Testing the Validity of the Pathways Model: A Latent Class Analysis of Potential Pathological Gambling Subtypes in a Non-Treatment Sample.

Latent class analysis (LCA) was used to test the validity of the Pathways Model in 285 subjects with DSM-IV pathological gambling (PG). In addition to identifying three subtypes that roughly correspond with those described in the model (Behaviorally Conditioned, or BC, Emotionally Vulnerable, or EV, Antisocial-Impulsivist, or AI), LCA identified a fourth class, termed the Antisocial Drinker, or AD, characterized by high rates of antisociality, conduct disorder, and alcohol use disorder. BC gamblers comprised 45% of the sample, followed by EV (24%), AD (22%), and AI (9%) gamblers. Women were more likely to be EV gamblers (OR = 1.89) and less likely to be AD gamblers (OR = 0.46). Those who had attempted suicide were more likely to be EV (OR = 3.06) or AI (OR = 3.05) gamblers and less likely to be BC (OR = 0.37) or AD gamblers (OR = 0.50). Greater childhood maltreatment was associated with AD (standardized OR = 1.81) and AI (standardized OR = 1.43) gamblers. Individuals with later PG onset were less likely to be AI gamblers (standardized OR = 0.48). Individuals who preferred slots were more likely to be EV gamblers (OR = 1.83) and less likely to be AD gamblers (OR = 0.33). The BC subtype was associated with better health outcomes, better social functioning, less childhood maltreatment, and less severe PG. The AI subtype was associated with worse health outcomes, worse social functioning, and higher PG severity. The findings provide a better understanding PG heterogeneity that could be relevant to clinical management.