Downer cows: a reanalysis of an old data set

Area of Science:

• Veterinary Medicine
• Epidemiology
• Biostatistics

Background:

• Downer cows, often unable to rise, present a significant challenge in dairy farming.
• Accurate prognostic models are crucial for guiding treatment decisions and managing herd health.
• The original Clark model (1987) for downer cow survival has been widely used.

Purpose of the Study:

• To compare the predictive performance of the original 1987 downer cow survival model with a new model.
• To evaluate the impact of modern data imputation techniques on prognostic model accuracy.
• To assess the continued clinical relevance of the 1987 prognostic model.

Main Methods:

• A new predictive model was developed using multiple imputation by chained equations on a 1987 dataset with missing values.
• The original and new models were tested on a separate, complete dataset from a different region.
• Key performance metrics including sensitivity, specificity, and accuracy were calculated for both models.

Main Results:

• The original 1987 model demonstrated slightly higher accuracy compared to the new model.
• Sensitivity for the original model was 0.85 (95% CI: 0.72-0.94) and specificity was 0.82 (95% CI: 0.7-0.91).
• The prognostic formula published by Clark et al. in 1987 performed comparably to the modern, imputed model.

Conclusions:

• The original prognostic formula for downer cow survival remains a valuable tool.
• The 1987 Clark model's performance is comparable to contemporary models developed with advanced imputation methods.
• The Clark model should remain integral to the clinical evaluation of downer cows by New Zealand veterinarians.