Twenty-year alcohol-consumption and drinking-problem trajectories of older men and women

Abstract

Objective: The aim of this study was to describe older adults' 20-year alcohol-consumption and drinking-problem trajectories, identify baseline predictors of them, and determine whether older men and women differ on late-life drinking trajectory characteristics and predictors.

Method: Two-group simultaneous latent growth modeling was used to describe the characteristics and baseline predictors of older community-residing men's (n = 399) and women's (n = 320) 20-year drinking trajectories. Chi-square difference tests of increment in fit of latent growth models with and without gender invariance constraints were used to determine gender differences in drinking trajectory characteristics and predictors.

Results: Unconditional quadratic growth models best described older individuals' within-individual, 20-year drinking trajectories, with alcohol consumption following an average pattern of delayed decline, and drinking problems an average pattern of decline followed by leveling off. On average, older men declined in alcohol consumption somewhat later than did older women. The best baseline predictors of more rapid decline in alcohol consumption and drinking problems were drinking variables indicative of heavier, more problematic alcohol use at late middle age.

Conclusions: The course of alcohol consumption and drinking problems from late middle age onward is one of net decline, but this decline is neither swift nor invariable. Gender differences in the timing of decline in drinking suggest that ongoing monitoring of alcohol consumption may be especially important for older men. Further research is needed to identify factors known at late middle age that prospectively explain long-term change in late-life use of alcohol.

Figure 1

Two-group simultaneous latent growth model: 20-year drinking trajectories of older men and women. DB0, DB1, DB4, DB10, DB20 = measured drinking behavior (alcohol consumption, drinking problems) variables assessed at baseline and 1,4, 10, and 20 years later; I = estimated average initial level of drinking behavior; S = estimated average linear growth rate; Q = estimated quadratic growth rate; X = baseline variables hypothesized to influence I, S, and Q growth parameters.

Figure 2

Twenty-year alcohol-consumption trajectories of older men and women

Figure 3

Twenty-year drinking-problem trajectories of older men and women