|
<<
|
|
IntroductionClients have a right to expect that feed testing laboratories: (1) provide true measures of chemical composition and nutritive value, (2) guarantee that results are consistent (repeatable) within a lab, (3) assure that results are comparable (reproducible) among labs and (4) insure that lab variation is not a major source of error in feed evaluation. Research labs have an obligation to insure that their results are comparable to other research and to information used in the field. Regulatory labs must be certain that their results are accurate and repeatable. Participation in the NFTA Proficiency Program provides some documentation that results from a feed analysis laboratory are acceptable. However, to validate the accuracy of routine analyses by a lab, certification samples must be analyzed by methods that are routinely used in that laboratory. Part of the value of the NFTA is to provide a mechanism whereby clients, researchers and regulators can be assured that laboratory results are valid. The current certification program is designed to evaluate the acceptability of a laboratory's results based solely on an allowable maximum deviation (bias) from reference values obtained on several samples. Deviations from reference values can be due either to systematic or random errors. Systematic errors among laboratories arise from differences in methods or in-house modifications of methods, but can also be due to poor method design or description, improper analytical techniques, instrumentation malfunctions, or biased calculations. Systematic errors should be identified and corrected to improve agreement among laboratories. Random errors are associated with unavoidable sampling variation, uncontrollable deviations in equipment function, environmental fluctuations, and variations in replicating timing and performance of procedural steps among and within technicians. Total accuracy is the closeness of repeated results to the true, expected or accepted value. Accuracy is the sum of variation associated with bias and precision. Bias is the systematic deviation of the average of repeated results from an accepted reference value. Precision is the agreement or repeatability among repeated observations made under the same conditions. Bias and precision are completely independent measures of accuracy. Close agreement of duplicate or triplicate analysis within a laboratory indicates only that the method is precise (can be repeated with little variation), not that it is unbiased or accurate. It is possible, and quite common, to be able to replicate an inaccurate result. Accuracy can only be determined by comparing results to an accepted reference value that is obtained using an accepted reference method. While many factors cause variation in analyses among laboratories, the greatest source of error (variation) among laboratories is probably associated with differences in routine methods. Routine methods often vary because they are poorly written and difficult to follow by less experienced technicians, instructions are passed from one technician to the next during training resulting in gradual modification of the original method, and methods are changed to suit the operation of the lab. These changes in methods, whether unintentional or not, often result in biases (differences) among labs that erode confidence in the value of feed analysis. Although modifications can cause errors in all methods, they are especially harmful when the component being measured is defined by the method (such as fiber). Any method that defines the component that is being measured must be followed exactly to guarantee accurate results. Since the method can affect results, it is evident that the reference value used to determine accuracy (and certification) among laboratories must be based on a reference method. These reference methods must: (1) have established reliability based on inter-laboratory collaborative studies, (2) be adopted by relevant organizations, such as AOAC, (3) be validated within runs by the use of appropriate standards and quality control samples, and (4) be measured directly in accordance with known chemical/physical principles. Based on these criteria, the following methods should be adopted for routine use when appropriate or be used to validate any method that is routinely used in the lab: Dry matter. The Karl Fischer method is the most accurate procedure for measuring true moisture (and therefore dry matter) in all types of samples when properly performed. This method is based on a chemical reaction of Karl Fisher reagent with water in the sample and is applicable to samples of moisture contents ranging from 0.1 to 100% water. Unfortunately, there is no AOAC approved method for the use of the Karl Fischer methods on feeds and forages. However, for samples >85% dry matter, oven drying at 135oC for 2 hr results in estimates of dry matter that are repeatable and closely agree with Karl Fischer methods. For this reason the reference method suggested for dry matter is:
Crude Protein. Based on long-standing tradition and numerous inter-laboratory collaborative studies, Kjeldahl nitrogen methods are suggested as reference methods for crude protein assuming that crude protein equals nitrogen times 6.25. Combustion methods have been validated against Kjeldahl methods in several studies. For this reason, the suggested reference methods for crude protein are:
Acid Detergent Fiber. Acid detergent fiber (ADF) is literally defined by the procedure used to measure it. Therefore, ADF can only be measured accurately if the AOAC method (973.18) is followed exactly. Any modification of the method that results in different values for feeds defines a new and different type of fiber that should not be called ADF. For this reason, the suggested reference for acid detergent fiber is:
Neutral Detergent Fiber. There is no AOAC approved method for neutral detergent fiber (NDF). The method for amylase-neutral detergent fiber (aNDF) described in this manual has been thoroughly tested on a diverse set of feeds and solves many of the problems associated with filtering residues. This method will be submitted to AOAC for a collaborative study in the near future. For this reason, the suggested reference method for neutral detergent fiber is:
|
|
|
<<
|
|
|
|
|