This was a cross-sectional study comparing a single-question screening tool to the standard 10-item AUDIT tool for assessing alcohol use (Babor et al., 2001). This study was nested in a community-based simulated vaccine efficacy trial among adults in two fishing communities along Lake Victoria, Uganda, between 2015 and 2018. The study was carried out in Nsazi and Kigungu, Lake Victoria Island fishing communities found in Mukono and Wakiso districts respectively. Briefly, this simulated trial enrolled HIV-uninfected, at-risk volunteers into a study that employed a licensed vaccine for hepatitis B and typhoid to simulate the procedures of a true clinical trial in these communities. The results of this study are not yet published.

We recruited male and female adults aged 18–49 years who were residents in these communities for at least 6 months and willing to comply with study procedures. Participants were excluded if they did not sign the informed consent and complete the baseline questionnaire, risk assessment questionnaire, and the AUDIT interview.

Community engagement officers worked closely with community leadership, peer leaders, and the community advisory board to identify and reach out to potential participants. Community information seminars were held in the study communities to provide study information. Those interested in participating were invited to the field clinic hubs in Nsazi and Kigungu for screening. A subset of participants was selected for enrolment into the immunology and alcohol sub-studies at screening. The baseline, HIV risk assessment (where the single-question on alcohol use was included), and AUDIT face-to-face questionnaires were administered respectively at the screening visit, after documented informed consent with passing of an assessment of trial understanding.

Study staff were trained on the protocol and data collection forms before the study began. Study participants received a complete physical examination and then completed a demographic questionnaire, HIV risk assessment behavioral questionnaire, and the AUDIT questionnaire respectively. Blood was collected for HIV testing per Uganda Ministry of Health algorithm. Participants were screened for HIV using Alere Determine. If participants tested HIV-positive with the screening test, they underwent a confirmatory test with Statpak. If the screening and confirmatory tests were inconsistent, Unigold was used as a tiebreaker test. All rapid HIV-positive results were confirmed using ELISA and participants were referred for care. Blood was also collected for storage. Individuals participated in face-to-face interviews using a paper-based AUDIT questionnaire, which uses scoring to determine risk levels for hazardous alcohol use. A score of 0–7 indicates mild drinking, a score of 8–15 indicates moderate drinking, a score of 16–19 indicates binge drinking, and a score of 20–40 indicates chronic drinking (Babor et al., 2001).

From the HIV risk assessment questionnaire, participants were asked “how often in the last three months they had a drink containing alcohol?” Responses were (1) daily, (2) weekly, (3) 1–3 times/month, (4) less than monthly, and (5) none, which were in turn graded as “little risk” (if the volunteer responded to not having consumed alcohol or consumed alcohol less than monthly, combining 4 and 5 above), “low risk” (if consumption was monthly), “medium risk” (if consumption was weekly), and “higher risk” (if consumption was daily). The collected data was checked for completeness, accuracy, and consistency and double entered in a Microsoft Access database.

Firstly, we performed descriptive statistics to summarize participants’ baseline characteristics. We computed t-tests with medians and interquartile ranges (IQR) to summarize continuous variables, whereas chi-squared tests were employed to summarize categorical variables.

Secondly, Spearman correlation coefficients were used to assess correlation between the single alcohol use screening question and the AUDIT tool. Specifically, we measured the correlation between the four categories from the AUDIT score on alcohol consumption and the four graded categories from the single question from the HIV risk assessment questionnaire on frequency of alcohol use.

Lastly, we calculated sensitivity and specificity of the single question on alcohol consumption compared to the reference AUDIT test. Participants were considered having a “positive” test result if they had an AUDIT score of 20 or higher and if they reported drinking daily in the previous 3 months. We also computed negative and positive predictive values with 95% confidence intervals (CI) for the sensitivity and specificity calculations. All analyses were done using Stata version 16 (College Station, TX, USA).

The study was approved by the Uganda Virus Research Institute Research Ethics Committee, reference number GC/127/15/07/439, and the Uganda National Council for Science and Technology, reference number HS 1850. All participants provided written informed consent.