Abstract
Background/Aims Investigators often assume that when patients are enrolled in a health plan, they have complete capture of utilization data from their health plan sources. This assumption may not always be true as patients have many incentives to choose care at multiple settings (convenience, price, residence, insurance type, drug coverage), which can give rise to missing data. Complete capture of medical data for population-based research is crucial to our ability to identify populations who have not had particular exposures or outcomes. In case-control designs, where the control group does not have a specified exposure, we measure this condition by absence of data. To address this problem, the VDW enrollment work group created a new enrollment variable, called “Outside_utilization”, designed to identify members suspected of incomplete capture of encounters or pharmacy fills. This work reports on a quality assurance analysis of the extent and nature of the data gaps at different sites.
Methods For V3, HMORN sites added new VDW variables in their enrollment file including the “Outside_utilization” variable. This variable identifies populations suspected of having incomplete health care utilization capture. Since the reasons for incomplete data capture vary among the sites, the methods for identifying members with incomplete data capture were determined by the local site data managers. The authors distributed a program that computed utilization rates for specific cohorts for “complete” and “incomplete” data capture populations. We also computed rates for those on high deductible plans. We compared differences in these rates by year and site. In addition, we conducted a survey to determine how the incomplete populations were identified for each participating site.
Results Some sites clearly identify populations that have incomplete capture of data using the “Outside_utilization” variable. At other sites, the difference in rates is less apparent. The sites that can distinguish patients with incomplete versus complete data have certain common definitions used when designing the “Outside_utilization” variable.
Conclusions The “Outside_ utilization” variable identifies populations with incomplete data capture at some sites. We recommend that projects use this variable to exclude populations suspected of incomplete data capture when computing population-based utilization rates.
