When a process has output which is within 3 standard deviations of the mean, then it can be said that the variation in the process is coming from common causes. But if the process begins to have values beyond 3 standard deviations from the mean, the logical conclusion is that it indicates a special cause is in effect.
When looking at the death rate due to cerebrovascular disease (stroke) for those in the age band of 55 to 59, what you find is that that process (dying from stroke) runs at a rate bounded between 23 deaths per 100,000 and 24 deaths per 100,000 — only a small amount of variation is found from the common causes of variation in that process.
But when COVID and COVID responses hit, things changed for those in this age band:
You can see what happens when only common causes create the variation in the death rate — by looking at the years from 2010 to 2019 — death rate values stay between 23 deaths per 100,000 to 24 deaths per 100,000. But in 2021 and 2022 (at 26.4 deaths per 100,000) the death rate due to strokes was 12 standard deviations above the mean.
Even using no assumptions about the distribution of values — i.e., applying a method (Chebychev Inequality Test) which always works, regardless of the background distribution of values that you have — that increase is statistically signficant. Any value beyond 5.0 standard deviations from the mean becomes significant.
The 2021 and 2022 death rates are beyond 12.0 standard deviations above the mean. There is a moral imperative, and I believe there is a legal one as well, for U.S. health officials to investigate this.