Failure of internal quality control in detecting significant reagent lot shift in serum creatinine measurement

Abstract Background Internal quality control (IQC) in clinical laboratories is carried out to monitor analytical stability. Usually, the satisfactory results of the IQC ensure the acceptability of the examination results. Here, we reported that patients' creatinine results are unreliable, although the internal quality control is satisfactory. Methods Creatinine levels were analyzed from two quality control materials and twenty patients' specimens using two different lots of reagents. Lot‐to‐lot comparison was performed. The daily median values of serum creatinine levels of patients were calculated from the test results recorded in our laboratory information system. Results Although IQC was consistent, serum creatinine concentrations were higher using lot B (median: 153 μmol/L; interquartile range: 122‐522 μmol/L) than using lot A (median: 133 μmol/L; interquartile range: 76‐508 μmol/L) for 20 patients (P = .001). The Deming linear regression showed a best fit of y = 0.9394 × x + 45.66. R 2 = .8919, and mean percentage difference between two lots was 34%. The new lot was considered unacceptable. Likewise, the median serum creatinine level from the 360 patients using lot B was 102 μmol/L, which was significantly higher than the daily medians of patients using lot A (median: 66 μmol/L; range: 61‐70 μmol/L) in the previous month. Conclusion The variations in creatinine concentrations proved to be due to different lots of reagents. However, IQC materials tested using both lots of reagent exhibited minimal variation. Therefore, IQC alone is insufficient for assessing laboratory analytical results. This finding prompts us to be vigilant in potential pitfall of interpreting test results based on satisfactory IQC alone.

risk management tool within the total testing pathway (TTP) that contributes to the overall objective of assuring the quality of results produced in medical laboratories. 5 IQC is primarily utilized in routine practice to monitor system performance (ie, make comparisons to what is expected under stable conditions), and allow analytical failures that affect performance to be detected. 5 Reliable tests are contingent on passing both IQC and external quality assessment (EQA).
Here, we report an incident in which a batch of creatinine test results was found to be unreliable due to a change in reagent lots, even though IQC results were satisfactory.

| Experimental design
Twenty patients' serum specimens with serum creatinine concentrations that span the reportable range of the test method were obtained. The serum creatinine was tested with two reagent lots (hereinafter referred to as lot A and lot B). Samples were tested within 4 hours after collection and were stored at room temperature before tested. After each lot reagents had been changed, the creatinine was recalibrated and IQC was performed. Lot-to-lot differences in creatinine results were compared.
Consecutive patients' serum creatinine data from October 16 to November 15, 2017, were extracted from the Laboratory Information System (Neusoft), which were assayed using lot A. The median of the 360 patients' results that had been measured with the new lot on November 16 was calculated and compared with daily medians using lot A. All these data were from patients, including outpatient clinics and hospital wards.

| Methods and equipment
Serum creatinine was measured by an enzymatic method on a Beckman Coulter AU5821 biochemical analyzer, using reagents and calibrators (lot 20170222 and lot 20170422) from Kehua Biological Products Co., Ltd, Shanghai. IQC materials were third-party controls from Bio-Rad Laboratories (lot 26400). To assess whether creatinine results are in control, the Westgard multiplication rules (1 3s , 2 2s , R 4s , 4 1s ) are used.

| Statistical analysis
Statistical analyses were performed with SPSS 19.0. The S-W test was used to evaluate the normality of distribution. Wilcoxon signedrank test was used for significance testing between groups of continuous data (data are not normally distributed). P values <.05 were considered statistically significant.
Lot-to-lot comparison was performed with the Deming regression analysis for estimation of the slope and intercept. Difference between paired samples, and mean percentage difference between the results obtained with the two reagent lots were evaluated too.

| Lot-to-lot differences
Serum creatinine levels of patients tested varied between the two reagent lots are shown in Table 1. The creatinine levels measured using two lots showed significantly different (P = .001). Although IQC was consistent and satisfactory, serum creatinine concentrations meas-  Table 1). The variation was remarkably higher (>10%, 10 of 11) for the specimens with creatinine concentration below 150 μmol/L.  lot B R1 B R2 were both obviously higher than those of lot A R1 A R2 when creatinine was less than 150 μmol/L (both P values were .003). The difference in creatinine results measured between lot A R1 B R2 and lot A R1 A R2 was not statistically significant (P = .050). Nevertheless, the quality control results measured using lot B R1 A R2 , lot B R1 B R2, and lot A R1 B R2 were satisfactory and showed no difference with using lot A R1 A R2 .

| Medians of patients' data
The daily medians of consecutive patients' creatinine results (analyzed using lot A) from October 16 to November 15, 2017, are shown in Figure 3.
The median number of daily serum creatinine tests on the instrument was 714 (range: 298-954). The median of daily medians was 66 μmol/L (range: 61-70 μmol/L). Figure 3 shows that the median of the 360 patients' creatinine concentrations analyzed using lot B was 102 μmol/L (the last dot point in Figure 3), which was significantly higher than the daily medians of patients' results using lot A in the previous month. able results for QC materials was frequent enough that the QC results could not be used to verify the consistency of results for patient samples when changing lots of reagents. 6 The results showed that IQC worked as expected in spite of big variations in specimen test results between two lots of reagent. The verification of new reagent lot performance is not only a routine but also an important laboratory task. 7 The Clinical and Laboratory Standards Institute (CLSI) EP26-A guideline provides a lot-to-lot verification protocol to detect significant changes in test performance. 7 Our laboratory's protocol for lot-to-lot verification consists of simultaneously testing five samples from both current and new lots. The relative difference for each sample is calculated. The new reagent lot is deemed acceptable only if the relative difference in at least four samples was less than the predefined rejection limit. Even so, verification of the new lot was overlooked by the technician and caused significant variations in test TA B L E 1 Lot-to-lot reagent differences for serum creatinine (unit: μmol/ L, rejection limit ± 12%) results. Fortunately, the issue was found before the patients' test reports were issued, and serum creatinine concentrations of these 360 patients were reanalyzed using the old reagent.

| D ISCUSS I ON
There are several reports on the use of patient results as a tool in monitoring analytical quality on the daily internal control. 3 with the reagent problems, the daily median of patient outcomes had changed significantly while IQC results were satisfactory, which suggested that the daily median could be a good complement to IQC. F I G U R E 3 Comparison of the median of serum creatinine results using lot B reagent with the daily medians using lot A in the previous month. The daily medians using lot A reagent (square points) ranged from 61 μmol/L to 70 μmol/L. The median of the data measured with lot B reagent was 102 μmol/ L (dot point), which was significantly higher than the daily medians using lot A reagent

E TH I C A L A PPROVA L A N D PATI E NT CO N S E NT
Ethical clearance for this study was obtained from the Ethics Committee at the First Affiliated Hospital of Nanjing Medical University. Because all the samples used in this study were collected from clinical residual specimen, written informed content from each patient was waived.