Estimating the impact of alcohol consumption on survival for HIV+ individuals

Abstract

Alcohol consumption is associated with decreased antiretroviral adherence, and decreased adherence results in poorer outcomes. However the magnitude of alcohol's impact on survival is unknown. Our objective was to use a calibrated and validated simulation of HIV disease to estimate the impact of alcohol on survival. We incorporated clinical data describing the temporal and dose-response relationships between alcohol consumption and adherence in a large observational cohort (N=2,702). Individuals were categorized as nondrinkers (no alcohol consumption), hazardous drinkers (consume > or =5 standard drinks on drinking days), and nonhazardous drinkers (consume <5 standard drinks on drinking days). Our results showed that nonhazardous alcohol consumption decreased survival by more than 1 year if the frequency of consumption was once per week or greater, and by 3.3 years (from 21.7 years to 18.4 years) with daily consumption. Hazardous alcohol consumption decreased overall survival by more than 3 years if frequency of consumption was once per week or greater, and by 6.4 years (From 16.1 years to 9.7 years) with daily consumption. Our results suggest that alcohol is an underappreciated yet modifiable risk factor for poor survival among individuals with HIV.

Figure 1. Influence diagram of selected constructs in computer simulation

Clinical characteristics are assumed to affect the probability of dying from HIV-related or non–HIV-related causes and also to affect viral genetic characteristics by altering viral replication rates and selection pressures for new mutations. Viral replication and selection pressures may give rise to HIV mutations. Each HIV mutation may or may not give rise to resistance to one or more HAART drugs. Adherence to HAART combined with resistance to HAART and other patient characteristics determine the level of HAART effectiveness. The effectiveness of HAART influences changes in viral load and CD4 count, and also feeds back to influence changes in the viral replication rate. Changes in CD4 and viral load influence the risk of death from AIDS. The risk of death from other causes may be influenced by patient characteristics including age, sex, and race. Alcohol use is assumed to impact adherence with HAART, and subsequently to influence HAARTeffectiveness, and ultimately the risk of death from AIDS. Although it is not depicted in this diagram, alcohol also was assumed to directly impact the risk of death from other causes during the ‘J curve’ sensitivity analyses.

Figure 2. Premature treatment failure from alcohol consumption

Lightly shaded regions denote the effect that is attributable to alcohol consumption itself, and darkly shaded regions denote the effect that is attributable to confounding factors rather than to alcohol consumption. Increased nonadherence among drinkers leads to more rapid accumulation of genotypic resistance and regimen intolerance, with corresponding decrements in the time to failure of antiretroviral drugs. While a substantial proportion of the decrement experienced by hazardous drinkers arises from confounding factors rather than from the hazardous drinking itself, the decrement attributable to hazardous drinking is greater than the decrement attributable to nonhazardous drinking at each frequency.

Figure 3. Premature death from alcohol consumption

Lightly shaded regions denote the effect that is attributable to alcohol consumption itself, and darkly shaded regions denote the effect that is attributable to confounding factors rather than to alcohol consumption. Decreased time to treatment failure among drinkers leads to decrements in the survival time. While a substantial proportion of the decrement experienced by hazardous drinkers arises from confounding factors rather than from the hazardous drinking itself, the decrement attributable to hazardous drinking is greater than the decrement attributable to nonhazardous drinking at each frequency.