Working Paper: NBER ID: w27780
Authors: Joshua S. Gans
Abstract: The most commonly used test for the presence of SARS-CoV-2 is a PCR test that is able to detect very low viral loads and inform on treatment decisions. Medical research has confirmed that many individuals might be infected with SARS-CoV-2 but not infectious. Knowing whether an individual is infectious is the critical piece of information for a decision to isolate an individual or not. This paper examines the value of different tests from an information-theoretic approach and shows that applying treatment-based approval standards for tests for infection will lower the value of those tests and likely causes decisions based on them to have too many false positives (i.e., individuals isolated who are not infectious). The conclusion is that test scoring be tailored to the decision being made.
Keywords: SARS-CoV-2; COVID-19; testing; infectiousness; infection
JEL Codes: D81; I18
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| sensitivity of tests for infection (C12) | likelihood of correctly identifying infectious individuals (C92) |
| time lag in obtaining PCR results (C41) | missed window of infectiousness (C41) |
| speed of testing (C91) | efficacy of public health responses (H12) |
| scoring for tests based on infection (I12) | higher rate of false positives (C52) |
| higher rate of false positives (C52) | unnecessary isolations (Y50) |
| test scoring should align with intended public health decision (C52) | reduced unnecessary isolations (I14) |