Cochrane on COVID-19: Ig tests best at least 3 weeks after symptom onset

  • Deeks JJ & al
  • Cochrane Syst Rev
  • 26 Jun 2020

  • curated by Liz Scherer
  • Clinical Essentials
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.

Takeaway

  • Antibody tests for SARS-CoV-2 have low sensitivity in the first week post-symptom onset but exceed 90% by week 3.
  • Antibody response duration remains unknown, and the value of testing in patients with mild or asymptomatic infections is unknown.
  • Future research should focus on desegregated sensitivity data (by time since symptom onset), independent evaluations.

Why this matters

  • Antibody testing might confirm SARS-CoV-2 exposure in asymptomatic patients who have not been RT-PCR-tested or with negative RT-PCR test results.

Key results

  • 54 studies (38 from Asia), 8526/15,976 cases of COVID-19.
  • Pooled IgG, IgM, and IgG/IgM sensitivity values (95% CIs) show the same pattern across weeks 1-3:
    • IgG:
      • Week 1: 29.7% (22.1%-38.6%).
      • Week 2: 66.5% (57.9%-74.2%).
      • Week 3: 88.2% (83.5%-91.8%).
    • IgM:
      • Week 1: 23.2% (14.9%-34.2%).
      • Week 2: 58.4% (45.5%-70.3%).
      • Week 3: 75.4% (64.3%-83.8%).
    • IgG/IgM:
      • Week 1: 30.1% (21.4%-40.7%).
      • Week 2: 72.2% (63.5%-79.5%). 
      • Week 3: 91.4% (87.0%-94.4%).
  • Specificity, week 3 (95% CIs):
    • IgG: 99.1% (98.3%-99.6). 
    • IgM: 98.7% (97.4%-99.3%). 
    • IgG/IgM: 98.7% (97.2%-99.4%).
  • No studies included direct assessment of seroprevalence.
  • Data were too limited for sensitivity estimates >35 days post‐symptom onset.
  • Authors found no "convincing differences in accuracy" among different tests. 

Study design

  • Cochrane review evaluating accuracy of antibody tests for determining SARS-CoV-2 exposure.
  • Funding: Cochrane Collaboration; National Institute for Health Research.

Limitations

  • Low-quality studies; preprint dominance.
  • Publication bias.
  • Cross-study comparisons lacking.
  • Small sample sizes.