What the response still had to learn
Published 13 October 2026
In April 2016, the state published another change to its outbreak count. Michigan Public reported that health officials now confirmed ninety-one cases and twelve deaths in Genesee County for 2014 and 2015. Three cases, including two fatalities, had been added for what the state described as clerical reasons. The report said about fifty-five percent of cases could be connected with a Flint hospital and described the river as a suspected source, not a confirmed one. This was a dated account of revised state figures. The January, March and April totals should be read as successive versions with their own boundaries, not separate tolls to combine. The revisions also explain why a later scientific study using a particular dataset may print a different case number without necessarily claiming that the earlier public release never existed. F018, michiganpublic.org, Death toll rises in Genesee County Legionnaires' disease outbreak, 2016-04-11 F015, michigan.gov, Increased cases of Legionnaires Disease Investigated in Genesee County, 2016-01-13 F014, content.govdelivery.com, FOR IMMEDIATE RELEASE CONTACT: Jennifer Eisner March 18, 2016 (517) 241-2112 MDHHS...
CDC's analysis, released online in June 2016 and in its July 1 weekly issue, used another set of periods and records, which should remain separate from Hurley's study. It examined children under six in the Flint service area before the switch, during the early river period, after the January 2015 advisory and after the return to the earlier supply. Surveillance records were cleaned and duplicates removed. When a child had more than one test in a period, the analysis used the highest result, with venous tests preferred. Among the analysed tests, the proportion at or above five micrograms per decilitre increased from 3.1 percent before the switch to 5 percent in the early period. The adjusted odds remained higher after accounting for age, sex, race and season. These are CDC's results for its stated surveillance analysis, not a replacement denominator for the published Hurley percentages. F070, cdc.gov, Blood Lead Levels Among Children Aged, 2016-07-01
The CDC report fixed its denominators before it reported any proportion. It estimated that 9,622 children under six lived in homes served by the Flint system, and that 7,306 of them received 9,672 blood lead tests across the study window. Of those tests, 250 came from a child who already had an elevated result earlier in the window, so they were set aside, which left 9,422 tests for analysis. A confirmed elevated level meant one venous result at or above five micrograms per decilitre, or two capillary results at or above that value drawn within twelve weeks of each other. The four periods were also dated: April 25, 2013 to April 24, 2014 before the switch, April 25, 2014 to January 2, 2015 for the early river period, January 3 to October 15, 2015 after the advisory, and October 16, 2015 to March 16, 2016 after the return to Detroit water. These are tested children in a surveillance system, not every child who lived in the service area. F070, cdc.gov, Blood Lead Levels Among Children Aged, 2016-07-01
The authors also explained why some children were more likely to test high. They named three factors that influence how much lead leaches into drinking water: whether the plumbing includes lead pipes, the pH of the finished water, and whether mineral scale has formed on the inside of older plumbing. They noted that very young children drink more water for their body weight than older children or adults, and that blood lead tends to rise in warm months. In the Flint data, tests from children aged one to two years were significantly more likely to reach five micrograms per decilitre than tests from children under one, and tests from summer and fall months were more likely than tests from winter. The early river period's unadjusted odds ratio was 1.65, with a confidence interval of 1.20 to 2.26, and the adjusted ratio was 1.46, with an interval of 1.06 to 2.01. F070, cdc.gov, Blood Lead Levels Among Children Aged, 2016-07-01
CDC gave one worked example of how water lead can reach a child's blood. For a child aged up to one year who drinks about 1,200 millilitres of water a day, it said a lead concentration of 20 parts per billion might be enough to raise blood lead from zero to five micrograms per decilitre when no other source is present. The report presented that as an illustration of scale and not as a measurement of any Flint child. It added that mineral scale protects against leaching only while it remains in place, and that when scale is removed or never develops, lead can come from solder even in plumbing described as lead-free. F070, cdc.gov, Blood Lead Levels Among Children Aged, 2016-07-01
CDC also set out limits on what the evidence could establish. The analysis was observational, could not assign the rise exclusively to the source change and did not measure every other source of lead exposure. Few infants under one had been tested, potentially underestimating effects in children receiving formula mixed with tap water. A later decline might partly reflect increased bottled-water use rather than the new municipal source alone. They prevent a causal claim or recovery claim from exceeding the study. The report still found the early river period independently associated with a higher probability of elevated blood lead after controlling for age and season. The evidence and its limits travel together, especially when a result will be used to describe the harm to an entire community. F070, cdc.gov, Blood Lead Levels Among Children Aged, 2016-07-01
On October 20, 2016, the EPA inspector general issued a management alert while its wider review remained underway. Its central finding was that Region 5 had both the authority and sufficient information to issue an emergency order as early as June 2015. The alert named the absence of protective treatment, residents' complaints and high household lead results among the information available. It said the region instead considered state action a jurisdictional barrier to federal intervention. The inspector general rejected that interpretation by reference to EPA's own 1991 guidance. State action, in that guidance, was not enough merely because it existed; it had to protect the public in a timely way. F093, epa.gov, EPA Office of Inspector General, Management Alert on Drinking Water Contamination in Flint, October 2016
The alert explained the two conditions for emergency action: contamination that could pose imminent and substantial danger, and failure by the appropriate state or local authorities to protect health. It said EPA could issue orders or begin civil action. Regional administrators held delegated administrative authority, while headquarters retained particular powers as well. The report's reading of the guidance allowed federal action when state measures were insufficient, including when state authorities intended to act. This mattered because a promise to install treatment or await another sample round was not the same as timely protection. The alert was not introducing an entirely new emergency power in October 2016. It was arguing that an existing power had been available earlier and that staff needed a clearer understanding of when to use it. F093, epa.gov, EPA Office of Inspector General, Management Alert on Drinking Water Contamination in Flint, October 2016
The alert revisited the information available by June 2015. It said EPA knew corrosion control was absent, that several homes had results above the lead action level and that complaints had accumulated since the switch. The state had first said treatment was in place and later admitted it was not, while still arguing that it was not required. The inspector general said state officials delayed while awaiting another monitoring round and the city did not act. Its account describes evidence sufficient for protective intervention even before the later Virginia Tech survey and children's results. That finding does not make the household samples representative of every tap. It concerns the threshold for emergency authority: the agency already had evidence of a danger and of a response that was not protecting residents quickly enough. F093, epa.gov, EPA Office of Inspector General, Management Alert on Drinking Water Contamination in Flint, October 2016
The alert also looked beyond Region 5. When headquarters enforcement staff were formally briefed in September, they recommended emergency action, but the region maintained its objection. Headquarters continued relying on that position as state activities unfolded. The inspector general's chronology included the return to Detroit water, a federal technical task force, a rule memo and supplementary corrosion control. Yet it said contamination remained unresolved, and that headquarters could have acted through the administrator's retained authority. The report's timeline sets out EPA's actions by date. Technical discussions and plans did not automatically establish that residents were protected. EPA's eventual January order came after months in which agencies were doing things, but the inspector general found that those things had not removed the basis for a stronger intervention. F093, epa.gov, EPA Office of Inspector General, Management Alert on Drinking Water Contamination in Flint, October 2016
The proposed remedy was more specific guidance and training. The inspector general observed that the 1991 document lacked examples of state actions that would be timely and protective and did not provide a checklist for considering emergency intervention. Its recommendations asked the enforcement office to update the guidance and train relevant drinking-water and enforcement staff. The alert identified an operational weakness in a law that already supplied power: people needed to recognise when circumstances justified its use. It did not say a checklist could replace professional judgment or prove the completion of Flint's recovery. It sought to make the choice to intervene less dependent on a disputed understanding passed between offices. The wider federal review would return to management controls in 2018. F093, epa.gov, EPA Office of Inspector General, Management Alert on Drinking Water Contamination in Flint, October 2016 F091, epa.gov, EPA Office of Inspector General, Management Weaknesses Delayed Response to Flint Water Crisis, Report 18-P-0221, July 19, 2018
A different report followed in February 2017. The Michigan Civil Rights Commission placed the crisis inside Flint's longer racial and institutional history. Its introductory letter described three hearings, neighbourhood visits and testimony on housing, the economy, environmental justice and emergency management. This was not another laboratory analysis of the river. It asked why an affected community had so little power to make government hear its concerns. The commission dated its report February 17 and said historical policies had helped produce disparate outcomes. That finding should remain identified as the commission's analysis. It does not require inserting an undated essay into the 2014 chronology: the report itself was a later public event, and its historical interpretation belongs here as part of the way the crisis came to be understood after the emergency response. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017
The commission acknowledged that government and volunteers had mobilised strongly once emergencies were declared. Water and filters were distributed, federal personnel arrived and state coordination increased. It nevertheless described timing as the problem. Its executive summary set that mobilisation against more than a year of complaints, the industrial corrosion warning, household sampling and the children's data. The report did not dispute that the response had eventually become substantial. It argued that late mobilisation left unanswered why earlier signs had not received the same urgency. The report treated the response as a question of how institutions behaved, not only of what was distributed. The report asked how public institutions decided whose concerns counted before a national emergency made ignoring them impossible. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017
The commission's explanation went back beyond the source decision. It described the legacy of housing segregation, restricted opportunity, employment patterns and the division between city and suburbs. Its argument was that these policies built upon one another, leaving Flint with a weakened tax base and residents with limited political leverage. It explicitly said it was not alleging that the people making the water decisions intended racial harm. The commission used systemic racism to describe outcomes produced through connected institutions, rather than simply labelling individual decision-makers. A finding about structure can coexist with the absence of evidence of conscious discriminatory intent. The report's historical account aimed to explain how racial disparities could persist within decisions described by their makers as neutral. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017
Environmental justice, in the commission's account, included both equal protection and a meaningful voice in decisions. It argued that Flint residents had received neither and that emergency management had privileged cost-cutting over welfare and participation. The report said a stronger environmental-justice plan or a different manager structure might have mitigated the crisis, while acknowledging that the counterfactual could not be known. This is a proposal about prevention, not a measured estimate of how many cases a particular plan would have avoided. The commission connected access to clean water with access to political decision-making. Its critique adds an institutional boundary to the technical story: a treatment decision can be unsafe partly because people affected by its results cannot readily force a review of the decision. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017
The report also warned that compensation, pipe replacement and changes in the emergency-manager law would not by themselves resolve the historical conditions it described. Those remedies could address important injuries and failures without eliminating the pattern that produced vulnerable communities. This is the commission's argument, not a reason to dismiss physical repair or individual claims. It places them on different levels. A lead line can be replaced at one address while housing, transport, employment and government representation remain unequally arranged. The report asked institutions to recognise those connections instead of assuming the last lawsuit would close the whole problem. The report separated what civil suits could deliver from the wider purpose of its own account. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017
In discussing implicit bias, the commission said it had no evidence that officials intended to poison residents or knowingly accepted danger because Flint was predominantly a community of colour. It nonetheless examined why complaints about visibly poor water were discounted and why tests were questioned instead of the supply. The report's distinction between intent and outcome prevents a false choice. It did not conclude that absent intent meant absent discrimination in institutional results. Nor did it present its analysis as proof of deliberate poisoning. Its central concern was that residents lacked political clout and had nowhere to turn, and that decisions might have differed in a community that looked more like Birmingham. F009, michigan.gov, Michigan Civil Rights Commission, The Flint Water Crisis: Systemic Racism Through the Lens of Flint, 2017