Throughout the pandemic the Office for National Statistics has been providing timely data and analysis of the deaths caused by Coronavirus (COVID-19). During 2021 an important part of this work is measurement of mortality by vaccination status. In this post Charlotte Bermingham explains why we use internationally-recognised methods to ensure comparability across all our releases and how the analysis so far should be interpreted.
The most recent ONS publication shows that between 2 January and 24 September the age-adjusted risk of deaths involving COVID-19 was 32 times greater in unvaccinated people than in fully vaccinated individuals. Due to various factors described in this blog this figure changes over time. It is also important to understand that it is not a measure of vaccine effectiveness.
The publications are an ongoing series which use internationally recognised age-standardised mortality rates (ASMRs) to show the risk of dying from COVID-19 for people with different vaccination statuses. They give an up-to-date picture of how mortality varies and has previously varied between people with different vaccination statuses, and how this is changing over time. This level of granular insight into mortality can support timely evidence-based public health policy.
Separate studies on the real-world effectiveness take into account factors that differ between the vaccination status groups and over time, as published by the ONS using the Covid Infections Survey. The ONS’s insight tool brings together data on vaccines, hospitalisations, deaths and a variety of themes associated with the pandemic.
Calculation of age-standardised mortality rates
As most causes of death vary significantly with people’s age, the use of standardised death rates improves comparability. Standardising for age is particularly important when looking at deaths by vaccination status due to the selective roll out of vaccinations, with older people having been prioritised. For information on how ASMRs are calculated, see our methodology report.
By calculating the ASMRs for the whole year to date we include people as they pass through different vaccination statuses and avoid choosing a specific time to compare (which can be done using the weekly ASMRs we also published). The population for these ASMRs is calculated in person-years to take into account the amount of time spent in each vaccination state. For example, 100 people who are unvaccinated for 6 months each during the period would together contribute 50 person-years to the unvaccinated population. The unvaccinated ASMRs will include a higher proportion of person-years earlier in the year than the vaccinated ASMRs.
Calculations for different periods will give different ASMRs due to changes over time and between the groups. We gave the overall comparison of the 2 January to 24 September as it was the full dataset available to us and provides a summary figure for 2021 so far. We will continue to update this as more data becomes available.
Not a measure of vaccine effectiveness
There are factors which change over the year and between the vaccination status groups meaning that no choice of time period is going to give a value of vaccine effectiveness and we describe clearly in our bulletin that this figure is not equivalent to vaccine effectiveness.
These factors include the changing underlying COVID-19 infection rate (which was high in January, when many people are unvaccinated, but is also high later in the year in mid-July, when many people have received vaccinations), changing levels of “natural immunity”, changing dominant variants and changes in the characteristics of the people in the vaccination status groups due to the selective roll out and differences in uptake.
Changes over time
During the second wave of the pandemic, we saw a large difference in the ASMRs for COVID-19 death between unvaccinated and vaccinated people. This difference decreased over time and remained lower (but substantial) during the summer.
Several factors can explain ASMRs change over time. First, COVID-19 mortality rates are strongly influenced by changes in COVID-19 infection rates. For example, if most people are unvaccinated while infection rates are high, but then the infection rates are lower when most people are vaccinated, this will increase the ASMR for the unvaccinated population compared to vaccinated. The same can be true for seasonal changes in mortality, which would be expected to affect the non-COVID-19 ASMRs.
If a more virulent strain is active for a particular period of the year, this can increase the mortality rates in this period. The ASMRs are therefore valid for the specific time period over which they are calculated but vary if calculated for different time periods.
Secondly, ASMRs may vary because the composition of the groups is changing over time. For instance, more vulnerable people and health and social care workers were vaccinated first, and as the vaccine rollout progressed, the group of people who had received one dose became more representative of the general population.
However, after most people had been able to receive two doses, this group becomes atypical, with people being too ill to receive their second dose becoming over-represented. As a consequence, we can see that the ASMR for non-COVID-19 deaths increases in the single-dose group in the later period.
Thirdly, the level of immunity due to past infection can change over time as more people have had COVID-19. This could explain why the difference in COVID-19 ASMRs between vaccinated and unvaccinated people remains relatively low in the third wave, despite a high level of infection.
Finally, the effectiveness of the vaccine could also wane over time, causing the mortality rate in fully vaccinated individuals to increase later in the year. It is challenging to untangle the effects that these different factors have on the ASMRs.
We will be updating our analysis next month, including more breakdowns by smaller age bands and rolling on the data to give a more up to date estimate of the age-adjusted risk of deaths involving COVID-19 by vaccine status.
Dr Charlotte Bermingham, Senior Research Officer at the ONS