If we don’t analyse statistics for a living, it’s easy to be taken in by misinformation about COVID-19 statistics on social media, especially if we don’t have the right context.
For instance, we may cherry pick statistics supporting our viewpoint and ignore statistics showing we are wrong. We also still need to correctly interpret these statistics.
It’s easy for us to share this misinformation. Many of these statistics are also interrelated, so misunderstandings can quickly multiply.
Here’s how we can avoid five common errors, and impress friends and family by getting the statistics right.
1. It’s the infection rate that’s scary, not the death rate
...We need to look at the infection fatality rate (IFR) — the number of COVID-19 deaths divided by all those infected (a number we can only estimate at this stage, see also point 3 below).
Second, and more importantly, we need to look at the basic reproduction number (R₀) for each virus. This is the number of extra people one infected person is estimated to infect.
Flu’s R₀ is about 1.3. Although COVID-19 estimates vary, its R₀ sits around a median of 2.8. Because of the way infections grow exponentially... the jump from 1.3 to 2.8 means COVID-19 is vastly more infectious than flu.
When you combine all these statistics, you can see the motivation behind our public health measures to “limit the spread”. It’s not only that COVID-19 is so deadly, it’s deadly and highly infectious.
2. Exponential growth and misleading graphs
A simple graph might plot the number of new COVID cases over time. But as new cases might be reported erratically, statisticians are more interested in the rate of growth of total cases over time. The steeper the upwards slope on the graph, the more we should be worried.
For COVID-19, statisticians look to track exponential growth in cases. Put simply, unrestrained COVID cases can lead to a continuously growing number of more cases. This gives us a graph that tracks slowly at the start, but then sharply curves upwards with time...
“Flattening the curve” is another way of saying “slowing the spread”. The epidemic is lengthened, but we reduce the number of severe cases, causing less burden on public health systems.
However, social media posts routinely compare COVID-19 figures with those of other causes of death that show:
3. Not all infections are cases
Then there’s the confusion about COVID-19 infections versus cases. In epidemiological terms, a “case” is a person who is diagnosed with COVID-19, mostly by a positive test result.
But there are many more infections than cases. Some infections don’t show symptoms, some symptoms are so minor people think it’s just a cold, testing is not always available to everyone who needs it, and testing does not pick up all infections.
Infections “cause” cases, testing discovers cases. US President Donald Trump was close to the truth when he said the number of cases in the US was high because of the high rate of testing. But he and others still got it totally wrong.
More testing does not result in more cases, it allows for a more accurate estimate of the true number of cases.
The best strategy, epidemiologically, is not to test less, but to test as widely as possible, minimising the discrepancy between cases and overall infections.
4. We can’t compare deaths with cases from the same date
Estimates vary, but the time between infection and death could be as much as a month. And the variation in time to recovery is even greater. Some people get really ill and take a long time to recover, some show no symptoms.
So deaths recorded on a given date reflect deaths from cases recorded several weeks prior, when the case count may have been less than half the number of current cases.
The rapid case-doubling time and protracted recovery time also create a large discrepancy between counts of active and recovered cases. We’ll only know the true numbers in retrospect.
5. Yes, the data are messy, incomplete and may change
Countries and even states may count cases and deaths differently. It also takes time to gather the data, meaning retrospective adjustments are made.
We’ll only know the true figures for this pandemic in retrospect. Equally so, early models were not necessarily wrong because the modellers were deceitful, but because they had insufficient data to work from.
Welcome to the world of data management, data cleaning and data modelling, which many armchair statisticians don’t always appreciate. Until now.
As the United States struggles to track coronavirus fatalities amid spotty testing, delayed lab results and inconsistent reporting standards, a more insidious problem could thwart the country's quest for an accurate death toll.
Between 20 and 30 percent of death certificates nationwide were wrong before COVID-19, Bob Anderson, chief of the mortality statistics branch at the National Center for Health Statistics, said in an interview with the USA TODAY Network.
“I’m always worried about getting good data. I think this sort of thing can be an issue even in a pandemic,” Anderson said.
Experts said the inaccuracies are part and parcel of a patchwork, state-by-state system of medical examiners, coroners and doctors who have disparate medical backgrounds, and in some cases none at all.
The problem is likely to get worse as the pandemic inundates overworked and sometimes untrained officials who fill out the forms.
Accurate death certificates are paramount for health officials trying to determine where to focus resources to fight the spread of the coronavirus, said Umair Shah, executive director of the Public Health Department in Harris County, Texas, which includes Houston.
“That death represents an ecosystem of people,” Shah said.
Inaccurate death reporting is a long-standing problem.
A review of Missouri hospitals in 2017, for example, found nearly half of death certificates listed an incorrect cause of death. A Vermont study found 51% of death certificates had major errors. Nearly half of the physicians the Centers for Disease Control and Prevention surveyed in 2010 admitted that they knowingly reported an inaccurate cause of death.
Death certificates regularly lack enough details to accurately pinpoint the cause of death, Anderson said.
“For example, cardiac arrest is not an acceptable cause of death, because everybody dies of cardiac arrest,” Anderson said. “That just means your heart stopped.”
Lack of expertise
The widespread inaccuracy of death certificate information stems largely from the varying levels of expertise of those who complete the forms, experts said.
Physicians, coroners, medical examiners, and in some states, other medical personnel, such as nurse practitioners, can legally sign death certificates, said Sally Aiken, president of the National Association of Medical Examiners and a practicing medical examiner in Spokane County, Washington.
Coroners and medical examiners are responsible for certificates in homicides, accidents and suicides, Aiken said. Physicians fill out the form when natural deaths, such as those caused by COVID-19, occur in a hospital. Medical examiners and coroners do it if the person died at home or in another non-health-care setting.
Medical examiners are generally physicians specializing in forensic pathology who can perform autopsies.
Coroners are not always doctors. In Alabama and Georgia, the only requirement for coroners to be elected is that they be nonfelons of legal age.
Even those with medical expertise regularly get it wrong. In Vermont, there are no coroners. If a death is natural or happens in a hospital or out in the community, physicians, nurse practitioners or physician assistants fill out death certificates. The state medical examiner’s office, which investigates violent deaths, reviews about 5,000 certificates each year to find and fix errors.
When the state medical examiner’s office compared 601 death certificates completed from July 1, 2015, to Jan. 31, 2016, with medical records, it found 51% had major errors.
Lauri McGovirn, a medical examiner who worked on that review, said some physicians didn’t complete death certificates regularly, so they were unfamiliar with the process. Others viewed it as an administrative chore.
“It does make you wonder in other states where they don’t have the type of resources or the money to review every death certificate, what their error rate may be,” McGovirn said.
Shortage of workers
In addition to expertise gaps, there’s a severe shortage of medical examiners nationwide.
In a report to Congress, the Justice Department said as many as 700 more forensic pathologists are needed. The report noted that in addition to staffing, “budgets, resources and supplies are too inconsistent to ensure that death investigations are of the same quality across the United States.”
Dr. Ray Fernandez has been the chief medical examiner for Nueces County, Texas, for 19 years. He knows what the shortage means — a punishing workload.
Despite hiring another full-time pathologist and two part-time pathologists several years ago, he and his colleagues each perform 200 to 300 autopsies per year, regularly bumping up against the National Association of Medical Examiners’ recommendation of no more than 325 per year.
The organization has temporarily suspended that caseload limit amid due to COVID-19, but Fernandez said the more cases medical examiners take on, the greater the chance they’ll make mistakes.
“COVID-19,” he said, “is impacting the system at a time when it’s already in a crisis with a shortage of people doing the work.”
To further complicate efforts to curb the spread of coronavirus, many medical examiners and coroners refuse to attribute a death to COVID-19 without a positive test before the person died. Some medical examiners are doing post mortem testing if they have the means. But with tests in short supply, that’s not always possible.
Dr. James Gill, vice president of the National Association of Medical Examiners and the chief medical examiner for the state of Connecticut, said he’s sending his staff to funeral homes to swab the noses of the deceased, which are then analyzed by an outside lab.
The family of the deceased and the first responders who attended to them need the lab results to know whether they should self isolate or get treatment, Gill said.
“You have to remember, though, that even if we are doing a swab on a dead person, those results may affect the living,” Gill said.
The National Center for Health Statistics, where Anderson works, updated its website on April 1 to clarify that those filling out death certificates should record COVID-19 as the probable cause if testing isn’t possible and if the medical records or circumstances support that...
“The fact is, a lot of these deaths are not going to be autopsied and post mortem testing is not going to be done, so we’re going to have to rely on second-hand accounts and what the symptoms were,” Anderson said. “We may miss some as a result.”