Straight Talk from ex-CDC for the Long Slog Ahead

By Bruce G. Weniger, MD, MPH and Chin-Yih Ou, PhD, medical epidemiologist and research-laboratory specialist, respectively; both retired from CDC (link for free, direct, future access to this essay and updates of its epidemic curve) (link to Rachel Maddow Show interview about this essay) _ _ _ _ _ _ _ [The authors will retire updating of the two figures and the Analyzed Data Table upon the final interval of 2022–12–30 to 2023–01–01.] _ _ _ _ _ _ _ [CLICK FIGURES FOR INCREASED SIZE AND LEGIBILITY.]

FIGURE. “Natural Experiment: Polled Glory — The Red, Purple, & Blue”. Total confirmed cases and death rates of COVID-19, United States, by report date and by electoral preference, January 21, 2020 through November 17, 2022. _ _ _ _ * Includes reported probable cases beginning on various dates since April, 2020, in addition to confirmed cases. _ _ _ _ “Red states” (n=24) are those which voted three or four times for Republican presidential candidates in the last four contests in 2004, 2008, 2012, and 2016. “Blue states” (n=24+DC) voted similarly for Democratic candidates (includes District of Columbia since 23rd Amendment in 1961). “Purple states” voted twice for each party (FL, IA, OH). Non-voting jurisdictions excluded (AS, GU, MH, FM, MP, PW, PR, VI). _ _ _ _ Data sources: cases, politics, mid-year 2019 population, vaccine doses. _ _ _ _ See: Dirty Epidemic Curves and Natural Experiment, below. _ _ _ _ † Marked rise in Blue-state death rate in the 25–27 June 2020 interval attributable to New Jersey (NJ) reporting on 2020–06–25 an additional 1,854 deaths (66% of Blue total 2,823) deemed probably caused by COVID-19 upon review of death certificates since epidemic start. _ _ _ _ ‡ Marked rise in Red-state death rate (5.76 per 100k) in the 2–4 February 2021 interval attributable in part to Indiana (IN) reporting on 2021–02–04 an additional 1,507 deaths (21% of Red total 7,016) occurring in 2020 and prior 2021 intervals upon audit matching death certificates with positive tests. _ _ _ _ § Spike in Purple-state death rate in 11–13 February 2021 (13.82/100k) attributable to Ohio (OH) beginning to report approximately 4,000 “overlooked” deaths (~80% of Purple total 5,021) since October 2020 upon reconciliation of separate reporting systems. _ _ _ _ ‖ Negative death rate ([minus] -0.28) in Purple states in 1–3 March 2021 results from Ohio’s change in its original death-counting protocol to rely on CDC’s attribution to COVID-19 causation, thus reporting [negative] -596 deaths on 2 March, followed by zero on 3 March. _ _ _ _ ¶ Spike in Red-state death rates in 6–8 April 2021 resulted from Oklahoma (OK) reporting on 7 April 1,076 cumulative deaths since 1 April 2021 not previously reported. _ _ _ _ # Initial negative Red-state death rate for 21–23 April 2021 ([minus] -0.09/100k), once attributable to Texas (TX) updating on ≈2021–05–30 its original reports on 20 and 21 April 2021 of 52 and 84 fatalities, respectively, to 1,013 and (minus) -877, respectively, was retroactively corrected to 49 and 87 cases, respectively (0.70/100k), during interval starting 2021–06–14. _ _ _ _ ** Florida (Purple) on 4 June 2021 began weekly reporting only, withholding hospitalized numbers and excluding from statewide counts of cases and deaths among non-Florida residents diagnosed in and dying of COVID-19 in Florida. (Authors’ Note: This is malign statistical fudging as deceitful as not counting traffic fatalities and murder victims who die in Florida as part-year residents and visitors; such trauma and COVID-19 disease risks are relevant to location, not legal residency. See sources A, B, C, D, E. Florida’s once-weekly reporting explains the subsequent volatility of Purple-state cases and death rates.) _ _ _ _ [Turkey footnote] During Thursday to Sunday of the Thanksgiving holiday and weekend of 2021, up to 42 states did not report cases and/or deaths for some or all of the five days (NY Times source anomalies), resulting in material undercounts in the Figure for intervals 23–25 and 26–28 November. _ _ _ _ [Christmas footnote] CDC announced 2021–12–23 that it will not update data on cases and deaths on 24, 25, and 26 December. Thus, only data for 23 December are included in the 3-day interval of 23–25 December in this graph. _ _ _ _ NOTE: The authors of this essay published on 3 May 2020 will continue to update this chart every three days as data become available. _ _ _ _ For free, direct, future access, copy and save this “friends” URL hotlink: https://medium.com/@bgweniger/straight-talk-from-ex-cdc-for-the-long-slog-ahead-f9d18a8502d1?source=friends_link&sk=449ec6deec0ec1595bdd9eeae23ea7a0 _ _ _ _ The authors receive no compensation for this essay, which may be distributed and re‑posted under Creative Commons Attribution-NoDerivatives 4.0 International License.
FIGURE 2. Vaccination Uptake: Public Health vs. Divisive Politics — Cumulative Percent of Persons Given ≥1 Dose(s) COVID-19 Vaccine per Analyzed Counties’ Total Populations, Stratified by Voting Preference in 2020 Presidential Election, December 14, 2020 through November 16, 2022 (figure initiated 2021–07–08). _ _ _ _ _ * Voting majorities in counties determined by 2020 Presidential election, counting only votes for Biden and Trump in 50 states and the District of Columbia (DC); excluding third-party candidates from denominators. _ _ _ _ _ Individual counties with missing or unavailable data on vaccination are excluded from analysis, or merged into a “catch-all county” in three states: _ _ _ _ _ In Alaska (AK), its 40 Presidential and state house election districts do not correspond geographically with AK’s county equivalents for population counts and dose reporting. Thus, AK is analyzed throughout Figure 2 as a single entity assigned to the Trump medium-red stratum of ≥55%-to-<65%. _ _ _ _ _ Hawaii (HI) does not report doses individually for its four counties and is analyzed throughout Figure 2 as one “catch-all county” assigned to the Biden dark-blue voting-majority stratum of ≥65%. _ _ _ _ _ Sources and analyses of Texas (TX) data differ by time period due to historical non-availability. Through 2021–08–15, TX was analyzed as one “catch-all county”, and was stratified to Trump light-red stratum of >50%-to-<55% (sources: until 2021–02–16, then through 2021–08–15). From 2021–08–16 through October 2021, the 254 counties in TX are analyzed individually using Texas-reported vaccine doses, and each stratified by their voting majorities. From 2021–11–01 onwards, TX county data as published by CDC are its source. _ _ _ _ _ See separate ANALYZED DATA TABLE for underlying numbers and details of method. _ _ _ _ _ County data sources: vaccination (except TX, see above); population (mid-year 2019); voting. _ _ _ _ _ On 2021–12–11, the New York Times noted a CDC footnote which acknowledged that an unknown proportion of administered “booster” doses may instead be counted incorrectly as “first doses” because of the inability to link individuals receiving subsequent doses at differing sites than for their primary-series dose(s). Thus, the percentages illustrated in Figure 2 may be artifactually higher than they really are for first doses. This is an example of the “dirty data” (see essay) with which epidemiologists must work, attributable to the nation’s haphazard and disorganized system accurately to monitor vaccination coverage, resulting from our federal structure, the 10th Amendment, and libertarian gubernaphobia. Nevertheless, separate trends for each voting majority remain informative. _ _ _ _ _ Nevertheless, separate trends for each voting majority remain informative. _ _ _ _ _ § Drop in cumulative percent of populations vaccinated in the 20–22 and 23–25 December 2021 intervals for some voting strata results from as-yet unexplained, anomalous doses-given reporting from some states to CDC. _ _ _ _ _ Discernible substantial increases and biologically-implausible decreases in cumulative dose-1 counts for intervals between 2022–01–19 and 2022–01–27 result from unexplained, anomalous discrepancies between the graph’s CDC data source and first-doses data from the State of Georgia dashboard and data from the State of California dashboard. Discernible decrease in Biden light-blue strata in interval ending 2022–03–20 results from New Hampshire (NH) reducing its cumulative counts of first doses in all its 10 counties. Decrease in Biden medium-blue strata in interval ending 2022–05–12 results from nine California (CA) counties reporting unexplained drops from prior interval in cumulative dose-1 counts totaling 599,284. _ _ _ _ _ ¥ CDC on 2022–06–13 discontinued daily reporting of vaccine doses administered nationally and began weekly reporting for the prior 7 days ending Wednesdays. These weekly cumulative percents are graphed on the X-axis in the 3-day interval encompassing the final day of the weekly report. _ _ _ _ _ [Turkey footnote] CDC announced 2021–11–24 not updating vaccination data (nor cases and deaths) for Thursday to Sunday of the Thanksgiving holiday and weekend of 2021, resulting in gaps in Figure 2 for intervals 23–25 and 26–28 November. _ _ _ _ _ [Christmas footnote] CDC announced 2021–12–23 that it will not update its data on vaccine doses given on 24, 25, and 26 December. Thus, only data for 23 December are included in the 3-day interval of 23–25 December in this graph.

Budget Wars and Ideology

The “Budget Wars” poster of the CDC Epidemic Intelligence Service (EIS) cohort of 1980–82.
The prevailing concern during 1980–1982 was the federal budget cuts in the first years of Ronald Reagan’s presidency. … Budget cuts were not limited to CDC — almost every health, social, and informational agency was targeted. The Reagan administration closed the U.S. Public Health Services hospitals, creating a reduction in strength of the Commissioned Corps (with more senior officers bumping more junior officers), and the Commissioned Corps itself was almost abolished. So our class was very concerned about funding of the agency itself and whether CDC would be able to hire any of us … .”

Early Skirmishes

Recent Battles

Casualties for COVID-19 Preparedness

Problems with the CDC Test

“Did protectionist pressures or scientific pride favor ‘our own’ tests … ?”

Challenges for Genetic Tests

Responsibility for Recruiting Industry

Accuracy in Testing

“… Senator Rand Paul’s coronavirus test … may have been a false positive …”

Dirty Epidemic Curves

Models Ignore Human Nature

Model for predicted daily U.S. deaths presented by Dr. Deborah Birx at the White House coronavirus task force briefing, March 31, 2020. Graph of COVID-19 forecasting model produced by the Institute for Health Metrics and Evaluation (IHME), University of Washington.

“The optimal titration is to achieve a roughly stable epidemic curve, which nowhere will be near zero.”

Metaphors for Mitigation and Relaxation

Vaccines? Not So Fast

The Banality of Guidelines

“ … how many would answer phones that display, say, ‘County Public Health’? ”

Mustering the Troops

Natural Experiment

“ … governors may … be Pied Pipers leading some constituents into the folkloric river to drown.”

Damaged Reputations

Disturbing political taint (transcript) in CDC investigation of COVID-19 outbreaks in meatpacking plants.

“One cannot have a ‘second wave’ without a trough after the current first one, which shows no sign of ebbing (Figure).”

--

--

Bruce G. Weniger, MD, MPH and Chin-Yih Ou, PhD are medical epidemiologist and research-laboratory specialist, respectively; both retired from CDC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Bgweniger

Bruce G. Weniger, MD, MPH and Chin-Yih Ou, PhD are medical epidemiologist and research-laboratory specialist, respectively; both retired from CDC.