Assessing the performance of EWMA and CUSUM algorithms in patient-based real-time quality control of electrolytes

Patient-based real-time quality control (PBRTQC) represents a useful complement to traditional internal quality control methods, particularly in the continuous monitoring of routine analyses. Sodium, potassium, chloride, calcium, phosphorus, and magnesium are critical for patient safety. This study aimed to evaluate the performance of the exponentially weighted moving average (EWMA) and cumulative sum (CUSUM) algorithms in patient-based quality control of electrolytes in clinical biochemistry.

A total of 6.000 consecutive patient results obtained over 30 days using the Roche Cobas platform were analysed. EWMA and CUSUM algorithms were applied to the individual analytical distributions. The performance of each method was compared in terms of sensitivity, error detection rate, and false alarm frequency. Statistical analyses were performed using QC Constellation software.

For Na, K, and Ca, both EWMA and CUSUM achieved 100 % sensitivity with ANPed values of 2-15. EWMA performance for Cl, Mg, and P was low (sensitivity 3.3-23.3 %), while CUSUM demonstrated high sensitivity (96.7-100 %) but longer detection delays (37-86 patient results). Minimum-Average Number of Patient results affected until error detection (ANPed) optimisation accelerated detection for all analytes but reduced specificity, particularly for P (from 100 % to 66.7 %).

CUSUM a patient outcome monitoring approach based on as a cumulative PBRTQC, enabled rapid and accurate detection of analytical errors in electrolyte analyses. While CUSUM offers high sensitivity and specificity in most parameters, the optimised EWMA has provided additional benefits in some tests. The findings suggest that analyte-specific PBRTQC strategies can enhance real-time quality assurance and thereby improve clinical decision safety.