Beyond the ubiquitous relapse curve: a data-informed approach

Relapse to alcohol and other substances has generally been described by curves that resemble one another. However, these curves have been generated from the time to first use after a period of abstinence without regard to the movement of individuals into and out of drug use. Instead of measuring continuous abstinence, we considered post-treatment functioning as a more complicated phenomenon, describing how people move in and out of drinking states on a monthly basis over the course of a year. When we looked at time to first drink we observed the ubiquitous relapse curve. When we classified clients (N = 550) according to drinking state however, they frequently moved from one state to another with both abstinent and very heavy drinking states as being rather stable, and light or moderate drinking and heavy drinking being unstable. We found that clients with a family history of alcoholism were less likely to experience these unstable states. When we examined the distribution of cases crossed by the number of times clients switched states we found that a power function explained 83% of that relationship. Some of the remainder of the variance seems to be explained by the stable states of very heavy drinking and abstinence acting as attractors.