A comment from Professor Sheila Bird on COVID-19 vaccination status, infection and hospitalisation data.
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
Comment on: COVID-19 vaccination-status (nil, 1st dose, 2nd dose) in England per age-group at 31 May 2021; prevalence of infection according to Office for National Statistics; assumed vaccine-effectiveness against delta variant; and reported hospitalizations by vaccine-status.
“England is shy to tell us per age-group how citizens received Oxford-AstraZeneca versus PfizerBioNTech or Moderna mRNA vaccines. This matters – on principle because it is in the public interest – and also be because vaccine-effectiveness after 2nd dose against the delta variant is reckoned to be higher for the mRNA vaccines although protection after 1st dose may be only one-third of vaccinees, irrespective of vaccine-type.
“Lacking information on vaccine-type per age-group, I am obliged to assume that one-third of 1st dose vaccinees and 70% of 2nd dose vaccinees – irrespective of vaccine-type – are protected against the delta variant, see Table.
“Latest ONS Infection Survey for England showed that modelled estimate for the proportion infected had doubled in a fortnight from 9 per 1000 (95% CI: 8 to 11) on 15 May to 21 per 1000 (95% CI: 17 to 26) by 29th May 2021.
“More importantly, the ONS Infection Survey ‘s modelled estimates are age-dependent: 20 (10 to 36) per 1000 aged 2-10 years, 46 (27 to 75) per 1000 aged 11-15 years, 23 (16 to 31) per 1000 aged 16-49 years but only 11 (8 to 16) per 1000 aged 50+ years.
“In the Table below, I have assumed recent exposure-risks of 33 per 1000 unvaccinated at ages 18-49 years but 11 per 1000 at 50+ years. The former may have been higher and the latter lower.
“The “Hospitalized” data in my Table were quoted yesterday in BBC: News bulletins. They appeared to relate to England may pertain delta-variant hotspots. Matt Hancock today reported broadly similar data for 114/126 recent hospitalizations: 83 unvaccinated (73%), 28 1st dose (25%), three both doses (3%).
“Suffice to say that my sums – with simplifying assumptions as detailed above – do not align neatly with the quoted hospitalization data. To beat the data into submission, I’d need exposure-risk to have been twice as high in the unvaccinated and half as high in those aged 50+ years who have had both doses of vaccine.
“Behavioural differences of the order required for “submission” are not off-the-wall. Besides, the ONS Infection Survey does not measure “intrinsic” risk as its participants too will have been offered immunization and risk will have been reduced for those doubly-immunized versus for those in the same age-group who declined.
“Finally, age-specific risks – for example, as measured by ONS Infection Survey – are not expected to remain 3-fold lower for those aged 50+ years as inter-generational mixing outside of bubbled-households is increasingly permitted.
“Naïve assumptions, as I have set out, help to understand some of the evolving complexities that more sophisticated models are grappling with but are no substitute for attention to detail.”
All our previous output on this subject can be seen at this weblink: