Abstract
The causal effect of hospital length of stay on crucial patient outcomes such as readmissions or mortality is under-investigated and therefore unknown for the vast majority of the US population. Existing evidence stems from association studies that are unable to draw causal conclusions. This study leverages Medicare's two-midnight (2MN) and three-day (3D) rules as two natural experiments to establish causal relationships between hospital length of stay (LOS) and patient outcomes. Using a quasi-experimental regression discontinuity design with data from a large US hospital, we find that the 2MN rule increases LOS by 0.10 days and the 3D rule by 0.21 days, confirming the validity of these rules as instruments for causal inference. However, despite these increases in LOS, there are no significant effects on 90-day mortality or 30-day readmission rates. These findings suggest that while the 2MN and 3D rules effectively extend hospital stays, they do not improve patient-related outcomes, indicating an inefficient use of hospital resources.