Convergent validity of DSM-III-R and DSM-IV alcohol dependence: results from the National Longitudinal Alcohol Epidemiologic Survey.

The purpose of this study was to separately examine the associations between ethanol intake and alcohol dependence, as defined in the DSM-III-R and DSM-IV, in conjunction with an identical set of external correlates, with a view toward assessing the convergent validity of the two diagnostic definitions of dependence. Although the sociodemographic, alcohol, drug and comorbid profiles of respondents classified as DSM-III-R and DSM-IV dependent were similar, the results of the linear logistic regression analyses differed for each of the diagnostic definitions. The risk of DSM-III-R dependence was reduced among blacks and increased among early onset drinkers. Given equivalent levels of ethanol intake, the risk of DSM-III-R dependence was greater for these respondents who had experienced a recent death of close relative compared to those who had not. In contrast, the risk of DSM-IV dependence was greatest among respondents with a current alcoholic spouse or partner relative to those without an alcoholic spouse or partner. Major differences in the content and structure of the two definitions of dependence were useful in predicting the observed discrepancies between the risk gradients associated with the DSM-III-R and DSM-IV classifications. Implications of these findings were discussed in terms of differential coverage of the DSM-III-R and DSM-IV diagnostic categories and the impact of the absence of complete convergent validity on epidemiologic research.