The excess mortality approach discussed in this thread is a promising approach not only for understanding the effects of COVID-19, but the impact of natural disasters and other public health crises. A recent report from the National Academies of Sciences, Engineering, and Medicine that I worked on discusses this approach as one estimation approach (based on sampling and other statistical methods) that can provide a useful complement to more common approaches to counting cases and deaths.
These methods, however, often have been used in a somewhat ad hoc fashion, with researchers making different choices about matters discussed in this thread and other issues. The report calls for research program that begins with a discussion of advantages and disadvantages and the documentation of researchers' and policymakers' experience with choices that have worked particularly well (or not) in the past. The research program could address such factors as: the spatiotemporal boundaries of the study; the specification of a comparison period or the handling of confounding or seasonal structure in the data; determination of an accurate sampling frame; development of appropriate standard survey questionnaires; crafting appropriate statistical models; and developing effective means of characterizing migration and population displacement, before, during, and in the immediate wake of the disaster.
Coming back to COVID-19, my Georgetown student, Sam Schlageter, and I have just published a White Paper describing excess mortality methods, as well as the benefits and challenges of using them to better understand the impact of the pandemic. Beyond total number of deaths, for example, excess mortality estimates can demonstrate how the pandemic has differentially affected racial and ethnic minorities and other population groups and provide insights about its impact on the healthcare system and economic disruptions. You can find the paper on the Georgetown Center for Global Health Science and Security's COVID-19 page. We'd welcome your comments!
Mike Stoto
------------------------------
Michael A. Stoto, PhD
Professor of Health Systems Administration and Population Health
Georgetown University and
Adjunct Professor of Biostatistics & Senior Preparedness Fellow
Harvard T.H. Chan School of Public Health
mike.stoto@gmail.com------------------------------
Original Message:
Sent: 04-14-2021 09:40
From: Brent Blumenstein
Subject: USA excess deaths, COVID-19
Now comes updated graphs based on an increment of data, but more important a correction to the titles so that age group is identified. Thanks to Elgin Perry for pointing out this error introduced when I changed from calendar year graphing to 52 week spans. And thanks to those making suggestions for modifications directed at the pandemic becoming > 1 year in duration. Stay tuned.
------------------------------
Brent Blumenstein
Original Message:
Sent: 04-13-2021 08:09
From: Brent Blumenstein
Subject: USA excess deaths, COVID-19
I have an affinity for excess death methodology as a means of conveying pandemic impact. Attached is a recent generation of my pandemic impact graphs (earlier versions posted previously). The innovation is comparison of cumulative counts (just another view). I am now puzzling about how to tweak my graph program due to the pandemic being > 1 year in duration (I recently transitioned from calendar year to 52-week spans).
------------------------------
Brent Blumenstein
Original Message:
Sent: 04-12-2021 16:58
From: John Major
Subject: USA excess deaths, COVID-19
Thanks to everyone who replied, especially Janet and Chuck for the pointers.
Brian: "Unarguable" is perhaps too strong; some people can argue with anything. But total death counts are more likely to be accepted by skeptics (my audience) who claim that (1) "confirmed" (via PCR test) case counts were until recently contaminated with an unknown number of false positives due to high cycle threshold and (2) COVID-19 specific death counts are ambiguous as to death-from-COVID vs death-with-COVID and possible bias from financial incentives. A death is a death and few skeptics are going to argue that those numbers have been distorted.
Stephen: In the Significance article (https://rss.onlinelibrary.wiley.com/doi/10.1111/1740-9713.01485) I mentioned, figure 1 bids us to compare the behavior of 2020 to 2017-2019. The weekly detail is of interest of course, but the point of the chart is a holistic comparison. We are essentially comparing 4 samples of a function and saying "Aha! This last one sure is different." COVID-19 is the obvious hypothesis to explain what is going on, but it would be good to see more years and what kind of variation there has been over the years, including other years with their own unique mega-events (like the Great Financial Crisis of '07-'08).
Regards,
------------------------------
John Major
Original Message:
Sent: 04-12-2021 10:44
From: Charles Coleman
Subject: USA excess deaths, COVID-19
CDC produces the National Death index, which has individual deaths by day, including state and cause. You have request a tabulation from CDC. You can find its information at https://www.cdc.gov/nchs/ndi/index.htm,
------------------------------
Chuck Coleman
Original Message:
Sent: 04-09-2021 11:50
From: John Major
Subject: USA excess deaths, COVID-19
The latest Significance magazine has an interesting article about measuring 2020 excess deaths in the USA, an unarguable metric of the impact of COVID-19. It appears to use the Weekly Counts of Death CDC data. However, that dataset only goes back to 2017. N=4 does not seem big enough to really understand what is going on. Does anyone know of a comparable source that goes back at least to, say, 2001? It doesn't have to be weekly, but sub-annual would be nice. Thanks in advance.
------------------------------
John Major
------------------------------