Inequalities in the Effects of Health Information Exchange on Hospital Choice - Evidence from Pennsylvania

Background.Health information exchange (HIE) is inevitable for health care reform. The existing HIE literature explores its vast potential, but leaves unanswered questions about which patients are more likely to take advantage of HIE, and for which hospitals patient demand may increase due to HIE availability.

Methods.This study evaluated the impact of HIE use on the demand for several common scenarios (heart attack, congestive heart failure (CHF) and hip or knee replacement) and further identified differential effects by hospital rurality. We linked data from multiple sources over 2005-2011: (1) Healthcare Cost and Utilization Project Pennsylvania State Inpatient and Outpatient Databases, (2) the American Hospital Association (AHA) Annual Survey, and (2) Area Health Resource Files. Merging this data set with the 2007 AHA Healthcare IT Database allowed us to estimate the probability that a patient uses a hospital and examined the specific effects of HIE on hospital choice.

We employed difference-in-difference estimation, using patient-level hospital choice data, with two discrete choice models – conditional choice model and nested logit model. The nested logit model consists of two selection steps: the choice of staying in local areas and the choice of a hospital in a choice market. We constructed choice sets as the inclusion from hospital h if 1) h is within 100 miles from patient i's location coordinate, 2) at least 10% of the patients in patient i’s zip code chose h, or 3) h is within 35 miles of any other hospitals closer to i. The maximum 20 hospitals are included in one’s choice set, as the marginal benefit of an additional further choice is close to zero. The models control for patients’ age, gender, race, insurance type, driving distance, hospital and market characteristics, hospital fixed effects and year dummies.

Principal Findings.The preliminary results document that odds of patients seeking CHF care were higher in HIE use versus non-HIE use hospitals. The number of admissions increased more rapidly in HIE-use hospitals than non-HIE use hospitals. Rural hospitals with HIE had a decreasing trend of average admissions for hip/knee replacement. HIE is a significant factor in hospital choice for all conditions examined. Patients traveling further would be more likely to admit to hospitals with HIE. Further work will test the robustness of these findings and will examine whether the magnitude varies by hospital predetermined attributes, patient’s age and insurance type. 

Conclusions. Certain hospital preconditions would affect the value of HIE use on the demand. Hospital management and policy makers may prepare and make specific changes prior to the HIE participation.