Assay Quality Control
It would seem to be a reasonable objective that results produced by a given laboratory for each assay are consistent over time and that results produced by different laboratories from the same samples are comparable. At least one published study has investigated within- and between-laboratory variation in Elisa results in the USA (Kreider 1991b). Results varied significantly and substantially among different laboratories (the greater source of variation) and among different days in the same laboratory. This suggests that single determinations on individual serum samples are not likely to give a reliable estimate of antibody titre. The large variability within laboratories further indicates the need for standard reference pools of positive serum to be included in assays in order to substantiate assay results.
Murray et al. (1993) have discussed in some detail the sources of variability of assay results. While these authors concentrated on biochemical assays the same basic principles apply equally well to serology. Each assay must be validated to identify and quantify sources of variation in results. We must also keep in mind that there are non-assay sources of variation. These can be grouped into those factors which precede the assay (how and when a sample is taken, how it is manipulated, stored, transported, and identified) and those which come after (for example transcription errors in report generation). In validating an assay the following areas need to be addressed :
1. Specificity - A highly specific assay will have a low tendency to show "false positive" reactions in animals exposed to a closely related pathogen. This can be tested by obtaining mono-specific sera raised against a range of other pathogens and including them in the test.
2. Sensitivity - This is a measure of the ability to detect clinically important but very low levels of antibody. There will sometimes be a trade-off in that the higher the sensitivity of an assay the lower its specificity. In poultry production it is common practice to apply a highly sensitive technique as a screening test in order to identify problem flocks or individuals. then apply more specific tests only to these or on repeat sampling of the flock. The sensitivity of a test can be evaluated by diluting known positive samples sequentially and determining the dilution at which the reaction is lost.
3. Accuracy - This is a measure of the ability of the test to measure purified amounts of the substance sought when it is added in measured amounts to a typical test sample. Rarely will we have purified antibody available for this type of study but neither is this required in that we will not be reporting results in "milligrams of antibody". If known positive field sera are available then they can be used as a pool in repeated assays. This will be most valuable if this pool is also submitted to a reference laboratory for testing using a recognized and already validated procedure. The alternative is to take purified mono-specific antiserum and use this to "spike" sero-negative field serum at different concentrations then use the spiked samples to establish a measure of the variability of the assay results within a given sample.
4. Precision - This is the ability of the assay to consistently reproduce a result when sub-samples are taken from the same specimen. Within-assay and inter-assay precision are two distinct measures of this can be made as part of the validation procedure. The formulae used for the calculation of CV% are slightly different from the conventional formula (Standard Deviation divided by the mean and multiplied by 100).
Within-assay precision - Assay 10 duplicated samples on the same plate (a total of 20 assays) and calculate an intra-assay coefficient of variation as follows:
Mean of the Standard Deviations of the Duplicates
------------------------------------------------------------ x 100
Grand mean of the duplicates
A figure of 10% or less is considered satisfactory (Murray et al 1993)
In this case the 10 runs on duplicate samples are run on different days. For each run the mean, the deviation, and the % C.E. are calculated. The interassay coefficient of variation is calculated from the formula:
Standard Deviation of the means of the duplicates
-------------------------------------------------------------- x 100
Grand Mean of the Duplicates
Much of the validation work described here will, in the case of commercially available antigens and test kits, be carried out by the manufacturer. A good first step in setting up a serological quality assurance system will be to request the manufacturer to provide data on with-run and between-assay precision. The laboratory can then quickly validate the assay using known positive and negative sera. Once an assay has been validated the continued satisfactory performance should be monitored by the use of a Quality control system.
This system should include a definition of criteria for assay acceptability, along with a means for identifying sources of variation and implementing corrective procedures. One component of such a system might be to repeat an Inter-assay Precision test. Alternatively, simply calculate the %CV for all assays of each standard control serum each month and plot the results over time. A reasonable target for %CV in routine testing is 10-15%.
The prime aim of in-house quality assessment is to confirm that the results generated are consistent over time. Laboratories may also participate in inter-laboratory quality assurance schemes with a view to ensuring uniformity of results between different laboratories. Accreditation schemes take this a step further by independently checking procedures and standards, often with a periodic inspection. The only scheme currently operated in the UK which has a component specific to avian serology is the accreditation scheme for Mycoplasma gallisepticum and M.meleagridis testing operated by the Ministry for Agriculture Fisheries and Food. For a detailed discussion of the concepts behind quality assessment and accreditation schemes the reader is referred to Manser (1994).
At every point in the validation and QA of assays it is important that staff be educated and motivated. It is equally important for those interpreting the results to understand the inherent variability of the results and to be able to express an opinion by comparing them with "population means". This may lead to a decision which will take into account information from a variety of sources.