Parent Study Overview
We conducted a secondary analysis of baseline data from the Increasing PrEP (iPrEP) study, a randomized controlled trial (RCT) investigating a survey-based behavioral intervention among cis Black women in two emergency departments (E.D.s) in Houston, Texas, led by Dr. Mandy Hill (Hill 2022; Hill 2023). The study protocol (HSC-MS-16-0892) was approved by the Institutional Review Board of the UTHealth Houston Center for Protection of Human Subjects.
Participants
Eligible participants were HIV-negative Black women, aged 18–55, with non-emergent health issues (emergency severity index, acuity level 3+) (AHRQ, 2020), a working mobile device, English literacy (Harris, 2019), and recent history of condomless sex and substance use in the last 30 days (Hill, Heads, Cardenas-Turanzas, & Grimes, 2020). Exclusion criteria included PrEP ineligibility based on the Public Health Service Clinical Practice Guidelines (CDC, 2017), male sex at birth, use of medications contraindicated for PrEP, and current PrEP use. Of the 2,490 participants screened via electronic medical records (EMR), 58 met the inclusion criteria, and intake screening confirmed 40 eligible participants (68%), all of whom consented to participate in the study, yielding a 100% response rate.
Setting
Study recruitment at two public hospitals spanned between August 27, 2019, and July 2, 2021. Recruitment was paused from March 2020 to March 2021 due to the COVID-19 pandemic but resumed from March 2021 through July 2022. Eligible participants completed an informed consent process and pre-test surveys on tablets using Qualtrics (Qualtrics, 2021), which covered sociodemographic factors, substance use, sexual behaviors, and PrEP-related factors.
Variables
Table 1 shows how the Andersen Behavioral Model factors were organized into one of three theoretical domains–predisposing factors, enabling/disabling factors, and need factors–to ensure adequate operationalization based on precedents from previous research using this model (Andersen, 1995; Andersen, 1973).
Table 1
Conceptual Framework of the Andersen Behavioral Model Factors Influencing PrEP Knowledge and Willingness
Andersen Behavioral Model Factors |
Predisposing | Enabling/Disabling | Need |
Demographic Variables: - Age | Enabling Factors: (Individual/family resources) - Monthly household income - Employment status | Not applicable (NA) (Presented in the ED for non-emergent healthcare) |
Social Structure Variables: - Education - Marital status - Sexual Orientation | Disabling Factors: (Key predictors) - Alcohol problems - Substance use problems - Condomless sex |
Health Belief Variables: - PrEP knowledge |
aTable 1 illustrates the organization of the Andersen Behavioral Model factors into three theoretical domains: predisposing factors (demographic, social structure, and health belief variables), enabling factors (individual and family resources), and disabling factors (key predictors of risk behaviors). The 'Need' category refers to participants presenting in the emergency department for non-emergent healthcare. This framework is based on previous research by Andersen (1995, 1973). |
Measures
Predisposing Factors
Sociodemographic data included age, education, marital status, and sexual orientation (D’Avanzo 2019; Chandran 2019; Seidu 2020). Age was assessed with the question, "My age (in years) is," and left continuous. Education was measured by the question "I have completed" and recoded into three groups: 0 (some high school education), 1 (high school graduate or some college), and 2 (college graduate). Marital status was determined using the question, "What is your marital status (select one)?” and recoded into two groups: 0 (unmarried) and 1 (married). Sexual orientation was determined by the question "My sexual preferences are (select one)?" with three answers: 0 (women who have sex with women (WSW]), 1 (women who have sex with men [WSM]), and 2 (women who have sex with women and men [WSWM]). PrEP knowledge was measured using a survey question explaining PrEP as a daily HIV prevention medication, coded as 0 (not PrEP-knowledgeable) or 1 (PrEP-knowledgeable) (Hill, 2022; Hill, 2023).
Enabling Factors
The enabling factors included monthly household income and employment status. Monthly household income was assessed by the question “My monthly household income is,” and was recoded into three categories: 0) less than $1,000, 1) between $1,001 and $2,000, and 2) $2,000 or more. Employment status, measured by the question “My employment status is,” had two response categories: 0) unemployed and 1) employed.
Disabling Factors
Hill et al. (2023) evaluated substance-using behaviors in three domains:1) alcohol use history, 2) substance use history, and 3) sexual history, which were determined by participants’ answers to questions regarding the frequency with which they experienced these behaviors in the past 30 days (Hill, 2020). The survey questions were specifically worded as follows; "How many days in the past 30 days have you experienced:1) alcohol problems? 2) substance use problems? 3) condomless sex?" (Hill, 2020). The response options for each measure are as follows: (0) none or zero, 1) 1–15 days, 2) 16–30 days (Hill, 2022), which are common approaches in similar studies (Floyd, 2007; Robinson, 2014; Metzger, 1992, 1993; Davies, 2016). This aligns with assessment tools such as Davie's willingness scale and the timeline follow-back method, which considers substance use and sexual risk behaviors within a specific timeframe (Davies, 1996; Bernstein, 2016; RAB, 2023). This approach is statistically valid, avoids arbitrary cutoffs, preserves ordinal data, and allows for meaningful comparisons between individuals with different engagement levels (Sobell 1996; Seidu 2020; Hill 2022).F
The 'need' category refers to participants presenting in the emergency department for non-emergent healthcare (Table 1).
Outcome Variable
The survey evaluated the willingness of individuals to uptake PrEP by posing the question, "Would you be willing to take PrEP to reduce your own risk of contracting HIV?" (Chandran, 2019; D'Avanzo, 2019; Hill, 2020). Originally designed to assess willingness to use alcohol in young adults (Davies, 2017), this scale was modified to evaluate willingness to use PrEP among Black women who reported substance use in the iPrEP study and was dichotomized as 0 (not willing [between 0–5]) and 1 (willing [between 6–10]) (Hill, 2022).
Statistical Analysis
We used descriptive statistics and Chi-square (χ^2) tests for categorical predictors and Mann-Whitney-Wilcoxon tests for continuous predictors in STATA (StataCorp. 2021) to analyze PrEP knowledge, willingness, and various variables. Alcohol problems, substance use issues, and condomless sex were converted into continuous scales to improve analytical accuracy and to manage data gaps. This included multiple imputations using chained equations to address missing data for several critical variables, thus maintaining the robustness of our model (Azur, 2011).
Linear regression models were imputed for continuous variables, with 40 observations per variable. Ten imputed datasets were utilized for reliable estimation with chained equations based on variable relationships to ensure theoretical coherence (Azur, 2011). Logistic regressions for PrEP willingness were conducted by calculating odds ratios (ORs) and 95% confidence intervals (C.I.s) using STATA (StataCorp, 2021), with statistical significance set at p < 0.05, for the predisposing, enabling, and disabling factors.
Moderation Analyses
Moderation analyses were conducted to examine the influence of the enabling/disabling factors on the relationship between PrEP knowledge and willingness. The predictor and moderator variables were mean-centered to address multicollinearity and simplify interpretation. Three logistic regression models were used to test the three research hypotheses, with PrEP willingness as the outcome variable.
Hypothesis 1
Alcohol problems will moderate the relationship between PrEP knowledge and willingness to use PrEP such that the relationship will be weaker among cis Black women with more days of alcohol problems.
Hypothesis 2
Substance use problems will moderate the relationship between PrEP knowledge and willingness to use PrEP such that the relationship will be weaker among cis Black women with more days of substance use problems.
Hypothesis 3
Engaging in condomless sex moderates the relationship between PrEP knowledge and willingness to use PrEP such that the relationship will be weaker among cis Black women who engage in more days of condomless sex.