Assessing subsets of analytes in context of detecting laboratory errors

Area of Science:

• Clinical chemistry
• Laboratory medicine
• Medical diagnostics

Background:

• Laboratory error detection is vital for patient care but current quality controls often fail to detect pre-analytic errors.
• Population- and patient-based detection methods are being developed to address these limitations.
• The optimal set of analytes for efficient error detection remains unclear.

Purpose of the Study:

• To identify the most effective analytes for detecting laboratory errors.
• To evaluate the impact of analyte subset size on error detection accuracy.
• To assess the utility of specific analytes in identifying erroneous measurements in renal kidney disease patients.

Main Methods:

• Utilized three distinct scoring functions to rank analytes based on their ability to distinguish erroneous measurements.
• Analyzed a dataset from renal kidney disease inpatients.
• Calculated the joint likelihood of top-performing analytes for error detection.

Main Results:

• Calcium, potassium, and sodium were identified as the top three indicators of erroneous measurements.
• Larger subsets of analytes did not consistently improve error detection accuracy.
• An Area Under the Curve (AUC) of 0.73 was achieved for error detection using the joint likelihood of calcium, potassium, and sodium.

Conclusions:

• Calcium, potassium, and sodium are highly effective indicators for detecting laboratory errors in the context of renal kidney disease.
• The study provides a method for ranking analytes to enhance laboratory error detection systems.
• Optimizing analyte selection, rather than simply increasing their number, is key to improving the accuracy of error detection.