Abstract
BackgroundPhosphatidylethanol (PEth), a phospholipid obtained from whole blood, offers superior sensitivity and detection windows over other biomarkers of alcohol use, and because it can be assayed from dried blood spots (DBS), it can be a suitable biomaker for virtual clinical trials. This study tested the feasibility of utilizing self-collected DBS to obtain PEth among men who have sex with men (MSM) in two virtual trials and examined associations between PEth and self-reported alcohol use.MethodsParticipants with hazardous drinking were included from Ending the HIV Epidemic (EHE) jurisdictions in the US South and mid-Atlantic who enrolled in two virtual trials from 2021–2024. One trial focused on MSM who were taking HIV pre-exposure prophylaxis (PrEP), and the other was focused on MSM who were not. PEth was analyzed in DBS, and online surveys collected Timeline Follow-back (TLFB) and AUDIT-C data at baseline, three, and six months. Mixed-effects models evaluated predictors of DBS return rates. Spearman rank correlations assessed associations between PEth and self-reported alcohol use, and mixed-effects models examined moderators of this association.ResultsA total of N=339 MSM were provided DBS collection kits. DBS return rates were 94% at baseline, declining to approximately 71-73% at follow-up. Fewer than 1% of samples were rejected for quality-related reasons. Agreement between PEth and self-report in classifying hazardous drinking across baseline and follow-ups was modest overall (62-66%). Spearman correlations were stronger for PEth and AUDIT-C (rs = 0.47) than PEth and TLFB (rs = 0.30). PEth and self-report was consistently weaker in the trial where participants were on PrEP, with the pattern of results suggesting underreporting of alcohol use on self-report measures among those on PrEP. ConclusionsSelf-collection of DBS for PEth analysis is feasible in virtual trials, although completion at follow-up remains challenging. Findings also suggested that brief self-report measures like the AUDIT-C may be the most practical complement to PEth in virtual alcohol research. Biomarkers should be incorporated into assessments of alcohol use in PrEP research, given that reliance on self-report measures may underestimate alcohol use.