Study population: Participants were enrolled in the Intervention for Seropositive Injectors- Research and Evaluation (INSPIRE) study, a 10-session secondary HIV prevention intervention conducted in Baltimore, Miami, New York City, and San Francisco (2001–2005). Participants were recruited from clinic and community settings. Eligibility criteria included being at least 18 years of age, confirmed HIV-positive serostatus, self-reported IDU in the prior year, sex with an opposite-sex partner in the prior three months, and willingness to provide oral and blood specimens. Assessments were administered by audio computer-assisted self interview. The CDC and local site IRBs approved the study protocol. A detailed description of the INSPIRE project has been reported elsewhere.8
The overall retention rates for randomized participants were 83% and 85% at 6- and 12-months follow-up, respectively, and 91% for either visit.
Measures: Measures were based on the findings of Metsch et al.14
and on the sampling distributions. Demographic variables included age, biological sex, educational attainment (having at least a high school degree), employment status, and homelessness (squat, homeless shelter, car, or street). Two questions, with a 5-point response option, assessed whether the participant thought that their friends lent needles or shared injection equipment, and two questions assessed whether the participant thought it was important to comply with friends’ expectations about lending needles or sharing injection equipment. A composite score was created from the 4 items. Mental health was assessed by combining 3 subscales (depression, anxiety, and hostility – 18 items) of the BSI (α=.95), which assessed psychiatric symptoms in the prior week.19
HIV primary health care visits were defined as “a visit to a doctor or medical provider to have a check up on how you’re doing with your HIV or AIDS, (which may include) discussion about HIV or AIDS medications, or blood test results.” Drug abuse treatment was measured by: 1) hospital admission for drug treatment; 2) outpatient drug treatment; and 3) methadone maintenance program attendance. The dependent variable, referred to in this study as “sharing” was defined as self-reported sharing of injection equipment (cookers, cotton, or rinse water) or lending a used syringe to individuals who were HIV-seropositive, HIV-seronegative, or of unknown serostatus at either the 6- or 12-months follow-up.
Missing data: A total of 89 (out of 966) participants did not return for assessments at the 6- or 12-months follow-up. Sensitivity analyses (p < 0.10), revealed that at baseline participants who were lost to follow-up did not differ from participants who were retained on most baseline measures, including study condition, homelessness, recruitment city, health care visits, or drug treatment. Those lost to follow-up were significantly less likely to have a high school education, an annual income below $5,000, younger, employed, male, and inject heroin and cocaine together.
The sample for this analysis (n=357) was restricted to participants who met the study enrollment criteria and reported injecting drugs in the prior 3 months during the 6 or 12-month follow-up assessments. There were 486 participants who reported no injection drug use in the 3 months prior on either follow-up assessment and 44 individuals with missing data who were excluded from the analyses. Participants were categorized as not sharing drug equipment (n=195) if they reported using injection drugs at one or more of the follow-up visits and reported not lending syringes or sharing of cookers, cottons, or rinse water.
Analyses: Bivariate analyses were conducted with logistic regression models. Subsequently, a multivariable logistic regression model was constructed to test which baseline variables were associated with sharing injecting equipment at follow-up. In addition to demographic factors of gender, age, and education, the multivariable model adjusted for recruitment city and study condition. Bivariate tests indicating statistical differences at p < 0.20 were included in multivariate models. Variables no longer significant (p > 0.10) in the multivariable model were removed if the removal did not change the value of the other covariates in the model by more than 10% and were not considered to be theoretically important confounders.