Escaping the COVID-19 Testing Paradox

6 Pages Posted: 8 Jun 2020

See all articles by Arjun Manrai

Arjun Manrai

Computational Health Informatics Program, Boston Children's Hospital; Department of Biomedical Informatics, Harvard Medical School

Kenneth D. Mandl

Boston Children's Hospital - Computational Health Informatics Program

Date Written: May 28, 2020

Abstract

As with all diagnostic tests, PCR and serology tests are imperfect, sometimes producing false negatives and sometimes false positives. While the sensitivity and specificity of these tests can be measured, their positive and negative predictive values are inextricably linked to the cumulative incidence of the disease, which remains unknown. Paradoxically, a false belief in a much lower prior probability of disease, now called into question by population-based studies, may lead to underestimating predictive value of serology testing. In order to implement testing strategies that will underpin opening up the economy, regular and repeated measurement on both individuals and populations is needed. Strategies using multiple tests can improve predictive value.

Note: Funding: Supported by grant U01HL121518 from the NIH/NHLBI.

Declaration of Interest: KDM is on the Scientific Advisory Board of Medal, Inc. KDM's program at Boston Children’s Hospital receives sponsored research support and philanthropy from Quest Diagnostics. AKM is a co-founder and adviser of XY.ai.

Keywords: COVID-19, infectious disease, positive predictive value, serological testing

Suggested Citation

Manrai, Arjun and Mandl, Kenneth D., Escaping the COVID-19 Testing Paradox (May 28, 2020). Available at SSRN: https://ssrn.com/abstract=3612991 or http://dx.doi.org/10.2139/ssrn.3612991

Arjun Manrai

Computational Health Informatics Program, Boston Children's Hospital ( email )

401 Park Drive
Landmark 5th Floor East
Boston, MA 02115
United States

Department of Biomedical Informatics, Harvard Medical School ( email )

10 Shattuck St
Boston, MA 02115
United States

Kenneth D. Mandl (Contact Author)

Boston Children's Hospital - Computational Health Informatics Program ( email )

300 Longwood Avenue
LANDMARK 5506, MAIL STOP BCH3187
Boston, MA n/a 02215
United States
6173554145 (Phone)
02115 (Fax)

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
67
Abstract Views
1,292
rank
423,275
PlumX Metrics