A cross-classified and multiple membership Cox model applied to calf mortality data

Area of Science:

• Veterinary Epidemiology
• Statistical Modeling
• Animal Science

Background:

• Calf mortality presents significant economic and welfare challenges in the cattle industry.
• Understanding factors influencing calf survival is crucial for improving herd management and productivity.
• Complex data structures in livestock research require advanced statistical approaches.

Purpose of the Study:

• To apply a novel statistical model to analyze calf mortality data in Western Canada.
• To identify key risk factors associated with calf mortality within a multi-level data hierarchy.
• To evaluate the performance of the chosen statistical methodology.

Main Methods:

• Utilized a cross-classified and multiple membership Cox model applied to 23,409 calves across 174 herds.
• Employed a 3-level hierarchical structure accounting for veterinary clinics and ecological regions.
• Fitted the model using Bayesian Markov Chain Monte Carlo (MCMC) estimation within a mixed-effects Poisson framework.

Main Results:

• Herd-level variation was a more significant factor in calf mortality than ecological regions or veterinary clinics.
• Identified significant fixed effects including twinning, calf gender, calving assistance, cow age, temperature, and herd management practices.
• Simulation studies confirmed the validity and reliability of the applied statistical approach for similar datasets.

Conclusions:

• The cross-classified and multiple membership Cox model effectively captures complex hierarchical influences on calf mortality.
• Herd-specific management and intrinsic calf/dam characteristics are primary drivers of calf survival.
• The methodology provides a robust framework for future epidemiological studies in animal health.