Managing Post-analytical Errors


Vol. 19 • Issue 20 • Page 42
Laboratory errors have a reported frequency of 0.012% to 0.6% of all test results.1Clearly, at least some of these errors will have a major impact on patient care.

Laboratory errors can be divided into three groups:

1. Pre-analytical errors. These are errors that occur due to patient misidentification, improper sample, improper sample handling and the amount of time to deliver the sample to the laboratory. These account for the vast majority of what we shall term laboratory errors.

2. Analytical errors. These occur once the sample has been logged into the lab and during the analysis of the sample and are usually due to instrument errors, reagent issues, calibration errors and errors in the interpretation of external control data or the instrument self-checks. Although studies have shown these to be quite rare, some have a profound effect when they do occur.

3. Post-analytical errors. These errors occur once the instrument has processed the sample. Some liberty has been taken with the term “post-analytical” as errors that may have occurred in the pre-analytical or analytical phase but weren’t detected until after the analytical run is complete.

figure

Click to view larger table

A Closer Look

Here, we’ll concentrate on post-analytical errors. Although uncommon (more so than analytical and far fewer than pre-analytical errors), these errors don’t necessarily get the attention they need. It has recently been demonstrated that even with new technologic facilities (e.g., online connection between laboratory and wards and between laboratory to physician’s office), the frequency of post-analytical errors can worsen rather than improve the communications between laboratories and clinicians without proper organization.2The Table lists some of the sources of post-analytical error.

Monitors to Avoid Errors

Once the sample has been processed by the instrument (analyzed) and a result is available, monitors should be in place to catch an error before it is reported, including:

External QC data. Most laboratories analyze two or more external controls at the start of the morning run. To be successful as monitors, a person must pay attention to any out-of-control conditions when they are announced by the control. Another concern regarding the external control is that even whilst the values from them are within the limits set by the laboratory, it is possible for the instrument-reagent system to fail once the controls have been analyzed and post-analytical errors could be reported. There has not been a consensus on whether to repeat the analysis of controls during the run or during the day.

Instrument checks. Today’s instruments perform dozens to hundreds of checks on themselves, similar to what your automobile does. These checks are often also performed by the operator before the run begins and the external controls analyzed. Each new instrument that comes on the market has more of these built-in monitors. As with the external controls, when the instrument identifies an error (a situation that exceeds the limits), it is necessary that a person intervene to correct the problem or, again, errors can be reported.

Delta checks. A delta check is a flag delivered by the LIS/HIS/middleware to the instrument operator signaling a change in a patient’s value for a test between one time and another. The magnitude of the change and the time interval are set by the laboratory. For example, if on Monday Mary Smith’s hemoglobin was 11.2 and on Wednesday it was 9.7, the LIS/HIS/middleware would flag that result. Arguably delta checks are not purely post-analytical errors; they are included here as the delta check is reported during or at the end of the analytical run and can prevent an error (e.g., patient ID) from leaving the laboratory even though the error occurred during the pre-analytical phase. It is worth noting that delta checks can identify an instrument malfunction during a run if there are enough patient delta checks within that run.

Average of normals (AoN). Like delta checks, the AoN quality tool has been available for decades; however, it had to wait for computers and software to be routinely useful. The idea of this error detection tool is that the mean of those patient values with certain limits (e.g., “normal” range) should be reasonably stable and, like the external control, detect errors in the analytical system. Unlike the external controls, the AoN can detect errors that occur during the run. If, for example, the temperature drops during the run, many of the values on patient samples will be low and the average for that run will be lower than usual. Like delta checks, it could be argued that strictly speaking the AoN detects analytical errors. It could also be argued that since the AoN value occurs after (post) analysis and before results are sent out, it is a post-analytical error detector.

Protocol Necessities

Error detection and correction are imperative responsibilities of the laboratory. Today there are a number of aids to help the laboratory manager not only identify errors, but correct them. Having a protocol for applying these tools is as important to adhering to it. Even though the laboratory does an exemplary job in minimizing errors that affect patient safety, there will always be errors for us to eliminate.

David Plaut is a chemist and statistician in Plano, TX.

Click here for references.

About The Author