This was a nonrandomized, single-arm clinical trial that used a pre-post comparison with extended follow-up. The majority of participants would be people with HIV but a subset of persons without HIV were also included as controls and to allow for subgroup comparisons. After confirming eligibility, all participants provided preintervention data and then received a CM intervention to stop or reduce drinking for 30 to 90 days, and a MI intervention 30 days after baseline. Detailed clinical and behavioral assessments were collected at baseline, 30 days, 90 days, and 12 months after enrollment. The first participant was enrolled in December 2017 and final participant data were collected in April 2022. The study procedures were modified after study initiation to include additional data related to the gut microbiome and to adapt to the COVID-19 pandemic in 2020-2022.

Ethics approval was obtained from the institutional review boards at Florida International University (FIU; IRBSITE00000291), the University of Florida (UF; CED000000011), and the University of Miami (UM, 20170396). The study was registered at ClinicalTrials.gov (NCT03353701) in November 2017.

Participants were recruited in the Miami Metropolitan region (Florida) from HIV clinics, community outreach, and a contract registry. Recruitment advertisements were placed on public transportation and in local HIV clinics. Potential participants were screened over the phone or in person and those initially eligible were invited to attend an enrollment visit to review procedures, obtain consent, and confirm eligibility.

Potential participants were informed that one of the major goals of the study was to examine changes in the body after drinking reduction, and therefore they were asked to participate in a “30-Day Challenge,” in which they would try to reduce or stop drinking for at least 30 days. After informed consent was obtained, participants completed a detailed assessment to determine whether they were eligible to continue with the study.

The inclusion criteria were age 45-75 years, 21 or more drinks per week for men or 14 or more drinks per week for women, confirmed HIV status (for those who reported being HIV positive), English speaking, willingness to participate in CM, and wear an alcohol biosensor for at least 30 days. Exclusion criteria included neurological disorders (eg, dementia, stroke, seizures, and traumatic brain injury); past opportunistic infection; major psychiatric disturbance (eg, severe major depression); unstable medical conditions (eg, cancer); magnetic resonance imaging (MRI) contraindications (eg, pregnancy, severe claustrophobia, metal implants, and physical impairment precluding motor response or lying still); inability to demonstrate an understanding of key aspects of the study; and currently participating in other alcohol research.

Additional assessments used to determine eligibility at this enrollment visit included the Montreal Cognitive Assessment (MoCA) [23], the Alcohol Withdrawal Symptom Checklist (AWSC) [24], and a 30-Day Drug and Alcohol Timeline Followback (TLFB) [25]. Participants with MoCA scores lower than 17 were discussed with the investigators and allowed to proceed if they could clearly discuss the study goals and purpose with the research assistants. Those with AWSC scores greater than 8 were excluded due to the high risk of alcohol withdrawal. The TLFB was used to determine the average number of drinks per week, and persons with less than 21 drinks per week (men) or 14 drinks per week (women) were also excluded from the clinical intervention at this point.

At the enrollment visit, those who appeared to be eligible had the SCRAM CAM biosensor placed on their ankle. The monitor strap has a specialized clip that prevents removal without breaking the clip or cutting the ankle strap (and any removal sends an alert to SCRAM Systems and is viewable by the research team). The purpose of participants wearing the monitor for this week was to confirm that they did drink (for at least 3 days), they could go at least 24 hours without drinking and without withdrawal symptoms, they could tolerate wearing the SCRAM CAM, and they would communicate as expected with the research assistant. During this pre-CM phase, participants were instructed to drink as they normally would, except for at least 1 day of required abstinence (for persons who drink every day). Participants were given instructions about the monitor, including not submerging the device in water, avoiding using alcohol-based items, and not wearing socks under the monitor. A research assistant called the participant every other day to collect information on self-reported drinking and compared this information to that obtained by remote download from a cloud server from the SCRAM website. This ensured that for each participant, the monitor could accurately distinguish drinking from nondrinking days. During this pre-CM test week, participants received incentive payments for providing self-reports but no incentive for drinking behavior itself. During this pre-CM test week, some participants chose to withdraw, mostly due to not wanting to wear the ankle biosensor, and some were excluded because they did not drink enough or were not able to follow study procedures. All other people were scheduled to attend an in-person baseline assessment and to choose a specific date on which they would start the 30-Day Challenge.

The baseline assessment included a study questionnaire, neurocognitive assessments, neuroimaging assessments, liver Fibroscan, and collection of blood, urine, and stool samples. The baseline questionnaire was completed either during the enrollment visit (before the test week) or at the baseline assessment. Participants were required to have a 0 breathalyzer reading in order to proceed with the baseline questionnaire. The primary domains of the measures and the administration schedule are in Table 1. The study questionnaire items are available from the research team upon request.

A comprehensive battery of neuropsychological measures was administered to all participants (Table 2). The National Institutes of Health (NIH) toolbox (cognition battery) was given to participants to obtain an estimate of their crystalized (2 assessments) and fluid intellect (5 assessments) [43]. Uncorrected summary scores were created for the NIH toolbox crystalized and fluid scores and further analyses of these index scores included demographic factors such as age, race, and education in the models. For follow-up cognitive assessments, we used different forms when available, because this helps reduce the magnitude of practice effects.

Due to reported participant fatigue in the early phases of the research, the initial battery was refined to be completed in 1 hour or less (we dropped the California Computerized Assessment Package [CALCAP], Wechsler Adult Intelligence Scale-Fourth Edition [WAIS-IV], Brief Visuospatial Memory Test-Revised [BVMT-R]). Research assistants met with the team neuropsychologist regularly to ensure the best practices to maintain rapport and participant engagement. Reminders were given that the tests were purposefully created to be difficult, breaks were offered when somnolence was observed, and the research staff documented when people appeared to be providing limited effort. A summary of the specific neurocognitive tests and neuroimaging assessments obtained at their respective timepoints is included in Table 2.

For data analysis, the primary outcome for neurocognitive assessments is the change in performance on the 5 measures comprising the Fluid Cognition index from the NIH toolbox, which includes memory, attention, cognitive flexibility, processing speed, and executive functioning. Secondary outcomes of cognition will include changes in the other neuropsychological measures.

Participants who had no contraindications underwent MRI neuroimaging at all 4 time points. Prior to the first MRI, participants were provided with instructions on the 2 functional MRI tasks. For example, they were administered the 2-back test, which required participants to view alphabets in English (ie, a stimulus) on a computer screen and indicate by clicking a button on a device held in their hand whether the currently displayed letter was the same or different from the letter that appeared 2 preceding times ago (ie, 2-back). The MRI protocol used for all 4 timepoints of this study is shown in Table 3. The total time to complete all the sequences in the MRI protocol was approximately 65 minutes.

The primary outcome from neuroimaging assessments is changes in brain inflammation from baseline to subsequent timepoints at the regional, tissue-type (ie, gray matter and white matter), and whole-brain levels. This outcome will be assessed from the quantitation of cerebral metabolite markers of neuroinflammation (ie, total choline and myo-inositol) and extracellular free water fraction (a measure determined from diffusion tensor imaging data). The above neuroinflammation markers will be quantified from the brain regions including the basal ganglia, thalamus, and frontal lobe that are primarily involved in HIV infection, alcohol use disorders, and their interaction [57-59]. Changes in brain function from functional magnetic resonance imaging and resting-state connectivity data will be measured for 5 major networks, that is, the default mode network, the dorsal attention network, the salience network, the limbic network, and the fronto-parietal control network.

Blood samples were collected at each timepoint. Part of these were sent to a commercial laboratory for measurements of complete blood count with differential, comprehensive metabolic panel, hepatitis C antibody, HIV antibody (to confirm HIV status), HIV-1 RNA (only for people with HIV), and CD4 lymphocyte count (only for people with HIV). A dried blood spot was collected to measure phosphatidylethanol, an alcohol biomarker. We used these samples and performed measurements of cytokines, inflammatory biomarkers, adhesion molecules, and markers of intestinal permeability and microbial translocation. Additional blood samples are stored in a biorepository at the University of Louisville, where there is planned testing related to gut microbiome and gut-derived metabolites (metabolomics). Urine tests were performed at each timepoint for ethyl glucuronide (an alcohol biomarker), drug screen, and (after approximately 1 year) urine specific gravity. Urine-specific gravity was collected to ensure any potential changes in free water–based neuroinflammation (Table 4) were not solely due to brain rehydration after abstinence from alcohol.

During the first year of the study, the research team received additional funding to add a collection of stool samples for gut microbiome assessment and to measure additional blood biomarkers related to systemic inflammation (U01AA026225). These assessments, together with a food frequency questionnaire [60], were added after 17 participants had been enrolled, but 11 provided this information at a follow-up visit. Stool samples were sent to the University of Louisville for 16S rRNA gene sequencing processing, taxonomic evaluation, and determination of longitudinal changes in bacterial composition and diversity.

A Fibroscan liver test was obtained on all participants (Fibroscan 502 Touch, EchoSens, Paris, and France with the XL probe). The 2 scores were calculated using an average of 10 assessments, fat or controlled attenuation parameter score and fibrosis or liver stiffness measurement.