Project 1: Effects of Alcohol Use

The effects of alcohol use on Alzheimer’s Disease and Alzheimer’s Disease Related Disorders (AD/ADRD) remain uncertain. Long hypothesized to have a J-shaped relationship, with the best cognitive outcomes among light drinkers, growing evidence suggests that benefits of even ”low-risk” drinking are unlikely. Yet, there is not clear quantification of how low-risk or moderate alcohol consumption (considered unhealthy but not constituting alcohol use disorder) affects AD/ADRD risk. Drinking is extremely common and modifiable via clinical and policy interventions. Estimates of the adverse effects of heavy drinking on AD/ADRD also vary widely. The recent Lancet Review estimation of the population attributable fraction (PAF) for AD/ADRD associated with alcohol use was based on a small number of studies that may not be generalizable. The population impact of alcohol use may be substantially underestimated. Individual studies of alcohol use are potentially biased due to confounding, reverse causation, or measurement error. We propose coordinated analyses across ten diverse clinical and population cohorts. Working with the TIME-AD Cores, we will triangulate evidence from doubly robust g-methods, genetic instrumental variables (IVs), and policy IV analyses to derive the best possible estimates on the effects of alcohol use on AD/ADRD risk. Kaiser Permanente Northern California electronic health record (EHR) data for over a million older adults are augmented with embedded surveys and genotyping. We additionally use data from the UK Biobank, All of US, and multiple cohorts with detailed, repeated assessments of alcohol use, confounders, and cognition. We adopt a systematic approach to interrogating potential biases by comparing patterns across populations, study designs, and identification approaches. Uncertainty in estimates will be characterized using quantitative bias analysis. We propose four aims: Aim 1. Assess how low-risk alcohol use, unhealthy alcohol use, and alcohol use disorder, compared to no alcohol use, affect AD/ADRD risk, triangulating across study designs and data sources. Aim 2. Evaluate heterogeneity in the effects of different quantities and patterns of drinking on AD/ADRD risk, evaluating type (e.g., beer, red wine, white wine, spirits), frequency-quantity pattern (e.g., drinks per month, binge drinking episodes), problem-drinking (e.g., ”blackout drinking”), and duration (e.g., years of consumption). Aim 3. Evaluate how the effects of alcohol use on AD/ADRD risk vary in relation to other risk factors for AD/ADRD risk, i.e., heterogeneity across characteristics of the person using alcohol, such as genetic risk, cardiometabolic comorbidities, gender, race/ethnicity, education, social isolation, cannabis use, or medication use. Aim 4: Evaluate the extent to which disparities in dementia incidence across social strata are attributable to differences in alcohol consumption or could be ameliorated by changes in drinking. Using PAFs or flexible transport estimators, we will estimate reductions in AD/ADRD and racial/ethnic, gender, or socioeconomic disparities in AD/ADRD that could be achieved through changes in alcohol use.