The Impact of Psychiatric Symptoms on Condom Self-Efficacy among People with Serious Mental Illness

Abstract

Objective:

To determine how psychiatric symptoms affect the self-efficacy of people with serious mental illness to protect themselves and their partners from human immunodeficiency virus (HIV) and other sexually transmitted infections (STIs) by using condoms.

Methods:

As part of a National Institute of Mental Health-funded study, people with serious mental illness (N = 467) were recruited in public psychiatric outpatient clinics in Rio de Janeiro, Brazil for an HIV prevention intervention. We examined the effects of psychiatric symptom severity on condom self-efficacy at baseline across four symptom clusters: affect, positive, negative, and activation.

Results:

Greater activation symptom severity (e.g., elated mood) was related to better condom self-efficacy, whereas greater negative symptom severity (e.g., blunted affect, emotional withdrawal) was related to worse condom self-efficacy.

Conclusions and Implications for Practice:

Our findings suggest that people living with serious mental illness who exhibit more-severe negative symptoms are less likely to perceive themselves as capable of using condoms, condom negotiation, and/or condom acquisition, whereas those with more severe activation symptoms are more likely to express confidence in their capabilities. Interventions to prevent HIV and other STIs among people living with serious mental illness should take into account the effects of these symptom clusters on condom skills acquisition and perceptions of self-efficacy in carrying out needed protective behaviors.

Introduction

Expressing sexuality and establishing intimacy is part of the recovery process for people with serious mental illness (SMI)—a term used to describe disorders (e.g., schizophrenia, schizoaffective disorder, and bipolar disorder) that most often include psychotic symptoms and functional disability— but sexual health remains a largely unaddressed issue in mental health services (Cook, 2000). Much of what has been learned in the past two decades about sexual expression among people living with SMI emerged from studies about their human immunodeficiency virus (HIV) infection rates and sexual risks, both of which are cause for concern (Meade & Sikkema, 2005; Otto-Salaj & Stevenson, 2001). Although a small number of studies has shown a link between psychiatric symptoms and sexual behaviors that increase risk of HIV and other sexually transmitted infections (STIs), none examined the role of psychiatric symptoms as potential mediators of the ability to engage in HIV prevention activities. This is the first study to examine one of the most important determinants of sexual risk behavior, condom self-efficacy, and its relationship to psychiatric symptoms.

Rates of HIV infection among people with SMI are higher than general population HIV rates in the United States (US) and worldwide (Hughes, Bassi, Gilbody, Bland, & Martin, 2016), and, due to their prevalence, the risk behaviors that have been the focus of prevention interventions have been sexual (multiple partners, partners of unknown HIV status, sex trading, and low rates of condom use) (Kelly, 1997; Otto-Salaj, Kelly, & Stevenson, 1998; Hobkirk, Towe, Lion, & Meade, 2015). However, studies examining the relationship between HIV risk behaviors and psychiatric symptoms among people with SMI are limited. It should be noted that while psychiatric diagnoses describe a person’s comprehensive picture as it presents longitudinally, psychiatric symptoms overlap across disorders and vary over time.

Previous research demonstrated that adults with SMI who experience more-severe excitement symptoms (i.e., activation) were greater than five times as likely to trade sex as those with less-severe excitement symptoms (McKinnon et al., 1996). Further, excitement symptoms (McKinnon, Cournos, & Herman, 2001), positive symptoms, and delusions increased the likelihood of having more sexual partners (Cournos et al., 1994). An early study showed that people experiencing manic symptoms were more likely to report increased sexual activity and less likely to use condoms (Sacks, Dermatis, Burton, Hull, & Perry, 1994).

Some studies showed that negative symptoms were associated with higher numbers of sex partners (McKinnon et al., 2001), yet others showed no association between negative symptoms (e.g., emotional withdrawal, blunted affect) and sexual risk indicators (Cournos et al., 1994; McKinnon et al., 1996). One study found that individuals who experience psychotic symptoms were more likely to have inconsistent condom use and trade sex (Tucker, Kanouse, Miu, Koegel, & Sullivan, 2003). Tucker and colleagues (2003), however, did not explore the differential impact of positive psychotic symptoms (e.g., grandiosity, hallucinations) and negative psychotic symptoms (e.g., blunted affect, emotional withdrawal) on skills to practice safer sexual behaviors.

McKinnon and colleagues (2001) found that depressive/anxiety symptom severity among people with SMI was associated with a history of sexually transmitted infections, but these symptoms were not found to be associated with sexual risk behaviors (McKinnon et al., 1996; McKinnon et al., 2001).

Taken together, there is evidence to suggest that positive, activation, and negative symptoms may be associated with increased sexual risk behavior among at least some people with SMI. It remains unclear, however, how depressive and anxiety symptoms affect sexual risk behaviors among people with SMI.

A person’s perceived self-efficacy in obtaining condoms, negotiating their use, and saying no to sex without condoms, i.e., condom self-efficacy (CSE), has been shown to be associated with sexual risk behavior (i.e., condomless sex) among adolescents and adults with SMI (Hadley et al. 2014; Kalichman, Malow, Dévieux, Stein, & Piedman, 2005). While the severity of overall psychopathology has been shown to be associated with less sexual self-efficacy (i.e., more difficulty implementing safer sex practices) (McMahon, Abbamonte, & Dévieux, 2017), the impact of particular symptoms on CSE has not been reported.

In this study, the effects of psychiatric symptoms on condom self-efficacy were examined for the first time. Participants in this study attended outpatient psychiatric treatment in Rio de Janeiro, Brazil. HIV prevalence among people with SMI in Brazil ranges between 0.8% and 1.6% (Almeida & Pedroso, 2004; Guimarães et al., 2009) compared to 0.6% of Brazil’s general population (UNAIDS, 2018). Similar to their US counterparts (Carey, Carey, Maisto, Gordon, & Vanable, 2001; Meade & Sikkema, 2005), high HIV infection rates among Brazilians with SMI appear to be due in large part to increased sexual risk behavior rather than to drug injection (Wainberg et al., 2008).

We hypothesized that positive symptoms (e.g., grandiosity, unusual thought content) would be associated with greater condom self-efficacy (Hypothesis 1). Likewise, activation symptoms (e.g., elated mood, excitement) also would be related to greater condom self-efficacy (Hypothesis 2). This may appear counterintuitive given the literature, which suggests that individuals who experience positive or activation symptoms are more likely to engage in risky behaviors (Cournos et al., 1994; McKinnon et al., 1996). However, grandiosity symptoms in people with mania have been shown to be associated with higher levels of general self-efficacy and invulnerability (Garety et al., 2013). Conversely, we hypothesized that negative symptoms (e.g., blunted affect, emotional withdrawal) would be associated with lower condom self-efficacy (Hypothesis 3). The literature suggests no association between sexual risk behavior and affect symptoms (e.g., anxiety, depression) among people with SMI; as such, we did not expect to find an association between this symptom cluster and condom self-efficacy.

Method

The current study analyzed baseline data from an intervention study described in detail elsewhere (Wainberg et al., 2016) to examine cross-sectionally the associations between psychiatric symptoms and condom self-efficacy. The intervention study was funded by the National Institute of Mental Health and conducted jointly by investigators from the New York State Psychiatric Institute and Instituto de Psiquiatria da Universidade Federal do Rio de Janeiro. Participants were referred to an HIV prevention intervention through posted flyers or by their providers in their outpatient psychiatric settings in Rio de Janeiro, Brazil. They were screened for eligibility and, if eligible, enrolled into the study. Extensively trained staff who were mental health care providers conducted all baseline interviews face-to-face at participants’ outpatient facilities in private offices.

Sampling

As part of an HIV prevention trial (R01 MH65163: Wainberg, PI), sexually active adults with SMI were recruited from eight public outpatient psychiatric clinics where they received care in Rio de Janeiro, Brazil. All study procedures were approved by Institutional Review Boards of both the New York State Psychiatric Institute and Instituto de Psiquiatria da Universidade Federal do Rio de Janeiro. To establish study eligibility, research staff conducted a Mini-International Neuropsychiatric Interview Plus (MINI-PLUS)—a widely used, cross-culturally validated diagnostic tool—with potential participants (Sheehan, Lecrubier, & Sheehan, 1998). Individuals with a primary diagnosis of substance use disorder are treated in a separate system of care in Brazil, and therefore were not present at our study clinics. The analyses presented here include only participants with a primary diagnosis consistent with an SMI disorder (N = 467), 72.9% of the total sample (N = 641).

Measures

The battery of assessments included a demographic questionnaire and measures assessing psychiatric symptom severity and condom self-efficacy. Racial/ethnic categories were self- described by participants as white, black, multiracial (known as pardo in Brazil), or other.

Psychiatric symptoms.

The Expanded Brief Psychiatric Rating Scale (BPRS) was used to measure presence and severity of psychiatric symptoms at the time of the baseline research interview (Lukoff, Liberman, & Nuechterlein, 1986). The Expanded BPRS is a widely used structured interview that consists of 24 symptom constructs. Interviewers recorded the presence and severity of symptoms on a scale as follows: 1 = Not present, 2 = Very mild, 3 = Mild, 4 = Moderate, 5 = Moderately severe, 6 = Severe, and 7 = Extremely severe.

Originally developed by Overall and Gorham (1962), the BPRS has undergone several updates, was adapted into many languages for international use including in Brazil (Zuardi, Loureiro, Rodrigues, & Correa, 1994) and has been shown to be an accurate measure of psychiatric symptoms among Brazilians with psychiatric diagnoses (Crippa, Sanches, Hallak, Loureiro, & Zuardi, 2002).

Condom Self-Efficacy.

Condom self-efficacy was assessed through the Sexual Risk Behavior Beliefs and Self-Efficacy Scales (SRBBS) (Basen-Engquist et al., 1999). The SRBBS is a widely used, valid, and reliable measure that includes a condom self-efficacy subscale that assesses an individual’s perceived ability to negotiate condom use (e.g., one’s ability to “Say no to sex if your partner won’t use a condom”). Participants answer each item using a four-point Likert scale as follows: 0 = Not sure, 1 = Somewhat sure, 2 = I’m sure, 3 = Absolutely sure. For the purposes of this study, we used 6 of the 8 items based on an exploratory factor analysis, which yielded a two-factor solution in which two items related to female condom use loaded onto a separate factor. Given the scarcity of female condoms in Brazil, the two items were dropped from the analysis. Internal consistency reliability analysis of the remaining six items resulted in an acceptable Cronbach’s alpha (α = 0.74). As with the Expanded BPRS, a translated version of the SRBBS was administered through face-to-face interviews with trained research staff members.

Covariates.

Participant interviews included a demographics questionnaire that recorded participants’ age, gender, race, socioeconomic status (SES), marital status, and education.

Because previous condom use and sexual risk behavior are likely to affect condom self-efficacy, we also controlled for participants’ proportion of condomless sex acts with all partners within the previous three months. Sexual risk behaviors were assessed through the Brazilian Sexual Risk Behavior Assessment Schedule (SERBAS-B) (Pinto et al., 2007; Wainberg et al., 2008). The SERBAS-B is an adaptation of the Sexual Risk Behavior Assessment Schedule (SERBAS), a semi-structured interview that elicits sexual practices and drug use in the previous three months (Meyer-Bahlburg et al., 1991). The SERBAS has been shown to have good internal and re-test reliability across studies focused on people with SMI (McKinnon et al., 1993; Sohler, Colson, & Meyer-Bahlburg, 2000) and has been adapted with demonstrated reliability for Brazilians living with SMI (SERBAS-B) (Pinto et al., 2007; Wainberg et al., 2008). The sexual risk behavior look-back period of the prior three months represents a temporal gap between current symptoms and CSE; therefore, this analysis included prior risk behaviors not as an outcome but instead as a covariate of CSE.

Data Analyses

Before examining our main hypotheses, we first conducted a Confirmatory Factor Analysis (CFA) to establish robust symptom clusters informed by previous literature (Dazzi et al., 2016). To examine how the severity of these symptoms affected condom self-efficacy, we conducted an Analysis of Covariance (ANCOVA) across clusters of symptoms (Hypotheses 1-3). In our model, the independent variables were the symptom clusters (as designated by the CFA in the previous data step), which are comprised of the average severity of Expanded BPRS symptoms within that cluster. The dependent variable was the average score of the condom self-efficacy measure. Using ANCOVA, we controlled for age, gender, race, reported sexual orientation, socioeconomic status, education, marital status, and previous sexual risk behaviors (number of partners and proportion of condomless sex acts). All analyses were conducted using SAS software, Version 9.4 of the SAS System for Windows. Copyright © 2017 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.

Results

Demographic characteristics of the sample are presented in Table 1. The mean age of participants was 41.39 years (SD = 9.7) and the vast majority (90.1%) identified as heterosexual. The study sample was racially diverse. The most common diagnoses were Schizophrenia (45.2%) and Bipolar Disorder (29.8%).

Table 1.

Sociodemographic characteristics (N = 467)

	Mean (SD)
Age	41.39 (9.7)
Household Assets Indicator (HAI)*	.45 (.4)
	Frequency	Percent
Male	214	45.8%
Marital Status
Married	211	45.2%
Single/Separated/Divorced/Widow(er)/Other	256	54.8%
Race
White	152	32.5%
Black	96	20.6%
Multiracial	172	36.8%
Other	47	10.1%
Sexual Orientation
Heterosexual	421	90.1%
Gay/Lesbian	13	2.8%
Bisexual	25	5.4%
Transsexual	3	0.6%
Other	5	1.1%
Education
First degree complete or lower	212	45.5%
High school incomplete or higher	254	54.5%
Primary Diagnosis
Schizophrenia	211	45.2%
Schizoaffective Disorder	20	4.3%
Bipolar Disorder	139	29.8%
Major Depression w Psychosis	66	14.1%
Psychosis Unspecified	31	6.6%

^*Household Assets Indicator (HAI) is a proxy for income calculated by the number of household items owned by the individual (e.g., refrigerator, car, television), where higher numbers indicate higher socioeconomic status. HAI was determined using the Latin American Public Opinion Project (LAPOP) wealth index (Córdova, 2009).

Confirmatory Factor Analysis

The 15-item symptom cluster model identified in Dazzi et al. (2016) resulted in significant factor loadings across all symptoms and met most criteria for a good fit (Maximum Likelihood Estimation; Adjusted Goodness of Fit Index [AGFI] = 0.91; Bentler-Bonett Normed Fit Index [NFI] = 0.88; Root Mean Square Error of Approximation [RSMEA] = 0.06; Comparative Fit Index [CFI] = 0.91). The accepted cut-off score for AGFI, NFI, and CFI goodness of fit indices is ≥ 0.90 and the RSMEA is < 0.08 (Bentler & Bonett 1980; Hooper, Coughlan, & Mullen, 2008). Distractibility (an activation symptom) yielded a considerably low lambda (λ = 0.24). This low lambda, coupled with Distractibility’s association with many symptom profiles (e.g. depression, anxiety), led to its being dropped from the model.

The resulting 14-item model yielded significant factor loadings across all symptoms and resulted in a better fitting model (Maximum Likelihood Estimation; AGFI = 0.93; Bentler-Bonett NFI = 0.91; RMSEA = 0.06; CFI = 0.94) (presented in Figure 1).

Figure 1.

Confirmatory factor analysis. The symptom clusters examined in this study are based on those identified in a meta-analysis of the Expanded BPRS (Lukoff et al., 1986) conducted by Dazzi and colleagues (2016) and are as follows: (a) Affect Symptoms comprised of Anxiety, Depression, Suicidality, and Guilt; (b) Positive Symptoms comprised of Grandiosity, Suspiciousness, Hallucinations, and Unusual Thought Content; (c) Negative Symptoms comprised of Blunted Affect, Emotional Withdrawal, and Motor Retardation; and (d) Activation Symptoms comprised of Elated Mood, Excitement, Distractibility (dropped from model), and Motor Hyperactivity. Rectangles are manifest variables, large circles are latent constructs (i.e., symptom clusters), and small circles are residual variances. Factor loadings are standardized and all are significant (p < .001).

Bivariate Correlations and ANCOVA of Demographic Variables

A correlational matrix of the primary variables of interest is presented in Table 2. Preliminary analyses showed significant relationships between several demographic characteristics and independent variables within our study. Of note, a higher proportion of condomless sex in the past three months was associated with less condom self-efficacy (r[458] = −0.29, p < 0.01), demonstrating that condom self-efficacy may be an accurate indicator of past sexual risk behaviors. In addition, race was associated with condom self-efficacy in the bivariate correlations (r[462] = −0.11, p < 0.05) and ANCOVA analyses (F[3,452] = 2.79, p = 0.04). A post-hoc analysis of variance showed that participants who self-identify as black (β = −1.19, t = −1.98, p < 0.05) or other (β = −2.04, t = −2.64, p < 0.01) reported significantly less condom self-efficacy than those who self-identify as white (ref = white). Identifying as multiracial was not associated with CSE. Sexual orientation was associated with number of partners (r[463] = 0.22, p < .01) and condomless sex (r[458] = −0.10, p < .05) in the past 3 months. Post-hoc intergroup comparisons on these sexual risk indicators were not feasible given the limited number of participants who identify with a sexual minority group (less than 10% of our sample). Further, there was not a significant association between sexual orientation and the outcome variable CSE. Given the small subsample of people identifying as a sexual minority group member and no significant association between sexual orientation and CSE, sexual orientation was not included in the ANCOVA model.

Table 2.

Correlations of sociodemographic characteristics, sexual risk behaviors, psychiatric clusters, and condom self-efficacy (CSE) (N = 467)

	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Gender	-
2. Age	−0.05	-
3. Race	0.01	−0.01	-
4. Marital Status	0.12**	0.25**	0.00	-
5. Sexual Orientation	−0.07	−0.13**	−0.07	−0.12**	-
			-
6. Education	−0.06	−0.06	0.15**	0.02	0.10*	-
7. HAI	0.00	−0.02	−0.03	−0.01	0.02	0.28**	-
8. # Partners^	−0.03	−0.10*	0.03	−0.06	0.22**	0.07	0.08	-
9. Condomless Sex^	0.19**	0.15**	0.04	0.15**	−0.10*	−0.06	0.05	−0.14**	-
10. Affect Symptoms	0.09*	−0.03	−0.03	0.02	0.05	−0.03	−0.06	−0.04	−0.01	-
11. Positive Symptoms	−0.10*	0.06	−0.01	0.03	0.01	0.02	−0.02	0.01	0.06	−0.34**	-
12. Negative Symptoms	−0.07	−0.04	0.02	−0.06	0.08	0.12*	−0.08	−0.02	−0.08	−0.15**	0.09	-
13. Activation Symptoms	−0.06	0.02	0.04	−0.03	0.03	0.16**	0.05	0.06	−0.07	0.06	−0.21**	−0.05	-
14. CSE	−0.06	−0.05	−0.11*	0.07	0.00	0.08	−0.02	0.05	−0.29**	−0.02	−0.06	−0.12**	0.09*

^*p < .05, two-tailed.

^**p < .01, two- tailed.

^{^}Sexual risk behaviors from past three months

Impact of Psychiatric Symptoms on Condom Self-Efficacy

To examine the relationship between symptom cluster severity and condom self-efficacy, we conducted an ANCOVA that included all proposed covariates (presented in Table 3). Results of the primary analysis did not support our first hypothesis, that those with greater positive symptom severity would be more likely to have greater condom self-efficacy.

Table 3.

ANCOVA examining the relationship between symptom cluster and condom self-efficacy after controlling for gender, age, race, SES, marital status, education, and past 3 months sexual risk behaviors (n = 403)

Covariate/Symptom Cluster	Type III Sum of Squares	df	Mean Square	F	β	P
Gender (ref = Male)	0.08	1	0.08	0.15	-	0.70
Age	0.47	1	0.47	0.91	0.00	0.34
Race (ref = White)	4.34	3	1.45	2.79	-	0.04*
Household Asset Indicator	0.24	1	0.24	0.47	−0.07	0.50
Marital Status (ref = Married)	1.08	1	1.08	2.09	-	0.15
Education (ref = Primary School or less)	0.03	1	0.03	0.06	-	0.81
# of Partners in Past 3 Months	0.04	1	0.04	0.07	0.00	0.79
Proportion of Condomless Sex	20.11	1	20.1	38.82	−0.53	<0.01***
Positive Symptoms	0.55	1	0.55	1.05	−0.01	0.31
Activation Symptoms	2.36	1	2.36	4.54	0.04	0.04*
Negative Symptoms	3.75	1	3.75	7.24	−0.05	<0.01**
Affect Symptoms	0.00	1	0.00	0.00	0.00	0.95

^*p < 0.05, two-tailed.

^**p < 0.01, two-tailed.

^***p < 0.001, two-tailed.

Parameter estimates (β) for continuous variables provided.

The findings did support our second hypothesis, yielding a significant effect of activation symptom severity on condom self-efficacy after controlling for all covariates (F[1,452] = 2.35, p = 0.03). The resulting parameter estimate for activation symptom severity (β = 0.04, t = 2.13, p = 0.03) indicated that those who experienced greater activation symptom severity are likely to have greater condom self-efficacy.

We also observed a significant effect of negative symptom severity on condom self-efficacy (F[1,452] = 3.75, p < 0.01). The resulting parameter estimate for negative symptom severity (β = −0.05, t = −2.69, p < 0.01) indicated that those who experience greater negative symptom severity were likely to have less condom self-efficacy, which is consistent with our third hypothesis.

We also examined whether affect symptoms were associated with condom self-efficacy. Our results showed a positive relationship between these constructs, though the findings were not significant.

In addition to our main findings, the model also shows that a higher proportion of condomless sex in the past three months is negatively associated with condom self-efficacy (F[1,452] = 38.82, β = −0.53, t = −6.23, p < 0.01). Of major importance, after controlling for the effect of previous sexual risk behaviors on condom self-efficacy, we still see significant associations between psychiatric symptoms and condom self-efficacy.

Discussion

This is the first study to show that, among people with SMI, psychiatric symptom clusters have differential effects on condom self-efficacy, and that some symptoms may pinpoint which patients have the greatest need for targeted interventions to improve CSE. Moreover, this is the first study to demonstrate that, among people living with SMI, prior condomless sex is associated with poorer condom self-efficacy. Even after controlling for past sexual risk behaviors, psychiatric symptoms and condom self-efficacy are related.

Positive symptoms and condom self-efficacy (Hypothesis 1) were not related in this sample. Although previous literature has shown positive symptoms to be associated with sexual risk, specifically with having multiple partners (Cournos et al., 1994), it appears that positive symptoms do not have an impact on condom self-efficacy.

As expected, severity of activation symptoms was predictive of greater condom self-efficacy (Hypothesis 2). The positive association between these symptoms and increased condom self-efficacy may seem counter-intuitive given the literature which showed that activation symptoms, particularly excitement, are associated with trading sex and number of partners and sex episodes (McKinnon et al., 1996; McKinnon et al., 2001). However, these studies did not find an association between condom use and psychiatric symptoms. We did not ascertain whether those with more-severe activation symptoms perceive themselves as capable of safer sex practices like condom acquisition or negotiation, but place less importance on utilizing these skills if they feel invulnerable to sexually transmitted infections and believe these practices are not necessary.

As predicted, increased negative symptom severity was associated with less condom self-efficacy (Hypothesis 3). This suggests that people living with SMI who exhibit blunted affect, emotional withdrawal, and/or motor retardation are less likely to perceive themselves as capable of using condoms, condom negotiation, and/or condom acquisition.

Results also showed no significant relationship between affect symptoms and condom self-efficacy, which is consistent with previously reported symptom-risk behavior findings (McKinnon et al., 1996; McKinnon et al., 2001). It is possible that risk behaviors not related to condom use may be more likely to be affected by this symptom cluster.

Our study yielded unanticipated findings showing race to be associated with condom self-efficacy; black individuals and those who identified as “other” than white or multiracial reported significantly less condom self-efficacy. Condom self-efficacy was not associated with being self-identified as multiracial. While no study has yet to examine the impact of race/ethnicity on condom self-efficacy in the Brazilian general population or among people with SMI, an ethnically diverse US sample of men and women found that Hispanic individuals had less condom self-efficacy than their white and Asian counterparts (Farmer & Meston, 2006). Another US-based sample showed African American women to have less condom self-efficacy than their European American peers, which was predictive of riskier sex practices (Schröder, Hobfoll, Jackson, & Lavin, 2001). The findings presented in the current study add to the sparse literature highlighting the potential impact of minority stress on psychosocial mediators of sexual risk. As the minority stress literature has demonstrated, persons who are members of a marginalized community report lower levels of overall self-efficacy (Crocker & Major, 1989; Meyer, 2003; Collins & Lightsey, 2001; Ouch & Moradi, 2019), and it is possible that the impact of minority stress extends to condom self-efficacy specifically.

This study sample was older (M = 41.39 years old) than non-psychiatric HIV risk study samples. However, this sample is representative of SMI outpatient clinical settings and is consistent with other SMI outpatient HIV study samples in the US (Bonfils, Firmin, Salyers, & Wright, 2015; Meade & Sikkema, 2007; Weinhardt, Carey, Carey, Maisto, & Gordon, 2001) and Brazil (Menezes & Ratto, 2004; Wainberg et al., 2008).

This analysis adds to the small global literature examining the relationship of psychiatric symptoms and safer sexual practices among people with SMI. Prior condomless sex appears to be a marker for low condom self-efficacy; even over and above those effects, some psychiatric symptoms appear to influence CSE. Our findings suggest that HIV prevention intervention strategies targeting condom self-efficacy may be particularly beneficial for people with SMI who experience more negative symptoms. Given the positive association between condom self-efficacy and activation symptoms, targeted interventions focusing on the motivation to use condoms in addition to perceived self-efficacy among people with activation symptom presentation could have a positive impact. HIV prevention interventions for people with SMI have been shown to increase HIV knowledge and condom self-efficacy (Weinhardt, Carey, & Carey, 1997) and improve safer sex communication and condom use intentions (Carey et al., 2004a). Existing efficacious HIV risk reduction interventions for people with SMI could be tailored to meet the needs of individuals in accordance with their particular symptom profiles.

This study was cross-sectional so we did not capture long-term causal relationships between symptoms and condom self-efficacy or their relationship to current or future sexual risk behavior. This study also did not include people with a dual diagnosis of SMI and the common comorbidity of substance use disorder (SUD). The results of this study therefore may not be generalizable to dually diagnosed people, and the findings presented here may be under or over estimates of CSE for people who are dually diagnosed. Comorbid SMI and SUD may increase the likelihood of sexual risk behavior (Carey et al., 2004b; McKinnon & Cournos, 1998; Meade & Sikkema, 2005) and should be considered in future research related to psychiatric symptoms and condom self-efficacy. Although the system of care in which this study took place did not serve people with co-occurring SMI and SUD, mental health clinics like those found in Rio de Janeiro can offer HIV services to many people with serious mental illness who might not otherwise be able to access such services.

The current study also examined symptom clusters rather than individual symptoms; it is possible that some symptoms within clusters may have differential impacts on condom self-efficacy and other sexual risk indicators. Furthermore, our findings may not generalize to all persons living with SMI; this sample varied in socioeconomic status, participants lived in an urban area of Brazil with access to public and private mental health treatment, and were already engaged in care. Symptom severity may not represent that of people living in rural areas with limited access to care or those not in treatment. Additionally, most of our sample self-identified as heterosexual, and there were not enough participants who identified with a sexual or gender minority identity for intergroup comparison. This may have resulted in under or over estimates of CSE among people who identify themselves as members of gender and/or sexual minority groups. Because men who have sex with men and transgender men and women may experience additional vulnerability to HIV infection, results of this study do not further our understanding of the ways in which sexual and gender identity affect condom self-efficacy.

Conclusions

Sexuality and intimacy are legitimate components of the recovery process for people with serious mental illness, yet sexual health remains a largely unaddressed issue in psychiatric treatment settings (Cook, 2000). Targeted efforts to tailor HIV prevention messages and interventions to these settings might also reduce the risk of other sexually transmitted infections, which have dramatically risen over the past decade (CDC, 2018), and enhance the recovery process for people with serious mental illness. Despite the established links between psychiatric symptoms and sexual risk behaviors (Cournos et al., 1994; McKinnon et al., 1996, 2001; Sacks et al., 1994), little was known about the impact of psychiatric symptoms on condom self-efficacy among people living with SMI. Findings from this study may assist providers and researchers in improving HIV and STI prevention interventions targeting condom use and condom self-efficacy by tailoring them to specific symptom profiles. Future studies should expand their inclusion criteria to be more inclusive of people with SMI who have comorbid substance use disorders and/or identify as members of sexual minority groups.

Impact and Implications.

This is the first study to examine the role of psychiatric symptoms on potential mediators thought to influence one of the most important skills for practicing safer sex: condom self-efficacy. As we work to include sexuality and intimacy as legitimate components of recovery from serious mental illness, these findings may assist providers and researchers in improving human immunodeficiency virus (HIV) and sexually transmitted infections (STI) prevention interventions by tailoring them to specific symptom profiles.