The time to do serosurveys for COVID-19 is now.

In their study in The Lancet Respiratory Medicine, Scott Pallett and colleagues assessed the performance of lateral flow serological assays and estimated the prevalence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection to be 10·6% (95% CI 7·6–13·6) in asymptomatic health-care workers and 44·7% (42·0–47·4) in symptomatic health-care workers at two hospitals in London, UK. This work raises important issues in the design of seroprevalence surveys, how they should be done, and, importantly, how to interpret and act on the results.

Seroprevalence studies, serosurveys for short, are important for determining the true extent of an outbreak, map its distribution, and identify hotspots and at-risk groups, such as health-care workers and older people. For serosurveys to inform public health measures and control strategies, they must have high sensitivity and specificity. Of these attributes in a serology test, high specificity is crucial, to avoid misclassifying people as having been infected when they have not (ie, false-positive results). This would give a false sense of security to individuals and governments, misleading public health interventions by overestimating the level of immunity in the population, and prematurely easing restrictions. Therefore, establishing the accuracy of lateral flow serological assays before doing a serosurvey is crucial to generate confidence in the results. Relatively small variations in test specificity and the prevalence of SARS-CoV-2 infection can heavily influence results.

Two lateral flow serological assays were evaluated in this study: the Encode SARS-CoV-2 IgM/IgG One Step Rapid Test Device (Zhuhai Encode Medical Engineering, Zhuhai, China) and the Onsite CTK Biotech COVID-19 split IgG/IgM Rapid Test (CTK Bitotech, Poway, CA, USA). The Encode assay had a sensitivity (compared with PCR-confirmed cases of SARS-CoV-2 infection) of 93·4% (95% CI 87·8–96·9) and a specificity of 99·0% (94·6–100·0), whereas the Onsite assay had a sensitivity of 88·2% (81·6–93·1) and a specificity of 94·0% (87·4–97·8). Having established which test to use, the next questions are identifying the study population and estimating an adequate sample size for the results to be generalisable. Pallet and colleagues focused on health-care workers, an essential resource in a pandemic response, and a group that is at particular risk when caring for patients, especially when personal protective equipment might be in short supply. Infection rates in this group are expected to be higher than in the general population. At a seroprevalence of 10·6%, a test with 99·0% specificity would have a positive predictive value (PPV) of 91·7%, meaning around 8% of the results might be false positives. However, if the authors had used the test with 94·0% specificity to do the serosurvey, the PPV would be 63·6%, meaning approximately a third of the results would be false positives, leading to an overestimate of prevalence in health-care workers by almost a third. If these tests were used to do serosurveys in the general population (estimated prevalence 2·7%), then the Encode assay would have a PPV of 72·2% and the Onsite assay 29·0%, leading to unacceptably high rates of false-positive results.

So far, 58 serological tests of varying performance have received emergency use approval from the US Food and Drug Administration. How could serosurveys be carried out in the general population? In a study of COVID-19 in households in Geneva, cumulative seroprevalence increased from 4·8% in week 1 to 10·8% by week 5. The authors mitigated against the effect of false-positive results by confirming all positive and indeterminate results with another assay.

Moving forward, research on better tests for serosurveys should include non-invasive sampling and assays measuring protective immunity. Large-scale studies can be efficiently done on high-throughput immunoassay systems, but the drawback is that they need serum or plasma, requiring phlebotomists to collect blood samples and processing in the laboratory. Immunoassays on non-invasive samples, such as oral fluids, would be a game-changer.

Investment in research on the correlates of protection and its duration will enable the development of serology tests that allow those who test positive to safely return to work or school. Striking a balance between public health and economic interests has been notoriously difficult during the pandemic. The World Bank issued a policy brief describing how two tests can contain the COVID-19 pandemic and save the economy. Serosurveys to inform the design of chemoprophylaxis and vaccine trials are also needed to provide reliable estimates of the risk of infection in the target population and calculate the appropriate population sample to detect the desired effect size.

Changing testing strategies as the pandemic unfolds has made it very difficult for countries to estimate the proportion of the population that has been infected. Yet, this information is crucial for developing evidence-based strategies to adapt public health measures and travel restrictions. Furthermore, studies have shown that as much as 44% of COVID-19 transmission can take place when individuals are pre-symptomatic or asymptomatic. As most infections result in individuals developing antibodies against SARS-CoV-2 regardless of symptoms, seroprevalence studies are the most useful means of understanding the true prevalence of the pandemic, monitor trends and geographical distribution over time, identify hotspots and at-risk populations needing special attention, and ultimately allow infection control programmes to assess the effectiveness of interventions. With most of the world still in some form of lockdown, and the prospect of a vaccine more than a year away, governments need a strategy to ease restrictions while ensuring that the country has a healthy workforce for its hospitals and care homes, as well as a means of monitoring safe environments for health-care facilities and other workplaces, schools, and mass gatherings.

The time to do serosurveys is now, but we need to proceed with care.