Drinking problems and mortality risk in the United States

Highlights

• Drinking-related problems are common even among light drinkers..

• Drinking-related problem behaviors increase risk of death, net of consumption level.

• Drinking status and drinking problems appear to have distinct influences on mortality.

• Researchers/policymakers should consider multiple dimensions of alcohol consumption.

Abstract

Objective

We examine the links between 41 problems related to alcohol consumption and the risk of death among adults in the United States.

Method

We use Cox proportional hazards models and data from the nationally representative prospective National Health Interview Survey-Linked Mortality Files (NHIS-LMF).

Results

Drinking problems are relatively common among moderate and heavy drinkers and these problems are associated with increases in the risk of death. The strongest associations between problem drinking and mortality involved cases in which physicians, family members, or friends intervened to suggest reduced drinking. Losing one's job because of drinking problems within their lifetime (HR = 1.36, 95% confidence interval [CI]: 1.11, 1.65) was strongly linked to mortality risk. Social risks were equally or more strongly linked to mortality than physiological consequences of alcohol abuse such as lifetime reports of needing a drink to stop shaking or getting sick (HR = 1.23, 95% CI: 1.09, 1.40). Most importantly, these associations were evident despite statistical controls for alcohol consumption levels and demographic, social, economic, behavioral, health, and geographic factors.

Conclusions

Our results highlight the independent and additive effects of alcohol-related problems and alcohol consumption levels on the risk of death. We recommend that studies examining the mortality risks of alcohol consumption take into account drinking status and also specific drinking-related problems, paying particular attention to social problems related to alcohol use or abuse.

Introduction

Although there is extensive literature on the association between alcohol use and mortality (Di Castelnuovo et al., 2006, Ronksley et al., 2011), few studies have examined the associations between specific alcohol-related problems and mortality. Dawson (2000a) provides some of the only evidence that alcohol dependence is linked to an increased risk of death above and beyond the level of alcohol consumption. This research is important because it shows that use and abuse, while certainly correlated with one another, denote independent pathways through which alcohol-related behaviors can influence mortality. To date, however, no existing research has examined each specific alcohol problem and corresponding DSM indicator as an independent mortality risk.

Dawson (2000a) examined the risk of death associated with alcohol dependence, classified according to the definitions given in the Diagnostic and Statistical Manual (DSM) at that time, the DSM-IV (APA, 1994), among adults aged 25 and older, using the 1988 National Health Interview Survey matched to the National Death Index (NDI) through 1995. She constructed her alcohol dependence measure from 17 of 41 alcohol-related indicators and showed that rates of alcohol dependence increased with increasing consumption; it was 9% among light drinkers, 24% among heavy drinkers, and 42% among very heavy drinkers. Compared to lifetime abstainers, dependent drinkers suffered increased risk of death. For example, moderate and dependent drinkers were 32% and very heavy and dependent drinkers were 65% more likely to die over the follow-up period. Compared to nondependent drinkers, dependent drinkers were more likely to have long and heavy drinking histories, drink greater volumes of alcohol, engage in heavy episodic drinking (HED), and have more health problems, including more major limitations, hospitalizations, and bed days (Dawson, 2000a). However, individuals can experience drinking problems without dependence. According to recent estimates, 90% of excessive drinkers are not alcohol dependent (Esser et al., 2014). Thus, calculating mortality risk for the complete list of 41 individual problems among all drinkers may capture a broader scope of drinking's health consequences.

Further, Dawson's informative results suggest that it may be useful to calculate mortality risk for the updated DSM-V, and over a longer follow-up period. The current DSM version lists 11 domains of DSM-V criteria for substance use disorder (SUD) that describe current or lifetime physiological, psychological, behavioral, and social consequences of drinking (American Psychiatric Association [APA], 2013b). While the DSM-IV defined abuse and dependence separately, the DSM-V defines one SUD with categorizations of mild, moderate, and severe. Meeting SUD criteria may indicate a cluster of drinking problems that differ from the individual problems. Health-damaging behaviors such as problem drinking earlier in life can increase risk of disease and death later in life (Graham, 2002, Kuh et al., 2003, Montez and Hayward, 2011) and many assume that the primary effects of alcohol consumption are physiological and damage specific organs. However, exposure to various risks early in life may initiate biological, psychological, and social chains of risk, which can be mediated by various factors, including socioeconomic status (Kuh et al., 2003). Problem drinking may reduce physiological or psychological resilience and increase susceptibility to death through a variety of causes, including heart disease, stroke, cancer, liver disease, chronic alcoholism, and external causes of death (suicide, homicide, motor vehicle accidents, and unintentional injury; Brady, 2006, Dawson, 2001, Dawson, 2011, Single et al., 1999, Skog, 2002).

In particular, social conflict related to drinking problems may be one of the main mechanisms linking excess alcohol consumption to increased mortality risk. Drinking can cause social problems at home, school, and work, and with social and sports activities (Ames et al., 1997), family, friends, and coworkers (Barnes and Farrell, 1992), and the legal system (Caetano, 1997). Social resources—including ties to family, friends, school, work, and the community—protect against the risk of death (Berkman and Syme, 1979, Seeman, 1996). Thus, social disruptions such as the end of a relationship, decreased social interactions, or the loss of support from colleagues because of alcohol-related problems may be mechanisms through which chronic alcohol use leads to elevated risk of death. Furthermore, drinking is linked to emotional problems, depression, and lack of personal control (Holahan et al., 2003, Nigg et al., 2006, Windle and Windle, 1996).

There are also beneficial physiological and psychological effects of light to moderate consumption. Light to moderate alcohol consumption can enhance mood, reduce stress, lower anxiety, and decrease the risk of depression, which can in turn contribute to a stronger sense of sociability, social integration, and cohesion (Peele and Brodsky, 2000). Light to moderate alcohol consumption can also reduce the risk of heart disease through increased levels of HDL cholesterol, apolipoprotein A1, and adiponectin, and reduced levels of fibrinogen (Agarwal, 2002, Brien et al., 2011). Because low and even moderate levels of alcohol have been linked to salutary health profiles in previous research, it is important to control for current and past drinking levels to examine the influence of specific alcohol problems above and beyond current drinking levels.

In identifying the relationships between drinking problems and mortality, this study seeks to determine whether drinking's health consequences depend on the social consequences of drinking (in addition to consumption patterns). This paper addresses four central aims. First, we describe the prevalence of specific drinking problems across drinking statuses and consumption levels among a representative sample of U.S. adults. Second, to examine the association between specific drinking problems and mortality, we are the first to use information on 41 specific alcohol-related problems from adults in 1988 and their subsequent mortality nearly 20 years later. Third, we examine the association between clusters of drinking problems, as defined by the DSM-V criteria, and mortality. Finally, we determine the mediating effects of SUD on the association between drinking status and mortality.