Summary

The Washington State Department of Health (DOH) recently announced it will subtract hundreds of deaths from its count of COVID-19 fatalities as a result of methodological improvements.

However, while the changes are welcome, the Freedom Foundation’s analysis indicates they may still not be enough to correct continued deficiencies in DOH’s reporting of deaths due to COVID-19.

Combining several data sources obtained from DOH, the Freedom Foundation has compiled and analyzed the cause of death information from the death certificates of the nearly 2,000 fatalities DOH had attributed to COVID-19 as of early September.

The analysis shows that 170 death certificates contained no reference to COVID-19. Another 171 only reference COVID-19 as a possible “contributing factor” and not part of the causal chain of events leading to death. Dozens of other death certificates indicated that deaths attributed by DOH to COVID-19 had only a questionable or minimal connection to the virus.

While delays in the ability to obtain information from DOH make it impossible to say how may of these questionable cases remain in the state’s tally, the information undermines the claims of state officials that their response to COVID-19 is being guided by the best available science and data.

DOH should either ensure that the deaths it attributes to COVID-19 are directly linked to the virus or begin providing additional details about the deaths it counts so the public can more accurately understand the significance of the state’s data.

DOH announces long overdue methodological changes to increase accuracy of COVID-19 death tally

On Thursday, amid a Freedom Foundation investigation into the data underlying its reporting, DOH announced it will remove more than 200 deaths previously attributed to the virus from its COVID-19 data dashboard.

The deaths will be removed as a result of several, long overdue methodological changes DOH says it will implement over the next two weeks, though Freedom Foundation analysis suggests the changes may not be enough to fully correct the state’s over-counting of COVID-19 fatalities.

First, since the start of the pandemic, DOH has attributed to COVID-19 the death of any person who tested positive for COVID-19 before their death. In May, the Freedom Foundation reported how this method resulted in the inclusion of deaths clearly unrelated to COVID-19. After admitting that it was attributing even deaths caused by things like gunshot wounds to COVID-19, DOH announced it would take steps to clean up its reporting process and removed some of the most obviously inappropriate deaths from its COVID-19 tally.

In its Thursday press release, DOH noted it would now “assume a death is caused by COVID-19” if a person tests positive “28 days prior to death,” a change it says will result in the removal of 68 deaths. While a definite improvement, this method could still result in the inclusion of some deaths not actually caused by COVID-19, as some people who test positive for COVID-19 will subsequently die as a result of other factors.

Second, DOH says it will no longer attribute deaths to COVID-19 before they are fully registered in state databases, a process that can take days or weeks. DOH says this will affect about 150 deaths, but that it expects some to be “added back once the cause of death is investigated and officially determined to be due to COVID-19.”

And, finally, DOH will begin to “exclude deaths in which the certifier (the person completing the death certificate) mentions positive COVID-19 tests, but later tests were negative and significant time elapsed between tests.”

If DOH actually follows through, these changes should increase the accuracy of the state’s data.

Back in June, after the Freedom Foundation’s original reporting on its methodological flaws, DOH promised it would “expand how we report deaths” to indicate whether: (1) “COVID-19 contributed to the death”; (2) “COVID-19 probably contributed to the death”; (3) “COVID-19 is suspected to have contributed to death”; or (4) “COVID-19 did not contribute to the death.”

But, six months later, DOH still hasn’t changed its practice of reporting a single, cumulative number of purported COVID-19 deaths in its data dashboard.

And even if the promised changes are implemented, Freedom Foundation research shows that DOH data on COVID-19 deaths remains far from perfect.

Obtaining DOH data on COVID-19 deaths

Getting timely COVID-19 data from DOH has been challenging. DOH does not provide detailed information about COVID-19 deaths on its website. Behind the scenes, DOH adds and subtracts deaths from its single dashboard number as it believes appropriate, but the underlying data is not accessible online. All the public sees and the press reports on is the single number of deaths on the dashboard.

However, after months of effort, the Freedom Foundation obtained enough information from DOH to review the cause of death information from the death certificates corresponding to most of the deaths reflected on the DOH dashboard.

While the Public Records Act (PRA) provides a mechanism for obtaining most government documents and records, death certificates are governed by separate, and more restrictive, state laws. Still, death certificate information, including cause(s) of death but excluding personally identifying information, can be purchased from the DOH Center for Health Statistics. Such data is made available quarterly. From June through December, the Freedom Foundation was able to purchase data for the first three quarters of 2020.

Having death certificate information alone, however, does not indicate which deaths DOH attributes to COVID-19 on its dashboard. Accordingly, on Sept. 4, the Freedom Foundation submitted a public records request to DOH for the unique “state file number” (SFN) for each death certificate corresponding to a death DOH attributed to COVID-19 in its dashboard. Initially, DOH estimated it could produce the list of SFNs by Sept. 30, but subsequently extended its estimate by eight months, to June 2021. After much back-and-forth, DOH finally produced the list of SFNs on Nov. 25 — 82 days after the request was submitted.

DOH provided the Freedom Foundation with the third quarter death certificate information on Dec. 8, just two days before DOH announced the changes to its methodology for counting COVID-19 deaths.

The Freedom Foundation subsequently matched the two files, allowing it to identify the death certificate information for all deaths attributed by DOH to COVID-19 as of Sept. 4.

Analysis shows DOH COVID-19 death toll continues to be questionable

The Freedom Foundation’s analysis of the information obtained from DOH suggests both that: (1) state officials have attributed many deaths to COVID-19 that appear to have little-to-no connection to the virus; and, (2) many death certificates are improperly completed, limiting the ability to draw overly definitive conclusions from the data.

As of Sept. 4, DOH attributed 1,953 deaths to COVID-19 on its dashboard. DOH provided the Freedom Foundation with the death certificate SFNs for 1,937 deaths reflected on the dashboard as of that date. The 16 death difference between reported deaths in the dashboard on Sept. 4 and the number of corresponding death certificates produced to the Freedom Foundation is likely because either: (1) DOH removed those deaths from its dashboard between Sept. 4, when the Freedom Foundation requested the list, and Nov. 24, when DOH provided the list to the Freedom Foundation; or (2) the dashboard on Sept. 4 included 16 deaths for which there was not yet a corresponding death certificate.

Of the 1,937 SFNs produced by DOH, all but one corresponded to a death certificate in the death data files provided by DOH covering January-September.

The Freedom Foundation’s analysis of the death certificates corresponding to the remaining 1,936 deaths attributed by DOH to COVID-19 shows the following.

1. 1,766 (91.2 percent) include some reference to COVID-19 as a cause of death or a condition contributing to death; 170 (8.8 percent) death certificates make no reference to the virus whatsoever.

The following are some examples of deaths attributed by DOH to COVID-19 that lack any mention of the virus on the death certificate:

• a 64-year-old male who died of “acute combined fentanyl, heroin, methamphetamine, and methadone intoxication”;
• a 65-year-old male who died from “alcoholic liver disease”;
• a 69-year-old male suffering from Parkinson’s and vascular dementia who died from malnutrition/dehydration after refusing to eat;
• a 73-year-old female with underlying health conditions who died after declining treatment for an intestinal abscess;
• a 75-year-old-male who died following a “pacemaker infection”; and
• a 99-year-old female who died after losing her balance and falling while trying to retrieve an item from the top of her dresser.

DOH directions explain that death certificates are to be completed according to the “best medical opinion” of the person completing the certificate (the “certifier”). However, these judgement calls are often made without access to perfect data, or by people with limited knowledge at their disposal, and inherently involve a degree of subjectivity, so it’s possible COVID-19 may have played a role in some of these deaths.

On the other hand, DOH’s method of attributing these deaths to COVID-19 solely on the basis of a prior positive test almost certainly results in counting at least some deaths not reasonably caused by the virus.

2. Of the 1,766 death certificates containing a reference to COVID-19, 1,595 (90.3 percent) reference the virus in Part I, which is intended to list the causal chain of events leading to death.

In many of these deaths, the role of COVID-19 is uncertain or doubtful. For instance:

• An 87-year-old female was counted as COVID-19 fatality even though her death certificate stated she “was asymptomatic from a Coronavirus-19 standpoint.”
• An 85-year-old female who died the same day she suffered a fall and broke her hip was counted as a COVID-19 death.
• An 84-year-old male who contracted COVID-19 in May and died in July after repeatedly testing negative for the virus several weeks prior was counted as a COVID-19 death.
• 22 death certificates mentioning COVID-19 in Part I state that its role was “possible,” “presumed,” “likely,” “probable,” or “suspected.”

In many other cases, the death certificates do not appear to have been completed properly according to government directions.

Centers for Disease Control and Prevention (CDC) guidance for completing death certificates explains that Part I of a death certificate:

“…is for reporting the sequence of conditions that led directly to death. The immediate cause of death, which is the disease or condition that directly preceded death and is not necessarily the underlying cause of death (UCOD), should be reported on line a. The conditions that led to the immediate cause of death should be reported in a logical sequence in terms of time and etiology below it. The UCOD, which is ‘(a) the disease or injury which initiated the train of morbid events leading directly to death or (b) the circumstances of the accident or violence which produced the fatal injury’,  should be reported on the lowest line used in Part I.”

As for COVID-19 specifically, the CDC explains:

“If COVID–19 played a role in the death, this condition should be specified on the death certificate. In many cases, it is likely that it will be the UCOD, as it can lead to various life-threatening conditions, such as pneumonia and acute respiratory distress syndrome (ARDS). In these cases, COVID–19 should be reported on the lowest line used in Part I with the other conditions to which it gave rise listed on the lines above it.”

So, a reasonable death certificate involving COVID-19 might read something like this: COVID-19 infection on line D as the UCOD, leading to viral pneumonia on line C, leading to acute respiratory distress syndrome on line B, and ending with acute respiratory failure with hypoxia on line A.

However, many of the death certificates that reference COVID-19 in Part I do not follow this guidance. For instance, around 100 death certificates list nothing other than “COVID-19” or some variation in Part I and provide no explanation of the virus’ role in the causal chain leading to death.

Additionally, 206 death certificates do not list the virus on the lowest line as the UCOD triggering the chain of events leading to death. Instead, some certifiers appear to use Part I as a place to list factors contributing to death in no particular order.

For instance, in one case involving the death of a 78-year-old male, line B of the death certificate indicates the UCOD was diabetes leading to “COVID-19 pneumonia” on line A. In another case involving the death of an 85-year-old male, line B lists emphysema — a lung condition common to smokers — as the UCOD leading to “COVID-19” on line A. But, as neither diabetes nor emphysema directly lead to COVID-19 infections, these conditions probably shouldn’t have been listed in the direct causal chain leading to death in Part I. Instead, the certifiers probably should have included them on Part II as “significant conditions that contributed to the death,” as directed by CDC guidance.

3. Of the 1,766 death certificates containing a reference to COVID-19, 171 (9.7 percent) reference COVID-19 in Part II, the “conditions” section, indicating the decedent may have been infected by or exposed to the virus prior to or at the time of their death, but it was not part of the direct chain of events leading to death.

Many of the death certificates in this category do little more than note that the person was positive for COVID-19 at some point. In other cases, such as the following, the death certificates suggest the virus probably didn’t play much of a role.

• A 35-year-old female whose UCOD was “methamphetamine abuse” was counted as a COVID-19 death because the virus was referenced without explanation in Part II of her death certificate.
• A 92-year-old male whose UCOD was a “traumatic” catheter insertion was counted as a COVID-19 death because the virus was listed as one of eight contributing factors on Part II of his death certificate.
• A 71-year-old female was counted as a COVID-19 death even though her UCOD was “blunt force injury of the head and neck” suffered in a fall.
• An 82-year-old female who died from “complications of hip fracture” sustained in a fall was counted as a COVID-19 death because she had “positive COVID-19 test results.”
• The deaths of a 79-year-old female and a 92-year-old female were attributed to COVID-19 even though the only reference to the virus on their death certificates noted that they had “recovered” from the virus before their passing.
• Similarly, a 76-year-old male and a 72-year-old female who died as a result of injuries suffered in falls were counted as COVID-19 deaths because of their “history of COVID-19 infection.”
• An 86-year-old man with other medical conditions who died of a “urinary tract infection” but was “COVID positive” was included in the tally.
• A 98-year-old man who died of bladder cancer but was positive for COVID-19 was counted by DOH as a COVID-19 death.

Conclusion

Any life lost or cut short — whether due to COVID-19 or other, less headline-grabbing causes — is tragic. Each death touches the living in some way and serves as a reminder of the brokenness of the world we inhabit. Reading thousands of death certificates is nothing if not sobering.

It is precisely because human life is so precious that state policymakers should do their utmost to be guided by and grounded in sound policy. The decisions they make, especially in times such as these, can save or cost lives and affect, for good or ill, the quality of life of millions of Washingtonians. Decisions made to repeatedly shutdown aspects of Washington state’s social and economic life, while perhaps made with the intention to save lives, do not come without consequences of their own.

Evaluating the costs and benefits of these responses to COVID-19 becomes more difficult without good data. While state officials repeatedly assure the public their decisions are guided by the best available science and data, the DOH’s handling of COVID-19 death reporting calls those assurances into question.

Because of DOH’s recent changes to its methodology and unwillingness or inability to release information in a timely fashion, it is not possible to know at this time whether or which of the above deaths continue to be attributed to COVID-19 in the state’s data dashboard. All that can be concluded for certain is that, at least eight months into the pandemic, DOH was counting many deaths that appear to have had, at best, a tenuous connection to COVID-19.

It also extremely disappointing that DOH has failed to follow up on its months-old promise to provide additional details about the deaths it attributes to COVID-19.

Perhaps most concerning is the lack of context surrounding the single, ever-changing tally of deaths presented by DOH in its data dashboard. As this is a “novel” virus, it makes sense that public health officials would want to capture as much data about it as possible and would, accordingly, flag any death that may have had anything to do with COVID-19.

The problem is when state officials represent this number to the public as the definitive number of deaths due to COVID-19, as they routinely do, without acknowledging the limitations of the information relied upon or providing context to help understand what the number really means.

Accordingly, DOH should either: (1) update its methodology to ensure the number of deaths it directly attributes to COVID-19 is accurate and defensible; or, (2) provide the additional information about COVID-19 deaths it has promised to start releasing since June so the public can better understand the meaning and significance of the numbers cited by state officials.

The cause of death information for each of the deaths attributed by DOH to COVID-19 as of Sept. 4 is available here. Cells referencing COVID-19 in some way are highlighted yellow.