Substance Use
Lifecourse frequency of heavy episodic drinking from adolescence to late midlife in NLSY79: a sequence and cluster analysis Jillian Hebert* Jillian Hebert Hebert Hebert Hebert Hebert Hebert Hebert Hebert Hebert Hebert Hebert Hebert School of Public Health, Washington University in St. Louis
Emerging research suggests that even small amounts of drinking affect health but this is complicated by reverse causation (“sick quitter” effects) and conflation of lifetime abstainers with former problem drinkers. Few data sources include longitudinal information on participants’ use across the lifespan. Evidence regarding the health effects of heavy episodic drinking (HED) has been hampered by the complexity of characterizing lifecourse alcohol use. Sequence analysis groups thousands of individual trajectories into similar clusters, retaining substantive differences between trajectories. We used this method to capture lifecourse alcohol use.
We used 1979-2018 data from the National Longitudinal Surveys 1979 study (NLSY79), which recruited U.S. participants 14-22 years old in 1979 (N=2,732 male; N=2,871 female). NLSY79 assessed alcohol use beginning in 1982, when participants were 18+ years old. We characterized lifecourse alcohol use trajectories into three mutually exclusive HED states based on frequency in the past 30 days ([6+ drinks on one occasion]: No HED, Occasional HED [1-3 times/mo], Frequent HED [4+ times/mo]) – from ages 18-61. We used the Hamming method to calculate dissimilarity between all individuals’ trajectories and agglomerative hierarchical clustering with a Ward linkage to cluster trajectories. All analyses were sex stratified.
Overall, from ages 18-61, males were more likely than females to engage in occasional (15% vs 8%) or frequent (11% vs 3%) HED. We found evidence for 5 clusters for males and 7 clusters for females, characterized by differences in timing of HED (Fig 1). While several clusters had similar patterns of HED after age 50 (Fig 1, females 1-2; 3-4), these clusters differed in early-life drinking patterns.
Confining analyses to current alcohol use may miss important lifecourse risk factors for health in older adults. Measuring lifecourse trajectories of alcohol use using sequence analysis can inform problems with reverse causation.
