Fact check: No, CDC has not adjusted U.S. coronavirus death toll downward

Posted at 11:51 AM, May 05, 2020
and last updated 2020-05-05 19:31:00-04

CINCINNATI — Despite recent social media chatter to the contrary, the Centers for Disease Control and Prevention has not revised its coronavirus death toll count downward, as of May 2.

Social media posts started circulating over the weekend pointing to CDC numbers showing a much lower number of deaths attributed to COVID-19. Many said the agency had adjusted its death count on Friday. Some used that data to perpetuate conspiracy theories of over-inflated death tolls, in some cases arguing the higher numbers had been used to stoke fear and to justify shutdown decisions.

On Saturday, WCPO asked the CDC directly if it had revised the death count and asked about the apparent discrepancy in the numbers. A spokesperson forwarded this response:

"The COVID-19 death count shown here includes deaths reported daily by state, local, and territorial health departments. This count reflects the most real-time information CDC has based on preliminary reporting from health departments.

"In contrast, provisional COVID-19 death counts from NCHS are based on information collected from death certificates. It can take several weeks for death certificates to be submitted, processed, coded and tabulated, so these data lag and may not include all deaths that occurred during a given time period, especially for more recent periods. Death counts from earlier weeks are continually revised and may increase or decrease as new and updated death certificate data are received."

As this statement indicates, the claim that the CDC adjusted its death count downward is not true. Rather, the perception of a revision is the result of comparing two different sets of numbers without proper context.

The CDC total COVID-19 death count stood at roughly 67,456 deaths as of May 4, 2020. At the same time, the provisional death count from the National Center for Health Statistics (NCHS) was reported at 38,576.

The agency said there can be a lag time in the reporting of provisional deaths because of the wait for death certificates. The provisional number updates weekly and can be adjusted up or down over time.

Meanwhile, the total count updates throughout the day and includes real-time preliminary reports from various health departments.

The CDC website also explains other possible count discrepancies:

  • Provisional death counts may not match counts from other sources -- like media reports or even local health departments.
  • CDC counts can track one or two weeks behind other counts because it takes time for local authorities to complete death certificates, and waiting for test results can create further delays.
  • Some states report deaths at different rates than others: Currently 63% of all U.S. deaths are reported within 10 days, but that varies widely.
  • It takes extra steps to designate a death as a COVID-19-related death: NCHS can code the vast majority -- 80% -- of deaths within 10 minutes, but COVID-19 deaths must be coded manually. That adds an average of seven days to the reporting process.
  • Other agencies' reporting systems might use different definitions or methods when counting deaths.

The agency recently provided extensive technical notes on its data reporting for COVID-19 on its website.

WCPO used internet archives to cross-reference the numbers, using both one- and two-week lag times. We wanted to see if the preliminary numbers proved accurate after the death certificate counts were reported, and we wanted to see if the count had been revised down at any point.

For April 24, one week from the May 1 reporting that sparked online outrage over the death count, the CDC reported a total of 50,439 COVID-19 deaths. The NCHS provisional death count was 24,555.

For April 17, the CDC reported a total of 33,049 deaths. That same day, the provisional death count was 13,130.

The agency reported a total count of 16,570 deaths on April 10, compared to 4,984 provisional deaths that day.

Using these archives, we can see that the provisional numbers are in fact trending higher than the totals from two weeks prior.