Commentary on Tuithofet al. (2013): Implications of the DSM-5 revision for the analysis of persistence/remission of alcohol use disorder

Abstract

The findings of Tuithof et al. [1] are consistent with studies of the course of DSM-IV [2] alcohol use disorder (AUD) in the general population, demonstrating that many episodes of AUD are of limited duration and many individuals who have experienced remission of AUD symptoms drink at levels that may increase the risk of relapse [3–7]. A prospective study based on just two data points is of limited utility for yielding valid inferences about a dynamic process such as persistence/remission of AUD, in that at least three data points are necessary to chart the chronic relapsing pattern that characterizes a substantial proportion of individuals with lifetime AUD in long-term prospective studies [8–11]. In fact, retrospective data, despite limitations related to recall error and selective survival, may actually tell us more about the recovery process than prospective data with single follow-up [3]. Despite the inherent limitations of its study design, this study nonetheless raises some interesting questions about how the persistence of AUD is related to remission, how has this changed under the DSM-5 and what new research opportunities may arise out of these changes. Under the DSM-IV, individuals who meet the criteria for alcohol dependence at t1 but do not satisfy the requirements for persistent dependence or full remission (cessation of all AUD symptoms) at t2 fall into the category of partial remission, comprising alcohol abuse and/or subclinical dependence symptoms. Individuals who do not satisfy the requirements for persistent abuse include both those who have progressed to dependence and those in full remission. Thus, the absence of persistence does not necessarily signify full remission. Retrospective studies of US population samples have shown that partial remission of dependence is common, peaking in prevalence at ≈40% 5–9 years after onset of dependence [4]. Individuals in partial remission drink more and have a lower quality of life than those in full remission, but drink less and have a higher quality of life than those with persistent dependence [12,13]. Because individuals in partial remission are distinct from those with persistent or fully remitted AUD, a clearer picture of the significance of factors associated with the course of DSM-IV AUD could be obtained by excluding partial remission cases from analyses or including them as a separate outcome category in multinomial models. The proposed DSM-5 revision [14] does not recognize partial remission as a course specifier for AUD [15]. Among individuals with AUD (2+ positive criteria) at t1, those with subclinical levels of AUD symptoms at t2 count as fully remitted, along with those who are asymptomatic. In addition, the newly added criterion of craving, which may persist even after discontinuation of drinking, does not preclude achievement of remission. Thus, full remission and persistent AUD are more complementary under the DSM-5 than they were under the DSM-IV; i.e., there are no longer any cases that fall between these categories. However, regardless of whether partial remission is formally recognized or subsumed under full remission, the residual presence of cases with a subclinical level of symptoms reduces the contrast between persistent and asymptomatic cases. Thus, challenges remain with dichotomous outcome measures when modeling correlates of remission or persistence. However, the DSM-5 definition of AUD as a unitary latent construct whose symptoms vary along a continuum of severity suggests new approaches to the study of persistence/remission that may overcome some of the limitations of a dichotomous outcome measure. One such approach is to examine correlates of increase/decrease in AUD severity, and to see whether a change of a given magnitude has different correlates depending on the base level of severity. That is, does a drop from 8 to 5 positive criteria between t1 and t2 have the same correlates as a drop from 4 to 1? Similarly, does the impact of a given decrease in severity on changes in consumption and quality of live vary based on initial severity level? Results of such an approach would at least be suggestive of factors that differentially affect the early versus later stages of recovery, and they would clearly distinguish factors associated with remission of AUD of varying severity levels. The addition of craving as a criterion for DSM-5 AUD also suggests new opportunities for analyses. Unlike most AUD criteria, which reflect or result from but do not cause heavy drinking, craving can reinforce excessive consumption, which may in turn lead to the persistence of other AUD symptoms. By studying the unique contribution of craving to the course of AUD, i.e., its impact net of the number of other AUD criteria, we may be able to draw valuable inferences related to the benefits of medical treatment for the alleviation of craving. In summary, the proposed DSM-5 criteria for AUD should facilitate multiple new analytic approaches that will enhance our understanding of factors associated with the course of AUD and implications for treatment.