Misclassification bias in estimating clinical severity of SARS-CoV-2 variants
LANCET(2022)
Abstract
Tommy Nyberg and colleagues1Nyberg T Ferguson NM Nash SG et al.Comparative analysis of the risks of hospitalisation and death associated with SARS-CoV-2 omicron (B.1.1.529) and delta (B.1.617.2) variants in England: a cohort study.Lancet. 2022; 399: 1303-1312Summary Full Text Full Text PDF PubMed Scopus (147) Google Scholar use an unvaccinated cohort to show differences between the intrinsic severity of the omicron (B.1.1.529) and delta (B.1.617.2) variants of SARS-CoV-2 without confounding by pre-existing immunity. They report an 80% reduction in the severity of the omicron compared with the delta variant, suggesting the possibility of living through the COVID-19 pandemic without social and economic disruptions. However, reliance on SARS-CoV-2 test positivity to identify cases of COVID-19 and on all-cause hospitalisations and deaths as outcomes could have introduced misclassification bias and residual confounding.Up to one in three SARS-CoV-2 infections are asymptomatic,2Sah P Fitzpatrick MC Zimmer CF et al.Asymptomatic SARS-CoV-2 infection: a systematic review and meta-analysis.Proc Natl Acad Sci USA. 2021; 118e2109229118Crossref PubMed Scopus (110) Google Scholar and this proportion was even greater during the omicron wave.3Garrett N Tapley A Andriesen J et al.High rate of asymptomatic carriage associated with variant strain omicron.medRxiv. 2022; (published online Jan 14.) (preprint).https://doi.org/10.1101/2021.12.20.21268130PubMed Google Scholar Studies that exclusively use test positivity as the case definition might report inflated hospitalisation and case-fatality rates. Misclassification is exacerbated by the higher prevalence of infection due to more transmissible variants and by the increased ratios of non-severe to severe cases, potentially attenuating the differences in severity between variants. In the appendix, we show the potential effects of three SARS-CoV-2 case phenotypes on apparent hospitalisation and case-fatality rates of SARS-CoV-2 infection with the delta and omicron variants. Misclassification could also differ by age, vaccination status, and comorbidities that influence susceptibility to infection and disease.4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar, 5Yek C Warner S Wiltz JL et al.Risk factors for severe COVID-19 outcomes among persons aged ≥18 years who completed a primary COVID-19 vaccination series—465 health care facilities, United States, December 2020–October 2021.MMWR Morb Mortal Wkly Rep. 2022; 71: 19-25Crossref PubMed Google ScholarThe use of other data streams might help to populate large datasets when clinical data are scarce or absent. For example, administrative coding could be used to identify reasons for hospital admission that are likely to be related (eg, pneumonia) or unrelated (eg, trauma) to COVID-19,4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar and to identify comorbid conditions for inclusion as covariates in comparative analyses.4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar, 5Yek C Warner S Wiltz JL et al.Risk factors for severe COVID-19 outcomes among persons aged ≥18 years who completed a primary COVID-19 vaccination series—465 health care facilities, United States, December 2020–October 2021.MMWR Morb Mortal Wkly Rep. 2022; 71: 19-25Crossref PubMed Google Scholar The delivery of therapeutics used specifically or most commonly for COVID-19 (eg, remdesivir and dexamethasone) could enrich for those hospitalised with the disease. Ultimately, applying a probabilistic approach to case definition might allow for estimates of confidence when identifying cases and associating outcomes.After correcting for misclassification bias, the intrinsic severity of the omicron variant of SARS-CoV-2 might be even lower than that suggested by Nyberg and colleagues.We declare no competing interests. Tommy Nyberg and colleagues1Nyberg T Ferguson NM Nash SG et al.Comparative analysis of the risks of hospitalisation and death associated with SARS-CoV-2 omicron (B.1.1.529) and delta (B.1.617.2) variants in England: a cohort study.Lancet. 2022; 399: 1303-1312Summary Full Text Full Text PDF PubMed Scopus (147) Google Scholar use an unvaccinated cohort to show differences between the intrinsic severity of the omicron (B.1.1.529) and delta (B.1.617.2) variants of SARS-CoV-2 without confounding by pre-existing immunity. They report an 80% reduction in the severity of the omicron compared with the delta variant, suggesting the possibility of living through the COVID-19 pandemic without social and economic disruptions. However, reliance on SARS-CoV-2 test positivity to identify cases of COVID-19 and on all-cause hospitalisations and deaths as outcomes could have introduced misclassification bias and residual confounding. Up to one in three SARS-CoV-2 infections are asymptomatic,2Sah P Fitzpatrick MC Zimmer CF et al.Asymptomatic SARS-CoV-2 infection: a systematic review and meta-analysis.Proc Natl Acad Sci USA. 2021; 118e2109229118Crossref PubMed Scopus (110) Google Scholar and this proportion was even greater during the omicron wave.3Garrett N Tapley A Andriesen J et al.High rate of asymptomatic carriage associated with variant strain omicron.medRxiv. 2022; (published online Jan 14.) (preprint).https://doi.org/10.1101/2021.12.20.21268130PubMed Google Scholar Studies that exclusively use test positivity as the case definition might report inflated hospitalisation and case-fatality rates. Misclassification is exacerbated by the higher prevalence of infection due to more transmissible variants and by the increased ratios of non-severe to severe cases, potentially attenuating the differences in severity between variants. In the appendix, we show the potential effects of three SARS-CoV-2 case phenotypes on apparent hospitalisation and case-fatality rates of SARS-CoV-2 infection with the delta and omicron variants. Misclassification could also differ by age, vaccination status, and comorbidities that influence susceptibility to infection and disease.4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar, 5Yek C Warner S Wiltz JL et al.Risk factors for severe COVID-19 outcomes among persons aged ≥18 years who completed a primary COVID-19 vaccination series—465 health care facilities, United States, December 2020–October 2021.MMWR Morb Mortal Wkly Rep. 2022; 71: 19-25Crossref PubMed Google Scholar The use of other data streams might help to populate large datasets when clinical data are scarce or absent. For example, administrative coding could be used to identify reasons for hospital admission that are likely to be related (eg, pneumonia) or unrelated (eg, trauma) to COVID-19,4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar and to identify comorbid conditions for inclusion as covariates in comparative analyses.4Woodruff RC Campbell AP Taylor CA et al.Risk factors for severe COVID-19 in children.Pediatrics. 2021; 149e2021053418PubMed Google Scholar, 5Yek C Warner S Wiltz JL et al.Risk factors for severe COVID-19 outcomes among persons aged ≥18 years who completed a primary COVID-19 vaccination series—465 health care facilities, United States, December 2020–October 2021.MMWR Morb Mortal Wkly Rep. 2022; 71: 19-25Crossref PubMed Google Scholar The delivery of therapeutics used specifically or most commonly for COVID-19 (eg, remdesivir and dexamethasone) could enrich for those hospitalised with the disease. Ultimately, applying a probabilistic approach to case definition might allow for estimates of confidence when identifying cases and associating outcomes. After correcting for misclassification bias, the intrinsic severity of the omicron variant of SARS-CoV-2 might be even lower than that suggested by Nyberg and colleagues. We declare no competing interests. Supplementary Material Download .pdf (.17 MB) Help with pdf files Supplementary appendix Download .pdf (.17 MB) Help with pdf files Supplementary appendix
MoreTranslated text
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined