests a number of veterans may be choosing not to disclose problems in sexual functioning with their healthcare providers because of embarrassment, discomfort, or lack of knowledge about treatment possibilities.

#### *4.1. Strenghts*

This study was primarily designed to assess SD in the population of veterans with PTSD. Veterans with PTSD were included regardless of their relationship status, as even those not in a current romantic relationship may engage in sexual behavior and are often overlooked in studies. Data related to military deployment, sociodemographic and relationship factors, psychiatric comorbidity, psychotropic and other medication, and medical conditions were systematically collected, as all these factors could be important contributors to SD. PTSD symptoms were assessed jointly, but, more importantly, the impact of each cluster of PTSD symptoms (according to DSM-5 classification) on sexual functioning was also assessed. In assessing SD, we applied a comprehensive approach covering a broader range of possible sexual health problems as well as perceived sexual satisfaction.

#### *4.2. Limitations*

This study has several limitations. Findings from this study may not be reflective of and generalisable to the broader veteran or nonveteran population. Because of the many variables tested, data analysis suffered from multiple comparisons, allowing for possible false positive effects/predictors. Health-care-seeking participants could suffer from more serious problems in each area covered by the research. Furthermore, the generalisability is limited by a gender-imbalanced sample, as only male veterans were included in the research. Because of the cross-sectional design, the temporality of the relationship between the different studied variables and sexual dysfunction could not be evaluated. The findings were based on self-reported symptoms from questionnaire measures. Self-reports of sexual activity and satisfaction may be under- or overreported because of stigmatization.

#### **5. Conclusions**

One of the most salient implications of the current study is the importance of sexual health assessment in veterans with PTSD. This study represents an advancement in our currently limited understanding of patterns of association of PTSD with different types of SD and of the predictors of that relationship. As veterans with PTSD are more likely to suffer from SD if they experience more D symptoms and if they are not in a relationship or are less satisfied with the relationship, future research should develop therapeutic interventions more focused on the negative appraisals, emotional numbness, and irritability and other negative cognitions and emotions, as well as interventions intended to improve relationship functioning with the aim to lessen the rates of SD in this population. Psychotherapy is strongly recommended as the first-line treatment approach in PTSD. Sex therapy is effective in the variety of the SD, and couple psychotherapy is an established approach for relationship problems and dissatisfaction. Psychotherapeutic treatments, which would comprehensively cover different aspects of the problems in patients with PTSD and SD comorbidity, could have greater compliance rates, less iatrogenic adverse effects, and better treatment effects. Psychotropic treatment with fewer adverse sexual effects and management of the treatment-emergent side effects are of utmost importance if pharmacotherapy is applied. Medical conditions, particularly those stress-related and frequent in study populations with diabetes and hypertension, carry an additional burden of increased

risk for SD. Appropriate prevention, screening for those conditions, and their active treatment could improve the sexual life of veterans with PTSD.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/4/432/s1, Material S1: Prevalence of comorbid disorders and drug use and association with SD, including Table S1: Comorbidity and drugs and differences according to the presence of sexual dysfunction; Table S2: Correlation coefficients; Table S3: Summary of the final step in hierarchical regression analysis for the overall sample; Table S4: Summary of the final step in hierarchical regression analysis for the subset sample of veterans in a relationship.

**Author Contributions:** Conceptualization, M.L.-C. and M.P.; methodology, M.L.-C., A.S., M.P., B.P., D.S.-Ž., and S.V.; software, A.S.; validation, M.L.-C. and A.S.; formal analysis, A.S.; investigation, M.P., M.L.-C., A.S., B.P., D.S.-Ž., and S.V.; resources, M.L.-C., A.S., M.P.; data curation, M.L.-C., A.S., M.P., B.P., D.S.-Ž., and S.V.; Writing—Original Draft preparation, M.L.-C., A.S., and M.P.; Writing—Review and Editing, M.L.-C., A.S., M.P., B.P., D.S.-Ž., and S.V.; visualization, M.L.-C. and A.S.; project administration, M.L.-C., A.S., and M.P.

**Acknowledgments:** The authors would like to thank the participants of the study, without whom this research could not have been undertaken. The authors would also like to thank their colleagues Tanja Franˇciškovi´c, Ika Ronˇcevi´c-Gržeta, Jasna Grkovi´c, Tanja Grahovac-Jureti´c, Tomislav Lesica, Sandra Blaževi´c-Zeli´c, Gordana Šiki´c, Zoran Šukovi´c, Nada Kauri´c-Raos, Tihomir Peri´c, Gavrilo Neškovi´c, Vlatka Franjkuti´c, Andreja Korpar, and Alma Kranjc for their cooperation and support.

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

#### **References**


© 2019 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/).
