Retrospective cohort study protocol to determine prenatal and infant lead exposure using novel dentition analysis in Flint, Michigan: Flint Tooth FAIRY (Flint assessment of in-utero and at-risk young)

Abstract

Introduction

The Flint water crisis (FWC) exposed thousands of in-utero fetuses and infants to lead-contaminated drinking water, yet the full extent of exposure remains unknown due to limitations in traditional biomonitoring methods. While blood lead levels provide momentary snapshots of exposure, novel dentition analysis using laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) enables detailed reconstruction of prenatal and early-life exposure through deciduous teeth (“baby” teeth). This protocol describes a retrospective study investigating prenatal and early life exposure to lead during the FWC using shed deciduous teeth.

Methods and analysis

The Flint Tooth Flint assessment of in-utero and at-risk young study aims to recruit approximately 364 children born in Flint, Michigan, between 1 January 2011 and 31 December 2015, through the Flint Registry. Recruitment began in July 2021 and the expected study period is through December 2026. Caregivers will complete questionnaires detailing residential history and infant feeding practices. Naturally shed deciduous teeth will be analysed via LA-ICP-MS to measure lead and other plumbing-related metal concentrations. Using dental growth patterns, measurements will be assigned temporal points to construct weekly exposure profiles. Statistical analyses will assess associations between tooth metal concentrations and potential risk factors, including residential water service line material, household water consumption patterns, infant feeding practices and residential history. This study focuses on exposure reconstruction only and does not include child health outcome assessment.

Ethics and dissemination

This study was approved by the Biomedical and Health Institutional Review Board (study ID STUDY00003045). Results will be disseminated through peer-reviewed publications and presentations to stakeholders. Individual exposure profiles will be shared with families using the Tooth Lead Report-Back Toolkit, created by researchers at The Institute for Climate Change, Environmental Health and Exposomic Research at the Icahn School of Medicine at Mount Sinai.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This is the first study to use laser ablation-inductively coupled plasma-mass spectrometry analysis of deciduous teeth to reconstruct prenatal and early-life exposure to lead during a community-wide drinking water contamination event.

• Integration with existing Flint Registry infrastructure enables efficient recruitment and comprehensive demographic data collection.

• Weekly-resolution exposure profiles will provide unprecedented temporal detail about metal exposure patterns.

• Sample size is limited by natural tooth shedding patterns and timing of the Flint water crisis.

• Potential recall bias in questionnaire responses about early life feeding practices and address history.

Background

Flint water crisis

In April 2014, while under state-appointed financial emergency management, the City of Flint, Michigan, switched its water supply from properly treated Detroit-supplied Lake Huron water to the untreated and highly corrosive Flint River.¹ This “cost-saving” measure resulted in further disintegration of already outdated water infrastructure and caused drinking water pipes to heavily corrode, leaching toxic lead and other metals into the water.² Although the water quality effects were almost immediate – consumers began to complain of foul-smelling discoloured water causing rashes within a few weeks of the water switch – the city did not return to Lake Huron water until 18 months later, after scientists demonstrated increases in blood lead levels (BLLs) in Flint children. After the city switched back to treated Lake Huron water, water lead levels and BLLs began to decline.^{2 3} Unless children had BLLs tested during or shortly after the water crisis, historical exposure to lead-contaminated drinking water is nearly impossible to determine.

Measuring exposure to lead-contaminated drinking water during the FWC has been a key obstacle in targeted public health intervention and assessing the extent of harm and resulting impacts. Young children not routinely screened during the 18 months show underestimated exposure during BLL surveillance, as BLLs decline when exposure decreases. Still, past exposures may be important indicators of harm.4,6 While BLL testing is easily implemented in a clinical setting and for broad surveillance, it cannot be used to assess cumulative or historic lead exposure reliably. Due to blood lead’s short half-life (28 days), childhood BLL testing represents only recent or ongoing exposures. Additionally, recommended ages for BLL surveillance (1 and 2 years) are not designed to detect lead-in-water exposure.⁷

Lead is toxic

With no safe level, lead is a powerful neurotoxin proven to have devastating effects on every organ system in the human body. Foetuses, infants and children early in development are the most vulnerable to lead toxicity, and there is no treatment to mitigate the permanent developmental effects of lead. Furthermore, symptoms of lead exposure may take years to develop. Lead is absorbed into the body by inhalation or ingestion and mimics calcium, allowing it to travel uncontested throughout the body. A significant proportion of what is absorbed circulates into the bloodstream with a half-life of about 1 month. The proportion deposited into calcified tissues has a half-life of 10–30 years.

Despite the increasing scientific consensus of lead’s irreversible neurotoxicity, young children continue to be exposed to lead through many pathways (water, dust, soil, paint, etc) with a disproportionate impact on poor and predominantly minority children. Unlike other sources of lead exposure, lead-in-water exposure impacts a younger and more developmentally vulnerable age group - specifically, in-utero and reconstituted formula-fed infants. A growing body of evidence suggests the most important standard for predicting adverse health outcomes is not recent lead exposure but cumulative lead exposure that occurs over many years.⁸

Teeth as biomarkers

Accurately measuring both the cumulative exposure and timing of exposure represents a major challenge in targeting public health intervention. Blood lead represents recent exposure.⁹ Similarly, urine lead mainly reflects recent lead exposures and shares many of the same limitations as blood lead for exposure assessment, though urinary lead is more sensitive to exposure changes than blood lead.¹⁰ Hair and fingernails have been described as medium-range biomonitoring agents, associated with exposure times from a few months to years. However, they are constantly contaminated by external agents, such as airborne dust, shampoo, etc. Also, they are unreliable markers of prenatal lead exposure and are not considered ideal indicators of exposure. Calcified tissues, such as bone and teeth, have a high affinity towards metal accumulation. However, human bone is not easily available for sampling and measurement and is subject to significant turnover during growth.^{9 11} Alternatively, teeth are not subjected to turnover and dental tissues are very hard, thus providing a permanent, cumulative record of past and recent environmental exposures to heavy metals.^{9 12 13}

Deciduous teeth (ie, “baby teeth”) begin to form and accumulate metals during the second trimester of gestation and continue to do so until the tooth is shed during childhood and adolescence.^{9 11 14} Teeth form incrementally, often likened to the growth rings of a tree, where each increment carries distinct temporal information.¹⁵ As teeth grow, chemicals circulating in the blood are deposited within the organic matrix which later undergoes mineralisation, preserving both the timing and intensity of exposure. Unlike BLLs which can only provide momentary snapshots of chemical exposures at one moment, deciduous teeth can be used as a retrospective biomarker, allowing us to reconstruct the history of chemical exposures throughout early development.^{16 17} By leveraging the incremental formation of teeth, we can even determine how much lead children were exposed to before they were born, beginning at seventeen weeks gestation. Therefore, analysing teeth of children living in Flint provides the best opportunity to understand and reconstruct exposure to lead-contaminated drinking water for the youngest children, which has not been previously possible.

Study aims and hypotheses

The Flint Tooth FAIRY (Flint assessment of in-utero and at-risk young) study aims to evaluate prenatal and early-life exposure to lead and other plumbing-related metals during the FWC by analysing deciduous teeth. The study aims to (1) Quantify prenatal and early-life exposure to lead and other plumbing-related metals during the FWC, (2) Examine associations between metal concentrations and drinking water exposure factors and (3) Establish exposure timing profiles to inform future intervention strategies. These analyses will provide unprecedented temporal resolution of metal exposure during critical developmental windows, enabling the identification of both exposure timing and intensity at weekly intervals from the second trimester through early childhood. This protocol focuses exclusively on exposure reconstruction and characterisation. We are not collecting or analysing child health outcomes as part of this study. These findings will address critical knowledge gaps regarding lead exposure during lead-in-water contamination events and may inform drinking water practices and policies, particularly for vulnerable populations.

Hypotheses of the study are:

1. Tooth dentin will show increased concentrations of lead and other plumbing-related metals following the April 2014 water source switch.

2. Children who resided in homes with lead service lines will demonstrate higher and more sustained metal exposure patterns compared with those in homes without lead service lines.

3. Metal exposure patterns will vary based on household water consumption practices and infant feeding methods.

Methods and analysis

Study design

The Flint Tooth FAIRY study is a retrospective cohort analysis to evaluate shed deciduous teeth of children to discern risk and exposure differentials based on factors like feeding type, dental lineage and geographical considerations. This study leverages data from the Flint Registry, an ongoing public health registry, to examine associations between early life environmental toxicant exposures to water crisis and paediatric health.¹⁸ The Registry has enrolled 21 492 members as of 31 July 2023, 6469 of which are children.¹⁸ Study recruitment began in July 2021 and the study will continue through December 2026. Survey recruitment and activities are conducted over the phone, online, and teeth are submitted via mail when teeth were naturally shed, but stored teeth are accepted as well.

Recruitment and sample size

Participants from the Flint tooth FAIRY study will be recruited from participants enrolled in the Flint Registry who previously indicated willingness to participate in additional research projects. The Flint Registry employed population-list identification (direct mail and telephone outreach using state health programme and hospital records), community outreach partnerships and marketing campaigns (television, radio, billboards, social media). The Registry enrolled over 22 000 members by the time enrolment closed in May 2025, representing approximately 16% of the estimated 134 788 eligible individuals. Full details of Flint Registry recruitment methods are described elsewhere (insert registry protocol citation).

Subjects are eligible if they: (1) Were born in Flint, Michigan between 1 January 2011 and 31 December 2015, (2) Were residents of Flint during April 2014 to October 2015, aligned with the utilisation of the Flint River as the primary water source and (3) Mothers or caretakers are at least 18 years or older and understand English. The selection of study birth years, 1 January 2011 and 31 December 2015, encompasses children who may have been exposed to water contaminated with lead during a crucial neurodevelopmental period, making them 5 to 10 years of age when study recruitment began and by the end of 2026, they were 11 to 15 years of age.¹⁹ To be eligible for participation, the legal guardian of potential subjects must have maintained custody of their child since the first month of life and express the willingness to provide up to three baby teeth from each enrolled child.

Contact information for potential study participants was obtained exclusively through a data request process through the Flint Registry. Individuals who met the study inclusion criteria and who had previously provided consent to be approached for additional research studies were referred to Flint Tooth FAIRY study staff for recruitment communication. All aspects of the Flint Tooth FAIRY recruitment process were remote. Personalised invitation letters were mailed to prospective participants to initiate the recruitment process and give participants an understanding of why researchers wanted to collect baby teeth, along with multiple planned telephone, email and text follow-up attempts. Participants who complete a survey will be sent a $100 gift card.

The sample estimation was based on the number of children needed to achieve adequate statistical power to draw comparisons to the full population of Flint children exposed to the water crisis. Based on 2020 5-year American Community Survey (ACS) survey census data, there are an estimated 6936 children in the 5-year-old to 9-year-old age bracket. To ensure sufficient statistical power, a sample size of approximately 364 is needed based on significance level 0.05 and a power of 0.80.

Data collection

Data will be collected using two tools: laser ablation-inductively coupled plasma mass-spectroscopy (LA-ICP-MS) and an in-depth retrospective address history questionnaire since pregnancy modelled after the previous IRISE study.²⁰ Additionally, the study will use baseline demographic and exposure data collected during Flint Registry enrolment, including information on potential confounding lead exposure sources such as maternal smoking status, parental occupational exposures, presence of deteriorating paint and home construction date.

Measuring lead and other metals in teeth

The novel dentition analysis of deciduous teeth using laser ablation-inductively coupled plasma-mass spectrometry has successfully estimated past exposure to toxic metals during critical windows of child development in communities with known lead exposure risks.^{13 21 22} It features a timely opportunity to further our understanding of lead exposure in children during the FWC. Measuring retrospective lead exposure, especially during early critical development periods such as in utero and early childhood, presents a significant challenge in public health intervention. Metals accumulate in tooth dentin beginning as early as the second trimester of pregnancy and continue until the tooth is naturally shed. Thus, teeth reflect a natural and temporally high-resolution record of exposure.

After children are enrolled in the FTF study, naturally shed deciduous teeth will be collected (which typically occurs around 4–8 years of age for most children) and shipped to Senator Frank R. Lautenberg Health Sciences Laboratory at the Icahn School of Medicine at Mount Sinai in New York for analysis. If multiple teeth are submitted for one child, the most intact tooth will be selected for analysis.

Teeth are examined under a stereo microscope and gross features recorded, including tooth type, degree of attrition, presence or absence of fillings, cracks in enamel or any unusual anatomical features. Dentin metal intensities and temporal exposure profiles are analysed using LA-ICP-MS following validated methods previously described.¹⁶

Briefly, teeth are sectioned longitudinally and analysed using a New Wave 193 nm excimer laser ablation system coupled to an Agilent 8800 ICP-MS.²³ Values were corrected to the National Institute of Standards and Technology NIST610 to account for day-to-day signal variation. Approximately 50 sampling points at 30 µm spot diameter are ablated along the enamel-dentine junction through primary dentine, representing weekly exposure from approximately 20 weeks gestation through 16 weeks postnatal age (approximately 32 time points per tooth). Temporal assignment of each sampling point is achieved through histological identification of the neonatal line (formed at birth) and daily incremental growth markings in dentine. Metals are background subtracted and normalised to calcium (²⁰⁸Pb:⁴³Ca ratios) to account for variations in mineral density across individuals. Quality control procedures include daily instrument tuning, monitoring oxide formation and analysis of certified reference materials.

Questionnaire

The study’s survey collects information on the child’s residential history since pregnancy. If participants cannot recall their exact addresses, the nearest cross streets are recorded to identify neighbouring water service lines. The survey includes questions about infant feeding and water usage during the study child’s first 4 years of life. Parents or legal guardians complete the self-administered study survey remotely. However, if a caregiver other than the biological mother desires to participate in the study, a research assistant surveys the participant via telephone, and pregnancy-related questions are omitted. Research assistants reviewed all completed questionnaires and conducted follow-up calls with all participants to address questions, inform participants about tooth collection kit mailing and clarify any inconsistencies or missing information.

Data management

The Flint Tooth FAIRY study conducts consent and data collection procedures using the digital survey interface (Qualtrics). Servers are hosted and secured by Michigan State University following the study’s Institutional Review Board (IRB) protocol. Once enrolled in the study, participants receive unique identification numbers. Steps were taken to remove all personal identifiers, but study design required the use of address history. Michigan State University IRB will review any modifications to the approved study protocol. Teeth are stored in a locked filing cabinet until shipped to Mount Sinai.

Data analysis

Data from completed surveys will be downloaded from the Qualtrics survey interface and imported to R for analysis.²⁴ We will employ multiple statistical techniques to test the robustness and sensitivity of our findings. Descriptive and exploratory analyses will be conducted to describe pre-switch and post-switch tooth metal intensities by child, home and demographic characteristics using R. We will calculate summary statistics and examine distributions of all predictors. Multilevel analysis models will be used to investigate the association between service line material and tooth metal concentration in prenatal and postnatal periods, particularly lead, to evaluate if tooth lead intensities increased during the FWC and if the presence of other heavy metals correlates with lead to determine if the water is the source of lead exposure.

For preliminary analyses of dentine element intensities, we will compute average intensities for the prenatal period (second and third trimesters) and postnatal period (0–1 years). Simple correlation plots and spaghetti plots will be used to assess the time-varying elemental intensities.

To test hypothesis 1 (temporal changes following water switch), we will use generalised additive models fitted using a simple cubic spline to examine time-varying lead concentrations before and after April 2014, implemented in R package mgcv.²⁵ This approach does not impose linearity assumptions and can identify inflection points corresponding to the water source switch.

To test hypothesis 2 (service line material effects), we will use generalised linear mixed-effects models with prenatal and postnatal periods modelled as time-varying exposures. For each metal, the R package lme4²⁶ and nlme²⁷ will be used depending on the outcome. Model parameters will be estimated with population-level fixed effects and individual-level random effects with a specialised intra-class and inter-class correlation matrix. These models will compare tooth metal concentrations between children with lead versus non-lead service lines.

To test hypothesis 3 (variation by water consumption and feeding practices), we will extend the generalised linear mixed-effects framework described above to include interactions between metal concentrations and household water consumption patterns and infant feeding methods (breastfeeding, formula with tap water, formula with bottled water).

Additionally, we will implement the lagged weighted quantile sum regression to examine the time-varying exposure mixture effect of multiple plumbing-related metals (lead, copper, zinc). This analysis will support source identification by examining coordinated temporal patterns consistent with water infrastructure as the primary source.

Patient and public involvement

Patients and members of the public were not involved in the development of research questions or study design. Results will be disseminated to study participants through the Tooth Lead Report-Back Toolkit, which provides individual infographic reports presenting personalised tooth lead data with population comparisons and educational materials on lead exposure reduction. Participants who consented to receive opt to receive individual results will have access to study staff for questions and support. Additionally, community sessions will be planned to review aggregate findings with participants as they become available.

Ethics and dissemination

Our research was approved by the Biomedical and Health IRB (study ID STUDY00003045). All prospective participants can access and review the consent form on the study website before enrolment. Caregivers are asked to communicate with their study child about the Flint Tooth FAIRY study in a manner that aligns with their family dynamics and cultural values to obtain child assent. After completing the consent process, participants can authorise researchers to access their study child’s BLL data from the Michigan Department of Health and Human Services Child Lead Poisoning Prevention Programme. Additionally, participants have the option to receive their tooth analysis results through a two-stage consent process. During initial enrolment, participants indicate interest in receiving individual results. When tooth results become available, interested participants are sent an updated consent form. Individual tooth lead concentrations are shared only with participants who complete and return this updated consent form explicitly authorising result release. This approach addresses both participant autonomy and the interpretive complexities of dental lead measurements. Unlike BLLs, which have established clinical reference ranges, validated dental lead thresholds do not currently exist. Individual tooth lead measurements are most meaningfully interpreted in the context of the population distribution rather than as standalone clinical values.

Personal tooth biomarker results will be communicated to participants in an accessible manner. To reduce potential communication barriers, we will create individual infographic reports and host community sessions to review the results. The Tooth Lead Report-Back Toolkit, created by researchers at The Institute for Climate Change, Environmental Health and Exposomic Research at the Icahn School of Medicine at Mount Sinai, was developed to provide data about tooth lead levels as markers of prenatal and early life exposure. The Toolkit was designed using a Research to Action framework and includes exposure reduction tips, community resources and strategies to promote advocacy related to practices and policies that lead to long-term, meaningful reduction in lead exposure at the community level. This toolkit will be adapted specifically for Flint Tooth FAIRY study participants. Reports will include individual and aggregate tooth lead data in multiple formats (graphic, numerical, text) as well as information about the health effects of lead exposure.

Participants receiving individual results are provided with study staff contact information to address questions and concerns. Individual reports present tooth lead data with comparison to the study population distribution and emphasise that tooth lead measurements represent cumulative past exposure. If tooth lead levels are of concern, families are referred to their healthcare provider for blood lead testing, and families without an established provider receive referral assistance.

Limitations

This study design may have potential practical or operational issues. First, the study’s success depends heavily on participants’ ability to mail in their children’s shed baby teeth. Given that the study is being conducted years after the start of the FWC and initial enrolment in the Flint Registry, participant retention may be challenging as families may have relocated or become difficult to contact. This timing issue could affect our ability to achieve the adequate sample size needed for meaningful statistical inference about the broader population of Flint children.

Potential physical sample issues may cause additional limitations as teeth may be damaged during transportation or may be too deteriorated by decay to be analysed. The study also faces potential recall bias, particularly regarding residential history and early childhood exposures. Research has shown that mothers tend to underreport adversities,²⁸ which could affect the accuracy of our survey data. Furthermore, some survey responses were incomplete or contained logical inconsistencies. Although the research team attempted to contact participants to complete missing information or clarify inconsistent responses, some participants were lost to follow-up, potentially affecting data quality.

Discussion

This study is the first to evaluate the impact of the FWC on in-utero fetuses and infants using deciduous teeth as biomarkers. While most studies focus on lead exposure from paint or soil, significant knowledge gaps exist regarding lead exposure through water and its impact on children. The Flint Tooth FAIRY study presents an innovative approach for assessing prenatal and infant lead exposure by examining shed teeth during the critical developmental stages. Examining the teeth of children who resided in Flint during the water crisis provides a more precise record of early lead exposure, particularly within high-risk populations. This method overcomes limitations associated with other biomarkers of lead exposure.

Blood lead surveillance during the FWC, while valuable for documenting acute exposure trends, had significant limitations for characterising individual exposure burden. Many children were never tested due to limited healthcare access or awareness, and blood lead’s short half-life (~28 days) means results represent snapshots of exposure at single timepoints. Tooth lead analysis addresses these gaps by providing retrospective measurement of cumulative prenatal and early postnatal exposure during critical developmental windows. This exposure reconstruction establishes which children and communities experience the highest cumulative exposures, information that could inform future longitudinal studies examining associations with developmental outcomes.

Additionally, our analysis of multiple plumbing-related metals provides a unique advantage in source identification. When lead is detected alongside other metals commonly found in plumbing systems, such as copper or zinc, it strengthens the evidence for water-based exposure rather than other environmental sources. This analysis is particularly valuable in urban settings where children may be exposed to lead from multiple sources, and the temporal alignment of these metal signatures can help distinguish water-related exposure during the FWC from ongoing background exposures.

Beyond source identification, these findings will address critical knowledge gaps regarding lead exposure from drinking water and inform public health action. Validating water infrastructure as the primary exposure source through the temporal alignment of lead with other plumbing-related metals strengthens the case for continued investment in water system remediation across affected areas. For infant feeding practices, findings can inform guidance on when formula-fed infants in communities with ageing infrastructure should use bottled water versus tap water, and how pregnant women and new parents can minimise lead exposure during critical developmental periods. More broadly, establishing tooth biomarkers as a surveillance tool enables retrospective exposure assessment for populations where real-time blood lead testing was incomplete or unavailable. This methodology provides a template for exposure reconstruction when drinking water is suspected as a lead source, particularly for assessing prenatal and early-life exposures that cannot be captured through conventional blood testing.