Note: The risk of bias by domain corresponds to the highest risk of bias among outcomes by domain.
The overall risk of bias corresponds to the overall highest risk of bias assessed among outcomes.
| Bias | Author's judgement | Support for judgement |
| Confounding |
Serious |
The analysis adjusted for age, race, sex and ethnicity. Uncontrolled confounding is likely, for example arising from socio-economic status, comorbidities and health-seeking behaviour. |
| Selection of participants into the study |
Moderate |
The study used a test-negative design. This has the potential to suffer from selection bias by being restricted to individuals getting a test, although the issues are not yet well understood. |
| Clasification of interventions |
Serious |
Vaccination status was based on a mixture of hospital records and patient/proxy recall. It is unclear in the paper how many participants had vaccination status recorded using hospital records versus patient/proxy recall. |
| Deviations from intervention |
Low |
No particular concerns in this domain - the study was observational. |
| Missing outcome data |
Low |
No particular concerns in this domain - there was minimal missing data. |
| Measurement of the outcome |
Serious |
It is unclear whether those in the second control group (who were hospitalised for non-COVID-19 like illness) would have been classed as cases if they tested positive. |
| Selection of the reported results |
Serious |
There is no evidence of a SAP or protocol. Further, multiple control groups were available but the study only reported results for all controls grouped together. |
| Overall risk of bias |
Serious |
|
| Overall comment | Concerns over uncontrolled confounding and potential for misclassification of vaccination status. |