COVID-19 research briefs: Tests for diagnosing SARS-CoV-2 lack sensitivity

Reference

Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in false-negative rate of reverse transcriptase polymerase chain reaction–based SARS-CoV-2 tests by time since exposure. Ann Intern Med 2020;173(4):262-267.

Synopsis

Research Brief #64: These researchers identified 7 published studies that evaluated the accuracy of reverse transcriptase polymerase chain reaction (RT-PCR) for the detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) at various points in time in relation to infection and symptom onset. They then fitted a Bayesian hierarchical logistic regression model using these data (the hierarchical part is necessary as they had 7 different sets of data). The authors estimated a median duration from infection to symptom onset of 5 days, and a pretest probability of infection of 11% based on studies of close household contacts in China. They then graphed the estimated rate of false negative results by day from infection. It begins at 100% (all infected cases are missed), then declines beginning on day 4 to 70%, then to approximately 40% at symptoms onset, bottoming out at 20% for the first 5 days of symptoms, then rising steadily to 50% by day 21. The Bayesian part of their model uses that pretest probability of 11% to estimate the post-test probability of infection given a negative rtPCR. That probability goes as low as about 3%, rising to 5% on day 21. The bottom line: Timing matters, and if a patient is clinically highly suspicious for COVID-19, repeat the test a day later, considering the false negative rate was lowest 8 days after exposure. On the other hand, PCR is highly specific, so if the result is positive, don't repeat the test to confirm it. You are likely to be misled by a false negative result if that second test result is negative. Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in false-negative rate of reverse transcriptase polymerase chain reaction–based SARS-CoV-2 tests by time since exposure. Ann Intern Med 2020;173(4):262-267. doi:https://doi.org/10.7326/M20-1495. Research Brief #65: We mostly know the Cochrane Collaboration for writing systematic reviews of the effectiveness of interventions, but they have recently begun to dip their toes in the water of diagnostic meta-analyses. This systematic review identified published and preprint studies that compared a rapid antigen test or a point-of-care molecular test to a reference laboratory rtPCR test as the reference standard. In a few studies, clinical criteria were used to define a positive case. Specificity was evaluated using pre-pandemic stored samples. The news is not good for rapid antigen tests, with a mean sensitivity of 56.2% (95% CI 29.5% - 79.8%) and a mean specificity of 99.5% (98.1% - 99.9%). Given a pretest probability of 5% (for example, a 5% prevalence in a hypothetical sample of 1000 patients), the corresponding positive predictive value (PV+) is 85% and negative predictive value (PV-) is 98%. If pretest probability is 10%, corresponding PV+ and PV- are 92% and 95%, respectively. Rapid molecular tests performed better, with a mean sensitivity of 95.2% (86.7% - 98.3%) and a mean specificity of 98.9% (97.3% - 99.5%). The PV+ of these tests was very similar to the rapid antigen tests over a range of pretest probabilities from 5% to 15%, but the PV- was much better (99% - 100%). With regard to individual point-of-care molecular tests, the Abbott ID Now was less sensitive than the Cepheid Xpert Xpress (76.8% vs 99.4%), but was slightly more specific (99.6% vs 96.8%). At a 10% pretest probability, that corresponds to a PV+ of 96% for Abbott and 77% for Cepheid, but a PV- of 97% for Abbott and 100% for Cepheid. Importantly, the Abbott ID Now will miss approximately 25% of infected persons based on its sensitivity. This test was used in the White House; the correct strategy would've been to first test using the highly sensitive Cepheid test, and then confirm any positives with the highly specific Abbott test. But they didn't ask me. Dinnes J, Deeks J, Adriano A, et al, for the Cochrane COVID-19 Diagnostic Test Accuracy Group. Rapid, point‐of‐care antigen and molecular‐based tests for diagnosis of SARS‐CoV‐2 infection. Cochrane Database Syst Rev 2020;8:CD013705. doi.org/10.1002/14651858.CD013705.