On the Limits of Cross-Sectional Retrospective Data for Characterizing the Course of Pathological Gambling and its Relation With Comorbid Psychopathology: A Reply to Afifi, Cox, and Sareen

To the Editor: Dr. Afifi and colleagues refer to an analysis of the National Epidemiologic Survey on Alcohol and Related Conditions data in which treatment specifically for alcohol use disorder or to improve mood disorder was included in the definition of treatment-seeking for pathological gambling. In doing this, it is important to realize that one is also studying a different subgroup of individuals—those with lifetime pathological gambling who also have a lifetime history of alcohol use disorder or major depression. Identifying subgroups of individuals (e.g., those with comorbid psychopathology) who are more or less likely to seek treatment and recover from pathological gambling was not fully explored in my article but certainly warrants further investigation. In my article, I noted that there was a strong association in the National Epidemiologic Survey on Alcohol and Related Conditions between the number of lifetime pathological gambling symptoms endorsed and the probability of seeking treatment for gambling problems, with treatment rates ranging from 5% to 76% for those endorsing five to 10 pathological gambling symptoms. Individuals with pathological gambling in the National Epidemiologic Survey on Alcohol and Related Conditions who also had a lifetime history of major depression were also 1.6 times more likely to seek treatment specifically for problems with gambling (12%) compared with those without a history of major depression (8%). Unfortunately, there is no way to identify contemporaneous disorders in the National Epidemiologic Survey on Alcohol and Related Conditions (other than disorders that co-occurred in the past year that are uninformative regarding recovery). Thus, it is not possible to determine when to include treatment specifically for alcohol problems or major depression as relevant for an episode of pathological gambling because these may have occurred at different points in the participant’s lifetime.

Afifi and colleagues conclude, on the basis of my finding that approximately 40% of individuals with a lifetime history of pathological gambling no longer had any pathological gambling symptoms in the past year (1) , that the remainder and majority (60%) must have had a chronic, persistent course. Although comparing past-year to lifetime (or to “prior to past year clustered” [ 1 , p. 299]) diagnoses allows one to rule-out a chronic/persistent course, it does not allow one to rule-in such a course. Those individuals who did not meet the criteria for “recovery” in my study could have had a variety of courses of gambling problems, including but not limited to a chronic/persistent course. In the article, I also present results of supplementary analyses from the National Epidemiologic Survey on Alcohol and Related Conditions that 62% of participants with “prior to past year clustered” pathological gambling reported experiencing only one episode of pathological gambling in their lifetime, that the mean number of episodes was 2.8, that the longest mean episode duration was 2.4 years, and that the most common course of pathological gambling was a single episode lasting 1 year or less. Thus, the conclusion stated in the title of the letter by Afifi and colleagues is incorrect.

I wholeheartedly agree with Afifi and colleagues that “firm conclusions regarding chronicity and persistence will ultimately require the use of longitudinal prospective data.” I am pleased to have this opportunity to highlight some of the limits of cross-sectional retrospective data for characterizing the course of pathological gambling and its relation to comorbid psychopathology and to clear up any possible misunderstanding of my study.