Background: Drug poisoning (overdose) is a public health crisis, particularly among people living with HIV and hepatitis C (HCV) co-infection. Identifying potential predictors of drug poisoning could help decrease drug-related deaths.
Methods: Data from the Canadian Co-infection Cohort were used to predict death due to drug poisoning within 6 months of a cohort visit. Participants were eligible for analysis if they ever reported drug use. Supervised machine learning (stratified random forest with undersampling to account for imbalanced data) was used to develop a classification algorithm using 40 sociodemographic, behavioural, and clinical variables. Predictors were ranked in order of importance, and odds ratios and 95% confidence intervals (CIs) were generated using a generalized estimating equation regression.
Results: Of 2,175 study participants, 1,998 met the eligibility criteria. There were 94 drug poisoning deaths, 53 within 6 months of a last visit. When applied to the entire sample, the model had an area under the curve (AUC) of 0.9965 (95% CI, 0.9941-0.9988). However, the false-positive rate was high, resulting in a poor positive predictive value (1.5%). Our model did not generalize well out of sample (AUC 0.6, 95% CI 0.54-0.68). The top important variables were addiction therapy (6 months), history of sexually transmitted infection, smoking (6 months), ever being on prescription opioids, and non-injection opioid use (6 months). However, no predictor was strong.
Conclusions: Despite rich data, our model was not able to accurately predict drug poisoning deaths. Larger datasets and information about changing drug markets could help improve future prediction efforts.
Keywords: HIV-HCV co-infection; drug poisoning; machine learning; overdose; random forest.
Drug poisoning, also known as overdose, is a public health threat that affects the lives of many Canadians. Drug poisonings have been on the rise since 2016, with a notable increase observed following the COVID-19 pandemic. Drug poisoning deaths have also been increasing among people who are co-infected with HIV and hepatitis C (HCV). In an attempt to identify those at higher risk of harm from drug poisoning, the goal of this analysis was to predict drug poisoning deaths within an HIV–HCV co-infected population. Data from the Canadian Co-infection Cohort were used, a study that follows over 2,000 people living with HIV–HCV co-infection. A machine learning method, known as random forest, was used to predict drug poisoning deaths using information collected on participants’ sociodemographic, clinical, and behavioural factors. While our model appeared to perform well when run on the entire data set, it was not able to accurately to predict drug poisoning deaths on a subset of data, and its predictive value was poor. Potential predictors were ranked by their importance, and the top five important variables were addiction therapy in the past 6 months, history of sexually transmitted infection, smoking in the past 6 months, ever being on prescription opioids, and non-injection opioid use in the past 6 months. However, none of these variables were very strong predictors. Although our study provides clues as to factors that might predict drug poisonings among people with HIV-HCV co-infection, accurately predicting these events was difficult. Working with larger datasets and having access to detailed information about changing drug markets could help improve future prediction efforts.
© Canadian Association for the Study of the Liver, 2025.