| dc.description.abstract | Opioids are commonly prescribed for chronic pain. However, even when titrated appropriately, opioids may result in opioid-related adverse events (OAEs). Previous studies have defined OAEs categorically (e.g., did/did not occur), but OAEs are often combinations of phenotypes. Differentiating such OAEs might facilitate population-level studies of OAE burden. Burden related to Mendelian disease has been measured in the electronic health record (EHR) by aggregating combinations of disease phenotypes (EHR diagnostic codes) into the phenotype risk score (PheRS). In this study, we developed a PheRS (PheRS OAE) for quantifying OAE burden using EHR data. We validated PheRS OAE through control studies, analyzed the association between PheRS OAE and metabolizer status, and generated detailed OAE profiles for each opioid. We conducted a replication study using data from the All of Us Research Program. In both datasets, PheRS OAE was sensitive to opioid exposure (p<0.001) and detected increased drug-related toxicity among individuals exposed to opioids with greater mu-receptor affinities (p<0.001). Associations between 1) morphine and methylnaltrexone prescriptions and 2) hydromorphone and methylnaltrexone prescriptions were replicated (both p<0.001). PheRS OAE is scalable and aggregates combinations of OAE phenotypes into one score, thus enabling the rapid comparison of different medications, genetics, and OAEs across a population. This method may reveal genetic and phenotypic targets conferring risk for OAEs and inform pharmacogenomic interventions improving drug safety. | |
