- metadata: - source: https://gettr.com/post/pztrv5cafc - people: [[Edward Dowd]] --- # 2022-03-14 MiloMac's reproduction of EdwardDowd's excess mortality **Background:** Edward Dowd has published a set of charts which show excess mortality by years 201.9-YTD 2022 for various age bands. The intent of this project is to reproduce and confirm Edward's work from the CDC source data. It is my understanding that Edward's development started with aggregate excess deaths, which required him to impute expected deaths by age group. For my work, I have used a CDC data file that already included expected and excess deaths by week for 2020 through YTD 2020 by five year age band, therefore, I did not need to impute expected deaths and have relied on CDC's methodology. https://data.cdc.gov/INCHS/AH-Excess-Deaths-by-Sex-Age-and-Race-and-Hispanic-/m74n-Lthbs Note, Edward included 2019 excess deaths in his development. Because my source file did not include 2019, I have set excess deaths in 2019 to 0. I did this so that I can publish graphs on the same time frame as Edward's to make comparisons easier. However, I don't believe this is material given 2019 is not a zone of interest. **Assumptions:** The CDC data file includes both weighted and unweighted deaths. "Weighted" deaths include projections for more recent periods where data is less complete. I have used "weighted" deaths so that I would not need to complete the data manually. For expected deaths, CDC has used their normal procedure to project expected deaths in the current periods based off prior period levels and seasonality, and I have relied on this development. **Results:** In my opinion, I have sufficiently reproduced Edward's work so as to verify the results. While there are slight variations in our numbers, the differences are not material to the interpretation and are likely due to methodology differences in imputing expected deaths. One key consideration that I have added to my analysis that was not present in Edward's is the contribution of COVID versus Non-COVID causes in excess deaths. I will present charts that split this out and will also be supplying interpretation. This is an important distinction and more work can be done here. In addition, the data set includes results by 5 year age band, gender and ethnicity. I have produced comparable charts by these breaks and will be posting relevant findings. **Caveats:** I am conducting this analysis as a private citizen and while I have made every effort to be diligent, this should NOT be considered actuarial work product in the context of the actuarial standards of practice. --- (https://gettr.com/post/pztrv5cafc) Presented here is a summary of my analysis of public CDC source data in an effort to reproduce the work done by [https://gettr.com/user/edwardowd](@edwardowd) See here: https://gettr.com/post/pzhvce33ce There are 4 images attached this post. The 1st is a brief description of background, assumptions, results and caveats. (text above ^ line) The 2nd is my version of Edward’s ages 25-44 excess death percentage chart. ![[gettr-b966162e8ef8adef2f66a1c382b2cf3d.jpeg]] The 3rd is the same chart but as a stacked bar showing the contributions of COVID and Non-COVID causes to excess deaths. I will be deep diving this. ![[gettr-cc8c94907069e96cda5327a94479a92d.jpeg]] The 4th image takes Edward’s original chart and overlays it on my chart to better see how they compare. ![[gettr-15d2316d7450a332f62ea7e7560c8190.jpeg]] I believe I have sufficiently reproduced Edward’s work. Additional presentations of analysis will follow. @drnaomirwolf @rwmalonemd