We explore the causal relationship between hospital length of stay (LOS) and re-hospitalization for Acute Myocardial Infarction (AMI) patients in Japan, where the average LOS is the longest among OECD countries. Using chart-based data, we address the endogeneity between LOS and re-hospitalization probability by using an exogenous variation based on Rokuyo (the six basic labels allocated to each day), which is found to be irrelevant to admission day but relevant to discharge day. While we do find a significant and positive association between LOS and re-hospitalization probability in the OLS estimation, we do not find a significant relationship once LOS is instrumented by Rokuyo in various instrumental variable estimations. This implies that additional stay that was induced owing to patient’s choice of preferred Rokuyo at discharge has no effect on re-hospitalization probability.
ASJC Scopus subject areas