Posted by 
Vaccinating older people probably did avert some deaths in 2021, but the effects were small. And even those small effects on mortality seem to have dissipated during the booster programme. That’s the conclusion of our new study, published in the European Economic Review.
COVID-related deaths decreased significantly in most of Europe and the US from the middle of 2021. Although this reduction coincided with the rollout of COVID vaccines, it has proved surprisingly difficult to identify the extent to which vaccination contributed to the drop in deaths.
Randomised controlled trials (the gold standard for testing new treatments) suggest COVID vaccination can provide significant protection against serious illness and death relative to unvaccinated people who have not previously been infected with COVID. But there are reasons the effect of vaccination on mortality may be lower when viewed outside of trials.
Early in the programme, there were hopes that vaccination would also prove highly effective in preventing the spread of COVID but it has since become clear that vaccination provides only limited and short-term protection against infection and transmission.
It is also well established that a previous infection provides protection both against reinfection and against serious illness and death in the event of reinfection that is at least as effective as vaccination. Having a previous infection significantly reduces the likelihood of being vaccinated meaning the vaccinated population will include a relatively high proportion of people without protection from prior infection. So even if vaccination provides protection at an individual level, we may still observe population-level mortality rates that are similar for vaccinated and unvaccinated groups.
The effectiveness of vaccination programmes may also be limited by people’s behaviour. For example, there is evidence that vaccinated people who get infected are more likely to have mild symptoms and this may cause them to take fewer precautions than others against spreading infection. As a result, vaccination may sometimes be associated with more rather than less transmission.
Taken together, even if vaccination reduces the risk on an individual basis, it does not necessarily follow that it will reduce deaths at a population level. Existing research reflects this ambiguity with some research finding very significant effects of vaccination on death while other findings conclude there was little or no effect at all.
Our new study attempts to improve our knowledge about the effect of COVID vaccination programmes by estimating the effect of vaccination take up on deaths in care homes. This is a particularly important group to examine. Given that the vast majority of COVID-related deaths occur in the elderly, any effect on deaths is highly likely to be seen in care homes.
Machine learning used to analyse the data
We examined deaths from COVID in care homes across nearly 150 local authorities in England from the start of the vaccine rollout in December 2020 until after the second booster dose in summer 2022. We tested whether higher rates of vaccination of staff and elderly residents led to fewer deaths both in total and from COVID.
One feature of our research is the use of machine learning (a type of artificial intelligence) to isolate the effect of vaccination from other factors that may also have affected mortality including levels of prior infection as well as demographic, economic and health differences among local authorities.
Machine learning is particularly adept at separating out the effects of a high number of potential explanatory variables, providing much better evidence of when associations represent true causal relationships. In contrast to some other research, we also use a measure of vaccination that takes account of the fact that effectiveness wanes over time.
We found that higher vaccination rates of residents (but not of staff) did indeed lead to fewer deaths, but the effect was relatively small. For example, an increase in the resident vaccination take-up rate of 10% in a local authority caused, on average, a reduction of 1% in the total care home mortality rate. That is equivalent to about 22 fewer deaths per week nationwide.
Of course, any reduction in deaths is welcome. But vaccination does not appear to be the key factor in reducing care home deaths from COVID. We also found that the reduction in deaths was restricted to the initial vaccination rollout.
From September 2021, when the booster vaccination programme started in England, higher vaccination rates of elderly residents do not seem to have led to any reduction in deaths. Based on these results, vaccination is unlikely to have been responsible for the sustained fall in COVID-related deaths.
Why then did Europe and the US experience large reductions in COVID deaths since 2021, even during times when infection rates have soared?
There are two explanations. The first is the growth of variants such as omicron that, although highly infectious, are less deadly than variants responsible for the early waves.
Second, is the rise in the cumulative number of people who gained protection from having had previous infections.
These explanations are consistent with the experience of places such as Hong Kong, New Zealand and Taiwan. All saw relatively low COVID infections and deaths in 2020, meaning only limited levels of natural immunity had been built up. All then experienced high mortality rates during 2022, well after most people in those places had been vaccinated.
For example, the seven-day average mortality rate in Hong Kong reached 40 deaths per million in March 2022, a rate far above the highest peak seen in the US during the whole pandemic despite cumulative vaccination rates at that time being similar.
Even though vaccination probably reduced care home deaths by a small amount in the early rollout period, there is little evidence that the booster programme had any significant effect on COVID-related deaths.
David Paton is a member of HART (Health Advisory and Recovery Team).
Sourafel Girma does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
