DSM-IV alcohol dependence and abuse: Further evidence of validity in the general population

Abstract

Background

In order to understand the validity of the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV) alcohol abuse and dependence diagnoses, studies are needed in both clinical and general population samples. The purpose of this study was to examine the construct and criterion-oriented validity of DSM-IV alcohol dependence and abuse in the general population with respect to factor structure and their relationship to family history of alcoholism, treatment utilization, and psychiatric comorbidity.

Methods

This analysis is based on data from the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), in which nationally representative data were collected in personal interviews conducted with one randomly selected adult in each sample household or group quarters. A subset (n = 26,946) of the NESARC sample (total n = 43,093) who reported drinking one or more drinks during the year preceding the interview formed the basis of analyses. Latent variable modeling was used to assess the concurrent validity of DSM-IV alcohol abuse and dependence symptom items.

Results

The latent variable modeling yielded one major factor related to alcohol dependence, a second factor related to alcohol abuse and a third smaller factor defined by tolerance. The validity of alcohol dependence in general population samples was further supported by statistically significant associations with family history of alcoholism, treatment utilization, and psychiatric and medical comorbidities.

Conclusions

The factor structure and relationship to external criterion variables observed in the study provide support for the further validity of DSM-IV alcohol dependence in the general population, whereas support for the validity of DSM-IV abuse was equivocal.

Introduction

The DSM-IV (American Psychiatric Association, 1994) criteria for alcohol abuse and dependence serve as the basis for diagnosis in clinical work, prevalence estimation in epidemiology, and on much of the etiologic research conducted in the United States. Thus, understanding the validity of the DSM-IV criteria for alcohol abuse and dependence and their relationship to one another is highly important. The origin of the criteria in the DSM was the alcohol dependence syndrome (ADS) of Edwards and Gross (Edwards and Gross, 1976), later generalized to all drug disorders as classified in the World Health Organization's International Classification of Diseases (ICD) (Room, 1998). The essential elements of ADS continue to be reflected in contemporary definitions of alcohol abuse and dependence as classified in the DSM-IV and ICD-10.

The ADS was conceptualized as a combination of physiological and psychological processes reflecting impaired control over use, constituting what is now recognized as a “complex” disorder. Consistent with this, DSM-IV dependence criteria reflect both physiological and psychological symptoms, a syndrome that is now known to be highly heritable (Prescott and Kendler, 1999a, Prescott et al., 1997). The dependence syndrome was considered one “axis” of problems, differentiated from another axis consisting of problems resulting from substance use such as substance-related injuries, social, or legal problems (Edwards and Gross, 1976). This second axis corresponds to DSM-IV substance abuse. Importantly, the two axes in the “bi-axial” dependence syndrome distinction were not considered “orthogonal”. Instead, they were defined as different types of substance-related problems expected to co-occur in some but not all cases. Since alcohol dependence is described as increasing loss of control over drinking despite the occurrence of negative consequences, while abuse consists of those consequences, the two concepts differ theoretically in their etiology but are likely to be correlated in practice. The fundamental difference between the DSM-IV and its underlying ADS conceptualization is that the DSM-IV is categorical whereas the ADS is definitively dimensional in nature. More validation research is needed in order to understand the underlying dimensional construct underlying both the DSM-IV and ICD-10 categorizations of alcohol use disorders.

However, few such studies have been conducted. Partial support for construct validity is found in studies that explore the factor structure of criteria and the individual symptom items included within the criteria. In one such study, Muthén et al., 1993a, Muthén et al., 1993b, Muthén et al., 1993c examined the dimensionality of the relevant criteria in both the DSM-IV and its precursor, the Diagnostic and Statistical Manual of Mental Disorders, 3rd ed. Revised (DMS-III-R: American Psychiatric Association, 1987) across three population subgroups defined by progressively greater risk for alcohol dependence: current drinkers, heavy drinkers, and high-risk drinkers. Data were derived from the National Institute on Alcohol Abuse and Alcoholism's (NIAAA) Alcohol Supplement to the 1988 National Health Interview Survey (NHIS: Grant et al., 1991) and individuals were assigned to subgroups based on background characteristics predictive of alcohol problems (e.g., family history of alcoholism). Analysis yielded a two-factor model that applied consistently to all three subgroups and performed identically in each validation sample. Similar analysis at the symptom level (Muthén, 1995) produced a two-factor solution that best described the item correlations and bore a strong resemblance to the criterion-level solution reported by Muthén et al. (1993a). One factor was represented by “persistent efforts to cut down or control drinking,” “giving up important activities in favor of drinking,” and “continuing to drink despite physical, psychological or social problems,” “neglect of roles in favor of drinking” together with symptom items related to “tolerance” and “withdrawal”; this factor may be interpreted as corresponding to alcohol dependence. The second factor was represented by two criteria: “drinking larger amounts over a longer period of time” and “recurrent drinking in situations where alcohol use is physically hazardous”; this factor appeared to correspond to alcohol abuse. These studies provide some support for the validity of DSM-IV alcohol dependence criteria in general populations, although the distribution of criteria between the two factors did not exactly match that defined in DSM-IV. For example, the factor suggestive of dependence included items related to the DSM-IV criteria for alcohol abuse (e.g., continued drinking despite social problems and recurrent drinking resulting in failure to fulfill role obligations). Similarly, the factor suggestive of abuse included some DSM-IV criteria for dependence (i.e., “drinking larger amounts or over longer periods of time”).

Another approach for establishing the validity of DSM-IV alcohol dependence criteria, other than construct validation, in the general population involves criterion-oriented validation that relates the category of dependence to a set of external criterion correlates (i.e., variables that potentially indicate an alcohol use disorder but which are not included in the diagnostic criteria themselves). Hasin et al., 1997a, Hasin et al., 1997b identified heavy drinkers from among a general population sample and divided them into three groups: those with a DSM-IV diagnosis of alcohol dependence, those with a DSM-IV diagnosis of alcohol abuse, and those with no current alcohol use disorder diagnosis. A DSM-IV diagnosis of alcohol dependence, when compared with no diagnosis, was significantly associated with external criterion variables such as alcohol consumption, treatment seeking, family history of alcoholism, and suicidal ideation. DSM-IV alcohol abuse, compared with the no diagnosis group, was not associated with these variables. Drawing upon the 1992 NIAAA National Longitudinal Alcohol Epidemiologic Survey (NLAES: Grant et al., 1994), Hasin and Paykin (1999a) reported that a DSM-IV diagnosis of alcohol dependence, when compared with a diagnosis of alcohol abuse and no positive diagnosis, was significantly associated with each of the following external criteria: average daily ethanol consumption, suicidal ideation/attempts, treatment/help-seeking, blackouts, and family history of alcoholism. Comparisons between a diagnosis of alcohol abuse and no positive diagnosis yielded significantly greater odds of alcohol abuse for average consumption and blackouts, a marginal relationship for treatment/help-seeking, and no significant differences for suicidal ideation and family history of alcoholism.

Given that the two-factor solutions depart somewhat from DSM-IV conceptualizations of alcohol abuse and dependence, further studies of the construct validity of alcohol abuse and dependence in the general population are warranted. Harford and Muthén (2001) adopted a latent variable modeling approach in a secondary analysis of the National Longitudinal Survey of Youth (NLSY: Frankel et al., 1983), a representative sample of young adults at an age when the prevalence of alcohol use disorders is highest. Because alcohol use disorders cannot be observed directly with observations containing measurement error, the “true” value of these disorders is said to be latent, or hidden (see Muthén, 1992). In structural equation modeling, one or more continuous latent variables (e.g., alcohol dependence and abuse) are estimated from a number of indicators (i.e., symptom items) and regressed on other background or external validating variables. To avoid misclassification errors and the problem of choosing cut-points, researchers use a dimensional representation in which increasing factor values correspond to increased risk when compared with a dichotomous diagnosis. Findings from this study replicated earlier studies of the factor structure of alcohol symptoms. Background covariates, which were similarly related in the samples of current and heavier drinkers, provide further support for the criterion-oriented validity of DSM-IV alcohol dependence in general population studies. The findings indicated that alcohol dependence was significantly associated with positive family history of alcoholism, early age at onset of alcohol use, and patterns of current alcohol consumption — outcomes consistent with criterion outcomes used in other general population studies (Grant and Dawson, 1997, Hasin et al., 1997a, Hasin et al., 1997b, Hasin and Paykin, 1999a, Muthén, 1995, Prescott and Kendler, 1999a, Prescott and Kendler, 1999b). Demographic variables were also differentially related to alcohol dependence and abuse. Alcohol dependence was not related to age, and alcohol-dependent individuals were more likely to be black or Hispanic and to be school dropouts. Alcohol abusers were younger, more likely to be white, and less likely to be school dropouts.

This study is based on data from NIAAA's National Epidemiologic Survey on Alcohol and Related Conditions (NESARC: Grant et al., 2004a, Grant et al., 2004b, Grant et al., 2004c, Grant et al., 2004d). The major goal of this study is to examine the construct and criterion-oriented validity of DSM-IV alcohol abuse and dependence. The main objectives of the present study were to examine the factor structure of symptom items used to configure DSM-IV diagnoses of alcohol use disorders, to assess associations between symptom items and demographic covariates, and to examine associations between the obtained factors (i.e., latent variables) and the covariates. It extends previous work by Hasin and Paykin (1999a) by including a structural equation-modeling framework and by expanding the set of criterion variables to include type of treatment, psychiatric and medical conditions, in addition to demographic variables, parental history of alcoholism, and family history of antisocial personality disorder among biological fathers.