Quality assurance in agricultural laboratories in Southern Africa

The history of standardization of analytical procedures, especially for soil analysis in South Africa, is discussed, as is the evolution of the quality assurance schemes currently in use by the Agricultural Laboratory Association of South Africa (AgriLASA). Schemes currently comprise soil, fertilizer, and plant material (two samples of each on a monthly basis), animal feeds, growth media, and irrigation water (two samples of each on a bi-monthly basis). Reporting is done electronically, by tax and by post. Laboratories are encouraged to use at least two of these media to ensure timely receipt of results. The International Harmonized Protocol for Proficiency Testing of Chemical Analytical Laboratories, developed jointly by the AOAC International, ISO (1) and the International Union of Pure and Applied Chemistry (IUPAC), was adopted for statistical analysis of results using the following procedure: Laboratories do a single analysis or analyze in duplicate or triplicate as they would for a routine sample submitted to them. The mean of the results (or a single result) is reported. The results are scrutinized, and outliers are eliminated from the calculation of the mean and standard deviation. The rejected results are still evaluated with the other results. The criterion used for elimination is that, after ranking, results that are > 1.75 times larger or smaller than the neighboring results are rejected. No further outlier rejection is applied, and the robust mean and standard deviation of the remaining results are calculated. The robust technique gives less weight to results that are farther away from the center of the series of results and obviates the need for outlier rejection. By using the robust means and standard deviation, the z-score (z(i)) for each laboratory i is calculated from: z(i) = (x(i) - m)/s where xi is the result for the laboratory i, m is the robust mean, and s is the robust standard deviation. Statistical reports are prepared and distributed electronically to those participants who have requested them, and hard copies are still being posted to all participating laboratories and parties or organizations with vested interests. Certificates, based on two combined criteria (participation and performance), are issued annually to qualifying laboratories. Participation is evaluated at 90% and 75% of' analyses evaluated and performance at adjusted mean absolute z-values of < 1 or < 2. Summaries of these performances are also published in the agricultural and associated press. Potential users of analytical services are advised to request this information from laboratories they intend to use. Examples of how participation in the schemes has identified and addressed problem areas and led to improved performance are discussed.