Bayesian Model Averaging and Augmented Autoregressive Distributed Lag Approach for modeling hospital waste amount. Evidence from Dokuz Eylul University hospital in Turkiye

Medical waste management is a growing challenge for healthcare facilities worldwide, particularly in developing countries like Türkiye, where systematic econometric modeling for waste prediction is scarce. This study aims to address this gap by examining the factors influencing medical waste generation at Dokuz Eylul University Hospital (DEUH) from January 2019 to December 2023. The Bayesian Model Averaging (BMA) approach was used to select variables for the econometric model, while Augmented-ARDL analyzed co-integration, DOLS explored long-run effects, and the Fourier Todo-Yamamoto test examined causality relationships. The findings show that Bed Occupancy Rate (BOR), Bed Turnover Rate (BTOR), and Number of Days Hospitalized (NDH) have a statistically significant positive effect on medical waste in the long run. Conversely, Cycle Interval (CI) and Number of Inpatients (NI) have a negative but statistically insignificant effect. Causality analysis indicates a unidirectional relationship from BOR, CI, NDH, and NI to waste amount (WA), while there is bidirectional causality between BTOR and WA. These results suggest that increases in medical waste can impact hospital resource utilization and efficiency. The interaction between BTOR and WA indicates that high turnover may increase medical waste, while the increase in waste could affect hospital resource management. This study highlights the importance of BOR and BTOR in the long-term planning of medical waste management and hospital operations.