- Top of page
- Conflict of interest
Aims To investigate the prevalence and correlates of recovery from Diagnostic and Statistical Manual version IV (DSM-IV) alcohol dependence by examining the past-year status of individuals who met the criteria for prior-to-past-year (PPY) dependence.
Design Cross-sectional, retrospective survey of a nationally representative sample of US adults 18 years of age and over (first wave of a planned longitudinal survey).
Methods This analysis is based on data from the 2001–02 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), in which data were collected in personal interviews conducted with one randomly selected adult in each sample household. A subset of the NESARC sample (total n = 43 093), consisting of 4422 US adults 18 years of age and over classified with PPY DSM-IV alcohol dependence, were evaluated with respect to their past-year recovery status: past-year dependence, partial remission, full remission, asymptomatic risk drinking, abstinent recovery (AR) and non-abstinent recovery (NR). Correlates of past-year status were examined in bivariate analyses and using multivariate logistic regression models.
Findings Of people classified with PPY alcohol dependence, 25.0% were still classified as dependent in the past year; 27.3% were classified as being in partial remission; 11.8% were asymptomatic risk drinkers who demonstrated a pattern of drinking that put them at risk of relapse; 17.7% were low-risk drinkers; and 18.2% were abstainers. Only 25.5% of people with PPY dependence ever received treatment. Being married was associated positively with the odds of both AR and NR, and ethanol intake was negatively associated with both. Severity of dependence increased the odds of AR but decreased the odds of NR. The odds of AR (but not NR) increased with age and female gender but were decreased by the presence of a personality disorder. Treatment history modified the effects of college attendance/graduation, age at onset and interval since onset on the odds of recovery.
Conclusions There is a substantial level of recovery from alcohol dependence. Information on factors associated with recovery may be useful in targeting appropriate treatment modalities.
- Top of page
- Conflict of interest
These data from a nationally representative sample of US adults revealed substantial levels of recovery from DSM-IV alcohol dependence. Confirming previous studies that have reported similar findings, they provide evidence that alcohol dependence—at least when defined in terms of the DSM-IV criteria—may not preclude a return to low-risk drinking for some individuals. Typically, these might consist of people with less severe disorders who mature out of their drinking problems without treatment (Cunningham et al. 2000). The variation in past-year status over time suggests that a typical course of recovery might consist of continued drinking, accompanied by symptoms of alcohol use disorders, that would persist for 5–10 years before resolving into asymptomatic risk drinking and, ultimately, into either low-risk drinking or abstinence. However, such an extrapolation of the data would be risky for several reasons. First, it does not account for selective survival. Chronic alcoholics may be more likely to die than those who recover (Dawson 2000), inflating estimates of recovery in the later intervals since onset of dependence as the deceased become increasingly under-represented in the denominators of the recovery rates. Nor does such an extrapolation reflect the periodic relapses or shifts between AR and NR that have been observed in longitudinal studies (Skog & Duckert 1993; Vaillant 1995). Cross-sectional data do not necessarily reflect the course of recovery across time for any given individual. Subsequent waves of the NESARC will provide an opportunity to examine the natural history of alcohol dependence over time at the individual level in a large national sample.
The data reported in this paper show some interesting differences relative to earlier estimates of recovery based on the 1991–92 National Longitudinal Alcohol Epidemiologic Survey (NLAES). Using measures that were almost identical to those used in the NESARC, the NLAES findings indicated that 27.8% of people with PPY alcohol dependence were classified with either dependence or abuse in the past year—considerably lower than the estimate of 35.5% found in this study. The discrepancy was greatest at intervals of less than 5 years since onset of dependence (74.0% in the NESARC versus 57.1% in the NLAES). Among those with intervals of 20 or more years, the estimates were quite comparable, 14.7% in the NESARC versus12.4% in the NLAES (Dawson 1996).
The greater prevalence of past-year dependence or abuse in the NESARC reflects, in part, an increase in the prevalence of alcohol abuse from 3.03% in the NLAES to 4.65% in the NESARC (Grant et al. 2004a). However, the magnitude of this increase is too small for it to be a major explanatory factor. Rather, these findings indicate a trend towards less rapid remission of dependence over the past decade. There are no obvious explanations for why this might be the case. The change in age at onset of dependence between the two surveys (median age = 20 in 1991–92 and 21 in 2001–02) was not sufficient to explain the change. During both time periods, the first 5 years after onset of dependence typically encompassed the college and young adult ages, and college drinking patterns have remained fairly stable over the past decade (Wechsler et al. 1999, 2002). It has been argued that remission of early onset alcohol dependence often involves a spontaneous ‘maturing out’ of alcohol problems in association with taking on adult responsibilities such as full-time work, marriage and parenthood (Jessor et al. 1991). Perhaps changes in the economic and social climate have slowed this process, thereby indirectly slowing the rate of remission from alcohol problems. Again, data from Wave 2 of the NESARC should provide valuable information to address this issue.
Other than the discrepancy discussed above, the distribution of past-year status found in the current study was similar to that reported in the earlier analysis of the NLAES data, insofar as the data permit comparison. What cannot be determined from the earlier published data is whether there was any change over the ensuing decade in the ratio of low-risk drinkers to abstainers. However, this study‘s finding that low-risk drinking accounted for roughly half of all cases of full recovery is in line with the findings of the two Canadian general population studies in which it accounted for 38% and 63%, respectively, of all recovery (Sobell et al. 1996).
The factors associated with recovery in this study were generally similar to those observed in earlier studies, e.g. in the increased odds of recovery were increased among married individuals. As was the case in the analysis of NLAES data (Dawson 1996), severity was associated positively with the likelihood of AR and associated negatively with the likelihood of NR. (A significant interaction between severity and treatment history found in the NLAES analysis fell just short of significance in this study.) As in the NLAES, college education decreased the likelihood of abstinence, but only in the absence of alcohol treatment. That the results of the two analyses were so similar, despite the fact that each controlled for a somewhat different set of covariates, provides evidence of the robustness of these associations.
At the same time, this study yielded some interesting additional findings, for example, the roles of life-time tobacco and drug use in discriminating between types of recovery. Each of these defies obvious interpretation. Perhaps life-time smokers, many of whom were former smokers by the time of interview, were more inclined towards AR because smoking cessation required a similar all-or-nothing approach. Life-time non-dependent drug users may have tended towards NR because they were apparently able to use drugs without developing drug dependence and may have felt they could achieve non-dependent use of alcohol as well. This study‘s finding that individuals with a personality disorder had a reduced likelihood of achieving AR supports findings in clinical samples on the adverse effects of antisocial PD (Pettinati et al. 1999; McKay & Weiss 2001; Ciraulo et al. 2003). Recent research has shown that obsessive–compulsive, paranoid and antisocial PD are the most common personality disorders in the general US population and among people with alcohol dependence (Grant et al. 2004c). However, dependent, histrionic and antisocial PD are the most strongly associated with the odds of alcohol dependence (Grant et al. 2004d). Additional research to identify specific personality disorders that are implicated as impediments to AR should be helpful in tailoring treatment programs to the needs of alcohol-dependent individuals who have these disorders.
Several limitations of this study should be considered in the interpretation of its findings. First, age at onset and remission of dependence may have occurred many years prior to interview and might not be remembered accurately. Although errors in recalling these ages would not affect overall estimates of recovery, they could affect estimates within specific intervals since onset. Secondly, the classification of PPY dependence is dependent upon recall of whether multiple symptoms of dependence occurred at the same time. Errors in recall of co-occurrence that resulted in inaccurate estimates of PPY dependence (e.g. by including cases of borderline severity) might bias estimates of recovery. Finally, the rates of recovery presented in this study are higher than they would be had individuals with life-time rather than PPY dependence been examined (the proportion still dependent in the past year would have been 30.5% rather than 25.0%, data not shown). As discussed previously, this is because individuals with onset of dependence in the past year would by definition still be considered dependent in that period, thus lowering the proportions of individuals in the categories of remission and recovery.