Time Series Analysis for forecasting neonatal intensive care unit census and neonatal mortality

Abstract

Background: Neonatal intensive care units(NICUs) play a crucial role in caring for premature or critically ill newborns, but challenges persist in managing patient volumes and addressing mortality. This study aims to analyze the time series of the NICU admission numbers, hospitalization days, and mortality proportion. Methods: We used seven years of retrospective daily NICU census data for model development (March 2016 - December 2022, N=7,216 infants). Best-fitting models of ARIMA and SARIMA were applied for forecasting admission number, long stay and mortality proportion in STATA.14 and SPSS.20. The accuracy of forecasting approved by root mean squared error (RMSE), mean absolute percentage error (MAPE). Results: We observed a decreasing trend in mortality proportion in the NICU, with more pronounced seasonal patterns in admission numbers (which increased during the winter season) and length of stay (which decreased during the winter season). Our regression time series analysis showed that as the length of stay in the hospital increases, the mortality proportion also increases. Conclusion: More extensive and well-designed studies are required to investigate the risk factors for prolonged stays in the NICU and how to manage them. Research should also concentrate on interventions that can effectively reduce long NICU stays and improve short- and long-term outcomes for newborns.