Exploring the Decline in Vaccine Effectiveness Over Time

The development and mass administration of vaccines has enabled many governments to end the restrictive and costly measures adopted to help control the coronavirus disease 2019 (COVID-19) pandemic.

Although the vaccines were not 100% effective in preventing the disease, they significantly reduced the risk of developing severe symptoms, thereby reducing the pressure on hospitals and the number of deaths.

Study: Duration of protection against mild and severe disease by Covid-19 vaccines. Image Credit: BaLL LunLa/Shutterstock

However, scientists and healthcare workers have noted that the effectiveness of the vaccine appears to decline over time, as some countries have opted to provide their populations with a third “booster dose” to further protect them.

Researchers from the UK Health Security Agency have studied how much vaccine effectiveness is reduced over time, and how this varies with different vaccines and variants, and published their findings in the New England Journal of Medicine.

The study

The researchers used a test-negative case-control design to estimate vaccine effectiveness against symptomatic COVID-19, hospitalization due to COVID-19, and death due to the disease. For cases to be included, a positive PCR test must have been received. They stratified the analysis to better assess the efficacy of the vaccine against alpha and delta variants. The scientists used a logistic regression model adjusted for age, gender, deprivation, ethnicity, nursing home status, region, time of infection, clinical status and status of health/social workers to estimate vaccine effectiveness. When analyzing deaths, the sample size was significantly smaller than the other analyses, so the period was modeled using a cubic spline. To estimate the decline in vaccine effectiveness, the researchers assessed infections based on time intervals after vaccination, from one to twenty weeks.

More than 7 million eligible PCR tests have been evaluated, and approximately 6 million of these have been successfully linked to the National Immunization Management System (NIMS) database for vaccination status – approximately 85% of cases and witnesses. 1.7 million participants with a linked test showed a positive result, of which 540,000 were infected with the alpha variant and 1.2 million had the delta variant. 37,000 people were infected with another/unknown variant, and these were not included in the analysis. 4.3 million negative tests have been received. 2.3 million participants received two doses of ChAdOx1-S, 2.1 million received two doses of BNT162b2, and about 12,000 received mixed doses. Again, these were removed from the analysis. 22,000 with positive results were hospitalized and just over 6,000 died.

In general, mRNA vaccines than the ChAdOx1-S vaccine with respect to severe outcomes, as well as with the alpha variant versus the delta variant, and younger individuals versus older individuals. The scientists ensured that the results of the logistic regression for the all-ages analysis were included in the models. In general, vaccine efficacy against symptomatic delta-variant disease was greatest soon after the second dose. At 20 weeks post-vaccination, vaccine efficacy decreased to 44.3% for ChAdOx1-S and 66.3% for BNT162b2. As expected, vaccine efficacy declined faster and to a greater extent in older people.

A similar level of reduction in vaccine effectiveness was also seen in the level of protection provided against hospitalization. The ChAdOx1-S vaccine showed 80.0% protection against hospitalization at 20 weeks, compared to 91.7% for the BNT162b2 vaccine. This same level of waning protection could be seen against deaths for both vaccines. This situation was again exacerbated in the oldest age group for which information was available. The results of the sensitivity analysis of hospitalizations that had been coded as respiratory admissions showed similar results to the original analysis, as did the analysis examining only control participants. When the researchers stratified the data by risk group status, they identified a much greater than normal decrease from hospitalization with the delta variant in people over 65 who were also clinically extremely vulnerable. Clinical risk groups aged 40 to 64 were also at higher risk than those who were not clinically vulnerable.

The analysis restricted to people over the age of 80 who received BNT162b2 before January 2021 found lower vaccine efficacy in participants with a shorter interval between doses compared to those who waited longer. However, the confidence intervals overlapped between these two groups, which likely requires further investigation.


The authors emphasize that they have provided strong evidence against the decrease in protection against symptomatic disease of ChAdOx1-S and BNT162b2, but point out that protection against hospitalization and death remains at much higher levels for at least least 20 weeks after vaccination. Although the ChAdOx1-S vaccine appears to decline faster than the BNT162b2, this varies depending on who is receiving the vaccine. This information could help inform public health policy makers in the future and could influence the debate over a fourth dose.

Sara H. Byrd