The CDC numbers the epidemiologic weeks of the year using a method which meets two conditions:

—first day of the week is a Sunday

—first week of the year is the calendar week with 4-plus days in that year

Here is how the CDC describes it:

Notice how it is that December 29, December 30, and December 31 “sometimes” make it into the first week of the following year. Using the 313 weeks of death data available at CDC (2014-2019), you can form weekly death rates using the monthly population figures.

The weekly death rate values below are in Weekly Deaths per 100,000 (WDp100k).

After getting all 313 weekly death rates, you can get the average death rate for any given MMWR week. You can also get the standard deviation in weekly death rates, and also get the percentage deviation from the mean for any given week.

Here is a spreadsheet (click it to enlarge it) of all of that:

Week #53 (most likely seen during a leap year) was deliberately left out of these computations due to insufficient data — it only happened in a single year of these data.

# Z-Scores

The remaining 312 weeks of weekly death rates were standardized (converted to Z scores) in columns L to Q.

Only a single week (Wk 40 of 2014) in the entire set of 312 weeks had a Z score over 2.00, meaning that the weekly death rate of that week was over 2 standard deviations from the mean for that week.

The probability of finding a weekly death rate that is over 2 standard deviations away from expectations would be ( 1/312 = ) 0.0032 — or less than 0.5%.

The highest coefficient of variation for any individual week was 6.77% of the mean — indicating that “three-sigma violations” would be found whenever weekly death rates exceeded expectations by ( 6.77 * 3 = ) 20.31%.

Another way to say that it exceeded expectations is to refer to it as “20% weekly excess death” and it is a conservative estimate, given the P-scores computed below.

# P-Scores

In columns R to W, P-scores were found, where the values showing represent the percentage deviation away from the mean for the corresponding week in question. Only 2 weeks out of 312 of them (Wk 1 and Wk 2 of 2018) had “double-digit” P-scores.

This means that the probability of witnessing a week that has a death rate over 10% away from expected is ( 2/312 = ) 0.0064 — or less than 1%.

For large nations like the USA*, it is definitely the case that weeks where over 20% excess death are found are unusual and so they should be tagged for future investigation as to the causes.

In 6 full years of weekly death data for the USA, not one single week had 20% excess death in it. Even getting over 10% excess death was so rare it only happened less than 1% of the time.

*Investigation by Ioannidis et al. revealed that a conservative threshold for small-nation “three-sigma violations” in excess death would require ( 8.1 * 3 = ) 24.3% weekly excess death.

# Reference

**[313 typical weeks of all-cause death]** — CDC. **https://data.cdc.gov/NCHS/Weekly-Counts-of-Deaths-by-State-and-Select-Causes/3yf8-kanr**

**[monthly US population numbers] **— U.S. Bureau of Economic Analysis, Population [POPTHM], retrieved from FRED, Federal Reserve Bank of St. Louis; **https://fred.stlouisfed.org/series/POPTHM**

**[SD of weekly excess death during COVID] **— Ioannidis JPA, Zonta F, Levitt M. Variability in excess deaths across countries with different vulnerability during 2020-2023. Proc Natl Acad Sci U S A. 2023 Dec 5;120(49):e2309557120. doi: 10.1073/pnas.2309557120. Epub 2023 Nov 29. PMID: 38019858; PMCID: PMC10710037. **https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10710037/**

I follow you for weeks 1-52 from 2014-2019. The control is set. I can’t wait to see your weekly analysis from 2020-2023.