Construction of a consistency evaluation system for a nucleic acid testing kit: a preliminary study

Objective To evaluate the performance of a blood nucleic acid testing kit, construct a consistency assessment system, and analyze the influencing factors of the test results. Methods A total of 18 644 samples of voluntary blood donors in our center were collected, and 189 double positive samples of hepatitis B surface antigen (HBsAg) by ELISA were eliminated, and the remaining 18 455 samples were tested for HBV DNA in parallel with the test reagent and the control reagent (Roche Company). Samples with inconsistent results were confirmed with third-party reagents (Grifols Company, with the sensitivity of HBV DNA, HCV RNA and HIV RNA of 4.5 IU/mL, 2.4 IU/mL and 17.3 IU/mL by single reagent, which were significantly better than other reagents), and the consistency between the two methods was compared. Results No HCV RNA or HIV RNA positive samples were detected in this study. Among the 18 455 samples in initial test, 12 were positive with consistent results, 18 378 were negative with consistent results, and 65 were with inconsistent results between the test reagent and the control reagent, with positive concordance rate, negative concordance rate and concordance rate of the two reagents at 85.17%, 99.66% and 99.65%, respectively. In order to further clarify the infection of 65 inconsistent samples, five qualitative tests of hepatitis B and single NAT were conducted to comprehensively evaluate the infection. The results showed that there were 60 positive samples with consistent results, 18 371 negative samples with consistent results, and 24 samples with inconsistent results between the test reagent and the control reagent, with positive concordance rate, negative concordance rate and the concordance rate for the two reagents at 86.96%, 99.92% and 99.87%, respectively. The Kappa value was 0.83, and the Youden index was 0.40. The ROC curve was plotted with the false-positive rate (1-specificity) as the horizontal coordinate and the true-positive rate (sensitivity) as the vertical coordinate, and the area under the ROC curve was 0.705. 189 double-positive samples were selected for conformance testing, and the Ct values of the test reagent and the control reagent were analyzed and fitted linearly, with correlation coefficient r=0.972. A Bland-Altman model was constructed with the mean value of the logarithm of the quantitative values of the test reagent and the control reagent as the horizontal coordinate and the difference as the vertical coordinate, and 96.30% of the difference values of the test reagents were within the confidence interval. Conclusion The construction of the consistency evaluation system showed that the test reagent has similar detection efficiency with the control reagent, and is suitable for HBV DNA routine screening, which can effectively ensure the safety and effectiveness of blood screening.