Leveraging Administrative Health Data to Capture Rare Adverse Drug Reactions: Identifying Pediatric Patients With Trimethoprim‐Sulfamethoxazole Acute Respiratory Distress Syndrome

Rare adverse drug reactions (ADRs) are underrecognized and underreported in the electronic medical record (EMR). These events often require clinical review, making systematic identification challenging. The study aim was to develop an approach to prioritize identification and validation of a rare ADR to trimethoprim-sulfamethoxazole causing acute respiratory distress syndrome (TMP-SMX ARDS) across medical institutions. We developed a clinical phenotype based on 2 local TMP-SMX ARDS cases mapped to standardized elements of a national comparative healthcare database (PHIS). We validated identification of TMP-SMX ARDS at scale across medical institutions. A set of scoring criteria was created to prioritize cases and generate a center-specific top 10 list of candidate encounters. External validation at 3 PHIS contributing hospitals with a known TMP-SMX ARDS case was performed by reviewing the top 10 candidate list for the known case. The review period was January 1, 2012-January 1, 2025. EMR data extracted from 2 TMP-SMX ARDS cases included patients that both required extracorporeal membrane oxygenation for > 100 days, experienced air leak early in hospital presentation, required tracheostomy placement, and were hospitalized for > 440 days. Based on the TMP-SMX ARDS phenotype, the local cases ranked 1st and 3rd on the PHIS generated top 10 candidate list for internal validation. For external validation, known cases were identified on their respective hospital top 10 lists, ranking 3rd, 3rd, and 5th. Applying a TMP-SMX ARDS phenotype to a national health care database paired with clinical review of candidate cases may be an effective approach to identify underrecognized ADRs.