The Opinion titled “SARS-CoV-2 Serology: Much Hype, Little Data” (Clin Chem 2020;66:875–7), published in the July 2020 issue of Clinical Chemistry, contains errors. Figure 1 contains the following minor mathematical errors, none of which affect the interpretation of the figure itself.
In the example population, an error was made when calculating the PPV, increasing the PPV from 94.6% to 96.1% in the assay with 99% specificity.
The number of true positives and false negatives in the high prevalence population was also miscalculated for all three specificities, increasing the PPV in all groups by 0.5-0.7%.
In the low prevalence population, the number of false positives was erroneously entered as 36.7 as opposed to 30.6 in the 99.5% specificity group, but the PPV was correct.
The number of true negatives in the low prevalence, 95% specificity group was miscalculated as 5,948k instead of 6,008.2k, increasing the PPV from 3.2% to 4.7%.
The corrected figure is provided below. The authors regret the errors.
Fig. 1.
Effect of prevalence and sensitivity on positive predictive value.
Shown are the calculated positive predictive values (PPV) based on the incidence of a population. True negative (TN), false negative (FN), true positive (TP), and false positive (FP). Black portions of the pie charts on the center and right image indicate the estimated prevalence based on PCR-confirmed COVID-19 cases in the respective regions.