The Association between Medical Care Utilization and Health Outcomes: A Spatial Analysis

Publication date: Available online 9 March 2019Source: Regional Science and Urban EconomicsAuthor(s): Francesco Moscone, Jonathan Skinner, Elisa Tosetti, Laura YasaitisAbstractMany studies use geographic variation in health expenditures or utilization to estimate the marginal benefits of health care intensity. Nearly all fail to account for spatial autocorrelation, that adjacent regions are more similar than distant ones, leading to potential bias in coefficient estimates and confidence intervals. Using data on 897,008 heart attack (acute myocardial infarction) patients in the U.S. during 2007-11, we consider the importance of spatial autocorrelation in cross-section time-series models across Hospital Referral Regions (HRRs). We find substantial spatial autocorrelation in a parsimonious model, but conventional regression estimates are surprisingly insensitive to the inclusion of spatial adjustment methods. The estimates are more sensitive to risk-adjustment methods and the inclusion of specific types of care provided. Some regions of the U.S. exhibit both better outcomes and lower Medicare expenditures, suggesting that a focus on simple correlations between spending and outcomes may be misplaced.We are grateful for financial support to the National Institute on Aging Grant PO1-AG19783.

Source: Regional Science and Urban Economics - March 9, 2019 Category: Science Source Type: research