A New Testing Strategy for the Diagnosis of COVID-19 and Similar Pandemics
Mehmet Zahid Koçak · Turkiye Klinikleri Journal of Biostatistics · 2020
Objective: COVID-19 forced the entire world to close all borders yet to work together to contain the pandemic. One dimension of this joint work is sharing medical supplies, promising medications, and diagnostic tools. From the start of the pandemic, one difficulty was to obtain reliable diagnostic tools for this new virus that produces the results in a reasonable time window. We bring a new angle to this strife to increase the performance of a given diagnostic test. Material and Methods: In this research, we worked on improving the performance of the RT-PCR (Real Time-Polymerase Chain Reaction) COVID-19 diagnostic test. By obtaining the number of tests conducted and number of positive COVID-19 cases reported by the Ministry of Health of Turkey, using the Bayes' Rule, we predicted the prevalence, the number of false positives and number of false negatives, and we proposed several new testing strategies to improve the COVID-19 test. Results: We first presented the single test results. Then we showed that strong negative testing strategies would control the false negative successfully while inflating the false positives. On the other hand, strong positive testing strategy controls the false positives very well while inflating the false negative significantly. Following these results, we have also shown that three-test consensus call strategy perform the best in controlling both false negative and false positive rate. Conclusion: To contain COVID-19 or similar epidemics or pandemics, we propose a three-test consensus call strategy, which finds a reasonable balance between false positives and false negatives.