Probabilities and predictors of long-term continuing abstinence for heroin and methamphetamine users

Abstract

We examined the conditional probability of continuing abstinence given increasing durations of preceding abstinence and predictors of these probabilities among primary heroin and methamphetamine users sampled in two separate studies. Duration of abstinence from the individual's primary drug was measured for the first period of abstinence of at least 1 year. Receiver operating characteristic curves were used to examine probability of achieving a subsequent year of abstinence given the preceding years of continuous abstinence from designated drug. Logistic regression was used to examine predictors of each additional year of abstinence. Probabilities of continued abstinence were similar for both samples increasing steeply to .84–.85 through 3 years of abstinence, then tapering off and rising to .94 (heroin) and .91 (meth) by year 6. Attaining at least 3 years of continuous abstinence substantially decreases the likelihood of relapse for both heroin and methamphetamine users, although there were some variations in specific predictors across groups.