Appendix D: Analytical Methods 



guidelines to indicate the quality of the data. This chart is known as an analytical quality control 

 chart (Fig. D.2). 



Analysts are reminded that before a method is used routinely for samples it must have been 

 rigorously assessed to ensure that it will provide data of the required quality. The following 

 procedure is used to construct an AQCC: 



(i) Select the SRM to be analysed with samples on a regular basis. 



(ii) Analyse the SRM at least 10 times for the analyte(s) under examination. These 

 analyses should not be done on the same day but spread out over a period of time in an 

 attempt to ensure that the full range of random errors within and between batch analyses are 

 covered. 



(iii) Calculate the mean value (X), and the standard deviation (sd) and then plot the 

 following values on a blank control chart: X, X + 2sd (UWL), X + 3sd (UCL), X - 2sd 

 (LWL) and X - 3sd (LCL). 



Assuming that the analytical measurements for SRMs follow a normal distribution, 95% of 

 them (19 in every 20) should fall within the area between UWL (upper warning limit) and LWL 

 (lower warning limit). Similarly 99.7% of the results should fall within the area between UCL 

 (upper control limit) and LCL (lower control limit). 



The analyst should regularly analyze SRMs and plot the results of the analysis of the SRMs 

 after each batch of analyses to check where the data lie in relation to these limits, as shown in the 

 example of a plot given in Fig. D.2. 



The following guidelines can be used to assess whether the data for the SRMs and 

 consequently the data for the samples are of acceptable quality: 



(a) A single result which falls outside the warning limits need not require the analyst to 

 doubt the results or take any action provided that the next result falls within the warning 

 limits. 



(b) If the results fall outside the warning limits often, particularly if the same warning limit 

 has been crossed more than once on consecutive results, then the analyst needs to assess 

 the source of this systematic error. 



(c) If the results on more that 10 successive occasions fall on the same side of the X line 

 (either between X and UWL or X and LWL) then the analyst needs to check the analytical 

 procedure to determine the cause of this error. 



(d) If the result falls outside the UCL or LCL lines then the analyst should check the 

 analytical procedure to determine the cause of this source of error. 



107 



