Comprehensive Summary
This study conducted by Varghese et al. focuses on predicting opioid consumption after surgical discharge using an AI model to prevent excess opioid prescription. The researchers used patient data from a wide range of hospitals and countries. A wide range of variables were included in the dataset, such as smoking status, urgency of surgery and prior opioid use. The developed model performed well with internal testing having an AUC value of 0.84 and external validation having an AUC value of 0.77. Varghese et al. acknowledge that the patients’ opioid consumption can vary depending on the amount of opioid prescribed, and their external validation was limited. Nevertheless, their model could assist physicians in making a more flexible decision to prescribe safe quantities of opioids to patients.
Outcomes and Implications
Excess opioid prescription at surgical discharge is a frequent occurrence. Despite its intention being to help recovery by reducing postoperative pain, excess prescription is associated with continuous use of opioids and even opioid-related deaths. There is therefore a need for an AI model to predict opioid consumption to prevent overprescription. This AI tool can be distributed to physicians to guide their decision in prescribing an appropriate opioid dosage. However, the model needs further validation and studies on its clinical impacts to be used in practice.