Setting minimum clinical performance specifications for tests based on disease prevalence and minimum acceptable positive and negative predictive values: Practical considerations applied to COVID-19 testing

Tze Ping Loh, Sarah J. Lord, Katy Bell, Mary Kathryn Bohn, Chun Yee Lim, Corey Markus, Hernan Fares Taie, Khosrow Adeli, Giuseppe Lippi, Sverre Sandberg, Andrea Horvath

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Objectives: Several guidelines for the evaluation of laboratory tests for severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) infection have recommended establishing an a priori definition of minimum clinical performance specifications before test selection and method evaluation. Methods: Using positive (PPV) and negative predictive values (NPV), we constructed a spreadsheet tool for determining the minimum clinical specificity (conditional on NPV or PPV, sensitivity and prevalence) and minimum clinical sensitivity (conditional on NPV or PPV, specificity and prevalence) of tests. Results: At a prevalence of 1%, there are no minimum sensitivity requirements to achieve a desired NPV of 60%-95% for a given clinical specificity above 20%. It is not possible to achieve 60–95% PPV even with 100% clinical sensitivity, except when the clinical specificity is near 100%. The opposite trend is seen in high prevalence settings (60%), where a relatively low minimum clinical sensitivity is required to achieve a desired PPV for a given clinical specificity, and a higher minimum clinical specificity is required to achieve a desired NPV for a given clinical sensitivity. Discussion: The selection of laboratory tests and the testing strategy for SARS-CoV-2 involves delicate trade-offs between NPV and PPV based on prevalence and clinical sensitivity and clinical specificity. Practitioners and health authorities should carefully consider the clinical scenarios under which the test result will be used and select the most appropriate testing strategy that fulfils the a priori defined clinical performance specification.

Original languageEnglish
Pages (from-to)18-22
Number of pages5
JournalClinical Biochemistry
Volume88
DOIs
Publication statusPublished - Feb 2021
Externally publishedYes

Keywords

  • Evaluation
  • False negative
  • False positive
  • Negative predictive value
  • Performance specification
  • Positive predictive value
  • Prevalence
  • Sensitivity
  • Specificity

Fingerprint

Dive into the research topics of 'Setting minimum clinical performance specifications for tests based on disease prevalence and minimum acceptable positive and negative predictive values: Practical considerations applied to COVID-19 testing'. Together they form a unique fingerprint.

Cite this