Abstract
Background/Aims When a particular health event or condition is uncommon, researchers consider broadening the target population. All Group Health (GH) patients have an electronic medical record (EMR). However, GH EMR implementation has only occurred at GH-owned clinics. For studies requiring medical record review in addition to other data collection, the practice has been to include only enrollees at GH-owned clinics. However, perhaps EMR data are more widely available. We conducted a pilot study of: How many GH network enrollees (i.e. enrolled at non-GH owned clinics) have GH EMRs that contain clinical data? What kinds of visits generate rich data in the EMR for network enrollees?
Methods Our population was Group Health (GH) enrollees enrolled at network clinics in western Washington. Although this population is demographically comparable to enrollees at GH-owned clinics, the assumption has been that clinical data in the EMR is missing or meager, making them poor candidates for studies requiring chart review.We examined 4 indicators (patient use of the shared medical record, or presence of scanned documents, transcriptions, or vitals in EMR automated data). We chose these as likely markers of clinical information in the GH EMR, and randomly selected 5 enrollees for each indicator and reviewed their EMRs.
Results Among enrollees at network clinics, 20% (10,327/51,637) had at least one of the four selected indicators. Only 1 of the 20 records reviewed had no clinical EMR data. The best indicator of a rich EMR was presence of vitals.
Conclusion Enrollees at network clinics with visits to GH clinics (or with recorded vitals in EMR automated data) may thus be a suitable target population to add to studies that include chart review. However, this may only add 1/5 of all network enrollees to a study population. Future investigations might include:
-
Select enrollees at network clinics with hospitalizations to see what outside utilization is captured by the GH EMR;
-
Examine how health plan type affects availability of EMR data.
