How do we know the effect of boosters?

Last week, the UK Health Security Agency (UKHSA) reported encouraging research on booster doses. That analysis estimated that, compared with the preceding waned protection, the booster reduced the risks of symptomatic Covid-19 disease in people over 50 by around 85%.

Compared with not being vaccinated at all, a booster gave similar protection to that obtained soon after two Pfizer/BioNTech doses. Encouragingly, this reduction occurred regardless of the initial vaccine, implying Oxford/AstraZeneca vaccinees have more protection after the booster dose than ever.

Two types of study produce estimates about the effectiveness of boosters. First, the Office for National Statistics Covid infection survey of around 90,000 people a week showed the chance of testing positive after a booster was similar to after a recent second Pfizer/BioNTech jab. In contrast, the UKHSA used a “test negative case-control” design, comparing vaccination histories of 106,000 people who tested positive and 165,000 who tested negative. That design, often used in influenza vaccine studies, neatly controls for any differing tendencies to get tested between vaccinated and unvaccinated people. An earlier UKHSA study estimated two Pfizer/BioNTech doses gave around 99.7% (97.6% to near-100%) protection against Delta-infected hospitalisation, but after 20 weeks that effectiveness waned to 92.7% (90.3% to 94.6%). This estimated decline for people over 16 may not sound much, but if we look at it in terms of “lack of protection”, their estimated vulnerability relative to being unvaccinated went from 0.3% to 7.3%. That is a major, although uncertain, increase in risk.

Such “negative framing” can change impressions: “90% fat-free” sounds rather different than “10% fat”. But this framing might better reflect the importance of waning immunity, showing the power of third doses to lower hospital admissions, who have mainly had two doses.

All these risks are relative and the impact depends on the absolute risks of Covid-19. These viral risks are communicable and variable: between people and even from day to day in the same person, adding up over a year. Like seatbelts, vaccines make small risks lower, but also help others. These early analyses indicate booster doses are very effective.

David Spiegelhalter is chair of the Winton Centre for Risk and Evidence Communication at Cambridge. Anthony Masters is statistical ambassador for the Royal Statistical Society

Add comment