Are we using the right method to estimate the number of COVID-19 cases?

ABSTRACT

The great concern stemming from the current COVID-19 pandemic has been a challenge for governments and international organizations around the world. Likewise, the diagnosis has played an important role in the prioritization processes, in particular, for the identification and follow-up of cases. In this context, PCR and serological tests have become the international standard; however, some limitations, as well as the scope of these techniques, must be understood in terms of population numbers. This comment seeks to clarify the interpretation of the results of these tests from a public health perspective.

KEYWORDS:

• COVID-19
• Diagnosis
• Serological test
• Polymerase Chain Reaction

The pandemic caused by SARS-CoV-2 has been a great concern for worldwide public health. Governments in numerous countries have implemented mitigation measures (e.g. social distancing, lockdowns, and other restrictions) based on georeferenced disease data and diagnostic methods [1]. Therefore, it is crucial to evaluate the diagnostic accuracy of these methods. In this commentary, we discuss the effectiveness of serological tests that detect anti-SARS-CoV-2 immunoglobulins for estimating the true number of COVID-19 cases.

Clinically, the RT-PCR test has been used as the standard for diagnosing COVID-19. However, its shortcomings, such as the variability in its diagnostic accuracy depending on the stage of the disease, has driven the adoption of alternative methods. Serological tests are an alternative method for the diagnosis and surveillance of cases [1,2]. Compared with the RT-PCR, serological tests show high specificity rates and pooled sensitivity for IgG or IgM (66% [95% CI 49.3–79.3%] to 97.8% [46.2–100%]) [1].

Although seroprevalence of SARS-CoV-2 has been an alternative to determining and comparing the impact of COVID-19 at a local level among populations, it is possible that it has led to an underestimate of actual case numbers by a significant magnitude. We analyzed studies from Spain and Switzerland with a seroprevalence of 5% and 10.8%, respectively [3,4]. These figures are likely to include a non-negligible number of false results, leading to a difference in the total number of infected patients when comparing the seroprevalence as determined by RT-PCR positive rates with serological test rates at the same approximate date. Therefore, the adjusted number of cases was estimated using the serological test’s sensitivity of 84.3% (Table 1) [1]. In Switzerland, the ratio of positive results between IgM or IgG and RT-PCR was 4.46%, while in Spain it was 10.09% (Table 1).

Table 1. Estimates of IgM or IgG/RT-PCR ratio

Location	Seroprevalence	Sensitivity ELISA adjusted seroprevalence/0.843	Population	Adjusted number of cases expected
Switzerland	10.8%	12.81%	183,981	25,367
Spain	5%	5.93%	46,756,254	2,772,645

^aPopulation estimates by country and city, 2020 [6].

Minor differences or imprecisions in estimations can have significant consequences, as shown by the conversion of relative figures into absolute values. Indeed, in a population of one million inhabitants, a 1% false-negative rate indicates thousands of individuals with incorrect test results. Considering the speed of SARS-Cov-2 transmission (Ro = 3.54, 95% credible interval 3.40–3.67) [5] and the context of low-and middle-income countries with a shortage of health facilities and scarce access to RT-PCR tests, limited resources may seriously compromise the quality, timeliness, and availability of health services, amplifying the real impact of the pandemic.

Furthermore, IgG and IgM detection and the measures for controlling virus dissemination should not rely on serological tests because people who test negative for SARS-CoV-2 antibodies are still potential transmission vectors. Although serological tests play a role in estimating the number of cases, they should be used carefully for diagnosis and active case finding. Future research needs to be more specific about how data are compiled, clearly stating the precision of tests used, as well as the case number estimates.

Disclosure statement

No potential conflict of interest was reported by the authors.