Study design and setting
We conducted a cross-sectional analysis of PrEP awareness and associated factors measured from baseline data for a two-year cohort study. Between January and October 2019, 14-24-year-old AGYW were recruited from commercial hotspots (i.e., where sex work is prevalent) and slums of southern and northern Kampala, and invited to enroll into a cohort of AGYW at the Good Health for Women Project (GHWP) clinic located in a peri-urban community in southern Kampala (23). The clinic offered HIV prevention and care services, free general health care and reproductive health services to eligible high-risk women including female sex workers (FSWs).
Study population and sampling design
We consecutively enrolled participants into the parent cohort study that was assessing knowledge and preferences of biomedical HIV prevention methods, and uptake of oral PrEP among at-risk 14-to-24-year-old females in Kampala. AGYW were enrolled in the cohort if they were HIV negative, Hepatitis B immune (through prior infection or vaccination) or willing to be vaccinated if susceptible, sexually active in the past 3 months, willing to use effective contraceptive methods, undergo regular HCT, pregnancy testing, and screening for sexually transmitted infections (STIs). Participants were excluded if pregnant or planning pregnancy in the next 12 months; had known allergy to components of oral PrEP drugs, contraceptives or Hepatitis B vaccine; and reported or were diagnosed with an illness that would affect participation in the study.
Data collection
Trained research assistants collected data using standardized paper questionnaires administered face-to-face. We collected data on socio-demographic and risk behavior (i.e.: number of sexual partners; transactional, anal, group or forced sex; condom use with sexual partners; knowledge of sexual partners’ HIV status; and having STIs). Data were double entered into an Open Clinica database by trained data entrants. Discrepancies were resolved by referring back to the source documents.
Study variables
Primary outcome:
Awareness of Oral PrEP was the primary outcome in this study and was measured as a binary variable (Yes/No). As a secondary outcome, awareness of injectable PrEP (a hypothetical product still in clinical trials at the time of this study) was also measured as a binary variable (Yes/No).
Volunteers were also assessed on; 1) their knowledge about of oral and injectable PrEP (Pills were demonstrated using samples at site), 2) the primary source of information for those who reported prior knowledge about these methods and 3) willingness to use these methods after explaining their benefits and frequency of use.
Independent variables included the following: socio-demographics factors included age, highest level of education (below or above primary level), marital status (married, separated, widowed, or single) and main occupation (sex work, no job, other).
Behavioural risk factors included alcohol consumption risk level and drug use. Drug use was defined as use of any illicit drugs in the past 1 month. Alcohol consumption risk level was based on the score of the Alcohol use disorders identification test (AUDIT) questionnaire (24). Participants were categorised as low risk (AUDIT score <8), moderate risk/hazardous (AUDIT score 8-19), and high risk/ dependent (AUDIT score ≥20).
Sexual risk factors included age at first sex, number of sexual partners in the past 3 months, consistent condom use with sexual partners in the past 3 months, transactional sex (receiving and/ or giving money and goods for sex), history of STI symptoms (abnormal discharge and/ or genital ulcer), frequent traveling away from home in the past 3 months and current contraceptive use.
Statistical analysis
We analysed data using Stata version 15 (StataCorp, College Station, TX, USA). Baseline characteristics of the participants were summarised using frequencies and proportions. Chi-square and Fisher’s exact tests were used to assess associations with awareness of oral PrEP. Univariable robust poisson regression was used to estimate crude prevalence ratios (PR) and 95% confidence intervals (CIs). Variables significant at p < 0.1 and those hypothesized to be associated with the outcome based on prior knowledge (i.e., condom use, age and level of education) were selected for inclusion in a multivariable robust poisson regression model. We also tested for collinearity and for the variables that were found to be highly correlated with each other only one of these was selected and included in the final model. Tests for interactions between variables were conducted using the Wald test. The final model was selected based on variables that were found to be independently associated (P<0.05) with the outcome. After fitting the final model, adjusted prevalence ratios (aPR) and 95% CIs were obtained and reported.
Ethical considerations
Ethical approval was obtained from the Uganda Virus Research Institute Research Ethics Committee (GC/127/18/06/658) and Uganda National Council for Science and Technology (HS2435). Participants gave written informed consent before being enrolled.