[Validation of the usual values provided by a supplier without access to a control population: example of serum...

As part of the laboratory accreditation process, it may be required to determine the laboratory's own reference values. Thus, we have accredited in our laboratory, the serum amyloid A (SAA) assay for which the supplier gave the usual values. However, we did not have a reference population to check them.

We extracted from our laboratory information management system, the values of all SAA and C-reactive protein (CRP) assays performed simultaneously from 2014 to 2018. We selected all SAA-CRP couples with a CRP <5 mg/L and <3 mg/L assuming that these subjects were comparable to the general population. For each of the selected CRP thresholds, we quantified the number of subjects with SAA ≤6.4 mg/L (provider's data), then SAA ≤10 mg/L (threshold below which the risk of kidney complications is low). We compared annual averages of SAA-CRP couples (2014-2018) by analysis of variance (ANOVA).

For subjects with CRP <5 mg/L, we found SAA value ≤6.4 mg/L in 84.6% and ≤10 mg/L in 92.8% of cases. These results were respectively 89.6% et 95.6% when considering CRP <3 mg/L. Similar thresholds were observed for adults and children. The ANOVA analysis did not show any difference per year from 2014 to 2018 for both SAA and CRP in children, adults and the total population.

Our reference values were comparable to those given in the supplier's record and those issued from the literature both for children and adults. The analysis of the evolution of the biomarker average level as a function of time could be an additional relevant element.