Hospital closures and patient outcomes: Evidence from AMI emergency admissions in Italy

Abstract

In the European Union, cost containment strategies following the financial crisis in 2008 have been particularly targeted towards reorganization of the inpatient sector given that hospitals account for large shares of healthcare budgets. Despite the increasing numbers of hospital closures and mergers, few empirical studies have provided credible evidence on the consequences for patient health. This paper uses a difference-in-differences approach to study the causal effect of hospital closures on AMI patient outcomes including in-hospital mortality, readmission and length of stay. In particular, the outcomes of all AMI patients aged 65 and over in Italy admitted from 2008 to 2015 that were exposed to a hospital closure are compared with those that were not. Results show that hospital closures significantly increase in-hospital mortality and length of stay. There is no statistically significant effect on 30- and 90-day readmissions. The effect of hospital closure on in-hospital mortality and readmission is persistent across the post-closure years indicating that there is no short- to medium-term adaptation. These results are robust to a set of sensitivity checks, including the Callaway and Sant'Anna approach. Further, heterogeneity analyses show that the adverse effects are largely driven by women and patients above 80 years, and that the geographic North/South divide is particularly relevant for the mortality outcome. We further investigate travel time to the hospital as potential mechanism for the identified effects. We propose three different methods to disentangle the direct from the indirect effects of hospital closures: (1) closure as instrumental variable for travel time, (2) analysis of never-takers, and (3) blocking the travel time channel. Our results indicate that increased travel time following a closure explains most of the effect in-hospital mortality, but that other channels, such as bed congestion due to spill-overs, should be considered for other patient outcomes.