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, calendar time and comorbidies etc using time-dependent cox regression. However, uncontrolled confounding is likely because COVID-19 symptoms at the time of planned vaccination were not accounted for, and these are likely to be strong predictors both of whether an individual is vaccinated and of severe disease outcomes. |
| Selection of participants into the study |
Low |
No particular concerns in this domain. |
| Clasification of interventions |
Low |
No particular concerns in this domain - vaccination status is likely to be adequately measured (using electronic health records). |
| Deviations from intervention |
Low |
No concerns in this domain - the study was observational. |
| Missing outcome data |
Moderate |
The database studied was a national one and it is likely that data were nearly complete for the variables analysed. However, there is likely to be informative censoring when controls became vaccinated. Controls who became vaccinated are likely to be different from those who did not. Controls might not have got vaccinated if they developed COVID-19 symptoms, although the number of these may be relatively small. |
| Measurement of the outcome |
Low |
No particular concerns in this domain - differential bias between vaccinated/unvaccinated is unlikely to be an issue for severe/mortality outcomes. |
| Selection of the reported results |
Moderate |
A study protocol was approved but not available to check, however the results appear to be faithful to the analyses undertaken. |
| Overall risk of bias |
Serious |
|
| Overall comment | Concerns over uncontrolled confounding, particularly due to having COVID-19 symptoms at the time of planned vaccination. |