Why excess death analyses are such a powerful tool for newsrooms

Even if a public agency reports a number of deaths attributed to a given cause, that doesn’t mean the figures are accurate.

Andrew Ba Tran knows this well.

As an investigative data reporter for The Washington Post, he has written several stories detailing how the actual death toll of the pandemic is almost certainly far higher than what the government has reported.

He spoke to reporters at the 2021 Data Fellowship this week via Zoom about how they can tell stories about so-called excess deaths in their own coverage areas.

“There's so much potential to do this locally because the (Centers for Disease Control and Prevention) only provides data at the level of the state,” Tran said. “But if you're able to get the data from your local medical examiner, you can tell so many stories yourself that probably no one else has done yet.”

Tran was introduced to the concept of excess deaths in 2017, when The New York Times published a story about how Puerto Rican officials were likely vastly understating the number of deaths from Hurricane Maria (the country’s government eventually acknowledged as much). The newspaper had compared the death totals around the time of the hurricane to the same periods in 2015 and 2016.

Early in the pandemic, The Post teamed up with a research team led by the Yale School of Public Health to estimate the true number of COVID-19 deaths. Some states, like North Carolina and Pennsylvania, were slow to release provisional death counts, with some medical examiners still reporting deaths by paper.

Tran said experts are now estimating the virus may have already caused 10.2 to 19.3 million deaths worldwide (compared to about 5 million officially) and around a million in the United States (nearly 750,000 is the official death count).

The concept of excess deaths also includes fatalities that weren’t directly caused by COVID-19 — it could include someone who skipped regular medical care due to overrun hospitals and died from another disease, for instance.

But Tran said the discrepancies between official counts and the excess deaths estimates were also caused by the length of times it takes to finalize death certificates — as well as who’s filling out the forms. The Kansas City Star reported that in one Missouri county, the coroner would exclude COVID-19 from the death certificates upon the family’s request. STAT found that unreported coronavirus deaths were greatest in areas that supported former president Donald Trump, with the largest inconsistencies in counties with elected coroners.

Before the pandemic, the CDC provided data on deaths two years after the fact. The agency now releases provisional data on all deaths on a weekly basis.

Excess deaths analyses have caught on. News outlets have been taking a look at other mortality trends, such as spikes in deaths from heart disease and Alzheimer’s and dementia, the overall increase in deaths among communities of color, and the surplus death tolls of the Texas winter storm and Northwest heat wave.

The Post has also reported on undeclared COVID-19 deaths in Russia, while The Economist tracks excess coronavirus deaths in a variety of countries.

Tran said reporters don’t have to be data and coding experts to tell this story.

“You don't need to know how to model. You don't need to create a sophisticated or even simple model that comes up with a baseline that adjusts for population growth or seasonality. You don't need to do that,” he said “All you need is recent deaths data and historical death data. And you need something to compare it to in the past — it could be one point in time or an average.”

Since the CDC releases only state-level mortality data, reporters trying to drill down further will need to request data from local agencies, as some news outlets have already done in writing stories about at-home deaths and the Texas freeze.

This type of data reporting can be done for other types of stories as well. For example, The Post recently reported on the surge in firearms sales during the pandemic.

Tran pointed to a page from the book “The Little Prince,” in which the narrator drew a picture of a boa constrictor with an elephant in its belly, but the grownups in his life mistook the drawing for a simple hat.

“So as a journalist, when it comes to data, the underlying philosophical question is: Did you see a hat? Or do you see the elephants and the boa?” Tran said. “Do you only see what's in front of your face? Or do you have the ability or imagination to see what might be under the surface?”

**