# Impact of physician shift work implementation on mortality and length of stay in an emergency and critical care center: an interrupted time series analysis

## TL;DR

A study in Japan found that implementing a physician shift system did not increase patient deaths but caused a temporary rise in ICU length of stay.

## Contribution

The study provides observational evidence on the impact of physician shift systems on mortality and ICU length of stay in emergency and critical care.

## Key findings

- No significant change in in-hospital mortality after implementing the physician shift system.
- A transient increase in ICU length of stay was observed at the time of implementation.
- Adjustments for patient severity and case mix did not change these findings.

## Abstract

Shift-based physician work systems are increasingly being introduced in emergency and critical care settings in Japan. However, evidence regarding their effects on patient outcomes—particularly mortality and intensive care unit (ICU) length of stay (LOS)—remains limited.

This single-center retrospective observational study was conducted at the Emergency and Critical Care Center of Nippon Medical School Hospital in Tokyo, Japan. We evaluated whether implementation of a structured physician shift-based work system was associated with changes in in-hospital mortality and ICU LOS using an interrupted time series (ITS) design. All patients admitted between November 2017 and December 2024 were screened. Patients admitted during the COVID-19 pandemic period, those with cardiopulmonary arrest before or upon arrival, and elective surgical admissions were excluded. A structured two-shift physician work system was implemented in October 2023. The primary outcome was in-hospital mortality, and the secondary outcome was ICU LOS. Segmented ITS regression models were used to estimate level and slope changes before and after implementation. Additional models adjusted for monthly mean APACHE II scores and diagnostic category distributions to account for temporal changes in patient severity and case mix.

A total of 5,477 patients were included (pre-shift: 3,745; post-shift: 1,732). After adjustment, no clear change in mortality was identified at implementation (level change − 1.48%, 95% CI − 5.23 to 2.27) or in the subsequent trend (slope change + 0.05% per month, 95% CI − 0.26 to 0.36). LOS showed a pre-intervention downward trend (− 0.13 days per month, 95% CI − 0.35 to 0.09), followed by an immediate increase at implementation (+ 2.95 days, 95% CI 0.10 to 5.80). No meaningful change in post-intervention slope was observed (− 0.09 days per month, 95% CI − 0.33 to 0.15).

Introduction of a structured physician shift system was not associated with an increase in in-hospital mortality. A transient increase in ICU length of stay was observed at the time of implementation, even after adjustment for monthly APACHE II scores and diagnostic category distributions. These findings describe temporal changes in selected clinical outcomes following system implementation and should be interpreted with caution given the potential for residual confounding and the limitations inherent to observational time series analyses.

The online version contains supplementary material available at 10.1186/s12913-026-14226-6.