Design: retrospective cohort study using mixed-effects logistic regression on national electronic health record data
Population: 60,980 adults with autoimmune rheumatic conditions and incident COVID-19, December 2020 to August 2024, median age 62
Primary outcome: COVID-19-related hospitalisation
Effect: adjusted odds ratio 0.59 initial series, 0.29 booster, 0.31 updated vaccine versus unvaccinated; absolute risk reduction 5.6% for booster or updated versus initial series alone

Sixty thousand nine hundred and eighty adults with autoimmune rheumatic conditions and an incident COVID-19 infection between December 2020 and August 2024 were analysed from the National Clinical Cohort Collaborative, with vaccination status graded as unvaccinated, initial series, booster, or updated 2023-2024 vaccine, and mixed-effects logistic regression adjusting for demographics, clinical risk, social vulnerability and medications. Median age was 62; 76% were female.

Only 6,336 (10%) had received an updated vaccine. Adjusted odds of COVID-19-related hospitalisation fell progressively: 0.59 (95% CI 0.56 to 0.63) with the initial series, 0.29 (0.27 to 0.31) with a booster, and 0.31 (0.28 to 0.34) with the updated vaccine, all against no vaccination. Comparing those with only the initial series against those with a booster or updated dose, the absolute risk reduction was 5.6%.

The actionable number is 10%. Nine out of ten patients in a population defined by immune-mediated disease and immunosuppressive treatment had not had an updated vaccine, and the absolute benefit of moving them from an initial series to a current dose was 5.6 percentage points of hospitalisation — around 18 patients boosted to prevent one admission, in a group that tolerates admission badly.

The practical block is that vaccination rarely belongs to anyone in a rheumatology clinic. It is assumed to be primary care's job, primary care assumes the specialist is managing the immunosuppression and will advise, and nobody asks. Making vaccination status a recorded field at the annual review — alongside the drug monitoring that is never forgotten — is the change this supports. The same applies to pneumococcal, influenza and herpes zoster vaccination, where the argument is the same and the neglect is similar.

Add vaccination status to the standing annual review checklist alongside drug monitoring
Recommend a current dose rather than accepting that the initial series was completed
Time vaccination around rituximab and other B-cell depleting therapy where you can
Use live vaccine restrictions as a reason to plan, not as a reason to omit the inactivated ones
Cover influenza, pneumococcal and herpes zoster in the same conversation; the same neglect applies

Don't overread it

Observational data on a self-selected exposure — vaccinated patients differ from unvaccinated ones in ways adjustment cannot wholly remove.

The statistics, in plain English

An adjusted odds ratio of 0.29 for a booster against no vaccination is large, but the comparison that changes practice is the smaller one: initial series versus booster or updated dose, expressed as a 5.6 percentage point absolute risk reduction, which is around 18 patients to prevent one hospitalisation. This is observational, so healthy-vaccinee bias is a real concern — people who seek boosters differ from those who do not in ways adjustment cannot fully capture — and the 2023-2024 vaccine group is small at 6,336. The progressive dose-response across four vaccination categories is what makes confounding a less complete explanation.