Counting deaths without erasing their boundaries
Published 13 October 2026
In April 2019, FRONTLINE published the story of Jassmine McBride, who had survived Legionnaires' disease in 2014 but died in February 2019. The report described cardiac arrest as the immediate cause of her death and said the death certificate named neither Legionnaires' disease nor the water crisis. Her treating physician, Marcus Zervos, nevertheless linked her later kidney, heart and respiratory problems to complications of the earlier infection. A death certificate's immediate cause and a clinician's assessment of contributing history answer different questions. FRONTLINE did not report that McBride had been added to the state's official outbreak fatality total. It described a later death and attributed the connection with the earlier disease to people who knew and treated her. F020, pbs.org, Five Years In, the Flint Water Crisis Continues Its Deadly Toll, 2019-04-25
The same report challenged the public boundary of the toll. FRONTLINE said at least twenty of seventy-eight patients who initially survived had since died in the weeks, months or years after hospital discharge. It based its count on records, court documents, news reports and interviews. The state figure it described was twelve deaths, defined through death during hospitalisation or within thirty days after discharge. MDHHS said it could not confirm FRONTLINE's figures and had not tracked patients longer because the disease's role in later mortality was difficult to determine. These figures cannot be collapsed into a single verified outbreak death count. One was the state's surveillance measure; the other was a news organisation's follow-up of later deaths among survivors, with a contested causal interpretation. The report raises a question about lasting harm without resolving every later death's cause. F020, pbs.org, Five Years In, the Flint Water Crisis Continues Its Deadly Toll, 2019-04-25
FRONTLINE reported that many later death records named heart, lung or kidney disease. It also reported families describing survivors who never returned to their earlier health. In McBride's case, the programme sought a medical review from infectious-disease specialist Amesh Adalja, who considered the earlier critical illness a contributor without calling it the specific final cause. MDHHS emphasised that many outbreak patients had other serious conditions and that such conditions raised longer-term mortality risk. Neither point should displace the other. A person can have pre-existing illness and still be harmed by an infection that weakens an already vulnerable body. Equally, the later death of an outbreak survivor does not by itself establish that the infection caused it. The account preserves that difficult distinction instead of turning uncertainty into either a complete denial or a larger certified toll. F020, pbs.org, Five Years In, the Flint Water Crisis Continues Its Deadly Toll, 2019-04-25
Disagreement about the outbreak source also remained. FRONTLINE reported the state health department's rejection of the scientific team's conclusions as incomplete and inaccurate, including concern about missing samples and laboratory tests. The department's account focused on McLaren Flint hospital, while the hospital denied being the source. Specialists quoted by FRONTLINE questioned a sharp separation between a hospital and the municipal supply feeding it. These are attributed positions, not a resolution reached by counting how many experts favoured each side. The paper's statistical case and the state's objections concern different forms of evidence and different limits. The report demonstrates that a hospital exposure category need not settle the origin of the bacteria. It also shows why the chronology should not silently replace the state's earlier uncertainty with an unqualified later statement of consensus. F020, pbs.org, Five Years In, the Flint Water Crisis Continues Its Deadly Toll, 2019-04-25 F022, medrxiv.org, Excess Pneumonia Mortality During a Legionnaires' Disease Outbreak in Flint, Michigan
Later in 2019, Anne Chamberlain, Kristin Nelson and Zachary Binney presented a separate estimate of excess pneumonia deaths in a medRxiv manuscript. The supplied version is a preprint, and it should be named as such rather than treated as a final official surveillance report. Its abstract estimated seventy excess pneumonia deaths in Genesee County during the outbreak period, with a ninety-percent uncertainty interval from thirty-six to 103. That is not seventy individually confirmed Legionnaires' fatalities. The outcome was excess deaths attributed to pneumonia in a population-level analysis. The authors described the result as consistent with a larger outbreak than recorded, not as proof that every excess death had been caused by Legionella. The measure broadens the question of missed cases while retaining a different evidentiary status from the state's case count. F022, medrxiv.org, Excess Pneumonia Mortality During a Legionnaires' Disease Outbreak in Flint, Michigan
The authors used monthly death counts from CDC WONDER and a separate county death-record dataset for mapping. They compared Genesee with forty-five other medium-metropolitan counties in Michigan and nearby states. Some monthly counts were suppressed because they were small, and the analysis substituted an assumed value in those months. The mapping used residential census tracts from county records, while the national database lacked that detail. Those choices are not incidental: they determine what could be estimated and where. The authors also used an incomplete geographically identified list of outbreak cases for mapping, excluding non-residents and entries without required information. The manuscript itself notes that the state's final report listed 90 confirmed cases, suggesting the Frontline list was missing 23. A map from those records therefore cannot be read as a complete map of every confirmed patient. F022, medrxiv.org, Excess Pneumonia Mortality During a Legionnaires' Disease Outbreak in Flint, Michigan
To estimate excess deaths, the authors modelled monthly pneumonia mortality while accounting for population, location, calendar month and influenza-season year. They then estimated the deaths expected during the outbreak without the outbreak terms and compared those predictions with observed totals. Their selected outbreak period began in May 2014, before the first reported June case, to include possible earlier deaths. The result depends on that modelling framework and period. It is a counterfactual estimate, not an audit of seventy death certificates proving a common bacterial diagnosis. It can reveal a population pattern that case surveillance may miss, while leaving individual causes uncertain. The interval and the method belong beside the number wherever the estimate is repeated. F022, medrxiv.org, Excess Pneumonia Mortality During a Legionnaires' Disease Outbreak in Flint, Michigan
The manuscript discussed underdiagnosis and limitations in its available records. Its findings were consistent with missed Legionnaires' disease among pneumonia deaths, but the authors did not have complete patient testing that could retrospectively classify every case. The official twelve-death count they cited included ten county residents; the excess-mortality analysis concerned county residents and a broader pneumonia outcome. Adding the estimate to twelve would mix geography, outcome and method, and could count some people twice. Adding FRONTLINE's later survivor deaths would introduce a third time boundary and another causal standard. The record supports presenting these measures together only with those differences explicit. That is a more demanding account than a single dramatic total, but it keeps readers from mistaking a surveillance number, a follow-up report and a statistical estimate for three independent lists of the same kind. F022, medrxiv.org, Excess Pneumonia Mortality During a Legionnaires' Disease Outbreak in Flint, Michigan F020, pbs.org, Five Years In, the Flint Water Crisis Continues Its Deadly Toll, 2019-04-25