Error rates in SARS-CoV-2 testing examined with Bayes' theorem

Area of Science:

• Virology
• Epidemiology
• Medical Diagnostics

Background:

• The COVID-19 pandemic necessitates widespread SARS-CoV-2 testing for transmission control.
• Field performance of diagnostic tests, like RT-PCR, may differ from laboratory conditions.
• Accurate quantification of test error rates is crucial for public health strategies.

Purpose of the Study:

• To evaluate the accuracy of SARS-CoV-2 testing under real-world conditions.
• To develop methods for correcting bulk test results and assess societal testing needs.
• To determine the limitations of current and future diagnostic tests for clearing infection.

Main Methods:

• Literature review of SARS-CoV-2 reverse-transcription polymerase chain reaction (RT-PCR) studies.
• Construction of a clinical test confusion matrix.
• Analysis of test sensitivity and specificity requirements for various use cases.
• Sequential analysis of common testing scenarios.

Main Results:

• In some regions, a significant proportion of mild symptomatic individuals with positive test results may not be infected.
• Current and anticipated diagnostic tests are insufficient to definitively clear individuals of SARS-CoV-2 infection.
• Large-scale testing for infection tracing may not be viable in all regions due to laboratory capacity.

Conclusions:

• Regional authorities must monitor operational test characteristics before implementing large-scale testing programs.
• RT-PCR should not be the sole gold standard for scaled SARS-CoV-2 diagnosis; clinical assessment and diverse diagnostic tests are essential.
• Focused, capacity-aware testing strategies are recommended over broad infection tracing in resource-limited settings.