A parametric method for quality control and long term use in routine analytical laboratories is proposed. The method consists of: i) selection of a statistical level of significance for the analyses performed; ii) initial analysis of control samples with about 30 repetitions; iii) calculation of the mean and standard deviation for each of the control samples; iv) every time the control sample is routinely analysed the probability of obtaining the actual result for the sample is computed through the use of approximation to the normal distribution function, and further approximation to the Student's t distribution; v) if the probability is greater than the selected confidence coefficient, the result is rejected, otherwise it is accepted and used to compute a new mear and standard deviation for the sample. The proposed method maintains invariant the selected level of significance, preserves the natural heterogeneity of the sample, allows for easy distinction between sistematic and accidental errors, and is suitable for use with programmable calculators.
quality control; routine analytical laboratories; parametric method