Tasas de Caídas de Pacientes Internados Ajustadas por Riesgo como Indicadores del Desempeño del Sistema de Salud...

Background/Objectives: Inpatient falls are widely used patient safety indicators, yet their behavior under periods of large-scale health system stress remains insufficiently understood. This study aimed to evaluate whether risk-adjusted inpatient fall indicators can capture changes in hospital safety performance during such periods, using a prolonged system disruption as an empirical context. The study period was a priori divided into three phases (pre-pandemic, initial pandemic, and later pandemic) according to changes in COVID-19 admission burden and system stress intensity. Methods: We conducted a retrospective observational time-series analysis using daily inpatient fall events and census data from a Japanese acute care hospital between December 2018 and March 2023 (50,140 inpatients; 962 falls). Expected fall rates were estimated using a validated pre-disruption prediction model, and observed/expected (O/E) ratios were calculated to assess risk-adjusted safety performance. Ordinary least squares regression models adjusted for calendar month and seasonal Fourier terms were used to examine temporal associations between fall outcomes and indicators of hospital-level system burden. Results: Both observed and expected fall rates increased during the study period, whereas O/E ratios declined only in the later phase, indicating improvement in risk-adjusted safety performance despite rising intrinsic patient risk. Seasonal patterns in fall outcomes were disrupted during the early phase of system stress but re-emerged over time. Associations between system burden indicators and fall outcomes were most pronounced in the early phase and attenuated in later phases. Conclusions: Risk-adjusted monitoring of inpatient falls provides insight into dynamic changes in hospital safety performance during periods of large-scale system stress and subsequent adaptation. This indicator can also be interpreted as a benchmarking scale for future month-to-month and seasonal safety surveillance beyond crisis contexts.