Unraveling predictors of relapse in substance addictions
Relapse is a major problem in addictions. Several variables have been associated with higher probabilities of relapse, such as marital status, unemployment/employment problems, life events and duration of drug use.
Despite the investment and scientific advances in this area, predicting relapse remains a challenge to clinicians.
Our goal is to statistically identify predictors of relapse and sobriety in patients with substance addictions, as well as to develop a method to predict the probability of relapse.
Review of clinical files and related literature.
Our sample includes about 100 individuals, followed in ET de Matosinhos, a specialized treatment center for addictions in the north of Portugal. We only consider individuals with substance addictions, excluding behavioral addictions.
We follow each individual for at least two years and register the occurrence of relapse episodes at 6 months, 12 months, and 2 years after admission.
Using regression techniques (more specifically, logit regression), we estimate the probability of relapse using two categories of variables. On the one hand, we test as predictors of relapse the following variables evaluated at baseline: age, education, marital and employment status, criminal record, offspring, loneliness, personal history of drug use (age at onset of addiction, main drug, route of administration, initial drug and concomitant alcohol consumption), medical and psychiatric comorbidities and chosen treatment. On the other hand, we include several variables that were registered at follow-up, including life events, job and health changes, and treatment of choice.
Data was processed using STATA13.
From the total admissions recorded, while 26% relapsed within 6 months after admission, 61% stayed sober at least one year. For those who stayed sober at least one year, 70% maintained sobriety for more than two years.
We contribute to debate on this literature regarding the choice of treatment. More particularly, when comparing to other choices of treatment, we find that methadone significantly reduces the probability of relapse within 6 and 12 months.
Furthermore, psychiatric comorbidities were found to increase the probability of relapse within 12 months after admission.
Finally, our evidence suggests that the patient’s social environment significantly affects the probability of relapse. More specifically, we find that patients with a higher degree of social isolation have a higher probability of relapse at 6 months.
More results and specific findings will be presented at the congress.
When treating substance addictions, several factors have to be accounted for.
Our study evidenced some interestingly new findings that can be translated into clinical practice. More particularly, we proposed a novel forecasting model to help clinicians predict the patient’s probability of relapse. We argue that, in order to maximize treatment success, clinicians may customize treatment plans to address the patient’s latent risks.
We also point out that, when understanding the evolution of addiction throughout time, we may have to look at different variables.