We conducted a community-engaged research project in collaboration with syringe service programs (SSPs) in three urban US settings (Las Vegas, Los Angeles, and Atlanta) from May 2019 to February 2020. Our project included the delivery of an online tablet-based questionnaire which participants completed independently while seeking services at participating SSP locations. SSP location staff were involved in our recruitment efforts and facilitated relationship building with this population as our project staff did not have this established relationship. Data collection sites were chosen based on characteristics of the population of people using opioids, particularly those who may be vulnerable to HIV. The selected cities were Atlanta, Georgia; Los Angeles, California; and Las Vegas, Nevada. SSP’s were included in this project as project team leadership had previously developed collaborative partnerships with these entities through long-standing relationships between project leadership and SSP staff.
Sampling
Eligible participants included those who had accessed designated SSP’s services, reported using opioids in the previous six months, were 18–69 years of age, had English or Spanish language comprehension, and provided voluntarily consent to participate. A sampling target of 400 per city (total of 1200) was chosen to fit within funding restraints while providing an adequate sample size for city-specific and comparative analyses. This sampling target also ensured adequate samples of subpopulations such as LGBTQ + people. For completion of the questionnaire respondents were provided a nominal well-being item (e.g., hand sanitizer, hygiene kit, sunscreen, $5 McDonalds giftcard).
Study instrument
Survey data were collected using an interview administred, tablet-based online questionnaire, which enrolled participants completed on electronic tablets at SSPs. Our survey was developed with valid and reliable measures and was offered to participants in English and Spanish. A full explanation of the survey and protocol for data collection have been previously published [29].
Our outcome of interest was self-reported acceptability of 3-month LA injectable PrEP. Participants were asked the following question regarding LA injectable PrEP: “Please indicate how likely you would be to use injectable PrEP that could be taken every three months” with the following response categories: ‘Not very likely’, ‘Somewhat unlikely’, ‘Neither likely nor unlikely’, ‘Likely’ and ‘Extremely likely’. Response categories were collapsed such that ‘likely’ and ‘extremely likely’ were recoded as ‘likely’ while the remaining groups were recoded as ‘not likely’. We also asked participants about acceptability of one-month LA injectable PrEP, although this was not our primary outcome of interest.
We used the following independent variables: injection drug use, age, race/ethnicity, gender identity, sexual orientation, location of participants, U.S nativity, educational attainment, employment status, controlled environment, access to insurance, annual household income, SNAP benefits, and unstable housing/homelessness.
Substance use
Injection drug use in the past 6 months was the main independent variable. This was recoded by combining affirmative responses to injection (intravenous [IV] vs. non-IV) as the route of administration for any substances. We also captured data on other substance usage through questions which asked if participants had used substances (e.g., heroin, cannabis, opioids, amphetamines) in the past six months.
Race/ethnicity
Both race and ethnicity were self-reported by participants. The two variables, race and ethnicity, were recoded into a single variable so that Hispanic, non-Hispanic White, Non-Hispanic Black, and Other were the subsequent categories of this newly collapsed variable.
Educational attainment
The original eight response categories for highest educational attainment were recoded into three new categories: ‘Up to High School’; ‘High School to Associate’s’; and ‘Higher education’.
Relationship status
Response categories were recoded to reflect ‘Never married’, ‘Ever married’ and ‘Partnered’.
Employment status
Categories to describe employment status were coded as follows: ‘Unemployed and looking for work,Unemployed and not looking for work, Employed, and Out of Workforce.
Controlled environment
Responses to questions about living in a controlled environment in the last six months (i.e., jail, prison, alcohol or drug treatment center, psychiatric treatment center) were recoded as categorized as Yes, No, and Never.
Access to insurance
Participants who reported any access to insurance were recoded as ‘Yes’. Participants without reported access to insurance were recoded into ‘No access’, while those who were unsure about their insurance access were recoded as ‘Unsure’.
Gender identity/sexual orientation
Using the variables gender identity and sexual orientation, a new variable was constructed with four categories to understand differences between cisgender heterosexual men, cisgender heterosexual women, cisgender sexual minority men and cisgender sexual minority women. Participants not identifying as either ‘male’ or ‘female’ were not included in analysis, due to small sample size (n = 2).
Pre-exposure prophylaxis (PrEP)
Respondents provided answers to a series of questions pertaining to PrEP awareness, knowledge, acceptability, and use. Respondents were provided a brief description of what PrEP is and that it can “lower their risk of getting HIV”. Data derived from these responses were not recoded and data are presented with the original answer options. Questions asking about likelihood to use novel PrEP methodologies (e.g., 1-month injectable, 3-month injectable, lubricants, spermicides) utilizing a likert scale of likelihood with ‘Not Very Likely’ (1) and ‘Extremely Likely’ (5) on either end of the spectrum. Our instrument also asked participants if they had ever heard of or used PrEP and whether they were currently using PrEP. We also asked participants if they knew where they could get PrEP if they wished to start taking PrEP.
Data storage
Participants were given an identification number. We obtained demographic information during the screener and survey, which was collected on a Health Insurance Portability and Accountability Act (HIPAA)-compliant survey administration platform, Qualtrics, (Provo, UT). Qualtrics safeguards all data and uses secure data centers to ensure the highest protection per HITECH requirements. Previous publications have detailed our protocol, including participant inclusion criteria and sampling, field activities including informed consent, and analytic strategy [25]. Our project is among the first to capture a holistic view of health behaviors and to examine PrEP knowledge and LA injectable PrEP acceptability among PWID in these geographies.
Data analysis
SAS v.9.4 was used to conduct statistical analysis. In total, 1127 individuals consented and completed the survey. Fifty individuals who reported living with HIV were not included in the analysis. Among the 1077 individuals who responded to questions pertaining to PrEP, 80 individuals did not respond to the question regarding 3-month injectable-PrEP acceptability and therefore, were not included in the final analytic sample, which was comprised 997 participants across the three sites. We computed bivariate and multivariable logistic regressions to evaluate correlates of the outcome variable: acceptability of 3-month injectable-PrEP.
Based on contextual knowledge and bivariate analyses, a set of candidate variables were introduced for model specification to evaluate factors contributing to acceptability of LA injectable PrEP. A stepwise strategy was used to select the most parsimonious model. This model was chosen because it had minimum AIC compared to models with fewer variables.