In the United States, one in every 100 adults is currently incarcerated, the highest rate in the world.1 Racial and ethnic disparities in the criminal justice experience parallel those observed in HIV/AIDS, with non-Hispanic blacks (NHBs) and Hispanics representing about 12% and 16% of the general population, respectively,2 but about 41% and 22% of those incarcerated,3 and 49% and 20% of those recently diagnosed with AIDS.4 Better understanding the determinants of behavior that facilitate HIV transmission among the population that experiences incarceration may be critical for understanding both overall HIV/AIDS epidemiology and trends in racial/ethnic disparities in HIV/AIDS.
Incarceration affects the inmates' sexual partnerships, social networks, families, and communities, and these effects may extend throughout the period of community supervision. People on probation and parole are required to meet regularly with a parole or probation officer and abide by supervision conditions. In 2007, 5.1 million people were on probation or parole.5 Returning to the community may also entail a return to drug use and sexual behaviors and, for some, to an increase beyond preincarceration levels.6 Prisoners returning to the community may also have limited social and economic opportunities, often experiencing homelessness, poverty, and unstable living circumstances, which can contribute to vulnerability to HIV infection. People under community supervision can be especially vulnerable to population determinants of sexual behavior and may be more likely to engage in relationships with risky sex partners. For example, high rates of relationship dissolution, difficulties finding housing or employment on release, and relapse to drug use may encourage some ex-inmates to engage in sexual relationships in exchange for housing or money. The relative availability of partners in communities where inmates are released may facilitate or impede development of sexual relationships with multiple partners, or with partners that may be more likely to be infected with HIV or other sexually transmitted infections (STIs) themselves, or that may belong to a sexual network that includes infected individuals.
The male-female sex ratio and the male incarceration rate may represent structural or population factors outside of individual control that can affect sexual behavior and membership in sexual networks that may help explain the link between racial/ethnic disparities in HIV prevalence and in incarceration. The sex ratio is calculated as the number of men per 100 women. Although slightly more boys are born than girls, women tend to outnumber men in adulthood, as male mortality rate surpasses that of females throughout the lifespan. Incarceration, military service, and migration can also lower the sex ratio. These effects may be more pronounced in some geographic areas and racial/ethnic groups. For example, a high black male incarceration rate may have a large impact on a city's or county's sex ratio for blacks in the reproductive age range (ages: 15–49 years), but may not be noticeable at the neighborhood or state level, and may have no association with sex ratios of other racial/ethnic or age-groups. A low sex ratio, indicating a shortage of men, reflects an increased demographic opportunity for the men remaining in the community to accrue more heterosexual partners. For people on probation or parole, opportunities for sexual partnerships with low-risk partners may be limited by the assortative nature of sexual partnerships and by the perception among potential partners that ex-offenders are members of a high-risk group.
Although the incarceration rate may directly relate to HIV incidence through incarceration of injection drug users, it may also serve as an independent indicator of the rate of relationship disruption or of relationship vulnerability.6,7 Among men with primary partners, incarceration for longer than a month can lead to relationship dissolution and to having multiple sex partners on release.6 Having a partner who was recently incarcerated has been associated with having multiple, transactional, concurrent, and risky partners, as well as with having had a recent HIV or STI diagnosis.8,9 Partner characteristics are likely to reflect characteristics of risk networks. Thus, incarceration may decrease the ability of former prisoners to maintain safer social and sexual networks they may have had before incarceration. Few studies have examined sexual risk behavior during parole and probation,10–14 with most focused on individual knowledge and attitudes or relationship factors.
This study aimed to explore the influence, before and during community supervision, of 2 potential population determinants of sexual behavior thought to increase exposure to HIV and other STIs: the sex ratio and the male incarceration rate in the county of residence.
MATERIALS AND METHODS
The data came from the 2 completed studies with overlapping instruments that were part of the Criminal Justice Drug Abuse Treatment Studies-1 (CJDATS) cooperative agreement. One study, Transitional Case Management, tested whether case management during reentry increases participation in community drug abuse treatment, enhances access to social services, and improves outcomes during a 9-month follow-up. Eligible participants were (a) at least aged 18 years; (b) referred to a community-based substance abuse treatment program; (c) within 2 to 3 months of release; and (d) released to a metropolitan area where case management activities were being conducted. The second study, Step'n Out, examined the effect of collaborative behavioral management that integrated treatment counseling and community supervision versus standard parole on crime, drug use, and rearrest during 9-month follow-up. Parolees with preincarceration substance use disorders at moderate-to-high risk of recidivism constituted the target population. Inclusion criteria were (a) at least 18 years old; (b) English speaking; (c) probable drug dependence immediately before incarceration (i.e., Texas Christian University [TCU] Drug Screen II15 score ≥3 or mandated drug treatment); (d) substance use treatment as a mandated or recommended condition of parole; (e) moderate-to-high risk of drug relapse and/or recidivism (i.e., Lifestyle Criminality Screening Form [LCSF] score ≥716 or a history of ≥2 prior drug abuse treatment episodes or drug-related convictions). Neither of the studies aimed to alter HIV risk behavior, and intervention content did not directly address HIV. Study descriptions and findings are reported elsewhere.17,18
Data for the present analysis included 1287 drug-involved participants on probation and parole across 23 locations and up to 3 time points. Study sites were located in Rhode Island, Connecticut, Delaware, Virginia, Oregon, Kentucky, and Colorado.
Baseline self-reported age, race/ethnicity, gender and marital status (married/living as married with partner = 1, else 0), a time-varying variable, were assessed at each interview.
Using decennial census data, we estimated 3 population factors: sex ratios, incarceration rates, and poverty rates for the participants' counties of recruitment and follow-up, which were used as an indicator for county of residence. Poverty rate served as a control variable. County was assumed to approximate the catchment area of potential sexual partners. Two separate sex ratios were calculated for each racial/ethnic group. An assortative or matched sex ratio was calculated for the NHB, non-Hispanic white (NHW), and Hispanic populations. A disassortative sex ratio was also calculated for each race/ethnicity using the sum of the population in racial/ethnic categories other than that of the target racial/ethnic group. To facilitate comparisons, a median split defined categories of the population factors for each race/ethnicity and gender, where low sex ratios and high incarceration and poverty rates were dummy coded (1/0). In locations with sparse data for a given racial category, the population factor was set to missing, but was retained in the analysis as a separate category. Table 1 defines the population factor cut-points.
HIV/AIDS risk behaviors were collected by interview at baseline, 3- and 9-month follow-up using a modified version of the National Institute for Drug AbuseRisk Behavior Assessment.18 We analyzed 2 sexual risk behaviors reported in the past 30 days: sex with multiple partners (any gender) and unprotected sex with a risky partner. Sex with multiple partners was defined as reporting 2 or more to the following question: “In the past 30 days, how many different people have you had sex (vaginal, anal, or oral) with?” People who reported 0 or 1 partner were included in the reference group. Engaging in unprotected sex with a risky partner was defined as >0 response to any of the following questions: “How many times in the past 30 days have you had unprotected sex (vaginal, anal, or oral) with: (a) someone who is not your primary partner; (b) someone who shot drugs with needles; (c) someone who sometimes smokes crack/cocaine and/or methamphetamine; or (d) while trading sex for drugs, money, or gifts?” People who reported 0 to all of these questions or who reported no sex in the past 30 days were included in the reference group.
Because of known race and gender differences in criminal justice populations,1 drug use,19 health equity,20 and HIV risk behaviors,21 we stratified the analyses into NHB males, NHW males, Hispanic Males, NHB females, and NHW females. Hispanic females (N = 39) and all other race categories (N = 55) were excluded because of small population sizes. To adjust for similarities within site and across time for the same participant, generalized estimating equations (GEEs) with a log link (to estimate relative risks) were built. Because our goal was to generate population-average estimates of the association between population-level determinants and HIV sexual risk behavior and given their greater statistical flexibility, GEEs were preferred over nonlinear mixed models (e.g., SAS [version 9.2, Cary, NC] Proc GLIMMIX). Parallel analyses were conducted using Proc GLIMMIX as a sensitivity analysis. Similar findings for many of the analyses were detected but appeared less robust when there were smaller sample sizes with only 2 to 3 points per person, a known drawback of this modeling approach.22 Multicollinearity was checked for the full models using variance inflation factors and, when indicated, reduced parsimonious models are presented. All models controlled for individual-level factors and time (as a fixed effect). Individual-level factors included age, time, Criminal Justice Drug Abuse Treatment Studies-1 (Transitional Case Management or Step'n'Out), and marital status. Time interactions were tested but were either not significant or unstable (because of sparse data); therefore, we present only main effects.
This study was approved by the Rhode Island Hospital Institutional Review Board and, for the 2 underlying studies, by each site Institutional Review Board, and the federal Office of Human Research and Protection.
Sample Characteristics and Preincarceration HIV Risk Behaviors
Analysis included 1051 males and 236 females. Men were a median 34 years old, < one-third had completed high school, and a minority were married or living as married with a partner before incarceration (Table 2). Women were a median 35 years old and also had low educational attainment. Preincarceration HIV sexual risk behaviors were common. Having multiple sex partners in the past 30 days was reported by more than a third of men, and slightly under a third of women. NHB men had the highest prevalence of having multiple sex partners in the 30 days before incarceration. Both NHB men and women had the lowest prevalence of having unprotected sex with a risky partner in the 30 days before incarceration.
Population factors differed greatly across racial/ethnic groups (Table 1). NHB participants had higher median rates for incarceration and lower median sex ratio values compared with both NHW and Hispanic participants.
Social Determinants and HIV Sex Risk Behaviors During Community Supervision
Table 3 reports unadjusted and adjusted relative risk (ARR) ratios for the population determinants of interest, stratified by gender and race/ethnicity.
Non-Hispanic Black Participants
Among NHB male participants, both unadjusted (relative risk [RR]: 1.35 [1.03, 1.77]) and adjusted (ARR: 2.14 [1.39, 3.30]) models revealed significant associations of higher incarceration rate with increased risk of having unprotected sex with a risky partner. Adjusted (ARR: 1.76 [1.29, 2.42]) models detected significant associations of lower matched sex ratio with increased risk of having unprotected sex with a risky partner. Sex ratios and incarceration rates were not significantly associated with having multiple recent sex partners among NHB male participants.
In unadjusted analyses among NHB female participants, the matched sex ratio exhibited a strong association with the risk of engaging in unprotected sex with a risky partner (RR: 2.37 [1.27, 4.39]) and having multiple sex partners (RR: 2.14 [1.12, 4.09]). However, the adjusted model demonstrated multicollinearity between the poverty rate and sex ratio variables. Therefore, reduced models were fit that removed the variables in a stepwise manner. Reduced models without poverty rate confirmed the previously detected associations between the matched sex ratio and the risk of engaging in unprotected sex with a risky partner during community supervision (ARR: 2.48 [1.31, 4.73]) and with a low matched sex ratio and greater risk of having multiple sex partners (ARR: 2.00 [1.02, 3.94]).
Non-Hispanic White Participants
For the NHW males, a higher incarceration rate (ARR: 1.39 [1.05, 1.85]) was associated with both increased risk of engaging in unprotected sex with a risky partner and with having multiple sex partners (ARR: 1.92 [1.40, 2.64]).
Unadjusted models for the NHW females revealed associations between the low disassortative sex ratio (RR: 1.72 [1.03, 2.88]) and having multiple sex partners. Multicollinearity between poverty and both sex ratios precluded multivariable analysis of having multiple partners. Reduced adjusted models that removed poverty and age and included either sex ratio found an association between lower sex ratio and having multiple partners (matched ARR 1.71 [1.06, 2.75]) or disassortative sex ratio (ARR: 1.72 [1.06, 2.77]) (Table 3). GEEs with logit links were also fit to these data: low sex ratios (matched adjusted odds ratio [AOR]: 4.16 [1.55, 11.1]); disassortative sex ratio (AOR: 4.17 [1.57, 11.08]) were associated with increased odds of having multiple partners, controlling for age and poverty.
Among Hispanic males, a higher incarceration rate (ARR: 3.99 [1.55, 10.26]) and a lower disassortative sex ratio (ARR: 2.99 [1.63, 5.51]) were associated with increased risk of having unprotected sex with a risky partner in the adjusted model (Table 3). Sex ratios and incarceration rates were not significantly associated with having multiple recent sex partners in this group.
In this sample of people on probation and parole, sex ratios and incarceration rates in the county of residence predicted the quantity and HIV risk characteristics of recent sex partners. NHB male and female participants living in counties with low NHB sex ratios had greater risk of having unprotected sex with a risky partner. For these groups, the interaction between having been incarcerated and living in an area with a shortage of men may facilitate participation in or entrance into riskier sexual networks. This could partly explain elevated HIV and STI rates among blacks compared with whites. For NHB women, low NHB sex ratios were also associated with having multiple partners. The status of NHB men as ex-offenders may have limited their opportunities to acquire less risky partners.
Significant associations were also observed among NHW and Hispanic participants. High incarceration rates were associated with greater risk of having unprotected sex with a risky partner among Hispanic men and of having both unprotected sex with a risky partner and having more than one partner among NHW men. Hispanic men were more likely to have unprotected sex with a risky partner if the disassortative sex ratio was low. NHW women were more likely to have multiple partners if they lived in a county with low-matched or disassortative sex ratios. The associations of disassortative sex ratios among NHW women and Hispanic men may reflect more interracial/ethnic opposite sex partnerships with risky partners.
A recent national population-based study found that low sex ratios and high male incarceration rates were associated with a greater likelihood of having multiple partners in the past year among NHB men, but not among NHB women.23 Although the population and sexual behavior questions differ from those of the current study, the findings among NHB men are similar. One other study of the effects of sex ratios and incarceration rates on sexual behavior among blacks showed no association among men and a positive association among women.24 Further analyses indicated that this finding was the result of less sex work among women in low sex ratio areas. However, that study used sex ratio data from the census tract (a smaller geographic unit) in a single geographic area, and it is unclear how well this unit describes the area in which people seek partners.25 Geographic studies of mobility of drug-involved parolees suggest this population changes residences across moderate distances.26 Thus, county of residence may provide a reasonable estimate of the catchment area for potential sex partners.
Poverty was treated as a control variable in our analyses and was retained whenever possible in the models. Modeling problems arose when poverty rate was collinear with 1 or more of the exposures of interest. As this happened only in instances among the women, and sample sizes for the women were the smallest, it is not possible to entirely disentangle the effects of small cell sizes (a statistical problem) from sex-specific vulnerabilities (a sociological problem). That is, collinearity may have been because of high levels of the female probationer/parolee's particular vulnerabilities, linked to syndemics 27–29 of sex work, transactional sex, limited social support and family ties, proclivity to return to neighborhoods with concentrated poverty, high male incarceration rates, and fewer same-race male partners. Analyses to explore these associations in a larger sample of women under community supervision are needed.
Results underscore the fragility of the reentry period and the need to focus efforts on housing, work assistance, and job placement in the community. Ex-offenders are disqualified from many sources of public support and have difficulty finding employment,11 and people convicted of state or federal felonies involving the use or sale of drugs are banned from receiving cash benefits or food stamps.30 These diminished life opportunities may decrease ex-offenders' ability to develop and maintain committed sexual partnerships or marriage. This may be more pronounced among NHB males because of high levels of unemployment and incarceration.
Our findings have several limitations. First, data were self-reported and may have been affected by measurement error. Sexual risk behavior outcomes did not distinguish between vaginal, anal, and oral sex, and between same-gender and opposite-gender sex. Second, it was not possible to distinguish all instances of condom use from sex; thus, results may conflate influences on both frequency of sex and frequency of condom use. The sample may not have provided sexual risk or population data that were representative of all racial/ethnic populations on community supervision. The lack of a control group precludes analysis of incarceration as an individual factor. For the Transitional Case Management study, which recruited participants in prison, county of recruitment may be a poor proxy for county of residence. However, sensitivity analyses iteratively eliminating Transitional Case Management sites indicated no fundamental change in the magnitude or direction of the observed associations. The relatively small size of racial/ethnic subsamples precluded testing nonlinear associations with sex ratios, which are known to exist.23 We were unable to include data from counties with small populations. Finally, our exploratory analyses detected statistically significant and valid associations, but the use of GEEs with a log link returned unstable RRs in some strata, especially the women. Replication of these findings in a larger sample of women recently released from incarceration is recommended.
A lack of adequate primary and secondary HIV preventive interventions suggests that new approaches are needed to reduce HIV transmission to and from minority populations.31 To this end, efforts to address population factors may return greater HIV-prevention effects. For instance, interventions to reduce the incarceration rate in a community, such as through the use of drug courts32 or implementation research that realigns police and parole/probation officer interventions with public health goals,33 could help reduce sexual risk behavior and sexual network dynamics that facilitate HIV transmission. This kind of policy intervention could improve outcomes by both reducing the number of individuals experiencing incarceration and by reducing the contribution of high rates of incarceration to sex ratio imbalance. Improving access to treatment for drug-dependent inmates, before release and on reentry, can reduce recidivism, improve health outcomes,34 and, in time, may promote lifestyle and sexual partnership stability. Other policy interventions, such as lifting restrictions against people with criminal histories from receipt of public housing or other support, can influence community HIV risk by reducing a structural barrier that undermines committed partnerships.11 Public health may be served by helping incarcerated men and women maintain their marriages or other important partnerships, especially considering that most incarcerations last under 1 year,3 that many people are reincarcerated,1 and that relationship dissolution has been found to be related to subsequent sexual risk behavior.6
In conclusion, our findings suggest that ex-offenders living in counties with a low sex ratio or a high incarceration rate are particularly vulnerable to sexual behavior and partner characteristics that may put them at risk for HIV or other STIs, particularly for those living in high prevalence areas. Research is needed to determine whether racial/ethnic differences in sex ratios or incarceration rates contribute to racial/ethnic disparities in sexual transmission of HIV or other STIs.
1. Pew Project on the States. One in 100: Behind bars in America 2008, Pew Charitable Trust, Washington, DC, 2008.
2. U.S. Census Bureau, “Table 3. Annual Estimates of the Resident Population by Sex, Race, and Hispanic Origin for the United States: April 1, 2000 to July 1, 2009 (NC-EST2009-03),” June 2010.
3. West HC, Sabol W, Cooper M. Prisoners in 2008. In. Washington, D.C.: U.S. Department of Justice, Bureau of Justice Statistics 2009.
5. Glaze LE, Bonczar TP. Probation and parole in the United States, 2007 Statistical tables In. Edited by Statistics Bureau of Justice, 2008.
6. Khan MR, Behrend L, Adimora AA, et al.. Dissolution of primary intimate relationships during incarceration and associations with post-release STI/HIV risk behavior in a southeastern city. Sex Transm Dis 2011; 38:43–47.
7. Thomas JC, Torrone E. Incarceration as forced migration: The community health effects. Am J Public Health 2006; 96:1762–1765.
8. Khan MR, Epperson M, Mateu-Gelabert P, et al.. Incarceration, sex with an STI/HIV-infected partner, and STI/HIV infection status in Bushwick, Brooklyn, NY: A social network perspective. Am J Public Health 2011; 101:1110–1117.
9. Epperson M, El-Bassel N, Gilbert L, et al.. Examining the temporal relationship between criminal justice involvement and sexual risk behaviors among drug-involved men. J Urban Health 2010; 87:324–336.
10. Belenko S, Langley S, Crimmins S, et al.. HIV risk behaviors, knowledge, and prevention education among offenders under community supervision: A hidden risk group. AIDS Educ Prev 2004; 16:367–385.
11. Blankenship KM, Smoyer AB, Bray SJ, et al.. Black-white disparities in HIV/AIDS: The role of drug policy and the corrections system. J Health Care Poor Underserved 2005; 16:140–156.
12. Oser CB, Leukefeld CG, Tindall MS, et al.. Male and female rural probationers: HIV risk behaviors and knowledge. AIDS Care 2006; 18:339–344.
13. Martin SS, O'Connell DJ, Inciardi JA, et al.. HIV/AIDS among probationers: An assessment of risk and results from a brief intervention. J Psychoactive Drugs 2003; 35:435–443.
14. Knudsen HK, Leukefeld C, Havens JR, et al.. Partner relationships and HIV risk behaviors among women offenders. J Psychoactive Drugs 2008; 40:471–481.
15. Peters RH, Greenbaum PE, Steinberg ML, et al.. Effectiveness of screening instruments in detecting substance use disorders among prisoners. J Subst Abuse Treat 2000; 18:349–358.
16. Walters GD, McDonough JR. The lifestyle criminality screening form as a predictor of federal parole/probation/supervised release outcome. Legal Criminol Psychol 1998; 3:173–181.
17. Prendergast ML. Interventions to promote successful re-entry among drug-abusing parolees. Addict Sci Clin Pract 2009; 5:4–13.
18. Friedmann PD, Katz EC, Rhodes AG, et al.. Collaborative behavioral management for drug-involved parolees: Rationale and design of the step'n out study. J Offender Rehabil 2008; 47:290–318.
19. Van Etten ML, Neumark YD, Anthony JC. Male-female differences in the earliest stages of drug involvement. Addiction 1999; 94:1413–1419.
20. Committee on Quality of Health Care in America IoM. Crossing the quality chasm: A new health system for the 21st century. In. Washington, D.C, 2001.
21. Kaiser Family Foundation. Women and HIV/AIDS in the United States. In 2008.
22. Allison PD. Logistic Regression Using SAS: Theory and Application. Cary, NC: SAS Institute Inc, 1999.
23. Pouget ER, Kershaw TS, Niccolai LM, et al.. Associations of sex ratios and male incarceration rates with multiple opposite-sex partners: Potential social determinants of HIV/STI transmission. Public Health Rep 2010; 125 Suppl 4:70–80.
24. Senn TE, Carey MP, Vanable PA, et al.. The male-to-female ratio and multiple sexual partners: Multilevel analysis with patients from an STD clinic. AIDS Behav 2010; 14:942–948.
25. Fossett MA, Kiecolt KJ. A methodological review of the sex ratio: Alternatives for comparative research. J Marriage Fam 1991; 53:941–957.
26. Cahill M, LaVigne N. Residential mobility and drug use among parolees in San Diego, California, and implications for policy. In: Thomas YF, Richardson D, Cheung I, eds. Geography and Drug Addiction. New York, NY: Springer, 2008.
27. Romero-Daza N, Weeks M, Singer M. “Nobody gives a damn if I live or die”: Violence, drugs, and street-level prostitution in inner-city Hartford, Connecticut. Med Anthropol 2003; 22:233–259.
28. Singer M, Clair S. Syndemics and public health: Reconceptualizing disease in bio-social context. Med Anthropol Q 2003; 17:423–441.
29. Singer MC, Erickson PI, Badiane L, et al.. Syndemics, sex and the city: Understanding sexually transmitted diseases in social and cultural context. Soc Sci Med 2006; 63:2010–2021.
30. Allard P. Life sentences: Denying welfare benefits to women convicted of drug offenses. In: The Sentencing Project. 2002. Washington, DC.
31. Hallfors DD, Iritani BJ, Miller WC, et al.. Sexual and drug behavior patterns and HIV and STD racial disparities: The need for new directions. Am J Public Health 2007; 97:125–132.
32. Gottfredson DC, Kearley BW, Najaka SS, et al.. The Baltimore City drug treatment court: 3-year self-report outcome study. Eval Rev 2005; 29:42–64.
33. Davis CS, Beletsky L. Bundling occupational safety with harm reduction information as a feasible method for improving police receptiveness to syringe access programs: Evidence from three U.S. cities. Harm Reduct J 2009; 6:16.
34. Haig T. Randomized controlled trial proves effectiveness of methadone maintenance treatment in prison. Can HIV AIDS Policy Law Rev 2003; 8:48.