The accuracy of death certificates has never been more important

A worker walks down a hallway inside Phoebe Putney Memorial Hospital in Albany, Ga.
Officials at the Centers for Disease Control and Prevention issued guidance this spring aimed at improving the accuracy of death certificates as the novel coronavirus spread across the nation.
(Ben Roberts)

A death certificate is a final marker in someone’s life — an official accounting of the end.

The legal forms are used by families to settle estates and by public health officials to assess whether the healthcare system is addressing the actual reasons for our demise.

Despite their importance, study after study has shown that as many as half of them are wrong.


Faced with SARS-CoV-2, a highly contagious coronavirus that scientists had never seen before, federal officials at the Centers for Disease Control and Prevention issued guidance this spring aimed at improving the accuracy of what doctors identify as the cause of death.

On death certificates, doctors have often blamed heart disease and other chronic conditions for deaths actually caused by viruses, lethal bacteria and other infections — leading health officials to underestimate their role in American mortality.

The CDC’s guidance, in essence, reminded doctors to ask a basic question: Why did the patient die when they did? If the doctor believed COVID-19 had cut the patient’s life short, the disease should be written on the death certificate as the underlying cause of death, the rules said.

“If you know why people die, very often then you can develop programs to prevent people from dying from those causes,” explained Robert Anderson, chief of the mortality statistics branch at the CDC’s National Center for Health Statistics, which issued the guidance in April.

The April guidelines were needed, officials say, to trace the coronavirus and understand how to stop its spread. But the move has been attacked by critics who believe it has inflated the death toll of the coronavirus by wrongly blaming COVID-19 for unrelated fatalities.

Infectious disease experts say the new guidance is crucial to understand why overall U.S. deaths surged in April to more than 40% above what would otherwise be expected that time of year and even now remain elevated by about 10%.

Scientists are trying to determine how many of those “excess deaths” were caused by the virus and how many were from indirect consequences of the pandemic and stay-at-home orders. Recent studies have shown, for example, that some people delayed needed emergency medical care as officials ordered hospitals to be cleared to focus on virus victims.

And they say the CDC’s guidance could also be a model for capturing deaths from other serious infections, such as drug-resistant bacteria, that have long gone missing from death certificates.

“Death is sometimes a complicated thing,” said Michael Osterholm, director of the Center for Infectious Disease Research and Policy at the University of Minnesota. “You need to know all the different factors that come to play in it.”

Missing infections

Despite death certificates being the best documentation of the nation’s mortality, the CDC has not updated its guidelines for doctors on determining the cause of death since it published a handbook with instructions in 2003.


Among that handbook’s examples is a man suffering from heart disease and diabetes who dies from pneumonia. The book instructs doctors to select either heart disease or diabetes — not the infection — as the underlying cause of death.

Death certificates in the U.S. have so infrequently captured deaths caused by infections that the CDC doesn’t even use the forms to track another lethal threat: bacteria that have grown so powerful they can no longer be treated with antibiotics.

Instead the agency has tried to estimate the deaths from those drug-resistant superbugs with calculations that scientists have found to seriously understate the loss of life.

Last year, Dr. Jason Burnham, assistant professor at the Washington University School of Medicine in St. Louis, and two colleagues estimated that multidrug-resistant germs were annually killing as many as 162,000 Americans — more than four times the CDC’s estimate of 35,000.

When they looked beyond those superbugs and added in deaths from other infections, Burnham’s team estimated that more than 500,000 Americans could be dying each year from infectious diseases, making it the nation’s third-leading cause of death. On death certificates, Burnham said in an interview, most of those deaths are instead blamed on heart disease, cancer and other preexisting conditions.

Part of the problem, he said, is that doctors get little training on how to determine the cause of death.

“If we had a training session about how to write a death certificate, it was probably during the first week of our residency,” Burnham said. “That’s when you’re learning how to log in to the electronic medical record system, you’re trying to figure out how to keep people alive. If death certificate training was in there, I don’t remember it at all.”

In the guidance the CDC issued in April, the agency provided a reminder of the methods used for decades to determine the cause of death. The agency then applied those techniques to COVID-19.

A key lesson is to be specific. Writing that someone died of “old age,” for instance, is not allowed.

Another problem is imprecision: Doctors often blame deaths on hearts or lungs failing when that is simply what happens when people die. The question is: Why did the heart and lungs fail at that moment in time?

To help doctors be more definitive, the nation’s standard death certificate has long included four lines, if that many are needed, to list the sequence of events or medical conditions that led to the person’s demise. The doctor is expected to start with the immediate cause of death and work back to the root cause.

The guidelines give an example of a woman who tested positive for the coronavirus and dies after being on a ventilator in intensive care. The CDC suggested the lines in the certificate’s cause-of-death section would be filled out like this:

Immediate cause: Acute respiratory distress syndrome (Translation: She couldn’t breathe.)

Due to: Pneumonia (An infection in the lungs.)

Due to: COVID-19 (The disease caused by SARS-CoV-2.)

The final line becomes the underlying cause of death, which the CDC uses for its national mortality statistics.

The patient in the example was 34 years old with no known preexisting health conditions — not a typical victim of COVID-19. By far, the majority of those who have died have been elderly and had serious, chronic health conditions, making it harder to determine why they died.

The guidance also said that doctors could record a death as “probable COVID-19” if the patient had not tested positive for the virus but the doctor believed with a “reasonable degree of certainty” that the virus had killed them.

For decades, a similar standard has been included in autopsy guidelines, which suggest doctors use words like “probable” or “presumed” if they are not certain of the cause of death.

If the patient had preexisting illnesses that the doctor believed had contributed to the death but not ultimately caused it, the CDC said those conditions should be mentioned in “Part II,” a separate section of the death certificate not generally used for the nation’s primary mortality statistics.

Questions of overcounting

On April 16, the CDC hosted a national conference call to explain the guidelines. Doctors on the call had many questions.

One caller asked if the CDC was concerned that virus deaths would be overcounted by allowing words like “probable COVID-19” where there were no lab results to confirm the person was indeed infected.

A CDC scientist, Farida Ahmad, responded that just 10% of the certificates so far had used words like “probable” or “likely.”

“We know that testing isn’t always available before or after death,” Ahmad said. “We’re not worried that we’re overcounting.”

Another doctor asked: “If patients have multiple serious medical conditions and I am not sure which one is the underlying cause of death, how do I then report the cause of death?”

“This is a very common question,” Lee Anne Flagg, a CDC statistician, responded.

She explained that doctors should “use their best medical judgment” and “just pick the one cause” they believed “had the greatest impact on the death.”

Lawrence Muscarella, a Pennsylvania-based consultant who advises hospitals on infection control, told The Times he was struck by how differently the CDC was handling infections caused by the coronavirus compared to those from drug-resistant germs, including some known to kill 50% of those they infected.

He pointed to several CDC investigations of hospital outbreaks in the past where agency scientists had said they could not determine whether the drug-resistant germ killed the patients because they were already seriously ill with other conditions when they were infected.

“Why is there a different standard?” Muscarella asked.

In an interview, Anderson at the National Center for Health Statistics said the agency had followed the same principles in its April COVID-19 guidance as it had used for years and as laid out in its 2003 handbook.

“The physician has to figure out what the chain of events leading to the death was,” he said, “and what condition initiated that chain of events.”

“There’s no default cause,” he said. “You don’t say, ‘Oh, because they had COVID, COVID is the cause of death.’”

State and local governments have the legal responsibility to record deaths, which the CDC then compiles. Some public health departments have set definitions for deaths from COVID-19 that go beyond what the CDC outlined in April and that could include cases where it is uncertain whether the virus led to the person’s demise.

Los Angeles County public health officials said their tally of COVID-19 deaths includes any person who died from a heart attack, stroke or another ailment if they had tested positive for the coronavirus within the last 90 days.

That directive could end up including some people who would have died even if they were not infected.

“You can’t use an automatic time period cut off,” said Osterholm, the Minnesota infectious disease expert.

County officials explained that they believed that “only looking at people with a known COVID-19 diagnosis will undercount” the virus deaths. They said they were logging deaths that they couldn’t determine were directly caused by the virus as “COVID-19 associated.”

In Oregon, state public health officials said they include anyone who had tested positive within 60 days of death, including those who died from accidents such as automobile wrecks.

Oregon officials said the rules were needed to track COVID-19 and “understand the severity of disease in Oregonians.”

The CDC is also attributing more deaths to COVID-19 than agency officials had described in the April guidance and April 16 conference call.

Anderson said that if a doctor doesn’t directly blame a death on COVID-19 but includes it in Part II, the section of the death certificate for conditions contributing to the death, the agency is including it in the count of deaths due to the virus.

“It’s pretty standard procedure in a pandemic like this, you just collect everything, any mention of COVID on the death certificate, or whatever thing that you’re looking at,” he said. “You want to cast a slightly wider net when you’re doing surveillance.”

He said in its official statistics, the agency would present two categories of virus deaths — those where COVID-19 was the cause of death and those where it was a contributing factor. So far, only 8% of the death certificates mentioning COVID-19 listed it as a contributing factor, he said.

Indirect victims

Sorting out the lessons in the death certificates once they are finalized could take years.

Already there is evidence that it isn’t just the virus — attributed to more than 200,000 deaths in the numbers reported by states so far — that has ended American lives too soon.

Emergency room visits for heart attacks and strokes fell this spring, raising concerns that some people who were scared to visit a hospital needlessly died at home rather than calling for help. Other Americans delayed surgeries and necessary medical care such as chemotherapy as public health officials ordered hospitals to focus on treating virus victims. And doctors have warned that suicides and accidental overdoses have surged because of isolation, job losses and despair.

In a 219-page report last month, a committee at the National Academies of Sciences, Engineering and Medicine said that many direct deaths from COVID-19 have likely been missed, especially early in the outbreak when tests were not widely available.

The deaths caused indirectly by the pandemic will take longer to discover, the report said. For example, suicides, overdoses and heart attacks brought on by pandemic-related stresses could go on for years.

When it comes to recording deaths from disasters, the report said, there is often a “dramatic undercounting.”

Anderson said the CDC would be using death certificates to tally both direct deaths from the virus and those from the pandemic’s indirect effects, including the toll it has taken on the nation’s mental health.

Death certificates documenting suicides and overdoses, he said, take longer to finalize because those unnatural deaths must be investigated by coroners or medical examiners.

“Deaths could be attributed to the pandemic but not to the virus,” he said. “This is really hard to sort those out.”