Supplemental Oxygen Mini Study
Lessons from an undisclosed NHS Trust
An article entitled ‘No, vaccinated people are not ‘just as infectious’ as unvaccinated people if they get COVID’ in theconversation.com that was published back in Nov 2021 refers to results contained in a medRxiv paper by Po Ying Chia et al (Jul 2021) in which they claim:
‘Despite significantly older age in the vaccine breakthrough group, the odds of severe COVID-19 requiring oxygen supplementation was significantly lower following vaccination (adjusted odds ratio 0.07 95%CI: 0.015-0.335, p=0.001)’
In plain English what this is telling us is that vaccinated patients with breakthrough COVID-19 symptoms didn’t require supplemental oxygen treatment as often as unvaccinated patients. For those not familiar with odds ratios a ratio of 1.0 would mean no difference in oxygen use between patient groups, a ratio of 10.0 would mean vaccinated patients would be ten times more likely to need oxygen and a ratio of 0.1 would mean vaccinated patients are ten times less likely to need oxygen. A value of 0.07 thus puts us in the ten times less likely bracket and a p-value of p=0.001 means this result is unlikely to have arisen by chance. Whilst this is looking and smelling good, let us have a look at the small print…
Firstly, we must note that the vaccine breakthrough group excludes anybody who received just one dose and it also excludes anybody being admitted to hospital within 14 days of their second dose. In this instance unvaccinated really does mean unvaccinated. Secondly, we must note the small sample size of 218 patients of whom only 201 were included in the final analysis. In the pink part of the slide below I have summarised the findings of Po Ying Chia et al so we can see how fragile the numbers are. My crude odds ratio of 0.136 may be obtained by dividing 2.8% by 20.8% and is a little larger than the study estimate of 0.07 because they used multivariate modelling to refine things. Nevertheless we are very much in the same ballpark since their 95% confidence interval was 0.015 – 0.335. So far so good; small numbers maybe but statistically significant (I have run a comparison of proportions for good measure).
Let us now turn to the blue part of the slide. These are the figures we get by applying exactly the same selection criteria to admissions to the emergency department of a large UK NHS Trust over the period Jan – Sep 2021. Now that is what I call a difference! Po Ying Chia’s crude odds ratio of 0.136 has inflated into an odds ratio of 1.043. Before folk get excited and claim that breakthrough vaccinated cases have a higher demand for supplemental oxygen please do note the difference fails to reach statistical significance (p=0.8731).
So what does all this mean? It means we cannot go round using the results from a handful of cases in one region and make claims about vaccine efficacy in general. There’s a raft of prosaic reasons why use of supplemental oxygen may differ across consultants, across units, across regions and across healthcare systems, and all of these have nothing to do with disease severity. This doesn’t stop experts like Po Ying Chia and 18 other authors trying to polish the narrative.
