Excess deaths measure a work in progress as CDC revises metric
PARKERSBURG, W.Va. — The CDC has revised a metric that showed higher-than-expected death statistics around the country but whose accuracy was inconsistent from state to state.
A report released by the federal Centers for Disease Control and Prevention at the end of April showed approximately 66,000 more deaths nationwide than anticipated, based on historical data. Deaths over that predicted line are considered “excess deaths,” many of which – but not all – can be attributed to the novel coronavirus.
“Mortality has increased as a result of the pandemic,” said Bob Anderson, chief of the mortality statistics branch of the CDC’s National Center for Health Statistics.
Those deaths can generally be grouped in one of three categories: those caused by COVID-19 and recorded as such, those caused by the virus but attributed to something else and those indirectly connected to the circumstances. That could include people forgoing medical care or suffering a heart attack at home, alone, with no one to call for assistance.
“This is true with any sort of disaster or pandemic. … It wouldn’t have happened if the disaster hadn’t happened,” Anderson said. “I think we’re seeing this around the country, where people are actually afraid to go to the emergency room because they’re afraid they’re full of sick people.”
The Associated Press reported an excess deaths figure of nearly 24,000 in New York City alone between March 11 and May 2.
But the numbers raised questions elsewhere, like in West Virginia, where statistics posted as recently as Tuesday showed more than 1,900 excess deaths in the Mountain State from Feb. 2 to April 18. The total number of COVID-19 deaths reported by the state as of Friday was just 62.
Dr. Cathy Slemp, West Virginia state health officer and commissioner of the Bureau of Public Health, disputed the numbers when asked about them in a May 4 virtual press conference.
“When we saw the numbers, we actually did not think it resonated with the ground truth of what we see in our communities in terms of deaths,” she said.
Anderson agreed, saying an algorithm is used to account for the incompleteness of death data, as it may take more than a week for a death to be recorded. It’s based on past information, including how quickly a state has reported data in previous years.
“They’re still fairly slow, but they actually got much better in 2020,” Anderson said of West Virginia. “And so we ended up over-weighting for them.”
Allison Adler, director of communications for the West Virginia Department of Health and Human Resources, said the improved timeliness can be attributed to better staffing levels and a streamlining of the process, “as well as increasing the reporting frequency to the National Center for Health Statistics.”
The revised statistics show West Virginia with a range of excess deaths from 183 to an upper limit of 517.
Anderson said the lower number is closer to what the CDC was publishing before, albeit with the adjustments made to account for the state’s improved timeliness. The lower number is the more conservative estimate, taking into account a statistical “confidence interval” so as not to count deaths as excess until they exceed a level that couldn’t be explained simply as a random variation.
Earlier in the week, the statistics for neighboring Ohio showed fewer deaths so far this year than would have been expected under normal circumstances.
“We wouldn’t expect the total number of deaths to go down under the circumstances,” Anderson said.
The earlier numbers were “under-weighting” Ohio, he said. The revised formula shows Ohio with 1,735 excess deaths, seventh highest among states, including New York without New York City’s statistics.
“This is more like what we would expect to see from Ohio, given what we are hearing versus what we were seeing before,” Anderson said.
The Ohio Department of Health had reported 1,534 COVID-19-related deaths as of Friday afternoon.
A request for comment from an ODH spokesperson had not been answered as of Friday.
Some states show very few excess deaths, including Iowa and Kansas with zero through May 2. Those states’ upper limits are 135 and 160, respectively.
As of Friday, the Iowa Department of Public Health had reported 336 total deaths related to the virus, while the Kansas Department of Health and Environment listed 172 deaths as a result of it.
Anderson said it’s hard to know for sure what the CDC’s numbers for those states indicate.
“I think we just need to wait and see when more data are available,” he said.
The CDC statistics break down deaths in states by weeks, but the numbers do not immediately reflect the actual number of people who died that week.
Anderson said across most states, funeral directors typically provide personal and demographic data on the deceased while medical examiners and coroners are responsible for cause of death. Some states utilize electronic systems into which those individuals can enter data directly. Others provide the information on paper certificates, after which state employees enter it into a database.
West Virginia still uses paper certificates, Anderson said, which contributes to its slower data entry. Adler said the state is in talks with a contractor to develop an electronic death registration system.
Ohio has moved to an electronic system but it is not uniform across all 88 counties, he said.
“Obviously, the more data you get, the more accurate your picture,” Anderson said.
That’s why Anderson said it’s too early to draw conclusions from the excess death statistics.
“We need more data to come in before we can really say anything definitive,” he said. “But we are seeing excess deaths, so that is concerning.”