Comparison of molecular testing strategies for COVID-19 control: a mathematical modelling study

Abstract

Background: WHO has called for increased testing in response to the COVID-19 pandemic, but countries have taken different approaches and the effectiveness of alternative strategies is unknown. We aimed to investigate the potential impact of different testing and isolation strategies on transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

Methods: We developed a mathematical model of SARS-CoV-2 transmission based on infectiousness and PCR test sensitivity over time since infection. We estimated the reduction in the effective reproduction number (R) achieved by testing and isolating symptomatic individuals, regular screening of high-risk groups irrespective of symptoms, and quarantine of contacts of laboratory-confirmed cases identified through test-and-trace protocols. The expected effectiveness of different testing strategies was defined as the percentage reduction in R. We reviewed data on the performance of antibody tests reported by the Foundation for Innovative New Diagnostics and examined their implications for the use of so-called immunity passports.

Findings: If all individuals with symptoms compatible with COVID-19 self-isolated and self-isolation was 100% effective in reducing onwards transmission, self-isolation of symptomatic individuals would result in a reduction in R of 47% (95% uncertainty interval [UI] 32-55). PCR testing to identify SARS-CoV-2 infection soon after symptom onset could reduce the number of individuals needing to self-isolate, but would also reduce the effectiveness of self-isolation (around 10% would be false negatives). Weekly screening of health-care workers and other high-risk groups irrespective of symptoms by use of PCR testing is estimated to reduce their contribution to SARS-CoV-2 transmission by 23% (95% UI 16-40), on top of reductions achieved by self-isolation following symptoms, assuming results are available at 24 h. The effectiveness of test and trace depends strongly on coverage and the timeliness of contact tracing, potentially reducing R by 26% (95% UI 14-35) on top of reductions achieved by self-isolation following symptoms, if 80% of cases and contacts are identified and there is immediate testing following symptom onset and quarantine of contacts within 24 h. Among currently available antibody tests, performance has been highly variable, with specificity around 90% or lower for rapid diagnostic tests and 95-99% for laboratory-based ELISA and chemiluminescent assays.

Interpretation: Molecular testing can play an important role in prevention of SARS-CoV-2 transmission, especially among health-care workers and other high-risk groups, but no single strategy will reduce R below 1 at current levels of population immunity. Immunity passports based on antibody tests or tests for infection face substantial technical, legal, and ethical challenges.

Funding: UK Medical Research Council.

Figure 1

Infectiousness of SARS-CoV-2 over time since infection estimated from the serial interval and the reduction of transmission as a result of self-isolation after symptoms or a PCR-positive test result (A) Detection of presymptomatic SARS-CoV-2 infection and subsequent reduction in transmission through self-isolation after a positive PCR test. (B) Detection of asymptomatic SARS-CoV-2 infection and subsequent reduction in transmission through self-isolation after a positive PCR test. The shaded areas in these plots illustrate infectiousness that would be limited by PCR testing (grey) at 4 days after infection or self-isolation following symptom onset (red) at 6 days after infection. The area under the curves is equal to the reproduction number. (C) Percentage reduction in the reproduction number by self-isolation following onset of symptoms as a function of the proportion of infections that are asymptomatic and their relative infectiousness compared with symptomatic infections. (D) Additional percentage reduction in the reproduction number by a policy of repeated PCR testing at regular intervals with different timeliness from sample collection to isolation, assuming a third of infections are asymptomatic and that they are 50% as infectious as symptomatic infections. SARS-CoV-2=severe acute respiratory syndrome coronavirus 2.

Figure 2

Effectiveness of test-and-trace strategies as a function of the time from test to quarantine of contacts and coverage Coverage is defined as the proportion of symptomatic infections identified for contact tracing and the proportion of contacts successfully traced, assumed equal in these plots. The percentage reduction in the reproduction number of contacts for test-and-trace strategies (A) or test-trace-test strategies (B), assuming self-isolation on the basis of symptoms is already in place. This reduction is approximately equal to the reduction in the overall reproduction number if contacts who develop COVID-19 symptoms are also eligible to be an index case for further contact tracing.

Figure 3

Sensitivity and specificity of currently available antibody tests The circles indicate the reported test performance for different test platforms, with the size of the circle proportional to the total number of control samples tested and the colour indicating the test format. The lines indicate thresholds for test performance required to provide a 95% probability of correctly predicting the presence of antibodies (positive predictive value) for a prevalence of 5% (black line) or 25% (grey line) among those tested. Tests to the left of these lines meet this performance standard, although further clinical sample testing is required to confirm their specificity. The x-axis shows 1-specificity plotted with a square root transform to better show the high specificity threshold required. Only tests for IgG, or IgG plus either IgM or IgA, are included. Data are from the Foundation for Innovative New Diagnostics.