We Found Dozens of Uncounted Deaths During the Flint Water Crisis. Here’s How.

Scroll to Begin

Written by Sarah Childress, with reporting from Kayla Ruble, Jacob Carah and Abby Ellis.

Designed by Dan Nolan and built by Rob Archer.

The Legionnaires’ disease outbreak during the water crisis in Flint, Michigan was one of the largest in U.S. history, sickening at least 90 people, and killing 12, according to state data. But a FRONTLINE investigation strongly suggests the actual toll was much higher.

Legionnaires’ is a severe form of pneumonia that can be fatal if not properly diagnosed and treated, making public awareness critical to combating the disease.

Unlike other major Legionnaires’ outbreaks, the one in Flint happened largely in silence. Even as the county and state health departments tracked and recorded numerous cases through 2014 and 2015, they failed to notify the public until January 2016 — after the Legionnaires’ spike had subsided. The only notification to the medical community came halfway through the outbreak, in an email sent to 15 “ICPs” — infection control personnel — at three local hospitals.

At the same time, in the same areas of the city, people were also being diagnosed with — and dying of — pneumonia at unusually high rates. FRONTLINE analyzed death certificates in Genesee County, where Flint is located, and found that 115 people died from pneumonia during the Legionnaires’ outbreak in the city of Flint alone. Epidemiologists and infectious disease specialists told us that some of those people could have been Legionnaires’ disease cases that were left undiagnosed, untreated and, ultimately, uncounted.

We wanted to know whether the experts’ suspicions could be substantiated. So we commissioned a team of epidemiologists affiliated with Emory University in Atlanta, Ga. to conduct an independent analysis, which drew on a broader data set of county death rates from the U.S. Centers for Disease Control, and integrated a control group. Dr. Zach Binney and Dr. Kristin Nelson, along with Legionnaires’ expert Dr. Allison Chamberlain, then built a statistical model to determine whether the increase was statistically meaningful. Here’s what they found:

Pnuemonia Death Rates

Genesee County

Yearly quarters: Q1: Dec–Feb, Q2: Mar–May, Q3: Jun–Aug, Q4: Sep–Nov

The team plotted pneumonia deaths in Flint’s Genesee County beginning in January 2011 through the end of 2017, drawing on data from the U.S. Centers for Disease Control. “This is the gold standard data that’s used in these types of analyses all the time,” Binney said. They used only cases of non-viral pneumonia, since Legionnaires’ disease is caused by the bacterium Legionella pneumophila.

They also plotted pneumonia death rates from a control group, shown here in teal, composed of 45 counties in Michigan and surrounding states that are similar to Genesee County in population size and climate, such as Rockford, Illinois; Ft. Wayne and Evansville, Indiana; Youngstown, Toledo and Dayton, Ohio; and Lansing, Michigan.

Once they were plotted, the deaths tended to follow a seasonal trend: Pneumonia deaths are typically higher during the winter in Genesee County…

…and in the control counties.

By contrast, pneumonia deaths are typically lower in the summertime. That’s true every summer in the control counties.

When Genesee County and similar counties are mapped together, the lines generally follow each other, trending in the same pattern: down in the summertime, and up in the winter.

In the summer of 2014, however, they diverge sharply.

While deaths in the control group counties dipped that summer, as expected…

They increased that summer in Genesee County.

In fact, the spike was higher than any other summer from 2011 through 2017. “The fact that, during a time when pneumonia is not supposed to be rising, there was a multi-month rise in Genesee — but not in the other counties — that was, I think, what really drove it home for us,” Binney said.

The increase began in May 2014, shortly after the city of Flint began drawing its water from the Flint River.

The outbreak continued through the end of 2015. The same clear divergence in pneumonia deaths between the control group and Genesee County doesn’t appear that year, even though Flint was still using river water at the time. The Emory team didn’t have an explanation why, but noted that the pneumonia deaths were still on average higher in Genesee compared to the control counties in 2015. “It’s just very volatile, whereas in 2014 it was somewhat higher for a longer period,” Binney said.

The pneumonia deaths also coincide with the 90 cases of Legionnaires’ disease documented by the Michigan Department of Health and Human Services in 2014 and 2015.

Since there is no way to retroactively identify undiagnosed Legionnaires’ cases or deaths, it’s impossible to know the exact toll of the outbreak. But the Emory team’s statistical analysis offered an estimate.

They found 70 more pneumonia deaths than would have otherwise been expected in Genesee County during the 2014-15 Legionnaires’ disease outbreak. “Our conclusion is that it’s likely that the Legionnaires’ outbreak was bigger than that reported by official authorities,” Binney said.

The state’s death toll of 12 people included only individuals who were diagnosed with Legionnaires’ disease and died while hospitalized or within 30 days after leaving the hospital, which is a standard public health methodology. But a FRONTLINE analysis also showed that at least 20 people who survived their initial diagnosis died after that period, in the months and years since.

While it declined to comment on the Emory analysis, the Michigan Department of Health and Human Services told FRONTLINE previously that it had also noticed an increase in pneumonia deaths during the Legionnaires’ outbreak, which it attributed to influenza.

The Emory team also mapped the pneumonia death rates, and found they tended to cluster in the same areas as the diagnosed Legionnaires’ cases — in west Flint and northwest of the city — further evidence to them that at least some of the pneumonia cases were likely Legionnaires’.

The team presented their work to two independent reviewers — who offered suggestions that have been incorporated into the analysis — including Dr. Amesh Adalja, an infectious disease specialist at the Johns Hopkins University Center for Health Security. “It’s definitely very suggestive that there were much more cases of Legionnaires’ disease than what was officially reported,” he told FRONTLINE.

Dr. Emily Ricotta, an epidemiologist and statistician working in the Epidemiology Unit at the National Institutes of Allergy and Infectious Diseases in Maryland, also reviewed the work. She said that while she believed “that there were some excess deaths” she was hesitant to broadly attribute them. “I think that it’s a very complex situation, and I think that this analysis raises some questions and that would be important to interrogate further.”

She said one of the unresolved questions is why there wasn’t as dramatic an increase in pneumonia deaths in 2015 as there was in 2014. “Unfortunately there is a limit to what you can collect retrospectively,” she said, “and so sometimes you just hypothesize and say here’s the best evidence that we have. And you’ve got to go with it.”

The Emory team is currently preparing a paper for peer review and submission to an academic journal. You can take a closer look at their work, including more on the mapping analysis, here.

Support Provided By