September breakthrough data

DHS has released their latest update on breakthrough cases with data for the month of September. The risk reduction from the vaccines appears to be pretty much the same as last month.

September data by age group

Last month I wrote in detail on how to interpret these data, so for more background and August’s numbers please see that post. Here I want to share plots of this month’s data, and show how the age-adjusted risks are changing with time.

The bar plots below are a little complex, but I think helpful. Each compares vaccinated and unvaccinated groups for a different outcome: deaths, hospitalizations, and cases. The height of the bar indicates the rate of that outcome per 100,000 people. The width of each bar is proportional to the population of the group.

Finally, the plots are divided up by age group, because the risk of death and hospitalization is so dependent on age.

Deaths age stratified

Hospitalizations age stratified

Cases age stratified

In general, the patterns here are very similar to last month. Some differences are that numbers overall are higher, and that teens make up a larger share of cases. But the vaccines continue to provide very strong risk reduction against death and hospitalization, and pretty strong reduction against getting a case.

Estimated age-adjusted vaccine efficacy by month

Boiling this down to a single number, the table below shows estimates of vaccine efficacy based on this data for this month and last month.1 These numbers are age-adjusted, a process I also tried to explain in last month’s post.

Outcome August September
Cases 73% 78%
Hospitalizations 88% 89%
Deaths 89% 95%

Efficacy appears to have even improved a bit this month, but I would guess that is just some randomness in the data. But it is good news that while efficacy seems to have declined compared to the spring, it looks to be pretty stable right now.

  1. This table has slightly different August numbers than last month, because DHS updated its August data. But the difference is only 1 or 2 percentage points.