The relationship of smoking and unhealthy alcohol use to HIV care retention and viral suppression: findings from a multi-site cohort study

Objectives: Tobacco smoking and alcohol use may negatively influence HIV care, but associations have not been examined across cohorts. Design: Multisite international collaboration of cohort studies. Methods: People with HIV (PWH) were included from 11 cohorts; 5 North American and 6 Western European. Exposures were harmonized smoking and alcohol measures (2010–2018). Loss to care was defined as not having 2+ HIV care visits (HIV RNA and/or CD4 measurement dates) at least 60 days apart, within 12 months following alcohol measure date; HIV viral non-suppression was defined as >200 copies/mL. We calculated adjusted prevalence ratios (PRs) with modified Poisson regression, pooled effect estimates by random-effect meta-analysis, and variability (I$^{2}$). Results: Among 83,102 PWH (87.4% male, 46.1% white); 43.7% currently smoked, 44.5% reported low/moderate drinking, 6.9% heavy drinking, 48.6% did not drink. PWH who currently smoked had higher risk of loss to care than non-smoking PWH (pooled PR [95% CI]=1.12 [1.08–1.16], I$^{2}$ = 18.1%); those with heavy drinking had higher risk than those with low/moderate drinking (1.13 [1.03–1.25], I$^{2}$ = 57.8%). PWH who currently smoked had higher risk of viral non-suppression than non-smoking PWH (1.44 [1.25–1.67], I$^{2}$ = 90.6%); those reporting heavy drinking had higher risk than those with low/moderate drinking (pooled PR [95% CI]=1.18 [1.02–1.37], I$^{2}$ = 68.9%). PWH who reported heavy drinking and current smoking, in comparison to low/moderate alcohol use but no current smoking, had highest risk of viral non-suppression (pooled PR [95% CI] =1.74 [1.37–2.22]), I$^{2}$ = 81.8%. Conclusions: Smoking and unhealthy alcohol use were associated with HIV loss to care and viral non-suppression, with variability between cohorts.