A Critical Perspective on Statistical Issues in Estimating Nutrient Bioavailability in Animal Bioassays.

Abstract

Assessing nutrients' relative bioavailability value (RBV) in poultry nutrition has been a prominent subject in the scientific literature for several decades. This method of nutritional evaluation is commonly used to appraise emerging sources of trace minerals and amino acid chelates. References outlining the method for estimating RBV have been available since the 1970s. However, a simplified approach to RBV estimation using the slope-ratio method without preceding statistical considerations to ensure validity and meet fundamental requirements may yield misleading conclusions. Using the slope-ratio method, which involves dividing the regression slope of the test ingredient by that of the reference, can cause uncertainties regarding statistical significance if the model's probability is reported without confidence intervals (CIs) for the RBV estimates. Despite longstanding criticism regarding the misinterpretation and improper use of statistical tests and CIs, these issues persist in estimating RBV using the slope-ratio method. An additional concern is that the misuse of the slope-ratio method and the application of inappropriate statistical analyses can lead to the underestimation of the RBV of nutrients in poultry species. This means that improper application of these methods can cause inaccurately low RBV values, affecting the assessment of nutrient effectiveness. This review addresses the potential pitfalls in peer-reviewed articles within this field, with a particular focus on zinc bioavailability through a reevaluation of RBV data in broilers, laying hens, and honeybees.