Abstract
Introduction
Neighborhood characteristics shape sexual risk in HIV-uninfected adults in the United States (US). We assess relationships between census tract characteristics and sexual risk behaviors in a predominantly HIV-infected cohort of women living in the Southern US.
Methods
This cross-sectional multilevel analysis included data from 737 HIV-infected and HIV-uninfected women enrolled in the Women’s Interagency HIV Study. Administrative data captured characteristics of census tracts where women lived; participant-level data were gathered via survey. We used principal components analysis to condense tract-level variables into components: social disorder (e.g., violent crime rate) and social disadvantage (e.g., alcohol outlet density). We used hierarchical generalized linear models to assess relationships between tract-level characteristics and condomless vaginal intercourse (CVI), anal intercourse (AI), and condomless anal intercourse (CAI).
Results
Greater social disorder was associated with less AI (OR=0.63, 95% CI=0.43, 0.94) and CAI (OR=0.49, 95% CI=0.30, 0.80), regardless of HIV status. There were no statistically significant additive or multiplicative interactions between tract characteristics and HIV status.
Conclusion
Neighborhood characteristics are associated with sexual risk behaviors among women living in the Southern US, these relationships do not vary by HIV status. Future studies should establish temporality and explore the causal pathways through which neighborhoods influence sexual risk.
Keywords: HIV, Women, Residence Characteristics, Multilevel Analysis, Sexual Behavior
Introduction
The burden of HIV/AIDS in women in the United States (US) has grown substantially over time: rising from 8% of all newly diagnosed AIDS cases in 1983 to more than 19% of all new diagnosed HIV infections in 2014 [1, 2]. The Southern US now represents a significant proportion of the HIV/AIDS epidemic. In 2011, a group of nine states (Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, and Texas) reported higher HIV diagnosis rates than the US overall (24.5/100,000 vs. 18.0/100,000) [3]. Moreover, HIV-infected individuals living in this region experience the highest rates of morbidity and mortality in the US [3–5].
An emerging line of evidence indicates that several features of the social and built environment influence sexual risk and the transmission of HIV and other sexually transmitted infections (STIs) in HIV-uninfected populations, and in the Southern US in particular [3–5]. Ecologic studies have found that geographic areas with high levels of poverty, social disorder (e.g., violent crime), incarceration, or racial/ethnic residential segregation frequently have higher prevalences of HIV/AIDS and other STIs [6–10]. Multilevel studies extending this line of research to associations between neighborhood characteristics and sexual risk in individuals found that living in neighborhoods with low male:female sex ratios (i.e., fewer men than women), high incarceration rates, and high poverty rates is associated with non-monogamy, multiple sexual partners, condomless sexual intercourse, and risk discordant partnerships [11–18]. Conceptualizing neighborhoods as opportunity structures, residents of neighborhoods with comparatively greater economic disadvantage (e.g., high poverty rates) or social disorder (e.g., more violent crime) may have less access to social and physical resources (e.g., employment) needed to engage in healthful behaviors, and greater exposure to hazards associated with negative health outcomes [19, 20]. For example, a high density of liquor stores in a neighborhood may contribute to greater community-level alcohol consumption, more sexual risk behavior, and connect women to risky sexual networks [6, 19, 20].
However, little is known about whether or how neighborhood characteristics shape sexual risk for women living with HIV infection. To our knowledge, only one multilevel study has explored relationships between neighborhoods and sexual behaviors in HIV-infected adults. This study found no association between neighborhood poverty rates, racial/ethnic composition, unemployment rates, and condomless sexual intercourse among a predominantly male clinic-based population in the Midwestern US [21]. It is possible that the magnitude and direction of relationships between neighborhoods and condom non-use vary by women’s HIV status, in part because HIV-infected women may be more motivated to protect their own health or that of their sexual partners [22]. An understanding of whether or how neighborhood characteristics and HIV status shape sexual risk behaviors among women can inform the design of interventions to improve women’s sexual health and to reduce the transmission of HIV and other STIs.
The present analysis addresses this critical research gap by exploring relationships between neighborhood characteristics and sexual risk behaviors among a predominantly HIV-infected cohort of women living in the Southern US. We seek to:
Characterize relationships between neighborhood characteristics and sexual risk behaviors, and;
Test whether the magnitudes and directions of relationships between neighborhood characteristics and sexual risk behaviors vary by HIV status.
Materials and Methods
Participants
The Women’s Interagency HIV Study (WIHS) is a multisite, prospective study designed to investigate the progression of HIV among HIV-infected women and the incidence of HIV among women who are at high risk of HIV infection [23–25]. This cross-sectional analysis utilizes baseline data from women who were newly enrolled at WIHS clinical research sites in Alabama, Georgia, Mississippi, Florida, and North Carolina between October 2013 and September 2015. WIHS eligibility criteria included being between 25–60 years old. In addition, HIV-infected women were antiretroviral therapy (ART) naïve or started highly active antiretroviral therapy (HAART) after December 31, 2004; had never used didanosine, zalcitabine, or stavudine (unless during pregnancy or for pre- or post-exposure HIV prophylaxis); had never been on non-HAART ART, and had documented pre-HAART CD4 counts and HIV viral load. HIV-uninfected women reported at least one personal characteristic (e.g., illicit drug use) or male sexual partner characteristic (e.g., injection drug user) associated with increased risk of HIV acquisition within past 5 years. Participants were recruited using a variety of methods, including clinic and community-based organization referrals. Institutional Review Board approval was obtained at each of the collaborating institutions and written informed consent was obtained from each participant prior to initiation of study procedures. Methods are described in more detail elsewhere [23–26]. This analysis is restricted to WIHS participants who provided written informed consent to collect and geocode their residential address.
Data Collection and Measures
WIHS collected demographic and behavioral data using interviewer-administered surveys. Participant residential addresses were geocoded to 2010 census tract boundaries. We used existing data sources to construct census tract variables that captured neighborhood social and physical environments (e.g., US Census).
Primary Outcomes
Outcomes included condomless vaginal intercourse (CVI), anal intercourse (AI), and condomless anal intercourse (CAI) in the past 6 months. CVI was defined as reported inconsistent condom use during vaginal intercourse (binary: never or sometimes vs. always). AI was defined as a report of any anal sex (binary: yes/no). CAI was defined as reported inconsistent condom use during anal intercourse (binary, as defined above).
Census Tract-Level Exposures
Measures describing the social and physical environments of the census tracts where participants lived were constructed using existing data sources (e.g., US Census) (Table 1). In order to capture underlying constructs and to avoid multicollinearity in multivariable models, we used principal components analysis (PCA) with orthogonal rotation (varimax) to condense tract-level variables into components (Supplemental Digital Content 1). The PCA produced two components with eigenvalues >1.0: 1) “social disorder” (i.e., vacant housing units, violent crime rate, STI prevalence, poverty, unemployment) and 2) “social disadvantage” (i.e., renter-occupied housing and alcohol outlet density). Standardized continuous principal component scores were used as predictors in models. For each component, higher scores are indicative of greater than average social disorder/social disadvantage than the sample.
Table 1.
Census tract measures, definition, data source, and year
| Measure | Definition | Data Source | Year |
|---|---|---|---|
| Social disorder component | |||
| Percent vacant housing units | Percent vacant residential housing units | US Department of Housing and Urban Development and United States Postal Service | 2013 |
| Violent crime rate | Total murder, non-negligent manslaughter, forcible rape, robbery, and aggravated assaults per 1,000 tract residents1 | Law Enforcement Agencies (i.e., police department, Sheriff’s Office) | 2013 |
| Sexually transmitted infection (STI) prevalence | Number of newly reported STIs (i.e., Chlamydia, gonorrhea, and primary and secondary syphilis) per 1,000 tract residents aged 15–642 | State Department of Health | 2013 |
| Percent poverty | Percent residents with annual income below poverty level | American Community Survey (ACS) | 2009–2013 |
| Percent unemployment | Percent unemployed residents ≥ 16 years old | ACS | 2009–2013 |
| Social disadvantage component | |||
| Percent renter-occupied housing units | Percent renter occupied housing units | ACS | 2009–2013 |
| Alcohol outlet density | The number of businesses with a license to sell beverages containing alcohol (e.g., liquor, beer, wine) for off-premise consumption per tract square mile1,3 | State Licensing Agencies (e.g., Department of Revenue, Alcoholic Beverage Control Commission) | 2014 |
Addresses were obtained from state agencies and geocoded to tracts; addresses within a 100-foot buffer of the tract boundary were included in the tract’s calculation.
In Alabama, the number of newly identified STIs was available by ZIP code, but not census tract. ZIP-level STI counts were allocated to tracts based on the proportion of residential population using the 2015 boundaries USPS-HUD ZIP to tract crosswalk file. Twelve ZIP code-census tract combinations were not included in the crosswalk file. For these 15 participants (17% of participants with available census tract data at site), ZIP code STI prevalence was assigned to the participant census tract. We conducted sensitivity analyses, removing these participants from the analytic data set, to explore potential bias introduced by this substitution. The rounded odds ratio estimates for Final Model A with and without these 15 participants were the same.
In Mississippi, off-premise liquor licensing data were available (liquor can only be purchased at package/liquor stores), but licensing data for sale of beer and wine off-premise were not publically available. As a proxy, we used non-restaurant businesses with permits to sell eggs or milk (e.g., convenience stores, pharmacies) under the oversight of the Mississippi Department of Agriculture and Commerce because these types of businesses would have refrigerated display cases and likely have the capacity to sell beer and wine.
Participant-Level Characteristics
WIHS classified participants as HIV-infected if they had a reactive serologic enzyme-linked immunosorbent assay test and a confirmed positive western blot or detectable plasma HIV-1 ribonucleic acid.
Covariates included participant-level characteristics that might confound or modify relationships between tract-level characteristics and sexual behaviors. These a priori variables are classically included in analyses exploring associations of participant-level characteristics and sexual risk behaviors [27–29]. Covariates captured demographic characteristics and behaviors in the past 6 months and were binary unless otherwise noted: age in years (continuous, mean-centered), married or cohabitating, non-Hispanic African American, annual household income ≤$18,000, self-rated quality of life (QOL) as measured using an abbreviated Medical Outcomes Study Scale (continuous, mean-centered; scores ranged from 0 to 100 with higher scores indicative of better QOL) [30], alcohol or illicit substance use (>7 drinks in the past week, any injection or non-injection use of crack, cocaine, heroin, marijuana, hallucinogens, club drugs, methamphetamines, or recreational prescription drug use in the last 6 months), exchange of sex for drugs, money or housing, and homeless (currently living in a rooming or halfway house, shelter, welfare hotel, or on the street).
Analysis
We used descriptive statistics to explore distributions of participant and census tract characteristics. We compared characteristics by HIV status and for participants who did and did not provide geocodable address information using t-tests and chi-square tests. All bivariate and multivariable relationships were modeled with hierarchical generalized linear models (HGLMs) using a logit link function with random effects for the intercept, thus allowing for participant-level variation across census tracts [31]. All HGLMs had two levels: participants (Level 1) were nested in census tracts (Level 2). The modeling process had four phases.
In Phase 1, we used an unconditional model to assess the proportion of variance in sexual risk behaviors due to clustering within census tracts (i.e., intra-class correlation [ICC]).
In Phase 2, we modeled bivariate relationships between each tract- and participant-level characteristic and sexual risk behavior accounting for nestedness.
In Phase 3, we modeled multivariable relationships between tract-level characteristics and sexual risk behaviors, controlling for potential participant-level confounders. In order to determine whether the magnitudes and directions of relationships between tract characteristics and sexual risk behaviors might vary by HIV status, we tested statistically for multiplicative and additive interactions between neighborhood characteristics and each outcome by HIV status. In Phase 3A, we tested for interaction between tract characteristics and HIV status on the multiplicative scale by entering cross-level interaction terms for HIV status and tract-level variables (e.g., HIV status*social disorder), retaining interaction terms with p<0.05 in the multivariable model (Final Models). In Phase 3B, we tested for interaction between neighborhood characteristics and HIV status on the additive scale by fitting separate models using a binomial distribution and identity link, controlling for participant-level confounders [32, 33]. Interaction terms with p<0.05 were considered statistically significant on the additive scale.
Participant-level covariates traditionally included in models evaluating sexual risk outcomes (e.g., sex exchange) may lie in the causal pathway between tract characteristics and sexual behaviors. Including these variables in the full model would attenuate relationships between tract characteristics and study outcomes if they did indeed lie on the causal pathway. In Phase 4 (Reduced Model), we reran the final multivariable model, excluding variables that might lie on the causal pathway between neighborhood characteristics and sexual risk behaviors (i.e., income, QOL, alcohol and substance use, sex exchange, and homelessness). We compared odds ratio (OR) estimates for all tract-level variables in the Final vs. Reduced Model; >10% differences in magnitude suggested that excluded variables may attenuate relationships between neighborhood characteristics and sexual risk behaviors.
HGLMs were fit using PROC GLIMMIX using Newton Raphson optimization and Gauss-Hermite quadrature approximation in SAS 9.4. Estimates with p<0.05 were considered statistically significant.
Results
A total of 845 women were enrolled at WIHS’s southern sites (Table 2). Eighty seven percent of enrolled women both provided consent to collect and geocode their residential address and provided information that could be geocoded to census boundaries. In the analytic sample (N=737), participants were on average 44 years old (standard deviation [SD]=9.3), 83.3% identified as non-Hispanic African American, 66.8% reported annual household incomes of ≤$18,000, and 71.9% of participants were HIV-infected (Table 3). Forty-two percent of participants reported CVI, 6.8% of participants reported AI, and 4.3% of participants reported CAI in the last 6 months. On average, participants lived in census tracts where 16.1% of residents were unemployed (SD=8.0), 29.1% were living in poverty (SD=13.6), and where roughly half of housing units were renter-occupied (SD=21.7).
Table 2.
Women’s Interagency HIV Study enrollment and availability of participant census tract identifier, overall and by site
| Site | Alabama | Florida | Georgia | Mississippi | North Carolina | Total |
|---|---|---|---|---|---|---|
| Enrolled | 112 | 146 | 275 | 114 | 198 | 845 |
| Census tract identifier available | 100 (89.29%) | 90 (61.64%) | 246 (89.45%) | 112 (98.25%) | 189 (95.45%) | 737 (87.22%) |
| Reasons for missing census tract identifier | ||||||
| No consent for geocoding | 0 (0%) | 48 (32.88%) | 17 (6.18%) | 0 (0%) | 0 (0%) | 65 (7.69%) |
| Living on street or in a residential treatment facility1 | 10 (8.93%) | 7 (4.79%) | 0 (0%) | 0 (0%) | 5 (2.53%) | 22 (2.60%) |
| Address could not be geocoded to census tract boundary | 1 (0.89%) | 0 (0%) | 5 (1.82%) | 1 (0.88%) | 0 (0%) | 7 (0.83%) |
| Unknown | 1 (0.89%) | 1 (0.68%) | 7 (2.55%) | 1 (0.88%) | 4 (2.02%) | 14 (1.66%) |
The Women’s Interagency HIV Study, with the exception of the site in Georgia, did not geocode address data for participants living on the street or in a residential treatment facility.
Table 3.
Distributions of participant and census tract characteristics among 737 women enrolled in the Women’s Interagency HIV Study Southern Sites
| Characteristics of participants and census tracts | Overall n (%) or Mean (SD) | HIV-infected n (%) or Mean (SD) | HIV-uninfected n (%) or Mean (SD) | |
|---|---|---|---|---|
| Outcomes | ||||
| Condomless vaginal intercourse1 | 312 (42.3) | 168 (31.8) | 144 (69.6) | |
| Anal intercourse | 50 (6.8) | 30 (5.7) | 20 (9.7) | |
| Condomless anal intercourse1 | 32 (4.3) | 16 (3.0) | 16 (7.8) | |
| Census tract-level characteristics | ||||
| Social disorder component | ||||
| Percent vacant housing units | 7.8 (6.3) | 7.6 (6.3) | 8.3 (6.3) | |
| Violent crimes per 1,000 residents1 | 13.7 (13.4) | 12.8 (12.1) | 16.0 (16.0) | |
| Percent poverty | 29.1 (13.6) | 28.6 (13.3) | 30.3 (14.5) | |
| Percent unemployed | 16.1 (8.0) | 15.7 (7.7) | 16.9 (8.5) | |
| Sexually transmitted infections per 1,000 residents2 | 19.1 (13.3) | 18.1 (12.5) | 21.7 (14.8) | |
| Social disadvantage component | ||||
| Percent renter-occupied housing units | 51.9 (21.7) | 50.7 (21.6) | 54.9 (21.7) | |
| Alcohol outlet density3 | 4.8 (7.6) | 4.7 (7.4) | 5.0 (8.1) | |
| Participant-level characteristics | ||||
| HIV-infected | 530 (71.9) | -- | -- | |
| Age in years | 43.7 (9.3) | 44.3 (9.1) | 42.7 (9.7) | |
| Married or living as married | 244 (33.1) | 176 (33.3) | 68 (33.0) | |
| Non-Hispanic African American | 614 (83.3) | 438 (82.6) | 176 (85.0) | |
| Annual household income of $18,000 or less | 492 (66.8) | 365 (70.8) | 127 (64.1) | |
| Quality of life index | 67.1 (20.5) | 67.6 (20.6) | 65.8 (20.2) | |
| Alcohol or illicit substance use | 279 (37.9) | 182 (34.4) | 97 (46.9) | |
| Sex exchange1 | 42 (5.7) | 16 (3.0) | 26 (12.6) | |
| Homeless1 | 47 (6.4) | 23 (4.4) | 24 (11.7) | |
Comparison by HIV staus p<0.05.
In Alabama, the number of newly identified STIs was available by ZIP code, but not census tract. ZIP-level STI counts were allocated to tracts based on the proportion of residential population using the 2015 boundaries USPS-HUD ZIP to tract crosswalk file. Twelve ZIP code-census tract combinations were not included in the crosswalk file. For these 15 participants (17% of participants with available census tract data at site), ZIP code STI prevalence was assigned to the participant census tract. We conducted sensitivity analyses, removing these participants from the analytic data set, to explore potential bias introduced by this substitution. The rounded odds ratio estimates for Final Model with and without these 15 participants were the same.
In Mississippi, off-premise liquor licensing data were available (liquor can only be purchased at package/liquor stores), but licensing data for sale of beer and wine off-premise were not publically available. As a proxy, we used non-restaurant businesses with permits to sell eggs or milk (e.g., convenience stores, pharmacies) under the oversight of the Mississippi Department of Agriculture and Commerce because these types of businesses would have refrigerated display cases and likely have the capacity to sell beer and wine.
A comparison of participant and census tract characteristics by HIV status indicated that HIV-uninfected women reported more CVI (69.6% vs. 31.8%), CAI (7.8% vs. 3.0%), sex exchange (12.6% vs. 3.0%), and homelessness (11.7% vs. 4.4%) than HIV-infected women (p<0.05). HIV-infected women, on average, lived in neighborhood with less violent crime (12.1 vs. 16.9 violent crimes per 1,000 residents) than HIV-uninfected women (p<0.05). All other participant and tract characteristics assessed were comparable by HIV status. Participants missing geocoded address data, as compared to participants with geocoded address data were more likely (p<0.05) to report annual household incomes ≤$18,000 (83.2% vs. 69.0%); alcohol and substance use (48.1% vs. 37.9%); and sex exchange (17.6% vs. 5.7%). We included these variables in the full multivariable models in order to minimize potential confounding [34].
Relationships between census tract characteristics and CVI
The unconditional model ICC was 1.5% (random intercept=0.05, p=0.41). In bivariate analyses (Table 4), tract-level social disorder (OR=1.12, 95% Confidence Interval [CI]=0.96–1.32) and social disadvantage (OR=0.98, 95% CI=0.84–1.15) were not associated with CVI. In the final multivariable model controlling for participant-level characteristics, tract-level social disorder (OR=0.99, 95% CI=0.82–1.20) and social disadvantage (OR=1.00, 95% CI=0.84–1.19) were not associated with CVI. There were no statistically significant interactions between census tract characteristics and HIV status on the multiplicative or additive scale (p>0.05). Odds ratios for tract-level characteristics in the final model, as compared to the reduced model were within 10% and were not statistically significantly associated with CVI.
Table 4.
Bivariate and multivariable relationships of census tract characteristics to the odds of condomless vaginal intercourse among women enrolled in the Women’s Interagency HIV Study’s Southern Sites (n=736)1
| Characteristics of participants and census tracts | Bivariate OR (95%) | Final Model aOR (95%)2 | Reduced Model aOR (95%)2 |
|---|---|---|---|
| Census tract-level characteristics | |||
| Social disorder component3 | 1.12 (0.96–1.32) | 0.99 (0.82–1.20) | 1.08 (0.90–1.29) |
| Social disadvantage component4 | 0.98 (0.84–1.15) | 1.00 (0.84–1.19) | 0.98 (0.82–1.16) |
| Participant-level characteristics | |||
| HIV-infected | 0.18 (0.12–0.26) | 0.20 (0.14–0.29) | 0.12 (0.11–0.26) |
| Age in years | 0.96 (0.94–0.98) | 0.96 (0.94–0.98) | 0.96 (0.95–0.98) |
| Married or living as married | 2.13 (1.52–2.98) | 2.32 (1.59–3.38) | 2.32 (1.61–3.35) |
| Non-Hispanic African American | 0.91 (0.59–1.41) | 0.96 (0.58–1.56) | 0.88 (0.55–1.42) |
| Annual household income of $18,000 or less | 0.75 (0.54–1.06) | 0.80 (0.54–1.18) | -- |
| Quality of life index | 0.99 (0.99–1.00) | 0.99 (0.99–1.00) | -- |
| Alcohol and illicit substance use | 2.07 (1.50–2.86) | 1.78 (1.22–2.59) | -- |
| Sex exchange | 3.92 (1.91–8.02) | 1.81 (0.80–4.09) | -- |
| Homeless | 2.37 (1.22–4.59) | 2.02 (0.96–4.25) | -- |
| Model fit | |||
| Random intercept variance (p- value) | -- | 0 | 0 |
| -2LL | -- | 741.10 | 761.75 |
| AIC | -- | 765.10 | 775.75 |
| BIC | -- | 818.90 | 804.28 |
One participant missing outcome.
Multivariable models restricted to participants with no missing predictors (n=654).
Component generated from principal components analysis (PCA) including tract-level percent vacant housing units, violent crime rate, sexually transmitted infection prevalence, percent poverty, and percent unemployment.
Component generated from PCA including tract-level percent renter-occupied housing and alcohol outlet density
Relationships between census tract characteristics and AI
The unconditional model ICC was 7.8% (random intercept=0.28, p=0.37). In bivariate models (Table 5), tract-level social disorder (OR=0.76, 95% CI=0.53–1.09) and social disadvantage (OR= 0.97, 95% CI=0.69–1.38) were not associated with AI. In the final multivariable model controlling for individual characteristics, tract-level social disorder was inversely associated with AI (OR=0.63, 95% CI=0.43–0.94). Notably, a one standard deviation higher social disorder component was associated with a 37% lower odds of AI. Tract social disadvantage (OR= 1.00, 95% CI=0.70–1.42) was not associated with AI. There were no statistically significant interactions between census tract characteristics and HIV status on the multiplicative or additive scale (p>0.05). Odds ratios for tract-level characteristics in the final model, as compared to the reduced model were within 5% for the social disadvantage component. The estimate for the social disorder component was 24% higher in the Reduced Model and was no longer statisically significantly associated with AI.
Table 5.
Bivariate and multivariable relationships of census tract characteristics to the odds of anal intercourse behaviors among women enrolled in the Women’s Interagency HIV Study’s Southern Sites
| Anal Intercourse (n=736)1 | Condomless Anal Intercourse (n=733)2 | |||||
|---|---|---|---|---|---|---|
| Characteristics of participants and census tracts | Bivariate OR (95%) | Final Model aOR (95%)3 | Reduced Model aOR (95%)3 | Bivariate OR (95%) | Final Model aOR (95%)4 | Reduced Model aOR (95%)4 |
| Census tract-level characteristics | ||||||
| Social disorder component5 | 0.76 (0.53–1.09) | 0.63 (0.43–0.94) | 0.78 (0.56–1.10)* | 0.58 (0.37–0.92) | 0.49 (0.30–0.80) | 0.58 (0.36–0.92)* |
| Social disadvantage component6 | 0.97 (0.69–1.38) | 1.00 (0.70–1.42) | 0.96 (0.68–1.34) | 0.98 (0.66–1.45) | 1.00 (0.69–1.47) | 0.95 (0.66–1.36) |
| Participant-level characteristics | ||||||
| HIV-infected | 0.52 (0.27–1.02) | 0.71 (0.33–1.50) | 0.54 (0.27–1.07) | 0.35 (0.16–0.76) | 0.40 (0.17–0.94) | 0.33 (0.15–0.72) |
| Age in years | 0.95 (0.92–0.99) | 0.95 (0.91–0.99) | 0.96 (0.92–0.99) | 0.97 (0.93–1.01) | 0.96 (0.92–1.01) | 0.98 (0.94–1.02) |
| Married or living as married | 1.33 (0.47–1.92) | 0.86 (0.41–1.84) | 0.89 (0.44–1.79) | 1.33 (0.45–2.31) | 0.84 (0.35–2.02) | 0.98 (0.43–2.26) |
| Non-Hispanic African American | 0.42 (0.19–0.92) | 0.44 (0.18–1.08) | 0.43 (0.20–0.94) | 0.63 (0.25–1.61) | 0.77 (0.28–2.12) | 0.69 (0.26–1.83) |
| Annual household income of $18,000 or less | 0.91 (0.45–1.82) | 0.80 (0.36–1.74) | -- | 0.74 (0.33–1.65) | 0.53 (0.21–1.35) | -- |
| Quality of life index | 0.99 (0.97–1.00) | 0.99 (0.97–1.01) | -- | 0.98 (0.96–0.99) | 0.97 (0.95–0.99) | -- |
| Alcohol and illicit substance use | 3.60 (1.73–7.47) | 3.21 (1.45–7.12) | -- | 4.15 (1.78–9.67) | 3.94 (1.60–9.72) | -- |
| Sex exchange | 4.43 (1.71–11.52) | 2.81 (0.92–8.58) | -- | 3.02 (0.98–9.28) | 1.69 (0.46–6.18) | -- |
| Homeless | 2.31 (0.73–7.29) | 2.08 (0.61–7.07) | -- | 2.76 (0.90–8.45) | 2.60 (0.73–9.23) | -- |
| Model fit | ||||||
| Random intercept variance (p-value) | -- | 0.75 (0.26) | 0.48 (0.31) | -- | 0 | 0 |
| -2LL | -- | 279.01 | 300.33 | -- | 186.08 | 209.5 |
| AIC | -- | 305.01 | 316.33 | -- | 210.08 | 223.5 |
| BIC | -- | 357.99 | 348.93 | -- | 263.82 | 254.8 |
One participant missing outcome.
Four participants missing outcome.
Multivariable models restricted to participants with no missing predictors (n=654).
Multivariable models restricted to participants with no missing predictors (n=651).
Component generated from principal components analysis (PCA) including tract-level percent vacant housing units, violent crime rate, STI prevalence, percent poverty, and percent unemployment.
Component generated from PCA including tract-level percent renter-occupied housing and alcohol outlet density.
>10% difference between Full and Reduced Model.
Relationships between census tract characteristics and CAI
Random intercept components in unconditional models for CAI were estimated to be 0. In bivariate analyses (Table 5), tract-level social disorder (OR= 0.58, 95% CI=0.37–0.92) was inversely associated with CAI. Social disadvantage (OR=0.98, 95% CI=0.66–1.45) was not associated with CAI. In the final mutivariable model, social disorder was inversely associated with CAI (OR=0.49, 95% CI=0.30–0.80). Specifically, a one standard deviation higher social disorder component was associated with a 51% lower odds of CAI. Social disadvantage (OR=1.00, 95% CI=0.69–1.47) was not associated with the CAI. There were no statistically significant interactions between census tract characteristics and HIV status on the multiplicative or additive scale (p>0.05). The estimate for the social disorder component was 20% higher in the Reduced Model. Odds ratios for tract-level characteristics in the final model, as compared to the reduced model were within 10% for the social disadvantage component.
Discussion
Our analyses reveal that neighborhood characteristics are associated with sexual risk behaviors among HIV-infected and high-risk HIV-uninfected women living in the Southern US and that these relationships vary by type of sexual intercourse. Specifically, greater social disorder was associated with less AI and CAI. Neighborhood characteristics were not associated with CVI in our sample.
Receptive CAI confers high risk STIs, including HIV [28, 35]. Studies exploring individual-level predictors of CAI among women suggest that CAI may be influenced by complex social and economic factors, yet no multilevel studies have explored relationships of neighborhoods characteristics to CAI specifically [28, 29, 36]. In this analysis, social disorder was inversely associated with CAI. The direction of this relationship was unexpected. Elements of our neighborhood disorder component, including STI prevalence and violent crime, are associated with partner concurrency, STI acquisition, and greater perceived sexual partner risk [16, 37, 38]. It is possible that women living in neighborhoods with greater social disorder perceived their partners to be riskier, and that these perceptions discouraged engaging in AI or promoted condom use during AI [39–41]. The reduced models which tested associations between tract-level social disorder, AI, and CAI, excluding participant-level characteristics that may lie on the causal pathway (i.e., income, QOL, alcohol and illicit substance use, sex exchange, and homelessness), found that excluding these participant-level characteristics attenuated relationships towards the null. Individual perceptions of neighborhood social disorder are influenced by individual- and network-level characteristics and relationships of social disorder to sexual risk are complex [42, 43]. Women who are homeless or engaged in high risk activities (e.g., sex exchange) may be more acutely aware of their neighborhood environments and consequently may perceive their neighborhoods to be more socially-disordered [43]. Additional research is needed to explore the direct and indirect pathways through which neighborhood social disorder influences sexual risk.
To our knowledge, only three multilevel studies have explored relationships of neighborhood characteristics to condomless sex in heterosexual adults; none of these studies distinguish between types of sexual intercourse (e.g., CVI vs. CAI) [14, 18, 21]. Our finding that the relationship between neighborhood characteristics and sexual risk behaviors varies by type of intercourse underscores the importance of evaluating these outcomes independently. Two of these three studies found no relationship between neighborhood economic characteristics (e.g., median income, unemployment) and condomless sexual intercourse [18, 21]. This is consistent with our own finding that the social disorder component, which included measures of tract-level poverty and unemployment, was not associated with CVI. A third study found that sex ratios were associated with condomless sexual intercourse among women [14]. In our predominantly African American sample, the vast majority of women lived in tracts with ratios of men to women well below equity and as a result, we were unable to test relationships between sex ratios and sexual behaviors. Past research has cited the challenges of exploring these relationships in predominantly African American populations in light of persistent social inequities (e.g., incarceration) which contribute to a shortage of male partners [11, 44].
These findings should be interpreted in light of the study’s limitations. Although WIHS provides a large, high-quality sample of women living with and at increased of HIV infection in the Southern US, study participants agree to indefinite, long-term study follow-up and may not be representative of the broader population. The majority of HIV-infected participants were recruited from clinic-based populations and may not be representative of HIV-infected women who are not connected to HIV care [24, 45]. Participant-level data were collected using interviewer-administered surveys, which may be subject to social desirability bias [46]. Alternative forms of data collection (e.g., computer-assisted) would be preferable for future research. Participants who were excluded from the analysis reported individual characteristics associated with increased sexual risk (e.g., sex exchange) and may have lived in qualitatively different neighborhoods. However, participants with and without address data were not statistically different with respect to study outcomes and we included these variables in the full multivariable models in order to minimize potential confounding [34]. Residential census tracts may fail to capture the activity spaces in which sexual risk behaviors most frequently occur. However, studies have found that individuals living with and at increased risk of HIV tend to select sexual partners in closer proximity than lower risk populations [47, 48]. WIHS did not capture data on the length of stay in the baseline census tract. Due to the cross-sectional nature of our study, we are unable to draw conclusions regarding the causality of relationships between tract characteristics and sexual behaviors.
Conclusion
This is the first multilevel study to test relationships between neighborhood characteristics and AI and CAI and the first to explore relationships between neighborhoods and sexual risk behaviors by women’s HIV status. Collectively, these findings support past research on the importance of neighborhood environments in shaping sexual risk among women living in the US. Additional longitudinal and qualitative studies are needed to establish the causality of these relationships and to better understand the pathways through which neighborhood characteristics shape sexual risk, and inform the development of future multilevel interventions designed to improve women’s sexual health and reduce HIV/STI transmission.
Supplementary Material
Acknowledgments
Funding Sources
This work was supported by the National Institute of Mental Health of the National Institutes of Health [F31MH105238], the Surgeon General C. Everett Koop HIV/AIDS Research Grant, the George W. Woodruff Fellowship of the Laney Graduate School, the Emory Center for AIDS Research [P30 AI050409], the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health [K01HD074726], and the Centers for Disease Control and Prevention [U01PS003315] as part of the Minority HIV/AIDS Research Initiative. Participant data in this manuscript were collected by the Women’s Interagency HIV Study (WIHS): UAB-MS WIHS [U01-AI-103401] Principal Investigators (PIs): Michael Saag, Mirjam-Colette Kempf, and Deborah Konkle-Parker; Atlanta WIHS [U01-AI-103408] PIs: Ighovwerha Ofotokun and Gina Wingood; Miami WIHS [U01-AI-103397] PIs: Margaret Fischl and Lisa Metsch; UNC WIHS [U01-AI-103390] PI: Adaora Adimora; WIHS Data Management and Analysis Center [U01-AI-042590] PIs: Stephen Gange and Elizabeth Golub. The WIHS is funded primarily by the National Institute of Allergy and Infectious Diseases, with additional co-funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute on Mental Health. Targeted supplemental funding for specific projects is also provided by the National Institute of Dental and Craniofacial Research, the National Institute on Alcohol Abuse and Alcoholism, the National Institute on Deafness and other Communication Disorders, and the National Institutes of Health Office of Research on Women’s Health. WIHS data collection is also supported by UL1-TR000454 (Atlanta CTSA)
The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health. The North Carolina Department of Health and Human Services does not take responsibility for the scientific validity or accuracy of methodology, results, statistical analyses, or conclusions presented.
The authors thank the Women’s Interagency HIV Study participants for sharing their time and experiences. The authors also acknowledge the efforts and dedication of WIHS study staff, with special thanks to Ighovwerha Ofotokun, Jess Donohue, Christine Alden, Erin Balvanz, Sarah Sanford, Deja Er, Rachael Farah-Abraham, Andrew Edmonds, Carrigan Parrish, Zenoria Causey, Venetra McKinney, and Lisa Rohn. In addition, the authors express sincere thanks to the State Health Departments and law enforcement agencies who provided data needed to construct census tract predictors.
List of Abbreviations
- AI
Anal intercourse
- AIDS
Acquired immune deficiency syndrome
- ART
Antiretroviral therapy
- CAI
Condomless anal intercourse
- CI
Confidence interval
- CVI
Condomless vaginal intercourse
- HAART
Highly active antiretroviral therapy
- HGLM
Hierarchical generalized linear model
- HIV
Human immunodeficiency virus
- ICC
Intraclass correlation
- OR
Odds ratio
- PCA
Principal components analysis
- QOL
Quality of Life
- SD
Standard deviation
- STI
Sexually transmitted infection
- US
United States
- WIHS
Women’s Interagency HIV Study
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.McDavid K, Li J, Lee LM. Racial and ethnic disparities in HIV diagnoses for women in the United States. J Acquir Immune Defic Syndr. 2006;42(1):101–7. doi: 10.1097/01.qai.0000199353.11479.08. [DOI] [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention. HIV Surveillance Report, 2014. 26 Published November 2015 [Available from: http://www.cdc.gov/hiv/pdf/library/reports/surveillance/cdc-hiv-surveillance-report-us.pdf. [Google Scholar]
- 3.Reif S, Pence BW, Hall I, Hu X, Whetten K, Wilson E. HIV Diagnoses, Prevalence and Outcomes in Nine Southern States. J Community Health. 2015;40(4):642–51. doi: 10.1007/s10900-014-9979-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Adimora AA, Ramirez C, Schoenbach VJ, Cohen MS. Policies and politics that promote HIV infection in the Southern United States. AIDS. 2014 doi: 10.1097/QAD.0000000000000225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Reif SS, Whetten K, Wilson ER, McAllaster C, Pence BW, Legrand S, et al. HIV/AIDS in the Southern USA: A disproportionate epidemic. AIDS Care. 2013 doi: 10.1080/09540121.2013.824535. [DOI] [PubMed] [Google Scholar]
- 6.Cohen DA, Ghosh-Dastidar B, Scribner R, Miu A, Scott M, Robinson P, et al. Alcohol outlets, gonorrhea, and the Los Angeles civil unrest: a longitudinal analysis. Soc Sci Med. 2006;62(12):3062–71. doi: 10.1016/j.socscimed.2005.11.060. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zierler S, Krieger N, Tang Y, Coady W, Siegfried E, DeMaria A, et al. Economic deprivation and AIDS incidence in Massachusetts. Am J Public Health. 2000;90(7):1064–73. doi: 10.2105/ajph.90.7.1064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Biello KB, Kershaw T, Nelson R, Hogben M, Ickovics J, Niccolai L. Racial residential segregation and rates of gonorrhea in the United States, 2003–2007. Am J Public Health. 2012;102(7):1370–7. doi: 10.2105/AJPH.2011.300516. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cohen D, Spear S, Scribner R, Kissinger P, Mason K, Wildgen J. “Broken windows” and the risk of gonorrhea. Am J Public Health. 2000;90(2):230–6. doi: 10.2105/ajph.90.2.230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Thomas JC, Torrone EA, Browning CR. Neighborhood factors affecting rates of sexually transmitted diseases in Chicago. J Urban Health. 2010;87(1):102–12. doi: 10.1007/s11524-009-9410-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Adimora AA, Schoenbach VJ, Taylor EM, Khan MR, Schwartz RJ, Miller WC. Sex ratio, poverty, and concurrent partnerships among men and women in the United States: a multilevel analysis. Ann Epidemiol. 2013 doi: 10.1016/j.annepidem.2013.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ford JL, Browning CR. Neighborhood social disorganization and the acquisition of trichomoniasis among young adults in the United States. Am J Public Health. 2011;101(9):1696–703. doi: 10.2105/AJPH.2011.300213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Jennings JM, Taylor RB, Salhi RA, Furr-Holden CD, Ellen JM. Neighborhood drug markets: a risk environment for bacterial sexually transmitted infections among urban youth. Soc Sci Med. 2012;74(8):1240–50. doi: 10.1016/j.socscimed.2011.12.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Green TC, Pouget ER, Harrington M, Taxman FS, Rhodes AG, O3Connell D, et al. Limiting options: sex ratios, incarceration rates, and sexual risk behavior among people on probation and parole. Sex Transm Dis. 2012;39(6):424–30. doi: 10.1097/OLQ.0b013e318254c81a. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pouget ER, Kershaw TS, Niccolai LM, Ickovics JR, Blankenship KM. 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. doi: 10.1177/00333549101250S411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cooper HL, Linton S, Haley DF, Kelley ME, Dauria EF, Karnes CC, et al. Changes in Exposure to Neighborhood Characteristics are Associated with Sexual Network Characteristics in a Cohort of Adults Relocating from Public Housing. AIDS Behav. 2014 doi: 10.1007/s10461-014-0883-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cooper HL, Haley DF, Linton S, Hunter-Jones J, Martin M, Kelley ME, et al. Impact of public housing relocations: are changes in neighborhood conditions related to STIs among relocaters? Sex Transm Dis. 2014;41(10):573–9. doi: 10.1097/OLQ.0000000000000172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bluthenthal RN, Do DP, Finch B, Martinez A, Edlin BR, Kral AH. Community characteristics associated with HIV risk among injection drug users in the San Francisco Bay Area: a multilevel analysis. J Urban Health. 2007;84(5):653–66. doi: 10.1007/s11524-007-9213-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Cohen DA, Mason K, Bedimo A, Scribner R, Basolo V, Farley TA. Neighborhood physical conditions and health. Am J Public Health. 2003;93(3):467–71. doi: 10.2105/ajph.93.3.467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Truong KD, Sturm R. Alcohol outlets and problem drinking among adults in California. J Stud Alcohol Drugs. 2007;68(6):923–33. doi: 10.15288/jsad.2007.68.923. [DOI] [PubMed] [Google Scholar]
- 21.Shacham E, Lian M, Önen NF, Donovan M, Overton ET. Are neighborhood conditions associated with HIV management? HIV Med. 2013;14(10):624–32. doi: 10.1111/hiv.12067. [DOI] [PubMed] [Google Scholar]
- 22.Marks G, Crepaz N, Senterfitt JW, Janssen RS. Meta-analysis of high-risk sexual behavior in persons aware and unaware they are infected with HIV in the United States: implications for HIV prevention programs. J Acquir Immune Defic Syndr. 2005;39(4):446–53. doi: 10.1097/01.qai.0000151079.33935.79. [DOI] [PubMed] [Google Scholar]
- 23.Bacon MC, von Wyl V, Alden C, Sharp G, Robison E, Hessol N, et al. The Women's Interagency HIV Study: an observational cohort brings clinical sciences to the bench. Clin Diagn Lab Immunol. 2005;12(9):1013–9. doi: 10.1128/CDLI.12.9.1013-1019.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hessol NA, Weber KM, Holman S, Robison E, Goparaju L, Alden CB, et al. Retention and attendance of women enrolled in a large prospective study of HIV-1 in the United States. J Womens Health (Larchmt) 2009;18(10):1627–37. doi: 10.1089/jwh.2008.1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al. The Women's Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998;9(2):117–25. [PubMed] [Google Scholar]
- 26.Hessol NA, Schneider M, Greenblatt RM, Bacon M, Barranday Y, Holman S, et al. Retention of women enrolled in a prospective study of human immunodeficiency virus infection: impact of race, unstable housing, and use of human immunodeficiency virus therapy. Am J Epidemiol. 2001;154(6):563–73. doi: 10.1093/aje/154.6.563. [DOI] [PubMed] [Google Scholar]
- 27.Adimora AA, Hughes JP, Wang J, Haley DF, Golin CE, Magnus M, et al. CHARACTERISTICS OF MULTIPLE AND CONCURRENT PARTNERSHIPS AMONG WOMEN AT HIGH RISK FOR HIV INFECTION. J Acquir Immune Defic Syndr. 2013 doi: 10.1097/QAI.0b013e3182a9c22a. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Jenness SM, Begier EM, Neaigus A, Murrill CS, Wendel T, Hagan H. Unprotected anal intercourse and sexually transmitted diseases in high-risk heterosexual women. Am J Public Health. 2011;101(4):745–50. doi: 10.2105/AJPH.2009.181883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Justman J, Befus M, Hughes J, Wang J, Golin CE, Adimora AA, et al. Sexual Behaviors of US Women at Risk of HIV Acquisition: A Longitudinal Analysis of Findings from HPTN 064. AIDS Behav. 2015 doi: 10.1007/s10461-014-0992-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bozzette SA, Hays RD, Berry SH, Kanouse DE, Wu AW. Derivation and properties of a brief health status assessment instrument for use in HIV disease. J Acquir Immune Defic Syndr Hum Retrovirol. 1995;8(3):253–65. doi: 10.1097/00042560-199503010-00006. [DOI] [PubMed] [Google Scholar]
- 31.Raudenbush SW, Bryk AS. Hierarchical linear models : applications and data analysis methods. 2. Thousand Oaks: Sage Publications; 2002. p. xxiv.p. 485. [Google Scholar]
- 32.VanderWeele TJ. Explanation in causal inference : methods for mediation and interaction. New York: Oxford University Press; 2015. p. xvi.p. 706. [Google Scholar]
- 33.Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol. 2005;162(3):199–200. doi: 10.1093/aje/kwi188. [DOI] [PubMed] [Google Scholar]
- 34.Allison PD. Handling Missing Data by Maximum Likelihood, Paper 312-2012. 2012 Available from: http://www.statisticalhorizons.com/wp-content/uploads/MissingDataByML.pdf.
- 35.Baggaley RF, White RG, Boily MC. HIV transmission risk through anal intercourse: systematic review, meta-analysis and implications for HIV prevention. Int J Epidemiol. 2010;39(4):1048–63. doi: 10.1093/ije/dyq057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Livak BS, Prachand NG, Benbow N. Anal intercourse and HIV risk among low-income heterosexual women: findings from Chicago HIV behavioral surveillance. Open AIDS J. 2012;6:142–8. doi: 10.2174/1874613601206010142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Jennings JM, Taylor R, Iannacchione VG, Rogers SM, Chung SE, Huettner S, et al. The available pool of sex partners and risk for a current bacterial sexually transmitted infection. Ann Epidemiol. 2010;20(7):532–8. doi: 10.1016/j.annepidem.2010.03.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Haley DF, Kramer MR, Adimora AA, Haardörfer R, Wingood GM, Ludema C, et al. Relationships between neighbourhood characteristics and current STI status among HIV-infected and HIV-uninfected women living in the Southern USA: a cross-sectional multilevel analysis. Sex Transm Infect. 2017 doi: 10.1136/sextrans-2016-052889. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.He F, Hensel DJ, Harezlak J, Fortenberry JD. Condom Use as a Function of Number of Coital Events in New Relationships. Sex Transm Dis. 2016;43(2):67–70. doi: 10.1097/OLQ.0000000000000390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Anderson JE, Warner L, Macaluso M. Condom use among US adults at last sexual intercourse, 1996–2008: an update from national survey data. Sex Transm Dis. 2011;38(10):919–21. doi: 10.1097/OLQ.0b013e31822545d9. [DOI] [PubMed] [Google Scholar]
- 41.Sullivan K, Voss J, Li D. Female disclosure of HIV-positive serostatus to sex partners: a two-city study. Women Health. 2010;50(6):506–26. doi: 10.1080/03630242.2010.516697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Latkin CA, Curry AD, Hua W, Davey MA. Direct and indirect associations of neighborhood disorder with drug use and high-risk sexual partners. Am J Prev Med. 2007;32(6 Suppl):S234–41. doi: 10.1016/j.amepre.2007.02.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Latkin CA, German D, Hua W, Curry AD. Individual-level influences on perceptions of neighborhood disorder: A multilevel analysis. J Community Psychol. 2009;37(1):122–33. doi: 10.1002/jcop.20284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Senn TE, Carey MP, Vanable PA, Urban MA, Sliwinski MJ. The male-to-female ratio and multiple sexual partners: multilevel analysis with patients from an STD clinic. AIDS Behav. 2010;14(4):942–8. doi: 10.1007/s10461-008-9405-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Cunningham WE, Sohler NL, Tobias C, Drainoni ML, Bradford J, Davis C, et al. Health services utilization for people with HIV infection: comparison of a population targeted for outreach with the U.S. population in care. Med Care. 2006;44(11):1038–47. doi: 10.1097/01.mlr.0000242942.17968.69. [DOI] [PubMed] [Google Scholar]
- 46.Ghanem KG, Hutton HE, Zenilman JM, Zimba R, Erbelding EJ. Audio computer assisted self interview and face to face interview modes in assessing response bias among STD clinic patients. Sex Transm Infect. 2005;81(5):421–5. doi: 10.1136/sti.2004.013193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Gindi RM, Sifakis F, Sherman SG, Towe VL, Flynn C, Zenilman JM. The geography of heterosexual partnerships in Baltimore city adults. Sex Transm Dis. 2011;38(4):260–6. doi: 10.1097/OLQ.0b013e3181f7d7f4. [DOI] [PubMed] [Google Scholar]
- 48.Cooper HL, Bonney L, Luo R, Haley DF, Linton S, Hunter-Jones J, et al. Public Housing Relocations and Partnership Dynamics in Areas With High Prevalences of Sexually Transmitted Infections. Sex Transm Dis. 2016;43(4):222–30. doi: 10.1097/OLQ.0000000000000419. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.