Accurate and physiologically relevant in vitro models are needed to predict nutrient digestibility and hindgut fermentation in pigs, as conventional in vivo trials are resource-intensive and raise animal welfare concerns. This study evaluated and compared the predictive performance of three in vitro digestion approaches—shaking (S), dialysis (D), and a combined shaking plus dialysis (SD) method—for estimating in vivo apparent total tract digestibility (ATTD) and fermentation characteristics across weaning, growing, and finishing pigs. Commercial diets were subjected to simulated gastric and small-intestinal digestion using S, D, or SD, followed by fecal inoculation to model hindgut fermentation for 12 and 48 h. During the gastrointestinal phase, crude protein digestibility was highest with D (>75%), intermediate with SD, and lowest with S (50–60%), indicating that product removal by dialysis mitigated enzyme inhibition from metabolite accumulation. After 48 h of fermentation, all methods showed strong linear correlations with in vivo ATTD (r > 0.93), but only D achieved high absolute agreement (Lin’s CCC > 0.95 for dry matter and crude protein). Moreover, D and SD at 48 h closely reflected in vivo fecal profiles of skatole, indole, and microbial enzyme activities, with D at 12 h showing an especially strong correlation for protease (r = 0.98). While D provided the most precise predictions of absolute values, the SD method offered an optimal balance between physiological relevance and operational efficiency, supporting its use as a robust, high-throughput platform for porcine feed evaluation and fecal nitrogenous odorant prediction.
Chen et al. (Sat,) studied this question.