Troubleshooting can be time consuming and frustrating for any laboratory professional. Results may be called into question when they appear to be inconsistent with a patient’s clinical picture, or flagged due to a large delta change or other anomalous internal quality control (QC) metric. Modern analyzers now usually identify common interferents such as hemolysis, icterus, or lipemia, but when the problem goes beyond these familiar issues, further investigation becomes more daunting.
Investigating potential interferences systematically streamlines the overall process (Figure 1). One excellent way to structure this type of troubleshooting is to borrow an approach from the information technology field: Is the problem in an individual component or at the interface between components?
Applying this approach to troubleshooting lab test results, the two components to consider are the assay method and the sample itself. Problems may lie within the assay or the sample, or involve a specific interaction between the assay and the sample. Luckily, to narrow down the possibilities labs need to explore just two factors: how many samples are affected, and the repeatability of any aberrant results.
Systemic Assay Error
If there have been concerns about a specific test for more than one patient, the problem likely involves the assay itself. Assay issues may be due to systemic or random errors. A look at recent QC data and all patient means helps determine whether a systemic bias has crept into an assay. Small but clinically significant shifts sometimes are difficult to see simply in QC material means or external quality assessment results.
Another approach to assess whether there has been a clinically significant shift in the distribution of results is to evaluate the proportion of tests with results above and below the normal ranges before and after the complaints began. Clinicians often will notice and comment on a subtle increase in the number of flagged results even when laboratory QC doesn’t reveal a shift. Some laboratories have developed methods to monitor patient means and/or the proportion of abnormal patient results to assist in detecting these issues early.
Random Assay Error
Random error such as carryover or instrument errors—often in the pipetting or washing functions—also affects assays. These random errors can affect more than one assay on a platform, but may be difficult to detect on routine QC as the problem may affect QC results only infrequently. If operators simply repeat out of bounds QC rather than take note or troubleshoot the problem, the repeat likely will be acceptable and the error may go overlooked. In contrast, random errors are easy to find if a single result is in question. In this case, repeating the analysis on the original sample will provide significantly different results if the original analysis was affected by a random error such as carryover. Delta flags also alert to physiologically questionable changes possibly caused by random error.
Assay Interface Issues
If a repeat analysis on the original sample provides the same result and only a single sample appears to be affected, the issue probably doesn’t lie solely with the assay or platform. In this circumstance, the next step is to repeat the analysis using an alternate method. The available alternate method will depend on the test in question. Ideally the lab will be able to use an entirely different assay method or platform; doing so might require sending the sample in question to a different laboratory.
If the alternative method produces a starkly different result, the issue may lie specifically at the interface between the sample and the assay used. This problem surfaces most often in immunoassay methods, with interferences inherent in the sample such as heterophile antibodies or a macroprotein complex. It also affects some spectrophotometric methods if the sample contains colorimetric interferents, such as certain medications. The investigation into these issues will depend on the specific assay in question and the suspected problem.
If a second independent method returns results consistent with the original analysis, the problem most likely involves just the sample itself. Of course, one explanation might be that there was no problem at all, meaning the result was analytically accurate. In discussing this issue with the clinician, further thought should be given to the patient’s clinical presentation. Alternatively, there might have been a preanalytical problem involving sample labeling, collection, or handling that has affected the sample in question. Recalling the patient and collecting a fresh sample for analysis is the final step in this case.
Janet Simons, MD, FRCPC, is a medical biochemist at St. Paul’s Hospital in Vancouver, British Columbia. +Email: Janet.Simons@providencehealth.bc.ca