BY ANDREW BOSTOM, M.D.
John Graunt’s pioneering 1662 treatise “Natural and Political Observations on the London Bills of Mortality” has established him, appropriately, as the father of statistics, especially vital statistics, which he dubbed “political arithmetic.” Graunt recognized—almost 360 years ago—that accurate tallies of then-cataclysmic “bubonic” plague (Yersinia pestus) deaths, “the numbers which die of the Plague,” “cannot well be done,” absent comprehensive, “rational” account of all deaths.
An October 12, 2020 Florida House of Representatives report, “Analysis of COVID Death Data, March 5-September 16, 2020” (the report’s introductory memo and full report PDFs are linked below), alluded to Graunt’s seminal 1662 work in the report’s introduction, adding:
“Although greatly modernized since the 1600s, the foundation of vital statistics remains the same—it is the official source for counts and causes of death. The promise of valid data rests on an established set of rules for producing complete and consistent information about the cause(s) and conditions of each death.”
The Florida House of Representatives report relied “exclusively on death certificates.” Apropos to this reliance, the investigators provided a very lucid overview of proper death certificate recording, consistent with Centers For Disease Control and Prevention (CDC) guidelines, including an illustrative example:
“The medical information captured in a death certificate is organized in two parts. PART 1 is the place for identifying all conditions that directly caused the death, including the immediate and underlying causes. These are diseases and conditions with a direct causal relationship to the death. PART 2 is the place for identifying other contributing conditions, that is, conditions and diseases that made the death more likely but not the actual cause of death. The following example helps to illustrate this distinction.
Example of how the death certificate describes the chain of events.
PART 1 consists of four lines labeled a-d. Certifiers are directed to complete these lines in a way that indicates the sequence or chain of events that ended in the death. Line a. in PART 1 should be the immediate and direct cause of death. If only Line a. is completed, the information should indicate a disease or condition sufficient in and of itself in causing a death. Only one cause should be listed on each line. If multiple morbid conditions are present, the certifier should construct a logical sequence of clinical events and identify other significant conditions in PART 2.”
A 57pp CDC monograph, “Physicians Handbook on Medical Certification of Death,” highlighted these critical distinctions between Part 1 vs. Part II for precise, unambiguous recording of the cause of death on a death certificate:
“A specific cause of death should be reported in the last entry in Part I so there is no ambiguity about the etiology of this cause. Other significant conditions that contributed to the death, but did not lead to the underlying cause, are reported in Part II.”
The CDC’s handbook on death certification places such proper recording within a physician’s broader responsibility to individual families and communities:
“The physician’s principal responsibility in death registration is to complete the medical part of the death certificate. In fulfilling the role of the certifier (i.e., person completing the medical part of the death certificate), the physician performs the final act of care to a patient by providing closure with a well-thought-out and complete death certificate that will allow the family to close the person’s affairs. At the same time, the physician performs a service for the larger community.”
As a prominent example of this community responsibility, the CDC handbook noted:
“City and county health departments use data derived from these public records [i.e., death certificates]…to follow up on infectious diseases.”
When the CDC issued its “Vital Record Criteria” for COVID-19 in April of 2020, its own death certification standards were turned on their head. Imprecise language about death certificate Part 1 causation was added—”specific cause” was expanded to “a cause.” A further, even more inconsistent, feature was the allowance for Part 2 death certificate “contribution(s) to” death to also be used to define the cause of deaths, effectively nullifying the standard distinction between Parts 1 and 2:
“A death certificate that lists COVID-19 disease or SARS-CoV-2 as a cause of death or a significant condition contributing to death.”
The Florida House of Representatives death certificate analysis and report, in large measure, examines the impact of the CDC’s “expansive” guideline on COVID-19 death certification in Florida.
On September 23, 2020, the Florida Department of Health gave 13,920 individual death certificate records to State House staff for review. These death certificates (see also figure, below) included:
—11,460 records that listed COVID-19 as the immediate or underlying cause of death in Part 1.
—1,204 records that listed COVID-19 as one of the causes in the final sequence of events, but not the underlying cause of death.
—1,254 records listing COVID-19 as a condition that contributed to death in Part 2, but where COVID-19 was not listed as a cause of death in Part 1.
—2 records that did not list COVID-19 in either Part 1 or Part 2 [(car accident; dementia)–One must assume these individuals “tested positive” for the virus].
The House investigators wrote, “This breakdown suggests that the reported COVID deaths are increased by as much as 10% (i.e., presumably 1256/13,920) when contributing conditions are mingled with underlying causes.”
More granular analysis of the 11,460 records designating COVID-19 as the Part 1 immediate or underlying cause of death, revealed:
—8058 of these records listed COVID-19 in Line a. and listed no other causes in any other line of Part 1.
—3697 of these 8058 records specified COVID-19, solely, as both the immediate and underlying cause of death (Line a.), even though most of those (all but 246) had other contributing conditions.
—The remaining 4361 of these 8058 records reported multiple immediate and underlying causes, including COVID-19, in a single line (Line a.): 3,756 listed pneumonia; 265 listed respiratory failure; 218 listed pulmonary complications; 175 listed an assortment of immediate and underlying causes unrelated to COVID-19 (dementia, cancer, stroke, diabetes, etc.)
—1494 records identified COVID-19 as the underlying cause, or starting point, for the chain of events leading to death resulting from one immediate cause.
—1669 records listed COVID-19 as the starting point for the chain of events leading to death resulting from two other causes.
—239 records listed COVID-19 as the starting point for the chain of events leading to death resulting from three other causes.
—Nearly 80% of the 11,460 patients died in a hospital. 1,345 of the 11,460 (~12%) died in hospice care, “suggesting an advanced disease state unrelated to but possibly aggravated by COVID.”
—Over 81% of the 11,460 deaths occurred in patients aged 65 years or older.
The Florida House investigators stressed that “nearly 60% [8,058/13.920] of the records classified by the Department of Health,” had “errors” or were “recorded in a manner inconsistent with state and national guidance.” They also provided evidence, as noted earlier, that ~10% [1,256] of these alleged COVID-19 deaths were mis-classified.
I maintain that this latter ~10% estimate is far too conservative, and moreover that it reflects compliance with the CDC’s April 2020 COVID-19 death certificate coding guidelines, which destroyed the logical firewall between Part 1 “specific cause” of, and Part 2 “contribution(s)” to, individual deaths. Accordingly, to the 1,256 records where COVID-19 is not even listed as the final cause of death, one could add the 175 records listing non-COVID-19 immediate and underlying causes, 1,345 deaths occurring in hospice care, and the [3,697-246=] 3451 that somehow classified COVID-19, alone, as both the immediate and underlying cause of death despite acknowledging contributing co-morbidities, or “conditions.” This aggregate mortality total indicates, plausibly, that up to 45% (6,227/13,920) of Florida’s death-certificate-recorded “COVID-19 deaths” may not merit that classification.
A cover memo (available in full, below) written by Speaker José R. Oliva to introduce the Florida House report warned, appositely,
“…national guidelines drive the [COVID-19 death] count up. CDC has determined that the death count should include persons with COVID even if COVID is not an underlying cause of death.”
The Speaker concluded:
“I believe that these two categories are fundamentally different. They should both be noted, but not combined.”
All honest public health and healthcare professionals, as well as local, state, and national political leaders, should agree with—and abide by—Speaker Oliva’s wise, plainspoken observations. Cynical rejection of this counsel, due to modern “political arithmetic,” will distort evidence-based assessment of the COVID-19 outbreak and compromise rational efforts to resolve it.
Andrew Bostom is an associate professor of family medicine at Brown University and a trained clinician, epidemiologist, and clinical trialist.