Transitioning from First Drug Use to Dependence Onset: Illustration of a Multiparametric Approach for Comparative Epidemiology.

Studying transitions from first drug use (DU) to drug dependence (DD) onset, we estimate a parsimonious set of parameters based on epidemiological data, with plans for future longitudinal research on newly incident drug users and with tracking of self-administration frequencies and DD clinical features. Our expectation is a distinctive sigmoid pattern with one asymptote for lower DD probability (when DU is insubstantial), upturning slopes of DD risk beyond a middle value (PD50), and eventual higher DD risk asymptotes at higher DU frequencies. We illustrate this novel approach using cross-sectional data from the United States National Surveys on Drug Use and Health, 2002-2011. Empirical DD probabilities observed soon after newly incident use are estimated across DU frequency values, using parametric Hill functions and four governing parameters for differential comparison across drugs and DU subgroups. Among drug subtypes considered, cocaine shows larger estimates, especially among females (estimated P(min)=7% for females vs 3% for males; p<0.001), for whom PD50 is shorter by 8 days of use (p=0.027), conditional on the same rate of use in the past 30 days. Clear alcohol male-female differences also are observed (eg, female PD50 < male PD50; p=0.002). Although based on cross-sectional snapshots soon after DU onset, this novel multiparametric statistical approach for comparative epidemiological DD research creates new opportunities in planned studies with prospectively gathered longitudinal data. The cross-sectional estimates provide starting values needed to plan future longitudinal research programs on transitions from initial DU until formation of a DD syndrome.