SUMMARY
A collaborative study was conducted to evaluate the crude protein (CP) content in feeds in seven feed analysis laboratories from institutions linked to the National Institute of Science and Technology in Animal Science (INCT-CA). Were evaluated four samples of different feeds (sugarcane, corn silage, soybean meal and corn grain) and four nitrogen compounds assumed as standards (urea, monobasic ammonium phosphate, ethylene diamine tetraacetic acid and ammonium sulfate) to measure the accuracy of the estimates. Each laboratory analyzed the samples for six days, consecutive or not, with three replicates per sample per day of evaluation. The variances among laboratories and within of laboratories (repeatability) represented 34.5 and 10.3% of the total variation, respectively. The evaluation of nitrogen recovered from the standards showed that 25% of observations presented a significant bias (P<0.001). The overall assessment of the samples demonstrated significant variations (P<0.001) in function of the day of analysis, interaction of material with laboratory, and among laboratories. Additionally, it was found that the variation among laboratories corresponded of 43.6 to 68.2% of the total of random variation, while the repeatability (intra-laboratory variation) represented of 11.9 to 42.8% of reproducibility. The CP contents estimated are dependent of the interaction effect of the evaluated material and laboratory, which reflect the differences between the procedures utilized by each laboratory.
Keywords:
feed analysis; laboratory; nitrogen; repetibility; reprodutibility
