High and Medium Drinking--vs Low Drinking--Boost Death Risk in Big HIV Group
CROI 2020, March 8-11, 2020, Boston
Mark Mascolini

A study 33,000 people with HIV in the United States and Europe linked medium or high alcohol drinking to a higher death risk than low-level drinking [1]. Nondrinkers also ran a higher death risk than the low-level group

Because excessive alcohol drinking can threaten health and survival, researchers leading two big HIV collaborations (ART-CC and COMpAAAS) assessed the impact of drinking on viral load and mortality in 5 cohorts that measured alcohol use with the AUDIT-C tool [2]. AUDIT-C classifies people as low-, medium-, or high-level drinkers or as nondrinkers. Follow-up continued from a person's first AUDIT-C test after antiretroviral therapy (ART) began and continued until death or loss to follow-up (failure to keep study appointments).

The researchers used Cox survival models to estimate (1) chance of an undetectable viral load at baseline, (2) risk of virologic failure in people with an undetectable viral at their baseline visit and (3) risk of all-cause mortality in the whole study group. They adjusted these models for baseline CD4 count, age, gender, and HIV transmission risk group. They also adjusted the mortality analysis for baseline viral load.

The study group included 9475 nondrinkers (28.5% of the whole group), 19,585 low drinkers (59%), 3298 medium drinkers (9.9%), and 848 high drinkers (2.6%). Men made up a lower proportion of low drinkers than women (57% versus 77%) but higher proportions of nondrinkers (30% versus 14%), medium drinkers (10% versus 7%), and high drinkers (3% versus 2%). Among people who died during follow-up, 45% were nondrinkers, 44% low drinkers, 7% medium drinkers, and 4% high drinkers.

Compared with low-level drinkers, medium- and high-level drinkers had significantly higher odds of having a detectable viral load at the baseline visit (adjusted odds ratio [aOR 1.14, 95% confidence interval [CI] 1.05 to 1.24 for medium drinkers; aOR 1.57, 95% CI 1.36 to 1.83 for high drinkers). Among the 22,834 people with an undetectable baseline viral load, medium and high drinkers rebounded to a detectable load significantly faster than low drinkers (aOR 1.13, 95% CI 1.02 to 1.26 for medium drinkers; aOR 1.60, 95% CI 1.35 to 1.90 for high drinkers).

During 183,683 person-years of follow-up, 4056 study participants died. Compared with low-level drinkers, each of the other three groups ran an independently higher risk of dying during the study, at the following adjusted hazard ratios (aHR):

-- Nondrinkers: aHR 1.33, 95% CI 1.24 to 1.42
-- Medium drinkers: aHR 1.20, 95% CI 1.06 to 1.36
-- High drinkers: aHR 1.99, 95% CI 1.70 to 2.33

The researchers speculated that the higher death risk in nondrinkers than low-level drinkers may indicate that a portion of nondrinkers had quit high-level drinking but endured substantial health damage during their drinking days. The investigators called for "interventions to reduce unhealthy alcohol use" in people with HIV.

The higher risks of a detectable viral load and death with medium-level drinking than with low-level drinking may be sobering to some. Clinicians and people with HIV can rate themselves on the AUDIT-C scale to see if they fall into the low group (AUDIT-C score 1-4), the medium group (score 5-8), or the high group (score 9-12) [2]. A person who drinks 2 to 4 times a month (2 points) but has 5 or 6 drinks when they do drink (2 points) and had 6 or more drinks less than monthly in the past year (1 point) would be a medium-level drinker.

1. Trickey A, Ingle S, Zhang L, et al. Associations of alcohol consumption with viral suppression and all-cause mortality. Conference on Retroviruses and Opportunistic Infections (CROI). March 8-11, 2020. Boston. Abstract 895.
2. U.S. Department of Veterans Affairs. Alcohol use disorders identification test. https://www.hepatitis.va.gov/alcohol/treatment/audit-c.asp