Zero-augmented accelerated spatial failure model for modeling hospital length of stay data

Area of Science:

• Biostatistics
• Health Services Research
• Epidemiology

Background:

• Hospital length of stay (LOS) is a key metric for hospital efficiency and resource management.
• Traditional statistical models often struggle with the zero-inflated and highly skewed nature of LOS data.
• Geographical variations in health service utilization necessitate spatial considerations in LOS modeling.

Purpose of the Study:

• To develop and validate a novel statistical model capable of handling zero-inflated and skewed hospital length of stay data.
• To incorporate spatial smoothing to account for geographical variations in health service utilization.
• To investigate health district-level factors influencing the capacity for longer hospital stays.

Main Methods:

• Development of a zero-augmented accelerated frailty model to address data skewness and zero values.
• Application of conditional autoregressive priors for spatial smoothing across hospital health districts.
• Bayesian inference using Markov Chain Monte Carlo (MCMC) simulation techniques.
• Demonstration using patient data for chronic lower respiratory disease in Saskatchewan, Canada.

Main Results:

• The proposed model effectively handles the complexities of hospital length of stay data, including zeros and extreme skewness.
• Spatial smoothing revealed geographical patterns in health service utilization and hospital stays.
• The model identified health districts with consistently higher or lower capacities for extended patient stays.

Conclusions:

• The developed zero-augmented accelerated frailty model provides a robust approach for analyzing hospital length of stay.
• Incorporating spatial components enhances the understanding of geographical disparities in healthcare resource utilization.
• This methodology offers valuable insights for optimizing hospital resource allocation and patient management strategies.