In vitro analysis, an accurate tool to estimate dry matter digestibility in rabbits. Intra- and inter-laboratory variability

The aim of the present study was to determine the intra- and inter-laboratory variability of an enzymatic system of in vitro analysis for estimating dry matter (DM) digestibility in rabbits and validating the predicted nutritive value of 4 complete diets and 4 raw materials during three different periods of time. Chemical composition, DM digestibility and digestible energy (diets only) were known. In vitro DM digestibility (DMdinv) of all samples was determined by 4 laboratories (triplicate analysis) at different times with an interval of one month between analyses. DMdinv variability and chemical parameters were measured in terms of repeatability (SR: intra-series variability within each laboratory), reproducibility (SL: intra-series variability among laboratories) and reliability (SF: variability through time within each laboratory). Both the laboratory and sample affected DMdinv values (P<0.001). The period of time also had a significant effect (P=0.002) on mean DMdinv values (67.4, 66.8 and 67.0% for the 1st, 2nd and 3rd month, respectively). Significant laboratory×sample, time×laboratory and time×sample interaction effects were also observed. Repeatability, reproducibility and reliability values for the diets were better than those obtained for the raw materials (by 2.0, 1.9 and 2.4 times, respectively). Repeatability values were also better than the values obtained for reproducibility and reliability (by 2.2 and 3.6 times, respectively). Repeatability and reproducibility values were consistently worse for raw materials than for complete diets (by 1.5, 4, 2.9 and 1.3, 4.3, 2.8 times for SR and SL in period 1, period 2 and period 3, respectively), and were also worse in period 1 with respect to the other two periods (by 2.1 and 2.2 times for SR and SL, respectively). Finally, the in vitro method always showed better coefficients of variation of repeatability (CVR) and reproducibility (CVL) than those of the chemical parameters frequently used as predictors of dietary energy value (acid detergent fibre and crude fibre) (1.73 vs. 2.41 and 3.88 for CVR and 3.24 vs. 3.70 and 5.17 for CVL, respectively). In conclusion, the proposed in vitro methodology showed adequate repeatability and reproducibility, being suitable for predictive purposes.