Love, Trust, and HIV Risk Among Female Sex Workers and Their Intimate Male Partners

Abstract

Objectives. We examined correlates of love and trust among female sex workers and their noncommercial male partners along the Mexico–US border.

Methods. From 2011 to 2012, 322 partners in Tijuana and Ciudad Juárez, Mexico, completed assessments of love and trust. Cross-sectional dyadic regression analyses identified associations of relationship characteristics and HIV risk behaviors with love and trust.

Results. Within 161 couples, love and trust scores were moderately high (median 70/95 and 29/40 points, respectively) and correlated with relationship satisfaction. In regression analyses of HIV risk factors, men and women who used methamphetamine reported lower love scores, whereas women who used heroin reported slightly higher love. In an alternate model, men with concurrent sexual partners had lower love scores. For both partners, relationship conflict was associated with lower trust.

Conclusions. Love and trust are associated with relationship quality, sexual risk, and drug use patterns that shape intimate partners’ HIV risk. HIV interventions should consider the emotional quality of sex workers’ intimate relationships.

While public health interventions continue to be grounded in tenets of risk and disease, there is growing recognition of the importance of emotions such as love and trust in shaping the health behaviors of diverse populations. Unprotected sex allows for transmission of HIV and other sexually transmitted infections (STIs) but often carries emotional meanings of trust and commitment that impede condom negotiation even among individuals with high knowledge and self-efficacy.^1–4 Research has consistently found female sex workers’ condom use to be less frequent with their noncommercial male partners than with clients, which has been linked to intimacy, love, and trust.^5–7 Similarly, drug-using couples may share syringes and engage in additional known HIV risk behaviors to convey support and emotional closeness.^8,9 Although much of this research is qualitative, some quantitative work has confirmed associations between perceptions of intimacy and condom use among sex workers and clients¹⁰ and men who have sex with men.¹¹

Quantitative researchers have attempted to operationalize emotions into theoretical constructs for empirical research.¹² Sternberg’s “triangular” theory posits that love comprises 3 components: intimacy (feelings of emotional closeness between partners), passion (physical attraction and sexual consummation), and commitment (decisions that partners make about maintaining relationships).^13,14 Taken together, these components combine to generate different kinds of love.^13,15,16

Another important aspect of intimate relationships, trust is conceptualized as a belief in the integrity of a partner. Trust may be based on individuals’ assessments of their partners’ benevolence and honesty¹⁷ and subjective feelings of security within relationships.^18,19 To measure trust quantitatively, Larzelere and Huston developed a dyadic trust scale that correlated with love and relationship quality.¹⁷

A rich body of ethnographic scholarship has highlighted the importance of companionate relationships and love in diverse global contexts,^20–22 but quantification of emotional constructs are largely absent from settings outside the Western middle-class experience. Most quantitative studies of love and trust have focused on privileged populations who do not experience the burden of disease that disproportionately affects socially marginalized couples. Existing studies among US populations have also suggested that women and men value components of these emotions differently,^23–25 yet the implications of these potential gender differences for HIV prevention among vulnerable couples remains underexplored.

Our study of love and trust among sex workers and their noncommercial partners is situated in 2 Mexican border cities: Tijuana, Baja California, south of San Diego, California (population 1.6 million), and Ciudad Juárez, Chihuahua, adjacent to El Paso, Texas (population 1.3 million). Both cities have areas where sex work and drug use are concentrated,²⁶ lending to the conceptualization of the border as an HIV risk environment²⁷ characterized by poverty, unemployment, violence, and availability of illicit drugs.²⁸ Research has documented substantial overlap in risky sexual and injection drug use behaviors among marginalized female sex workers²⁹ who often turn to sex work as a last option for supporting themselves and their dependents.³⁰

Drawing on Connell’s conceptualization of gender as a relational process influenced by structural and sociocultural factors,³¹ our previous qualitative and ethnographic inquiry into sex workers’ intimate partnerships in these cities proposed that emotions are shaped by the HIV risk environment of the border. Structural conditions bring partners together into relationships that provide critical material and emotional support. In light of limited economic opportunities, couples form tacit agreements regarding women’s engagement in sex work. In the process, women emerge as primary wage earners, alienating men from traditional gender roles as providers and shifting gender dynamics. To cope, partners mostly avoid communicating about sex work or disclosing HIV risk to protect the emotional integrity of the relationship.³²

We also found that these couples’ relationships range in emotional intensity from friendship and care to transformative experiences of love. Regardless, most couples demarcate their primary relationships from sex work by engaging in unprotected sex. Among couples who inject drugs, sharing syringes together (but not with others) often reinforces trust and care.³³ Concerns of mistrust and infidelity also reduced couples’ interest in alternative HIV-prevention technologies such as pre-exposure prophylaxis.³⁴

In light of the central role of emotions in our qualitative research with these couples, we attempted to quantify constructs of love and trust and identify associated relationship traits and HIV risk behaviors in our larger cohort. We hypothesized that markers of relationship quality (e.g., higher satisfaction, lower conflict) and unprotected sex within primary relationships would be associated with higher love and trust, whereas extradyadic HIV risk behaviors (e.g., concurrent sexual partners, syringe sharing with others) would be associated with lower love and trust.

METHODS

We drew from Proyecto Parejas (Couples’ Project), a prospective study of HIV and other STIs within sex workers’ intimate relationships in Tijuana and Ciudad Juárez that began in 2010.³⁵ Fieldworkers used targeted and snowball sampling^36,37 to recruit women from known areas of sex work and drug use in these cities. Eligible women were aged 18 years or older, had ever used hard drugs, reported sex work in the past month, were in a noncommercial relationship for at least 6 months, and were not determined to be at risk for life-threatening intimate partner violence as a result of participating. Eligible women were invited to bring male partners to study offices to assess men’s eligibility (aged ≥ 18 years) and verify relationships.³⁵ Eligible couples provided written consent to complete quantitative surveys and HIV and other STI testing every 6 months for 2 years. Each partner was reimbursed US $20 per visit.

Data Collection

For this cross-sectional study, we drew from 12-month follow-up quantitative assessments. Of the original cohort of 214 couples enrolled in 2010–2011, 161 couples (n = 322 individuals who remained in the relationship and were not incarcerated or lost to follow-up) returned for 12-month follow-up visits from 2011 to 2012. Interviewers administered 1- to 2-hour-long computerized questionnaires to individuals. Questionnaires assessed demographics and personal factors, including depression (Center for Epidemiologic Studies Short Depression Scale [CESD-10]; range = 0–30 points; α = 0.826)^38,39 and lifetime and recent (past 6 months) sexual and drug-related HIV risk behaviors, including unprotected sex, concurrent sexual partners, and syringe sharing. We created relationship-level variables such as relationship duration by averaging both partners’ responses within couples.⁴⁰ We measured relationship satisfaction with a 5-item scale (range = 1–20 points; α = 0.907).⁴¹ We determined prevalence of past-year verbal and physical conflict (perpetration and victimization) by using affirmative responses to the Revised Conflict Tactics Scale subscales for psychological aggression, physical assault, injury, or sexual assault.^42,43

To quantify the emotional constructs that emerged as important in our qualitative research,³³ we adapted 2 scales for inclusion in 12-month follow-up assessments: a 19-item Triangular Love Scale (range = 19–95 points; α = 0.914)¹⁶ and an 8-item Dyadic Trust Scale (range = 8–40 points; α = 0.845).¹⁷ Five Likert-type response options ranged from “strongly disagree” to “strongly agree,” for a maximum of 5 possible points per item. Total scores represent summations of items in each scale and were used as dependent variables in our analyses; higher scores represent higher levels of love and trust within relationships.

Data Analysis

Quantitative descriptive statistics summarized and assessed gender differences for all measures including our 2 dependent variables, love and trust. To identify relationship and behavioral characteristics independently associated with love and trust, we used seemingly unrelated regression with bootstrapped SEs,⁴⁴ a generalization of linear regression that models women’s and men’s outcomes separately but simultaneously (i.e., using 2 separate equations), allowing error terms to be correlated to account for interdependence within couples.⁴⁰ We conducted cross-sectional dyadic regression analyses to examine hypothesized associations between our independent variables and love and trust. We built multivariable regression models by using a hierarchical block approach to grouping variables into theoretically relevant domains.⁴⁵ We assessed plausible interactions and confounding via changes in other estimates,⁴⁶ and compared the fit of nested models by using the Akaike information criterion.⁴⁷

Notably, one hypothesized correlate of reduced love, concurrent sexual partners, had a small sample size but strong predictive results among men. Because of minimal variation in the low love scores reported by men with concurrent partners, we retained concurrency in one final model because it may indicate a trend. However, because of the possibility that this binary concurrency variable could affect other estimates and explain much of the variation in love, we ran a second final model excluding concurrency. Results from both models are discussed.

RESULTS

Among 161 couples (n = 322 individuals), median age was 35 years (interquartile range [IQR] = 30–41), median relationship duration was 4 years (IQR = 2.8–6.2), and 96% of couples were married or in common-law unions (Table 1). Relationship satisfaction was relatively high (15 of 20 possible points), but conflict was also common, with 55% reporting perpetrating physical or psychological abuse in the past year and 50% being victimized. Most couples (89%) never used condoms in the past month. Overall, 5% of partners had “steady” (ongoing) concurrent sexual partnerships, which was more common among women than men (8% vs 2%; P < .05), because of regular clients being captured by this measure. Drug use in the past 6 months was prevalent, with 48% using heroin, 22% using methamphetamine, and nearly half (47%) injecting any drugs.

TABLE 1—

Characteristics of Female Sex Workers and Their Intimate Partners (n = 322; 161 Couples) in Tijuana and Ciudad Juárez, México: 2011 to 2012

Variable	Women (n = 161)	Men (n = 161)	Overall (n = 322)
Sociodemographics and personal factors
Age, y, median (IQR)	34 (28–40)	37 (32–44)	35 (30–41)**
Income < 2500 pesosa/month, no. (%)	57 (35)	27 (17)	84 (26)**
Children aged < 18 y currently live with participants, no. (%)	55 (34)	51 (32)	106 (33)
Depression score,b median (IQR)	8 (5–12)	3 (2–7)	6 (3–10)**
Relationship factors
Married to or in common-law union with partner, no. (%)	. . .	. . .	310 (96)
Median steady relationship duration in years (IQR)c	. . .	. . .	4.0 (2.8–6.2)
Never used condoms for vaginal sex within the steady relationship, past month,c no. (%)	. . .	. . .	236 (89)
Median relationship satisfaction scored (IQR)	15 (13–15)	15 (14–15)	15 (14–15)
Perpetrated conflict within relationship, past 6 mo,e no. (%)	99 (61)	79 (49)	178 (55)*
Victim of conflict within relationship, past 6 mo,e no. (%)	83 (52)	77 (48)	160 (50)
Drug abuse behaviors, past 6 mo, no. (%)
Used heroin	80 (50)	74 (46)	154 (48)
Used methamphetamine	35 (22)	35 (22)	70 (22)
Used cocaine	10 (6)	8 (5)	18 (6)
Used crack	17 (11)	10 (6)	27 (8)
Injected any drugs	80 (50)	72 (45)	152 (47)
Shared syringes within steady relationship	20 (12)	19 (12)	39 (12)
Shared syringes outside steady relationship	9 (6)	11 (7)	20 (6)
Sexual behaviors, no. (%)
Has other “steady” concurrent partners, past year	13 (8)	4 (2)	17 (5)*
Male partner had any casual outside sexual partners, past 6 mo	. . .	45 (28)	. . .
FSW rarely or never used condoms with clients, past month	56 (35)	. . .	. . .

Note. FSW = female sex worker; IQR = interquartile range.

^a< US $200.

^bCenter for Epidemiologic Studies Short Depression Scale: 10 items, 30 possible points.

^cDyad average (uses information from both partners’ responses within a given couple).

^d5 items, 20 possible points.

^eConflict Tactics Scale subscales (combined) for any past-year psychological aggression, physical assault, injury, and sexual coercion within relationships.

*P < .05; **P < .01 for differences between women and men (from regression with clustered SEs within couples).

Love and Trust Within Sex Workers’ Intimate Relationships

Love and trust scores were relatively high among couples. The median love score was 70 out of 95 possible points (IQR = 65–73 points; Table 2). Although there were no gender differences in total love or trust scores, there were statistically significant gender differences in responses to specific passion and commitment items: men were more likely than women to agree or strongly agree with the statements, “sex is important in your relationship” (P < .001), “you feel attracted to other possible partners” (P < .001), and “you will probably have another love relationship later in your life” (P < .001). More women agreed with the statements, “you see your relationship with your partner as never ending,” and “you would rather be with your partner than with anyone else,” although these differences only approached statistical significance (P < .1).

TABLE 2—

Love and Trust Among Female Sex Workers and Their Intimate Partners (n = 322; 161 Couples): Tijuana and Ciudad Juárez, México, 2011–2012

	Responses, No. (%)
Item	Strongly Disagree	Disagree	Neutral	Agree	Strongly Agree
Love scalea
Intimacy
You and your partner share personal information with each other.	3 (1)	22 (7)	29 (9)	244 (78)	15 (5)
You can tell everything to your partner.	5 (2)	46 (15)	39 (12)	208 (66)	15 (5)
You and your partner tell each other about private thoughts and feelings.	4 (1)	32 (10)	33 (11)	228 (73)	16 (5)
Some things you can only tell to your partner and to no one else.	4 (1)	20 (6)	26 (8)	243 (78)	20 (6)
Your partner understands how you feel.	4 (1)	28 (9)	33 (11)	230 (73)	18 (6)
You and your partner think the same about a lot of things.	5 (2)	27 (9)	34 (11)	228 (73)	19 (6)
Most of the time you feel very close to your partner.	4 (1)	28 (9)	34 (11)	228 (73)	19 (6)
Passion
You feel a strong attraction toward your partner.	3 (1)	18 (6)	28 (9)	248 (79)	16 (5)
You often feel aroused when your partner is with you.	5 (2)	25 (8)	30 (10)	238 (76)	14 (4)
You and your partner are very passionate together.	3 (1)	24 (8)	31 (10)	239 (77)	15 (5)
You find your partner sexually attractive.	1 (< 1)	24 (8)	28 (9)	242 (77)	18 (6)
You and your partner show each other a lot of affection.	1 (< 1)	23 (7)	30 (10)	240 (77)	19 (6)
Sex is important in your relationship.**	3 (1)	90 (29)	39 (12)	165 (53)	16 (5)
Commitment
Sometimes, you would rather be single than in a relationship.b	14 (4)	163 (52)	56 (18)	70 (22)	10 (3)
You feel attracted to other possible partners.b,**	16 (5)	147 (47)	42 (13)	102 (33)	6 (2)
You see your relationship with your partner as never-ending.	4 (1)	170 (54)	47 (15)	77 (25)	15 (5)
You will probably have another love relationship later in your life.b,**	17 (5)	133 (42)	63 (20)	96 (31)	4 (1)
You think that your relationship with your partner will last forever.	4 (1)	47 (15)	53 (17)	196 (63)	13 (4)
You would rather be with your partner than with anyone else.	3 (1)	23 (7)	33 (11)	237 (76)	17 (5)
Trust scalec
Your partner is more interested in his or her welfare than in yours.b	11 (4)	195 (62)	46 (15)	52 (17)	9 (3)
There are times when your partner cannot be trusted.b	10 (3)	136 (43)	53 (17)	102 (33)	12 (4)
Your partner is honest and truthful with you.	3 (1)	36 (12)	64 (20)	199 (64)	11 (4)
Your feel that you can trust your partner completely.	4 (1)	56 (18)	76 (24)	163 (52)	14 (4)
Your partner is truly sincere in his or her promises.	6 (2)	43 (14)	46 (15)	205 (66)	13 (4)
Your partner does not show you enough consideration.b,**	8 (3)	121 (39)	40 (13)	134 (43)	10 (3)
Your partner treats you fairly and justly.*	4 (1)	33 (11)	36 (12)	228 (73)	12 (4)
You can count on your partner to help you.	3 (1)	16 (5)	28 (9)	251 (80)	15 (5)

^aFrom the 19-item Triangular Love Scale (range = 19–95 points; α = 0.9137). The median total love score = 70 points (interquartile range = 65–73).

^bItem was reverse-coded for calculating total score.

^cFrom the 8-item Dyadic Trust Scale (range = 8–40; α = 0.8453). Median trust score = 29 points (interquartile range = 24–31 points).

*P < .05; **P < .01 for differences between women and men (from regression with clustered SEs within couples).

The median trust score was 29 of 40 possible points (IQR = 24–31; Table 2). There were no gender differences in total median trust scores, although men were more likely than women to agree or strongly agree with the statement, “your partner does not show you enough consideration” (P < .001), but also “your partner treats you fairly and justly” (P < .05). Both love and trust were highly correlated with relationship satisfaction.

Factors Associated With Love and Trust

Unadjusted regression analyses identified associations between relationship and behavioral factors and love and trust (Table 3). Men with lower income (i.e., earning less than US $200 per month) reported lower trust (P < .05). In unadjusted analyses, depression and perpetrating or being victimized by relationship conflict were associated with reduced trust for both women and men; depression and victimization were further associated with reduced love for men.

TABLE 3—

Factors Associated With Love and Trust Among Female Sex Workers and Their Intimate Partners in Tijuana and Ciudad Juárez, México (n = 322; 161 Couples), 2011 to 2012

	Love		Trust
Characteristics	Women, Unadjusted Coefficient (SE)a	Men, Unadjusted Coefficient (SE)a	Women, Unadjusted Coefficient (SE)a	Men, Unadjusted Coefficient (SE)a
Sociodemographics and personal factors
Age in years	0.058 (–0.105, 0.220)	−0.002 (–0.143, 0.140)	0.013 (–0.090, 0.116)	0.009 (–0.050, 0.069)
Income < $200 per month	−0.812 (–4.117, 2.493)	–3.896 (–8.211, 0.418)	−0.333 (–2.092, 1.427)	–1.612* (–3.064, −0.160)
Children aged > 18 years currently live with participants	−2.655 (–6.636, 1.326)	1.083 (–1.935, 4.101)	−1.673 (–3.667, 0.322)	−0.994 (–2.600, 0.613)
Depression score (CESD: 10 items, 30 possible points)	–0.364 (–0.745, 0.017)	–0.392* (–0.755, −0.029)	–0.426** (–0.628, −0.224)	–0.267* (–0.474, −0.060)
Relationship factors
Married to or in common-law union with partner	7.340 (4.86, 9.82)	3.411 (–30.12, 37.02)	10.515 (8.87, 12.16)	6.642 (2.21–, 11.07)
Duration of steady relationship in years	0.171 (–0.217, 0.559)	0.078 (–0.148, 0.304)	0.008 (–0.186, 0.202)	0.031 (–0.081, 0.143)
Never used condoms for vaginal sex within the steady relationship, past month	12.811* (0.053, 25.569)	4.333 (–11.942, 20.608)	4.96 (–0.149, 10.070)	1.960 (–3.399, 7.319)
Relationship satisfaction score (5 items, 20 possible points)	2.687** (2.012, 3.363)	1.955** (1.176, 2.734)	1.253** (0.861, 1.645)	0.948** (0.622, 1.273)
Perpetrated conflict within relationship, past 6 mob	−0.791 (–4.315, 2.734)	−0.789 (–3.388, 1.809)	–2.577** (–4.241, −0.913)	–1.439* (–2.712, −0.167)
Victim of conflict within relationship, past 6 mob	–3.247 (–6.584, 0.089)	–2.952* (–5.507, 0.396)	–3.013** (–4.669, −1.358)	–2.094** (–3.350, −0.839)
Drug abuse behaviors, past 6 mo
Used heroin	2.844 (–0.320, 6.007)	−0.475 (–2.940, 1.990)	1.267 (–0.535, 3.070)	0.462 (–0.784, 1.709)
Used methamphetamine	–8.347** (–14.066, −2.628)	–8.521** (–13.134, −3.908)	–1.957 (–4.241, 0.328)	–3.270** (–5.225, −1.315)
Used cocaine	−2.327 (–8.675, 4.021)	0.545 (–3.402, 4.492)	−3.375 (–7.588, 0.839)	1.258 (–0.273, 2.790)
Used crack	3.400 (–2.981, 9.782)	0.678 (–1.881, 3.236)	1.279 (–2.292, 4.850)	0.432 (–1.569, 2.434)
Injected any drugs	1.998 (–1.189, 5.185)	−0.730 (–3.247, 1.788)	0.966 (–0.766, 2.699)	0.327 (–0.890, 1.543)
Shared syringes within steady relationship	0.243 (–4.141, 4.627)	−0.359 (–3.447, 2.729)	0.349 (–2.226, 2.925)	0.622 (–1.659, 2.903)
Shared syringes outside steady relationship	–11.977** (–18.260, −5.693)	−3.789 (–10.680, 3.103)	–3.749 (–7.978, 0.480)	–4.315** (–7.370, −1.260)
Sexual risk behaviors
Has other “steady” concurrent partners, past year	−6.467 (–14.539, 1.604)	–10.519* (–20.677, −0.362)	−1.758 (–5.586, 2.071)	−1.236 (–8.332, 5.860)
Male partner had any casual outside sexual partners, past 6 mo	–7.32 (–15.682, 1.026)	–6.839 (–11.130, −2.549)	−2.167 (–6.112, 1.779)	–2.888* (–4.767, −1.009)
FSW rarely or never used condoms with clients, past month	–4.561* (–8.461, −0.662)	−1.718 (–4.797, 1.361)	–2.269* (–4.214, −0.324)	−0.271 (–1.509, 0.968)

Note. CESD = Center for Epidemiologic Studies Short Depression Scale; FSW = female sex worker.

^aBootstrapped SEs.

^bConflict Tactics Scale subscales (combined) for any past-year psychological aggression, physical assault, injury, and sexual coercion within relationships.

*P < .05; **P < .01 for associations with dependent variables (love and trust).

Regarding HIV risk behaviors, unprotected sex within steady relationships was associated with higher love (P < .05) and approached significance with higher trust (P < .1) among women. Men with steady concurrent sexual partners had reduced love scores (P < .05), whereas women who rarely or never used condoms with clients had reduced love and trust scores (P < .05). Methamphetamine use was associated with lower love for both partners and lower trust for men. Sharing syringes outside steady relationships (i.e., with other people) was associated with lower love among women and lower trust among men (both P < .01).

Factors Independently Associated With Love and Trust

Several factors were independently associated with love and trust in multivariable analyses (Table 4). Because of small numbers of men reporting steady concurrent sexual partners, we conducted 2 final regression models for love (first including and then excluding concurrency). In the first model, concurrency was associated with an 8.6-point reduction in love scores among men (95% confidence interval [CI] = −12.9, −4.2) and methamphetamine use approached significance (P = .084). In the second love model (excluding concurrency), methamphetamine use became significantly associated with reduced love among men (average 3.7-point reductions in love scores, 95% CI = −7.4, −0.06; P < .05). For women in both models, methamphetamine use was associated with reduced love (4.7- and 5.0-point lower scores, 95% CI = −8.4, −0.9 and −8.7, −1.2, respectively), whereas heroin use approached significance with increased love (P = .073).

TABLE 4—

Factors Independently Associated With High Love and Trust Among Female Sex Workers and Their Intimate Partners (n = 322; 161 couples) in Tijuana and Ciudad Juárez, México, 2011 to 2012

Variable	Women, Adjusted Coefficient (SE)c	Men, Adjusted Coefficient (SE)c
Love (model 1a)
Used heroin, past 6 mo	1.900 (–0.209, 4.009)	−0.653 (–2.287, 0.981)
Used methamphetamine, past 6 mo	–4.661* (–8.420, −0.902)	–2.962 (–6.324, 0.400)
Had steady concurrent partner, past year	−1.846 (–7.378, 3.686)	–8.561** (–12.915, −4.208)
Love (model 2b)
Used heroin, past 6 mo	1.962 (–0.185, 4.109)	−0.544 (–2.225, 1.137)
Used methamphetamine, past 6 mo	−4.962* (–8.743, −1.182)	−3.744* (–7.429, 0.058)
Trust
Shared syringses outside steady relationship, past 6 mo	1.146 (–2.662, 4.955)	–2.050 (–4.329, 0.224)
Victim of conflict within steady relationship, past 6 mod	–1.901** (–3.169, −0.633)	–0.915* (–1.804, −0.026)

^aModel included concurrency.

^bModel excluded concurrency.

^cBootstrapped SEs. Models controlled for relationship satisfaction, trust (in the love model), love (in the trust model), and correlation within couples.

^dConflict Tactics Scale subscales (combined) for any past-year psychological aggression, physical assault, injury, and sexual coercion within relationships.

*P < .05; **P < .01 for associations with dependent variables (love and trust).

In the single final model for trust, being the victim of verbal or physical conflict within relationships was associated with reduced trust among women and men (average 1.9- and 0.9-point reductions in trust scores, 95% CI = −3.2, −0.6 and −1.8, −0.1, respectively). Sharing syringes outside steady relationships approached significance with lower trust among men (P = .087).

DISCUSSION

To our knowledge, this study is the first to quantitatively assess love and trust among vulnerable couples in a resource-poor setting. Distinct from the higher socioeconomic status samples in which these scales have previously been implemented,^16,17 sex workers and their noncommercial partners in Mexican border cities face social and economic marginalization that heighten their vulnerability to HIV. We identified important associations between love and trust and indicators of relationship quality and HIV risk.

Similar to the theoretical relationships conceptualized in the literature¹⁶ and supporting our previous ethnographic and qualitative research,³³ love and trust scores were relatively high and strongly correlated with each other and with relationship satisfaction. Unlike previous studies that used this “triangular love” scale,¹⁶ we found no gender differences in overall scores, perhaps reflecting recent ethnographic work on the importance of companionate love relationships in which both partners equally value emotional and physical intimacy.^20,22 However, men had lower commitment scores than women because of specific items measuring sexual attraction to other partners and believing in the likelihood of having another relationship whereas women were more likely to view their relationships as lasting, patterns consistent with gender differences in intimate relationship literature.^24,48

Building on this finding, we identified a strong correlation between having steady concurrent sexual partners and reduced love among men. Though it was driven by a small number of men, this finding may reflect an invisible trend in this population. Ethnographic work from rural Mexico has found that emotional connections with spouses do not preclude men from engaging in extramarital relationships.⁴⁹ Indeed, in our sample, more men reported any outside sexual encounters (28%) versus steady concurrent sexual relationships (2%). We suggest that longer-term concurrent partnerships may reflect increasing emotional ties that threaten the stability of primary relationships, possibly leading to relationship dissolution. Although steady concurrent sexual partnerships could be a predictor of dissolution, we are unable to formally test this hypothesis because our cross-sectional sample was composed of complete couples who have remained together for at least 18 months. It was interesting that there was no significant association between concurrency and love among women, perhaps because these other sexual partners are typically clients representing financial gain.⁵⁰ Although women may “trust” their regular clients, these relationships are not generally ascribed the same intimate meanings as their primary relationships. HIV interventions should thus consider not only the presence of concurrent sexual partners, but also how external emotional attachment could compromise primary relationships and place all partners involved at risk for HIV.

Drug use was associated with love and trust in our final models in unique ways. Among men in the second love model (excluding concurrency), methamphetamine use was significantly associated with reduced love. Among women, methamphetamine use was also associated with reduced love, whereas heroin use approached a significant association with elevated love. Our findings may reflect the biobehavioral implications of these drugs. The depressant effects and intense malilla (physical withdrawal) associated with heroin use could strengthen women’s emotional dependence on their intimate partners, just as the act of sharing drugs and syringes could reinforce or deepen connections between partners, as observed in our ethnographic work.⁵¹ Conversely, methamphetamine can inhibit cues that normally control behavior, engendering heightened energy, agitation, interpersonal conflict, and physical arousal that may exacerbate sexual risk taking (e.g., unprotected sex, multiple partners).^52,53 Drawing on research on syndemics (interrelated epidemics) of stimulant abuse, intimate partner violence, and HIV risk,⁵⁴ our findings that partners who use methamphetamine report lower levels of love may reflect such relationship chaos and diminished emotional bonds. We also found that sharing syringes outside steady relationships approached significance with reduced trust among men, again supporting our previous ethnographic work and hypotheses that emotionally close partners engage in fewer known HIV risk behaviors outside their relationships.³³ Although it is uncertain from our cross-sectional analyses whether the associations with drug use reflect true emotional states or constitute externalizations of the psychoactive effects of the drugs themselves, our results nevertheless suggest that HIV interventions should address drug use in the context of intimate relationships.

Finally, being the victim of any past-year verbal or physical intimate partner violence was significantly associated with lower levels of trust among both women and men. Our previous work revealed high levels of conflict perpetration and victimization among both partners in these couples, which was associated with HIV/STI positivity and methamphetamine use among men, suggesting higher risk profiles and a propensity toward emotional instability and impulsivity.⁵⁵ We also documented partners’ difficulty disclosing sexual risks external to their relationships, especially within volatile couples, as a way to avoid conflict.³² Ongoing qualitative analyses suggest that men’s conflict is often grounded in drug withdrawal and feelings of emasculation and suspicion because of relying on women’s sex work, whereas women perpetrate conflict when they feel deprived of affection,⁵⁶ revealing how trust and conflict reflect the broader HIV risk environment and shifting gender roles on the border.

Implications for HIV Interventions

Our findings suggest that couples-based HIV prevention approaches require tailoring based on couples’ emotional dynamics. For couples with high love and trust, programs could build upon foundations of emotional commitment to motivate protection of intimate partners.⁵⁷ Alternative strategies for couples without such emotional attachment could address partners’ issues individually (e.g., drug addiction, impulsivity, concurrent sexual partners) as part of a process working toward sexual risk communication within relationships.⁵⁸ Interventions should also address gendered forms of conflict as a means to reduce HIV risk.^4,59

Attention to relationship emotions is particularly important in light of the increasing popularity of biomedical and combination approaches to HIV prevention.⁶⁰ Efficacy of many new biomedical prevention approaches (e.g., pre-exposure prophylaxis) is closely tied to adherence behaviors, which are embedded within social and interpersonal contexts.^61,62 Considering relationship factors could help improve combination intervention approaches, including addressing issues related to trust and communication. For structural interventions targeted at female sex workers, particularly efforts aimed at community mobilization and gender equality,^63,64 women may need and want their partner’s emotional support to benefit from empowerment approaches. Building on the supportive aspects of relationships may be a useful alternative to conceptualizing these relationships only as sources of risk.

Limitations and Strengths

Our study has limitations. Our sample included sex workers and their partners who had been together for at least 18 months, representing longer term, more supportive, and less conflictive couples. Additional work is needed to understand the importance of emotions in other relationship contexts. Although attrition may have had an impact on our sample size, we retained three quarters of our socially marginalized cohort after 12 months, likely reflecting our intensive retention strategies.³⁵ Because this was a cross-sectional study, we are unable to infer temporality from the associations we identified; longitudinal work should assess how the interrelationships among love, trust, relationship, and behavioral factors dynamically influence HIV risk over time.

Finally, our attempt to measure love and trust is both a strength and limitation. We acknowledge that love and trust may connote distinct meanings across diverse contexts and that scales impose an etic perspective rather than capturing an emic, or insider, experience that is better suited to ethnography. Although formal scale validity assessment was beyond the scope of this study, future work could use cognitive interviewing to better understand local interpretations of these scales.⁶⁵ Cultural domain analysis and cultural consensus techniques could also help construct culturally relevant emotional domains to measure how partners compare their relationships to idealized models.⁶⁶ However, supported by our ethnographic work and the literature, our use of these scales identified strong correlations between love, trust, and relationship satisfaction, suggesting the importance of emotions in these relationships.

Conclusions

Despite these challenges, our study represents a first effort to quantify the emotional constructs of love and trust within female sex workers’ intimate, noncommercial relationships. Intimacy within these relationships is rarely recognized by mainstream society, yet our results suggest that emotions are of fundamental importance to marginalized couples’ experiences and are interlinked with relationship quality, sexual risk, drug use, and conflict. These factors characterize an important context surrounding HIV vulnerability that prevention interventions should recognize. Researchers should continue to assess how love and trust shape HIV risk among diverse couples to inform prevention programming.