Evaluating variables affecting Pellet Durability Index (PDI) in pelleted corn-soy-based feeds for swine and poultry: A meta-analysis

Pelleting is one of the most widely used feed processing methods in poultry and swine production systems, as it enhances feed handling, improves animal production efficiency, and reduces feed waste. However, poor pellet quality remains a significant challenge for swine and poultry feed manufacturing, limiting economic returns at both the mill and farm levels. Predicting and optimizing pellet quality can be challenging due to the numerous factors in the feed manufacturing process, including diet formulation, production parameters and environmental conditions. A meta-analysis of published studies was conducted to investigate and quantify the impact of formulation and manufacturing factors on pellet quality and develop predictive equations for Pellet Durability Index (PDI), which represents an in-mill metric for pellet quality. Following a systematic review of the literature, a dataset comprising 16 variables and 280 treatment mean observations was extracted from 29 published studies. Models were developed while treating the study as a random effect. Eight factors were selected for model development, including feed composition factors comprising ingredient inclusions and nutrient content as well as key manufacturing parameters. All models were assessed using a 3-fold cross-validation approach, by placing entire studies into folds. The results showed that the bivariable model using Lipid (g/kg) and Protein (g/kg) , resulted in the best PDI prediction performance, with a Concordance Correlation Coefficient of 0.57 ± 0.076, a bias correction factor of 0.91 ± 0.092, and a Pearson Correlation Coefficient of 0.63 ± 0.137. This meta-analysis provides not only a pipeline for selecting models based on a set of statistical criteria, but also produced simple models that can be utilized in feed manufacturing to estimate PDI when data availability is limited.