A Dyadic Analysis of Criminal Justice Involvement and Sexual HIV Risk Behaviors Among Drug-Involved Men in Community Corrections and Their Intimate Partners in New York City: Implications for Prevention, Treatment and Policies

Abstract

People in community corrections have rates of HIV and sexual risk behaviors that are much higher than the general population. Prior literature suggests that criminal justice involvement is associated with increased sexual risk behaviors, yet these studies focus on incarceration and use one-sided study designs that only collect data from one partner. To address gaps in the literature, this study used the Actor Partner-Interdependence Model with Structural Equation Modeling (SEM), to perform a dyadic analysis estimating individual (actor-only) partner-only, and dyadic patterns (actor-partner) of criminal justice involvement and greater sexual risks in a sample of 227 men on probation and their intimate partners in New York City, United States. Standard errors were bootstrapped with 10,000 replications to reduce bias in the significance tests. Goodness of fit indices suggested adequate or better model fit for all the models. Significant actor-only relationships included associations between exposures to arrest, misdemeanor convictions, time spent in jail or prison, felony convictions, lifetime number of incarceration events, prior conviction for disorderly conduct and increased sexual risk behaviors. Partner only effects included significant associations between male partners conviction for a violent crime and their female partners’ sexual risk behaviors. Men’s encounters with police and number of prior misdemeanors were associated with their own and intimate partners’ sexual risk behaviors. Women’s prior arrest was associated with their own and intimate partners’ sexual risk behaviors. The results from the present study suggest that men on probation and their intimate partners’ criminal justice involvement are associated with increased engagement in sexual risk behaviors. It is necessary to conduct greater research into developing dyadic sexual risk reduction and HIV/STI prevention interventions for people who are involved in the criminal justice system.

Background

Approximately 6.6 million people are involved in the criminal justice system in the United States, of which 2.1 million are incarcerated, and 4.8 million are in community corrections programs [1]. Community corrections programs include probation, parole and alternative to incarceration programs. More than a quarter of people in community corrections programs—including probation, parole and alternatives to incarceration—are supervised due to illicit drug law violations [1, 2]. Rates of HIV among people in community supervision programs are as high as 12% of men and 17% of women [3]. Prior literature suggests that people who are in community supervision engage in greater rates of condom-less sex, sex under the influence of drugs, sex trading, and have multiple sexual partners compared to people in the general population [4–10]. Prior literature suggests rates of HIV testing of people in community supervision decline following periods of incarceration. Community corrections including courts and probation are opportune settings to conduct HIV prevention research as well as develop interventions to attenuate rates of HIV.

Criminal Justice Involvement and Sexual Risk in Intimate Partner Dyads

Community corrections involves close supervision by probation officers and courts which increases risk of police contact and arrest, potentially leading to incarceration. Probation and court supervision entails mandated services, reporting to officers which increases risk of re-incarceration for technical violations, and additional crimes. Formerly incarcerated people who are on probation return from spending time in jail to existing intimate relationships, may form new sexual partnerships or engage in a combination of sexual behaviors with new and existing partners [11–14]. Reincarceration of people who use drugs in community supervision programs destabilizes intimate partnerships resulting in increased risk of engaging in sexual risk behaviors [13, 15–18].

Prior literature found that individual exposures to incarceration, police questioning, arrest and conviction may increase one’s own [19–21] and their partners’ engagement in sexual risk behaviors [15, 22] and HIV infection [17]. Khan et al. [12] examined data from the National Survey of Family Growth and identified significant associations between incarceration in the past year and engaging in sex with multiple partners and condom-less sex among people who reported incarceration and illicit drug use. Khan et al. [11] examined data from North Carolina and found that men with incarceration in the past 12 months were more likely to report multiple sex partners, and transactional sex compared to men with no incarceration.

Interdependence in the Relationship Between Criminal Justice Involvement and Partners′ Sexual Risk Behaviors

In addition to shaping one’s own risk, criminal justice involvement of intimate dyads is interdependent in which the arrest and incarceration of an individual shapes their partners’ engagement in condom-less sex, sex trading, multiple sex partners and sex under the influence of drugs and alcohol [15]. Epperson et al. [15] identified significant associations between incarceration of male partners and their female partners’ engagement in condom-less sex using data from a sample of women involved in a methadone program and their intimate partners. Davey-Rothwell et al. [23] found that 50% of a sample of 175 women engaged in sexual risk behaviors with other partners during incarceration of their primary partner. Incarceration of intimate partners’ may remove critical emotional support resulting in the formation of other intimate partnerships [24]. Moreover, arrest and detention of intimate partners may place economic strains on partnerships resulting in increased risk of trading sex for food, drugs or money [24]. Intimate partners who are frequently arrested, detained and incarcerated may increase individuals’ risk of engaging in condom-less sex, and sex under the influence of drugs to fulfill needs for intimacy and coping with psychological distress. Dyadic analysis that considers interdependence can shed new insight into how criminal justice involvement of intimate partners shape the occurrence of sexual risk behaviors by considering the shared variance between members of couples [25–31].

SRB and HIV Risk Environment Among People in Community Corrections and Their Partners

The HIV legal risk environment framework examines relationships between ecological factors including criminal justice involvement and sexual risk behaviors within the contexts of intimate relationships [22, 32–40]. The legal risk environment is a multi-level framework that consists of encounters with law enforcement officers, exposures to arrests, length of prior incarceration, drug law enforcement and conditions of confinement. Research is lacking that incorporates interdependence of sexual risk behaviors of intimate partner dyads into the legal risk environment. Interdependence theories presume that social ecological factors including criminal justice involvement influence sexual risk behaviors of intimate partners by viewing the couple as a unit of analysis [41–45]. Conceptualizing the couple as the unit of analysis views experiences in the social environment including criminal justice involvement and sexual risk behaviors as inextricably intertwined. Prior research highlights differences in the legal risk environment between men and women. Women are more likely to be arrested and incarcerated for drug and property crimes, misdemeanors and other low-level crimes [46, 47]. Men are more likely to be arrested and convicted of violent crimes, serve longer sentences in jails and prisons compared to women [48, 49].

Studies are needed that examine the association between multiple kinds of criminal justice involvement and sexual risk behaviors among drug-involved populations on probation and other forms of community corrections. Dyadic analysis with a focus on interdependence sheds additional insights beyond what is currently known in extant literature by expounding upon how multiple forms of criminal justice involvement (police contact, arrest, conviction, incarceration) shape both partners’ engagement in sexual risk behaviors. Research is yet to examine relationships between types of criminal justice involvement and sexual risk behavior using dyadic data of men in community corrections and their intimate partners. Existing research relies on one-sided study designs consisting of self-reports from a single partner about their intimate partners’ sexual risk behaviors which introduces significant bias in existing studies [50]. Dyadic analysis offers an innovative strategy that includes more information about how one’s own and their intimate partners’ criminal justice involvement increase engagement in sexual risk behaviors. Findings from studies into interdependence within couples regarding relationships between individual and partners’ exposures to criminal justice involvement and sexual risk behavior could inform couples-focused HIV prevention interventions for drug-involved men in community supervision and their intimate partners. Prior dyadic research examining interdependence among people who use drugs examine measures of relationship quality including marital distress, satisfaction and conflict without including the criminal justice system [26, 51–58]. Moreover, research disproportionately focuses on how incarceration shapes sexual risk behaviors with few studies including other types of criminal justice involvement such as police contacts, arrests and types of convictions.

To address these gaps, the following study investigated associations between multiple types of criminal justice involvement including contacts with police, arrest, types of conviction as well as incarceration in jails and prisons and sexual risk behaviors of men in community corrections and their intimate partners in New York City. The following study applied an ecological framework to examine how criminal justice involvement shapes sexual risk behaviors among people on probation and their intimate partners. We hypothesized that drug-involved men in community corrections and their intimate partners who were exposed to criminal justice involvement consisting of contact with law enforcement officers, arrest, conviction, spending time in jail and incarceration), will report significantly greater sexual risk behavior (with any partner, other partners and study partner) compared to participants who are not exposed to criminal justice involvement (actor effects). At the partner and dyadic level, this study aimed to identify patterns of interdependence to explain the how partners’ criminal justice involvement is associated with increased risk of engaging in sexual risk behaviors. Guided by an interdependence framework, we hypothesized that exposures to criminal justice involvement would be significantly associated with intimate partners’ engagement in sexual risk behavior after adjusting for possible confounders (partner effects).

Methods

Study Procedures

Data and Sample

Data for this study come from the baseline assessment from Project PACT a biobehavioral randomized clinical trial of an HIV prevention intervention for 230 men on probation and their intimate partners conducted between 2013 to 2016 (n = 460). Men involved in community correction programs (probation, alternatives to sentencing, court diversion programs) were approached by research assistants who handed out study announcement flyers at supervision sites. Informed consent procedures were followed, and a subsequent screening interview determined eligibility for participation in the study. Male participants who met eligibility criteria were asked to invite their main female sexual partner to participate who were administered the same consent procedure and screening for eligibility. Eligible couples were scheduled for a baseline assessment at a date after the screening interview. Couples answered questions individually in a private space using audio computer-assisted self-interviews (ACASI) software. The total compensation for participation in all study elements was 265 dollars which was determined to not be coercive to the population based on review by the Institutional Review Board at Columbia University. Participants answered survey questions measuring sexual HIV risk behaviors, prior criminal justice involvement, prior drug and alcohol use, mental health problems and sociodemographic factors (race/ethnicity, age, education).

Inclusion and Exclusion Criteria

Participants were considered eligible to participate in the study if: (a) both partners were 18 years or older, (b) primary sexual partners of the opposite sex, (c) the total length of the relationship was at least 3 months, (d) one or more incidents of condom-less vaginal sex or anal intercourse with each other reported by at least one partner, (e) a minimum of one exposure to an outside HIV risk in the past year (i.e. condom-less sex with another partner, syringe sharing, HIV positive) or one partner suspected the other partner was exposed to an outside HIV risk, (f) the couple anticipated that they would stay together for at least a year, (g) use of illicit drugs or binge drinking or attending a substance use disorder treatment program in the past 90 days by the male partner, (h) participation by the male partner in community supervision consisting of alternative to incarceration or probation as documented by court records. Couples were excluded if either partner (a) exhibited impairment of psychiatric or cognitive faculties as demonstrated in the informed consent process, or (b) any self-reported safety concerns about participating in the study (i.e. order of protection reported by either partner in the past year).

The study screened 1167 individuals of which 258 were ineligible for the following reasons (1) no outside risk, (2) no unprotected sex or outside risk in prior 90 days, 3) no substance use or treatment in past 90 days, (4) female partner under the age of 18, (5) planning to move, (6) no probation or community supervision, (7) order of protection or felt unsafe with partner, (8) no partner, (10) not interested, (11) did not respond to mail or phone invitations for baseline, and (12) did not return for the first intervention assessment. Out of the male participants who screened eligible, 351 of their female partners did not consent to screening and 76 couples were excluded because of no outside risk, not interested, did not respond to mail or phone, and did not return for assessment in the intervention resulting in a sample size of 230 couples. Out of the total sample, 3 (1.3%) couples were excluded from analysis due to missing data resulting in a final sample size of 227 couples (n = 454).

Measures

Sexual HIV Risks

This study created an index of sexual behavior using the Risk Behavior Assessment of sexual and injection drug HIV risk behaviors that was validated in prior research [59, 60]. The decision to generate an index score of engaging in sexual risk behaviors was based on previously conducted studies [61–66]. The decision to generate an index score of sexual risk behaviors is aligned with the core hypotheses put forth in the study to estimate the actor-partner effects of criminal justice involvement on the quantity of sexual risk behaviors with other and study partners.

Two scales were created that summed question items to measure sexual risks with study and other partners. Scales were calculated for male (MP) and female partners (FP). Sexual risks with study partners (alpha: MP = .71, FP = .78) consisted of (1) condom-less vaginal sex, (2) condom-less anal sex, and (3) sex under the influence of drugs or alcohol. Sexual risks with other partners (alpha: MP = .74, FP = .81) consisted of (1) engaging in condom-less vaginal sex, (2) anal sex with other partners, (3) sex under the influence of drugs with other partners and (4) trading sex for money, drugs, shelter, or food. Twelve response categories were created for each of the continuous sexual risk variables that classified question items to reflect engaging in sexual risk behaviors consisting of: (1) 1–2, (2) 3–4, (3) 5–10, (4) 11–20, (5) 21–30, (6) 31–40, (7) 41–50, (8) 51–60, (9) 61–70, (10) 71–80, (11) 81–90 and (12) 90 + times.

Criminal Justice Involvement

Participants answered several questions to measure history of exposure to the criminal justice system that included dichotomous variables (1 = yes, 0 = no) of lifetime and 90-day exposure to (1) contacts with police, (2) arrest, (3) conviction type (misdemeanor or felony offenses) and (4) reasons for conviction. Participants were asked the number of (1) contacts with law enforcement, (2) arrests, (3) misdemeanor and (4) felony convictions. Type of conviction included (1) disorderly conduct, (2) violent crimes (3) drug-related offenses, and (4) other (public intoxication from alcohol or drugs, driving under the influence of alcohol or drug, forgery/fraud, buying and selling stolen goods, illegal gambling, prostitution, probation/parole violations, vandalism, property damage, burglary/attempted larceny/auto theft/carjacking, shoplifting/larceny/embezzlement). Incarceration included dichotomous question items measuring lifetime and 90-day exposures to spending a night in a jail or sentenced to prison. Continuous measures included number of jail episodes, times in prison as well as total time spent in prison.

Control Variables

Control covariates in Project PACT include relationship quality (MP alpha = .78, FP alpha = .81), black race, Hispanic ethnicity, drug use, binge drinking, and depression. Drug use severity included a scale that summed the number of days drugs were used in the past month (MP alpha = .89, MP alpha = . 86). Depression was measured using the depression subscale of the Brief Symptom Index (BSI) MP alpha = .88, FP = .86) a highly validated measure in prior research [67]. Binge drinking was measured by asking men if they had consumed 6 drinks or more and women if they had consumed 5 or more drinks in one occasion.

Statistical Analyses

Descriptive Analyses

Descriptive statistics (mean, percentages, median, confidence and standard deviations) for CJI, control variables drug use and sexual risk, illustrated the characteristics of the sample of intimate partner dyads of men on community corrections and their intimate partners. Mann–Whitney and Pearson correlations performed bivariate tests of differences between one’s exposures to criminal justice involvement and their own sexual risks and their partners’ sexual risks [68]. To avoid potential collinearity, significant variables at p < .05 in bivariate analyses were included in separate APIM models using SEM to test hypotheses about associations between different types of criminal justice involvement and sexual risks after adjusting for covariates of relationship quality, black race, Hispanic ethnicity, drug use, binge drinking, and depression.

Multivariate Analyses

The nested couple-level data warranted analytic techniques that accounted for interdependence of the data [20, 69–71]. This study used the actor partner interdependence model (APIM) to account for correlation between responses of each partner and identify several potential dyadic patterns [20, 28, 72, 73]. The Actor-Partner Interdependence Model (APIM) using Structural Equation Modeling (SEM) identified dyadic patterns by estimated the association between CJI and a scale of sexual HIV risks to identify underlying dyadic patterns after adjusting for potential confounders [69, 71–76]. The APIM with SEM correlates the independent variables and the error terms of the dependent variables to account for interdependence between responses of male and female partners [69, 70, 74–79].

Preliminary analyses identified significant correlations between sexual risk behaviors between male and female partners which confirmed interdependence between men on probation and their intimate partners requiring use of the APIM to account for the dyadic structure of the data (Table 1). Goodness of fit indices assessed model quality using Chi square (p value > .05), Root Mean Squared Errors of Approximation (RMSEA) (< .06), Comparative Fit Index (CFI) (< .90), Tucker Lewis Index (TLI) (< .90), and Standardized Root Mean Residual (SRMR) (< .08) [80–82]. Analyses were performed using SEM functions in MPLUS Version 8 [83–85].

Table 1.

Bivariate correlation tests for interdependence between partners’ sexual risk scores

	Partner
Sexual risks
Study	.48*
Other	.42*
Any	.52*

^*p < .05

Results

Descriptive Characteristics of Intimate Partner Dyads

Table 1 presents descriptive statistics of CJI and control variables of drug involved men in community corrections and their intimate partners. A majority of the sample was black accounting for 77.19% (n = 176) of male and 69.16% (n = 157) of female partners. The mean age of female partners was 34.53 years (SD = 13.15) and 35.66 years for male partners (12.45). The mean scores of the scale of sexual risks for study partners was 6.38 (SE = .35) for male and 5.94 (SE = .34) for female partners. The mean score of the scale of sexual risks with other partners was 2.41 (SE = .31) for male and 1.60 (SE = .27) of female partners (Fig. 1, Table 2).

Fig. 1.

Hypothesized actor and partner pathways between criminal justice involvement and sexual risks among men in community corrections and their intimate partners

Table 2.

Descriptive statistics of criminal justice involvement and control variables of drug involved men in community corrections and their intimate partners

	Male partner	Female partner
Criminal justice involvement
Stopped by police but not arrested
Lifetime	87.72 (200)	59.91 (136)
Number of times mean (SD)	29.42 (5.45)	3.74 (.63)
90 days %(n)	44.10 (101)	20.09 (46)
Number of times mean (SD)	3.13 (.51)	.83 (.27)
Arrest % (n)
Lifetime % (n)	96.93 (221)	62.11 (141)
Number of arrests mean (SD)	15.51 (1.36)	4.24 (.59)
90 days % (n)	28.95 (66)	8.81 (20)
Number of arrests mean (SD)	.82 (.39)	.12 (.03)
Conviction history
Misdemeanor
Lifetime % (n)	80.26 (183)	34.36 (78)
Mean	7.05 (.78)	1.70 (.35)
90 days % (n)	22.81 (52)	5.29 (12)
Felony
Lifetime % (n)	67.98 (155)	16.74 (38)
Number of convictions mean (SD)	1.69 (.17)	.33 (.07)
90 days % (n)	6.11 (14)	1.75 (4)
Type of conviction
Disorderly conduct
Lifetime % (n)	38.16 (87)	17.62 (40)
90 days % (n)	3.07 (7)	2.64 (6)
Drug law offences
Use or possession of illegal drugs
Lifetime % (n)	52.63 (120)	19.82 (45)
90 days % (n)	7.02 (16)	3.08 (7)
Possession of drug paraphernalia
Lifetime % (n)	22.81 (52)	14.10 (32)
90 days % (n)	2.19 (5)	2.20 (5)
Sale or distribution of drugs
Lifetime % (n)	35.53 (81)	12.78 (29)
90 days % (n)	.88 (2)	1.32 (3)
Public intoxication alcohol/drugs
Lifetime % (n)	12.72 (29)	5.29 (12)
90 days % (n)	2.63 (6)	1.32 (3)
Any drug law violation
Lifetime % (n)	58.33 (133)	23.79 (54)
90 days % (n)	9.65 (22)	3.96 (9)
Property law offenses
Buy/sell stolen goods
Lifetime % (n)	7.46 (17)	.44 (1)
90 days % (n)	.44 (1)	0 (0)
Vandalism/Property damage
Lifetime % (n)	5.70 (13)	1.76 (4)
90 days % (n)	0 (0)	.44 (1)
Burglary
Lifetime % (n)	14.47 (33)	.88 (2)
90 days % (n)	.88 (2)	0 (0)
Shoplifting
Lifetime % (n)	18.86 (43)	13.66 (31)
90 days % (n)	3.06 (7)	1.75 (4)
Any property crime
Lifetime % (n)	34.06 (78)	16.16 (37)
90 days % (n)	5.70 (13)	2.64 (6)
Violent crime
Assault
Lifetime % (n)	21.93 (50)	8.81 (20)
90 days % (n)	1.32 (3)	1.32 (3)
Manslaughter
Lifetime % (n)	2.19 (5)	.88 (2)
90 days % (n)	0 (0)	0 (0)
Any violent crime
Lifetime % (n)	22.71 (52)	9.17 (21)
90 days % (n)	1.32 (3)	1.32 (3)
Weapons offenses
Lifetime % (n)	16.67 (38)	2.20 (5)
90 days % (n)	2.19 (5)	.88 (2)
Other
Driving under the influence of drugs/alcohol
Lifetime % (n)	7.46 (17)	1.76 (4)
90 days % (n)	0 (0)	0 (0)
Forgery/Fraud
Lifetime % (n)	8.33 (19)	7.05 (16)
90 days % (n)	0 (0)	.44 (1)
Illegal gambling
Lifetime % (n)	1.75 (4)	.88 (2)
90 days % (n)	0 (0)	.44 (1)
Prostitution
Lifetime % (n)	0 (228)	5.73 (13)
90 days % (n)	0 (0)	.44(1)
Any other crime
Lifetime % (n)	15.35 (35)	12.33 (28)
90 days % (n)	0 (0)	.44 (1)
Probation/parole violation
Lifetime % (n)	29.82 (68)	8.81 (20)
90 days % (n)	2.19 (5)	.44 (1)
Incarceration
Jail
Ever %(n)	81.58 (186)	42.73 (97)
Number of days mean(SD)	13.63 (2.89)	7.54 (2.49)
90 days %(n)	22.37 (51)	4.85 (11)
Sentenced to prison
Ever % (n)	39.91(91)	10.13 (23)
Number of times mean (SD)	1.29 (.17)	.31 (.11)
Total time in prison days mean (SD)	32.19 (4.66)	6.46 (2.10)
Control covariates
Hispanic % (n)	31.14 (71)	29.52 (67)
Race % (n)
Black	77.19 (176)	69.16 (157)
White	9.21 (21)	12.33 (28)
Asian	.44 (1)	1.32 (3)
Native	3.51 (8)	6.17 (14)
Other	17.11 (39)	18.90 (86)
Age mean (SD)	35.66 (12.45)	34.53 (13.15)
Depression Mean (SD)	39.22 (.70)	35.83 (.82)
Drug use Mean (SD)	22.96 (5.04)	17.71 (2.61)
Binge drinking % (n)	33.33 (76)	23.79 (54)
Relationship satisfaction Mean (SD)	29.19 (.26)	28.45 (.25)
Relationship length years Mean (SD)	1.21 (.10)

Actor-Partner Interdependence Models of Associations Between CJI and Sexual HIV Risks

Table 3 presents findings from actor partner interdependence models of associations between criminal justice involvement and sexual HIV risks with study and other partners. Goodness of fit statistics show acceptable fit for each of the models using Chi-sq, RMSEA, CFI, TLI, and SRMR indices except for the model with arrest as an independent variable which suggested a poor fit.

Table 3.

Descriptive characteristics of sexual risk variables of drug-involved men in community corrections and their intimate partners

	Male			Female
	Mean	(SE)	Range	Mean	(SE)	Range
Condom-less vaginal or anal sex with SP
Number of times	51.48	(8.74)	0–1019	38.62	(5.54)	0–999
Categorical	4.68	(.27)	0–12	4.61	(3.91)	0-12
Anal sex with study partner
Number of times	21.19	(4.28)	(0–91)	11.12	(4.28)	(0–48)
Categorical	1.9	(.18)	(0–12)	1.2	(.18)	(0–12)
Vaginal sex influence of drugs or alcohol SP
Number of times	6.55	(.90)	0–90	5.52	(.93)	0–90
Categorical	1.60	(.14)	0–11)	1.33	(.14)	0–11
Condom-less vaginal sex with other partners
Number of times	2.75	(.93)	0–150	.82	(.31)	0–60
Categorical	.57	(.11)	0–12	.26	(.06)	0–8
Sex under influence of drugs or alcohol OP
Number of times	1.45	(.38)	0–70	.88	(.23)	0–36
Categorical	.48	(.08)	0–12	.32	(.06)	0–6
Exchanged sex for money, drugs or housing
Number of times	3.47	(1.63)	0–300	3.17	(1.16)	0–210
Categorical	.40	(.11)	0–12	.50	(1.70)	(0–12)
Number of sexual partners
Number of times	2.11	.25	0–20	1.10	.27	0–50
Categorical	.95	(.12)	0–7	.52	(.09)	0–7
Sexual risks scales
Sexual risks with all partners	8.53	(51)	0–52	7.43	(.45)	0–41
Study partner	6.38	(.35)	0–23	5.94	(.34)	0–23
Other partners	2.41	(.31)	0–33	1.60	(.27)	0–33

SP study partner, OP other partners

Hypotheses 1: Actor Effects

Contacts with Police

Results from the APIM identified statistically significant associations between contact with police in the past 90 days of male partners who were stopped and increased engagement in sexual risk behaviors with their study partners (a_m:β = .11, CI₉₅ = .00, .23, p < .05) and with other partners (a_m: β = .13, CI₉₅ = .00, .25, p < .05).

Arrest

Men’s exposure to arrest in their lifetime (a_m: β = .11, CI₉₅ = .00, .20, p < .05) and 90 days (a_m: β = .14, CI₉₅ = .02, .27, p < .05) was associated with greater engagement in sexual risk behaviors with other partners. Lifetime exposure to arrest of female partners was associated with increased engagement in sexual risk behaviors with their study partners (a_f: β = .14, CI₉₅ = .06, .23, p < .05).

Conviction History

Lifetime conviction for a misdemeanor of female partners was associated with increased engagement in sexual risk behaviors with their study partners (a_f: β = .10, CI₉₅ = .01, .21, p < .05) and other partners (a_f: β = .12, CI₉₅ = .00, .24, p < .05). The number of prior misdemeanor convictions of male partners was associated with increased engagement in sexual risks with any partner (a_m: β = .15, CI₉₅ = .04, .28, p < .05) and study partners (a_m: β = .13, CI₉₅ = .03, .25, p < .05). Female partners’ felony conviction was associated with increased engagement in sexual risk behaviors with their study partners (a_f: β = .21, CI₉₅ = .03, .38, p < .05). For male partners, the greater number of prior felony convictions was associated with increased engagement in sexual risk behaviors with their study partners (a_m: β = .20, CI₉₅ = .01, .34, p < .05) (Table 4).

Table 4.

Multivariable regression results actor and partner effects of criminal justice involvement and sexual risks with study and other partners of drug-involved men in community corrections and their intimate partners

	Study partner						Other Partners
	am		pm		af	pf	am		pm		af	pf
	β (CI95 %)		β (CI95 %)		β (CI95 %)	β (CI95 %)	β (CI95 %)		β (CI95 %)		β (CI95 %)	β (CI95 %)
Police contact 90 Days1	.07 (− .06, .22)		.11 (− .03, .26)		.04 (− .09, .16)	− .05 (− .16, .07)	.13* (.003, .25)		.20* (.08, .33)		− .04 (− .15, .07)	− .08 (− .20, .05)
Arrest
Lifetime2	.02 (− .08, .09)		.03 (− .18, .08)		.14* (.06, .23)	.10* (.00, .20)	.11* (.00, .20)		.10 (− .03, .20)		.03 (− .10, .16)	.10 (− .13, .12)
90 Days3	.05 (− .11, .18)		.16 (.05, .27)		.05 (− .10, .21)	− .09 (− .20, .033)	.14* (.02, .27)		.05 (− .09, .18)		.12 (− .02, .20)	− .03 (− .14, .89)
Conviction history
Misdemeanor
Lifetime4	.06 (.06, .16)		− .02 (− .13, .08)		.10* (.01, .21)	.05 (− .08, .18)	.12* (.00, .24)		.08 (− .05, .21)		− .03 (− .05, .21)	.11* (.01, .22)
Number of times5	.13* (.03, .25)		.00 (− .09, .10)		.04 (− .08, .24)	.09 (− .07, .27)	.11 (− .05, .26)		− .02 − .13, .10)		− .06 (− .14, .08)	− .10 (− .21, .03)
Felony
Lifetime6	.03 (− .11, .15)		− .04 (− .15, .06)		.21* (.03, .38)	.02 (− .09, .15)	.04 (− .07, .16)		.02 (− .11, .12)		− .01 (− .12, .13)	− .09 (− .19, .03)
Number of times7	.20* (.01, .34)		− .01 (− .07, .07)		.09 (− .04, .26)	.10 (.03, .29)	.01 (− .08, .10)		.03 (− .06,.15)		− .04 (− .15, .13)	− .07 (− .14, .04)
Lifetime conviction type
Disorderly conduct8	10* (.02, .22)		− .03 (− .14, .10)		.10 (− .03, .27)	− .04 (− .14, .09)	.17*** (.05, .29)		.15** (.03, .29)		.14 (− .06, .16)	.02 (− .09, .15)
Drug-related offenses9	.08 (− .04, .20)		.11 (− .03, .25)		.05 (− .09, .20)	.01 (− .11, .15)	.10 (− .01, .20)		.03 (− .07, .14)		.28* (.14, .42)	.11 (− .03, .27)
Violent Crimes10	.09 (− .05, .23)		.18** (.10, .28)		.13 (− .04, .31)	.07 (− .05, .22)	− .05 (− .17, .07)		.02 (− .09,.5)		− .04 (− .15, .08)	− .04 (− .16, .07)
Other11	.01 (− .10, .13)		.00 (− .11, .11)		− .01 (− .13, .13)	.00 (.11, .12)	− .01. (.13, .14)		.04 (− .10, .19)		.33** (.08, .59)	.12 (− .02, .26)
Jail
Ever12	.02 (− .11, .12)		− .04 (− .18, .11)		.14* (.01, .27)	.00 (− .12, .14)	.01 (− .10, .10)		.01 (.13, .15)		.07 (− .05, .22)	.12 (− .01, .23)
Sentenced to prison
Ever13	.08 (− .04, .21)		.01 (− .10, .14)		− .02 (− .14, .12)	− .05 (− .16, .07)	.15* (.02, .29)		.03 (− .08, .19)		.15 (− .05, .36)	.04 (− .09, .19)
Time in prison14	.07 (− .05, .18)		− .03 (− .12, .09)		− .06 (− .19, .10)	− .08 (− .17, .01)	.13 (.00, .28)*		.01 (− .15, .14)		.21 (− .04, .42)	− .06 (− .23, .06)
Total time inprison15	.05 (− .03, .15)		.02 (− .07, .13)		.04. (.10, .19)	− .04 (− .11, .03)	.10* (.01, .20)		.03 (− .05, .11)		.25* (.00, .43)	.03 (− .06, .14)
	df	χ2	RMSEA	CFI	TLI	SRMR	df	χ2	RMSEA	CFI	TLI	SRMR
1GOF	20	26.14	.037	.943	.928	.05	20	26.14	.037	.943	.928	.05
2GOF	20	20.38	.009	.997	.996	.037	20	20.38	.009	.997	.996	.037
3GOF	20	44.83*	.06	.842	.820	.06	20	44.83*	.06	.842	.820	.06
4GOF	20	19.21	< .001	1.00	1.00	.047	20	19.21	< .001	1.00	1.00	.047
5GOF	20	11.50	< .001	1.00	1.00	.031	20	11.50	< .001	1.00	1.00	.031
6GOF	20	28.87	.044	.923	.904	.05	20	28.87	.044	.923	.904	.05
7GOF	20	18.16	< .001	1.0	1.0	.037	20	18.16	< .001	1.0	1.0	.037
8GOF	20	12.68	< .001	1.0	1.0	.036	20	12.68	< .001	1.0	1.0	.036
9GOF	20	31.81	.051	.935	.920	.065	20	31.81	.051	.935	.920	.065
10GOF	20	28.23	.043	.932	915	.045	20	28.23	.043	.932	915	.045
11GOF	20	27.81	.041	.931	.913	.05	20	27.81	.041	.931	.913	.05
12GOF	20	30.72	.049	.91	.904	.049	20	30.72	.049	.91	.904	.049
13GOF	20	29.11	.045	.921	.901	.053	20	29.11	.045	.921	.901	.053
14GOF	20	11.62	< .001	1.0	1.0	.031	20	11.62	< .001	1.0	1.0	.031
15GOF	20	11.17	< .001	1.00	1.0	.033	20	11.17	< .001	1.00	1.0	.033

Supercript numbers reflect corresponding goodness of fit statistics for each criminal justice variable included in the APIM models

GOF Goodness of fit, df Degrees of freedom, χ² Chi square test, RMSEA Root Mean Square Error of Approximation, CFI Confirmatory Factor Index, TLI Tucker Louis Index, SRMR Square-Root Mean Residual

Bold indicates significance p < .05

Conviction Type

Male partners’ conviction for disorderly conduct was associated with increased engagement in sexual risks behaviors with their study partners (a_m: β = .10, CI₉₅ = .02, .22, p < .05) and other partners (a_m: β = .17, CI₉₅ = .05, .29, p < .05). Lifetime conviction for drug law offenses of female partners was associated with increased sexual risk with other partners (a_f: β = .28, CI₉₅ = .14, .42, p < .05) and any partner (a_f: β = .21, CI₉₅ = .06, 38, p < .05).

Incarceration

Spending a night in jail for female partners was associated with greater engagement in sexual risk behaviors with their study partners (a_f: β = .14, CI₉₅ = .01, .27, p < .05). Ever receiving a prison sentence for male partners was associated with increased sexual risks with other partners (a_m: β = .15, CI₉₅ = .02, .29, p < .05). Greater number of times receiving a sentence to prison for male partners was associated with increased engagement in sex risk behaviors with other partners (a_m: β=.13, CI₉₅ = .00, .28, p < .05). Greater months incarcerated in prison of male (a_m: β = .10, CI₉₅ = .01, .20, p < .05) and female (a_f: β = .25, CI₉₅ = .00, .43, p < .05) partners was associated with increased engagement in sexual risk behaviors with other partners.

Hypothesis 2: Partner Effects

Contacts with the Police

Contact with police for male partners were significantly associated with their female partner’s sexual risk behaviors (p_m: β = .20, CI₉₅ = .08, .33, p < .05) with other partners.

Arrest

For female partners, being arrested was associated with their partners’ engagement in sexual risk behaviors (p_f: β = .10, CI₉₅ = .00, .20, p < . 05) with study partners. Exposure to arrest in the past 90 days for male partners was associated with increased engagement in sexual risk behaviors with study partners for female partners (p_m: β = .16, CI₉₅ = .05, .27, p < .05).

Conviction History

Female partners’ lifetime conviction for a misdemeanor was associated with increased sexual risk behaviors with others for their male partners (p_f: β = .11, CI₉₅ = .01, .22, p < .05).

Conviction Type

Male partners’ conviction for disorderly conduct was associated with their female partners’ sexual risk behaviors with other partners (p_m: β = .15, CI₉₅ = .03, .29, p < .05). Male partners’ conviction for a violent crime was associated with their female partners’ sexual risk behaviors with study partners (p_m: β = .18, CI₉₅ = .10, .28, p < .05).

Discussion

This study addressed a gap in prior literature by examining the association between types of criminal justice involvement and one’s own and their intimate partners’ sexual risk behaviors in a sample of men in community corrections and their intimate partners. Findings supported several hypotheses that criminal justice involvement would be associated with increased sexual risk behaviors among men in community corrections and their intimate partners. Hypotheses 1 For male partners, path analyses identified actor-only patterns in which exposures to arrest, lifetime number of incarceration events, total time spent in prison, prior conviction for disorderly conduct, and increased engagement in sexual risks with other partners. An actor-only pattern was identified for male partners in which exposure to disorderly conduct was associated with increased sexual risks with any and study partners. For female partners, actor-only patterns consisted of associations between prior misdemeanor ever spending time in jail, total time in jail/prison, felony convictions and engaging in greater sexual risk behaviors with study and other partners.

Hypotheses 2 Partner-only patterns were observed in the association between male partners’ conviction for a violent crime and their female partners’ sexual risk behaviors with study partners and any partner. Findings from this study identified significant actor-partner patterns in which male partners’ encounters with police and greater number of exposures to misdemeanor convictions were associated with their own and their partners sexual risk behaviors with other partners as well as with any partners. For female partners, an actor-partner patternwas identified consisting of an association between engaging in sexual risk behaviors with study partners for both intimate partners.

Implications for Public Health Practice and Policy

Findings from this study provide several avenues for future research and implications for HIV prevention interventions and policy. This study expanded prior literature by focusing on a previously unstudied population of people on probation who are disproportionately at risk of HIV infection. Findings of actor and partner effects incorporated an ecological study design that contribute to extant literature on interdependence in partners’ criminal justice system involvement and sexual risk behaviors. This study emphasizes the importance of community corrections as critical settings to collaborate with public health to prevent HIV infection through testing and education among justice involved people and their intimate partners.

Couples-focused interventions deliver an HIV risk reduction intervention to both partners that aims to reduce condom-less sex by increasing confidence negotiating and insisting on the use of condoms and reducing relationship power inequities [86, 87]. Future research must investigate the potential benefits of expanding couples-focused interventions into community corrections settings. Finding from this study support existing literature calling for greater use of dyadic frameworks in models of HIV prevention and treatment [88]. Future research must expand on existing individual level research [89] to conduct analyses at the dyadic-level of the effects of partner’s incarceration on non-adherence to HIV antiretroviral therapy and HIV treatment.

Frequent police contact may undermine mechanisms of trust, intimacy and relationship satisfaction which could result in engaging in sexual risk behaviors with other partners. Actor-partner effects in the association between convictions for disorderly conduct and increased sexual risk behaviors suggest that exposures to ‘quality of life policing’ may disrupt intimate partnerships and undermine important relationship factors. Disorderly conduct consists of “quality of life” or “nuisance crimes” crimes including behavior that causes annoyance, alarm, unreasonable noise, or exhibits recklessness and results in a fine and potential jail sentence [90–92]. These findings also support future research to investigate the effects of reforming policies around low-level offenses such as quality of life offenses and broken windows policing on sexual risks of drug-involved men in community corrections and their intimate partners. Future research must examine more detailed aspects of police encounters including whether the contact involved questioning, physical force, or restraints.

Female partners with lifetime exposure to arrest may be more likely to select intimate partners with arrest histories because of the available suitable dating partners in their communities [94, 95]. Arrest in the past 90 days may disrupt intimate partnerships resulting in female partners having fewer resources available to negotiate condom use and other HIV risk reduction strategies. Findings from this study call for additional research to investigate if arrest in the past 90 days increases relationship-based power imbalances for women to negotiate HIV risk reduction sex practices with their intimate partners.

Felony and misdemeanor convictions often result in a period of time in which partners are separated from their intimate partners resulting in dissolution or breaks in intimate relationships that could result in greater engagement in sexual risks after release into the community. Findings from this study are aligned with prior research suggesting that conviction history particularly for felony crimes may strain intimate partnerships by imposing exclusions from employment, financial demands of legal representation, and detention that thereby increase sexual risk behaviors [93]. Additional research is necessary to identify if the financial demands imposed by conviction increases sexual risk behaviors for men in community corrections and their intimate partners. Male partners who commit violent crimes may coerce their intimate partners into sexual risk behaviors. Future research must investigate if men who are convicted of violent offenses are more likely to be controlling with their partners and establish power differentials that constrain women’s capacity to practice safe sex within intimate partnerships. Dyadic research must examine the association between aggressive behaviors, intimate partner violence as well as violence with others and the occurrence of sexual risk behaviors with study partners and other partners.

Findings from this study support future dyadic research examining the effects of providing HIV prevention interventions during the period immediately following incarceration on reducing HIV sexual risk behaviors. A greater proportion of women who are incarcerated spend time in jails compared to prisons for their offenses because of a greater likelihood to commit less severe crimes that result in sentences for less than a year. Findings from this study are consistent with prior research suggesting jails impose interruptions on intimate relationships, financial resources, and social support in the community [94, 95]. Future research must investigate if spending time in jail erodes the support structures that increase female partners’ agency and resources to engage in HIV sexual risk reduction practices.

Limitations

This study has several limitations. This study did not measure if criminal justice involvement occurred during or prior to the relationship. Future research must include measurements of criminal justice involvement within a wider timeframe of a year in addition to 90 days to establish a more accurate time frame for measuring significant associations between exposures to criminal justice involvement and sexual risks. Future dyadic research with larger samples of couples are needed to elucidate the associations between types of recent convictions and sexual risk behaviors. The variable of police contacts did not measure details of the contacts including use of force, misconduct or restraint. The non-random sampling design of this study selected high sexual risk couples of men in community corrections and their intimate partners thus restricting the generalizability of the sample to other samples of men in community corrections and their intimate partners in urban settings. This study selected couples based on male partners involvement in the criminal justice system in the past year and both partners’ high-risk sexual behaviors. This results in a different probability distribution of the criminal justice involvement variables between partners. This study is cross-sectional thus precluding causal inference of study findings. Two of the sexual risk variables included sex under the influence of drugs and alcohol and sex trading which are criminalized behaviors. This raises the possibility that associations may operate in the opposite direction for some of the criminal justice variables including drug crime conviction, arrest, and misdemeanor conviction. Therefore, caution is necessary not to imply directionality of the findings put forward in this study.

Conclusion

Limitations notwithstanding this study addressed a significant gap in the existing literature by examining associations between exposures to multiple factors of the HIV legal risk environment and engaging in sexual risks among men in community corrections and their intimate partners. Drug involved men in community corrections particularly African American men are a key affected population with rates of HIV that are much higher than the general population. Mass incarceration in the United States disproportionately impacts people who use drugs and their intimate partners and has contributed to disparities in rates of HIV and sexual health particularly among minority populations. It is critical that HIV prevention research incorporate intimate partner dyads into ecological research into factors of the HIV legal risk environment that shape the occurrence of sexual HIV risk behaviors among men in community corrections and their intimate partners.