A meta-analysis of passage rate estimated by rumen evacuation with cattle and evaluation of passage rate prediction models.

A meta-analysis of studies using the flux/compartmental pool method with indigestible neutral detergent fiber (iNDF) as internal marker was conducted to study the effect of extrinsic characteristics and forage type on particle passage rate (k(p)) in cattle. Further, the k(p) prediction equations in the National Research Council (NRC) and the Cornell Net Carbohydrate and Protein System (CNCPS) were evaluated. Data comprised 172 treatment means from 49 studies conducted in Europe and the United States. In total, 145 diets were fed to dairy cows and 27 to growing cattle. A prerequisite for inclusion of an experiment was that dry matter intake, neutral detergent fiber (NDF), proportion of concentrate in the diet, body weight, and diet chemical composition were determined or could be estimated. Mixed model regression analysis including a random study effect was used to generate prediction equations of k(p) and to investigate the relationships between NRC and CNCPS predictions and observed k(p) of iNDF. Prediction equations were evaluated by regressing residual values on the predicted values. The best-fit model when forage type was not included was k(p) (%/h) = 1.19+0.0879 × NDF intake (g/kg of body weight)+0.792 × proportion of concentrate NDF of total NDF+1.21 × diet iNDF:NDF ratio (adjusted residual mean square error = 0.23%/h). The best general equation accounting for an effect of forage type was as follows: k(p) (%/h) = F+1.54+0.0866 × NDF intake (g/kg of body weight) (adjusted residual mean square error = 0.21%/h), where F is the forage adjustment factor of the intercept. The value of F for grass silage, fresh grass, mixes of alfalfa and corn silage, and dry or ensiled alfalfa as sole forage component were 0.00, -0.91, +0.83, and +0.24, respectively. Relationships between predicted and observed k(p) were y = 0.53(± 0.187)+0.41( ± 0.0373) × predicted k(p) and y = 0.58(± 0.162)+0.46(± 0.0377) × predicted k(p) for the NRC and CNCPS models, respectively. Residual analysis of the NRC and CNCPS models resulted in both significant mean biases (observed--predicted) of -2.40 and -1.70% and linear biases of -0.59 and -0.53, respectively. The results from this meta-analysis suggest that ruminal particulate matter k(p) is affected by forage type in the diet. Further, the evaluation of NRC and CNCPS models showed that passage rate equations developed from marker excretion curves markedly deviated from observed k(p) of iNDF derived using the rumen evacuation technique.