Using linear regression and mixed models to predict health care costs after an inpatient event

[摘要] Gateway Health Plan® wanted to compare the before and after costs of a member who had an inpatient stay in a hospital which will allow them to evaluate costs in comparison trials. As part of my internship with Gateway Health Plan®, I was able to estimate a formula to evaluate difference in costs.Using Gateway Health Plan’s® internal data from the past three years, I used regression to evaluate the difference in cost for members before and after an inpatient event.I ran a simple linear regression model as well as a mixed effects model in order to look at the comparison of the before and after costs. Age and gender were also considered at as possible covariates in the prediction process because both of those factors are known to be associated with healthcare costs. The results showed that average cost before an inpatient event as well as gender were significant in estimating the average cost after an inpatient event. I found that females tend to cost less than males, and female patients cost less after the inpatient event compared to before the inpatient event, while men cost more after the event.Public health significance: This research will help Gateway Health Plan to evaluate interventions to assess whether they lower health care costs. Being able to evaluate if interventions are cost efficient will improve healthcare leading to an improvement in population health.

[发布机构] the University of Pittsburgh