Repeat testing – is it necessary?

Clinical laboratories usually have a policy for repeat testing when the test result is either grossly abnormal or unexpected in terms of recent validated test results for that particular patient. The rationale for this policy is that the first test result may represent a clinically significant analytical error, which is revealed by re-testing. Technological advance ensures that analytical errors are increasingly less likely. 

So, in the context of the modern highly automated clinical chemistry laboratory, is repeat testing really necessary anymore? This is the question addressed by a recent US study conducted at a large clinical chemistry laboratory processing close to 7 million tests annually. During a 2-month period in 2011, 855,000 results for 30 of the most common clinical chemistry tests were generated at the laboratory. 

Of these, 25,530 (3 %) were flagged as needing re-testing because the result was outside predetermined limits. Analytical error was deemed to have occurred if the difference between the initial and repeat measurement exceeded the “allowable error limit” for that analyte as determined by the College of American Pathologist/Clinical Laboratory Improvements Amendments (CAP/CLIA). 

By these measures just 668 (2.6 % of all repeated tests) initial test results were “errors”. Of these, only 102 (0.4 % of all repeated tests) related to initial values within the analytical range. The rest, i.e. majority (85 %) of the “errors”, related to initial values that were too high or too low to be within the measuring range of the analytical system. 

For 11 of the 30 analytes there were no “errors” and for most of the rest the “error” rate was less than 0.5 %. The analytes with the highest “error” rates in this study were sodium (6.2 %), chloride (5.3 %) and pO₂(a) (2.9 %). 

The authors conclude that their study results suggest that repeat testing for many automated chemistry tests is now unnecessary at their hospital. Adopting such a policy would allow critical value results to be delivered more rapidly and potentially save an estimated 2-3 % of reagent costs.