Identifying the Best Parametric Model of Survival Analysis to Determine the Factors Affecting the Length of Hospital Stay in Burn Patients in Mazandaran Province, 2013-2019

Background and purpose: Hospital Length of Stay (LoS) is a good indicator of the quality of hospital care. The purpose of this study was to identify the best standard models for survival analysis to determine the factors affecting the length of hospital stay in burn patients. Materials and methods: In this cross-sectional study, the hospital records of 460 burn patients hospitalized in Sari Shahid Zare Hospital, 2013-2019 were reviewed. Best model was introduced using Akaike information criterion (AIC) method. Multivariate survival analysis was done and factors affecting the LoS were determined. Statistical analyses were performed using R ver. 4.1.1 (survival package) and SPSS ver. 24. Results: The Burr parametric model had the lowest AIC value (3284.92) and was therefore selected as the best model. Comorbidities, including diabetes (HR= 1061, 95% CI: 0.28-0.67), location of accident (HR= 1.33, 95% CI: 0.07-0.49), burn percentage (HR=0.73, 95% CI: 0.28-2.35), and time interval between burn incidence and hospitalization (HR=1.69, 95% CI: 0.18-0.88) significantly affected the duration of hospitalization (P<0.05). Conclusion: The Burr model fits better than other standard parametric models of survival analysis to determine the factors affecting the length of hospital stay in burn patients.