Keywords: pragmatic trials, missing data, selection bias, weight loss interventions
Body weight is frequently measured as part of routine clinical care, so weights derived from the electronic health record (EHR) provide an opportunity to conduct research studies in real-world settings. Using EHR data for research requires care because EHRs were implemented to support clinical and billing systems, so the systematic measurement processes are unknown and may be informative (e.g., selection bias if more obese patients are measured more frequently). However, study-collected data in weight loss randomized trials also present measurement challenges because patients who lose less weight are often more likely to dropout. Augmenting study-collected weights with EHR weights can serve multiple purposes: 1) EHR data can help verify trial dropout assumptions; 2) EHR data can inform long-term trends beyond study-end; 3) study-collected data can enhance our understanding of EHR-data systematic measurement bias, an essential step in conducting larger-scale pragmatic trials. Methods will be illustrated across several weight loss maintenance trials that measured weights as part of the study protocol and, retrospectively, ascertained weights from the EHR for the study period.