College of Science and Health Theses and Dissertations

Date of Award

Spring 6-13-2014

Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Leonard Jason, Ph.D.

Second Advisor

Joseph Ferrari, Ph.D.

Third Advisor

La Vome Robinson, Ph.D.


Alcohol and other drug (AOD) abuse, treatment, and subsequent relapse and re-treatment are a common pattern faced by clients and AOD treatment providers. Relapse and re-treatment significantly contributes to the overall societal costs of substance use disorder (SUD) treatment, injury, and incarceration. Therefore, it might be useful to treatment providers to become aware of of the factors leading to AOD relapse after treatment. This dissertation examined the relationship of individual factors and AOD relapse among residents of self-run aftercare recovery homes called Oxford House over the course of a one-year study. To accomplish this, this study employed discrete-time survival analysis techniques that compared baseline hazard (risk of relapse) to hypothesized hazard models that included time-invariant and time-varying factors. This dissertation used archival data gathered across four time periods during a one-year study of 897 men and women participants living in 170 Oxford Houses from a U.S. national sample. This study selected a data subset of 268 men and women who entered the OH setting within 60 days of the start of the one-year study to limit eliminate pre-study effects. First, the baseline hazard rates and survival functions were calculated. Based on prior research, four models were constructed to test four hypotheses that were compared for statistically significant effects against baseline hazard rates. First, baseline demographics - age, marital status, and level of education­ will be significant predictors of hazard. Second, baseline addiction severity- length of sobriety upon entry into the OH, number of previous treatments, and AST alcohol, drug, and psychiatric severity composite scores- will be significant predictors of hazard. Third, baseline variables related to employment will be significant predictors of hazard. Fourth, changes in drug and alcohol abstinence self efficacy over the course of the one-year study will significantly affect hazard rates. Results indicated two of the tour hypotheses were supported. The overall model tor addiction severity was significant with more severe SUD and psychiatric problems significantly predicted higher hazard to relapse, specifically AST alcohol severity, AST psychiatric severity, and the number of previous drug and alcohol treatments. AST drug severity and days since last used substances did not contribute significantly to the model. The fourth hypothesis time-varying changes in abstinence self-efficacy across the one-year study did significantly affect hazard. Closer examination indicated that only increases in alcohol abstinence self-efficacy reduced hazard rates, and that changes in drug abstinence self-efficacy had no effects on hazard. Further, neither of the overall models for the two hypotheses involving baseline demographic and employment variables- age, marital status, level of education, employment income, days employed- significantly altered hazard rates compared to baseline. Implications based on these results are discussed. OHs could potentially reduce relapse by providing closer monitoring and referring additional services to residents who have more severe addiction severity prior to coming to live in an OH, and to enhance abstinence self-efficacy, particularly for alcohol. From a design and measurement standpoint, these results indicated that researchers should measure alcohol and drug abstinence self-efficacy separately rather than conflating them, or only measuring self-efficacy for drug of choice. In addition to these findings, this dissertation also represents the first use of survival analysis techniques to analyze relapse occurrence among Oxford House residents. This study also contributes to the literature by performing a rare survival analysis of relapse using a large national sample living in relatively homogenous settings over the course of one year. The author discusses additional future research and advocates research designs amenable to survival analysis techniques, and believes that further survival analyses of OHs are useful and warranted.