The sample consisted of women from the Maternal and Developmental Risks from Environmental and Social Stressors (MADRES) prospective pregnancy cohort. MADRES is an ongoing NIH-funded cohort consisting of over 1000 predominately low-income Hispanic women residing in urban LA. However, not all have provided data. A detailed overview of participant recruitment and data collection for MADRES is described elsewhere . Briefly, pregnant women were recruited beginning in 2015, from prenatal care providers in LA serving predominantly medically underserved communities. Eligibility criteria include being < 30 weeks gestation at the time of recruitment, being ≥ 18 years of age, and speaking English or Spanish fluently. Exclusion criteria include being HIV positive; having a physical, mental, or cognitive disability that would prevent participation or the ability to provide informed consent; current incarceration; and having a multiple gestation pregnancy . The current study includes data from 491 pregnant women in MADRES who had urine collected during the first, third or both trimesters of pregnancy. See Table S1 for a comparison of demographic characteristics between the study sample and all MADRES participants with available demographic data and see Figure S1 for a participant selection flow diagram.
Pre-pregnancy Body Mass Index (BMI)
Self-reported pre-pregnancy weight was ascertained through interviewer-administered questionnaires during pregnancy. If missing, then the first weight of the index pregnancy (obtained from maternal electronic medical records) was used in lieu of self-reported pre-pregnancy weight. Pre-pregnancy BMI, defined as [weight (kg) / height (cm)2] × 10,000, was measured both continuously and categorically. Categorical BMI was initially classified according to the CDC categories of “underweight”, “normal”, “overweight”, “class 1 obese”, “class 2 obese” and “class 3 obese”; however, for the current study, the 3 classes of obesity were collapsed into a single “obese” category. Therefore, the BMI variable was subsequently recoded into a 4-level variable of “underweight”, “normal”, “overweight” and “obese”.
Self-reported race and ethnicity were collected separately but were combined for this analysis into a four-category race/ethnicity variable: White non-Hispanic, Black non-Hispanic, Hispanic, and Multiracial or other non-Hispanic.
Maternal ethnicity by nativity
Participants were classified according to their self-reported ethnicity and nativity; non-Hispanic, US-born Hispanic, and non-US-born Hispanic.
Participants were asked, “what was the last grade in school you completed?” Their responses were classified as less than 12th grade (i.e., did not finish high school), completed grade 12 (i.e., high school), some college or technical school, completed 4 years of college, and some graduate training after college.
Participants were asked during at each trimester: “Excluding e-cigarettes, have you ever smoked cigarettes, cigars, or pipes?”. Their responses were coded as “yes” or “no”. A collapsed variable of any smoking during pregnancy and no smoking during pregnancy was then calculated.
Earliest ascertained income
Participants were asked during pregnancy, “in which of the following categories did your total household family income fall in the last year?” The categories included, don’t know, less than $15,000, $15,000 to $29,999, $30,000 to $49, 999, $50,000 to $99,999, and $100, 000 or more.
Mothers were asked what the birth order index of their child was at the time of pregnancy. Birth order was defined as first born, second born, third born, fourth-born, fifth born, sixth-born or later; however, this variable was collapsed into first born, second born and third born or later.
Single spot urine samples were collected from MADRES participants in 90-mL sterile specimen containers during their first and third trimesters of pregnancy. Women were asked to fast, if possible, prior to attending the study visit. During trimester one, only 46 women reported fasting, while in trimester three most participants (N = 385) reported fasting for at least 8 h. Urine samples were transported on ice to the laboratory within 1 h of collection. They were then aliquoted and stored at − 80°C in 1.5 mL sterile cryovials (VWR).
Urinary fluoride was measured at the Oral Health Research Institute, Indiana University, School of Dentistry. Fluoride content of urine samples was quantified using Martinez Mier et al. (2011) modification of the hexamethyldisiloxane (HMDS) microdiffusion method of Taves (1968). Each sample was dispensed into the bottom of a disposable 60 × 15 mm Petri dish and 2.0 mL of deionized water (diH20) was pipetted into each dish. After applying petroleum jelly to the inside of each Petri dish lid corresponding to each sample, 50 WI of 0.05N sodium hydroxide (NaOH) solution was placed in five equal drops on each dish lid. Each dish was then immediately tightly sealed. After burning a small hole into each lid with a soldering iron, 1.0 mL of HMDS-saturated 3N I-12S04 was pipetted in each hole and sealed immediately with petroleum jelly. During overnight diffusion at ambient temperature, fluoride was released and trapped in the NaOH. The trap was recovered and buffered to pH 5.2 with 25 WI of 0.1 M of acetic acid. The recovered solution was adjusted to a final volume of 100 pl with diH20. A similarly prepared standard fluoride curve was used to determine the fluoride content of each sample. Analyses of all standards/samples were performed using a fluoride ion-specific electrode and a pH/lSE meter. Testing included a standard check (using a fluoride standard traceable to NIST) performed with daily sample analysis.
All urine fluoride measurements were adjusted for specific gravity. Urinary specific gravity was measured using a zero-setting calibrated ATAGO@ Pen Refractometer under darkened conditions and was performed daily while setting up Petri dishes for fluoride analysis. Urine fluoride adjusted for specific gravity was derived from the unadjusted fluoride value and specific gravity of each sample using the Levine Fahy equation: [ConcentrationSG normalized = Concentrationspecimen (SGreference – 1)/(SGspecimen – 1)] where SGreference is the median SG for the cohort . After fluoride analyses was complete, data were reviewed and approved by quality assurance staff and the study director. Out of the 491 participants who had urine collected, all but one who had urine fluoride measured in trimester one also had urine fluoride measured in trimester three (n = 293 for urinary fluoride measured in trimester one; n = 490 for urinary fluoride measured in trimester three).
A detailed description of urinary metals measurement has been provided elsewhere [26, 27]. Briefly, urinary metals analyses were performed by NSF International in collaboration with the University of Michigan Children’s Health Exposure Analysis Resource (CHEAR) Laboratory Hub. Metals were measured in urine collected during trimesters one and three using inductively coupled plasma mass spectrometry (ICP-MS). Included in this panel were: antimony (Sb), arsenic (As), barium (Ba), beryllium (Be), cadmium (Cd), cesium (Cs), cobalt (Co), copper (Cu), chromium (Cr), mercury (Hg), manganese (Mn), molybdenum (Mo), nickel (Ni), lead (Pb), platinum (Pt), tin (Sn), thallium (Tl), tungsten (W), uranium (U), vanadium (V), and zinc (Zn). Specific gravity was also measured during the time of urinary metals analyses, and we adjusted urine metals for specific gravity using the Levine Fahy equation described above . Metals with concentrations below the lower limit of detection (LLOD) or in the undetectable range were imputed as LLOD/√2. There were four metals for which > = 80% of the sample was below the LLOD (i.e., Be, Cr, Pt, and V), and two metals for which > = 60% of the sample was below the LLOD (i.e., W and U), and therefore they were excluded from statistical analyses. All remaining metals for regression analyses of first trimester MUFsg and first trimester urine metals had < 30% of the sample below the LLOD except for Sb which had 33% below the LLOD. For regression analyses of third trimester MUFsg with third trimester urine metals, all urine metals had < 30% of the sample below the LLOD except for Cd which had 37% below the LLOD and Sb which had 49% below the LLOD.
Venous whole blood samples were collected from participants during the first and third trimesters of pregnancy during the same visit that the urine samples were collected. Blood metals were measured among a small subset of participants who also had urinary fluoride measured (n = 123 in trimester one and n = 90 in trimester three). Participants were asked to fast prior to attending the study visit when blood was collected; however, not all were able to. In trimester one, only 34 participants reported fasting for at least 8 hours, while in trimester three most participants with blood metals measured (n = 81) reported fasting. Collection time was not standardized. Using acid washed pipette tips, 50 μL of venous whole blood was spiked directly into 15 mL metal-free polypropylene centrifuge tubes (VWR, Atlanta, GA) and extracted in 1.5 mL of 5% ultrapure grade acetic acid and 0.01% ultrapure grade Triton X-100 (Fisher Scientific, Pittsburgh, PA) in 18.2 mΩ deionized water. Two hundred ppb of Au was added to amalgamate Hg to prevent analyte loss throughout the procedure (Inorganic Ventures, Christiansburg, VA). Five ppb of indium, bismuth, and yittrium were added to the extraction solution as internal standards (Inorganic Ventures, Christiansburg, VA). The blood extracts were then centrifuged at 3600 × g for 2 min and incubated for 90 min at room temperature on a shaker table at 300 rpm. As, Cd, Pb, and Hg were quantified in whole blood samples using ICP-MS, performed at the Quantitative Bio-element Imaging Center (QBIC) at Northwestern University. The following isotopes were quantified: 206Pd, 207Pb, 208Pb, 202Hg, 75As, and 114Cd, using previously developed methods [28, 29]. Collisional cell technology was used to eliminate interfering ions. All sample results were above the LLOD.
Statistical analyses were performed using IBM SPSS Statistics version 28. Descriptive statistics were calculated for fluoride, sociodemographic variables, and metals. Specific gravity adjusted maternal urinary fluoride (MUFsg) distributions were skewed, and therefore medians, standard errors, and interquartile ranges are reported. However, we also report arithmetic means and standard deviations for comparison with other fluoride studies. Spearman correlations examined associations of fluoride variables within and between trimesters, and an intra-class correlation coefficient examined consistency of MUFsg between trimesters. Kruskal–Wallis, Mann–Whitney U tests, and Spearman correlations examined associations of MUFsg with sociodemographic variables. Linear regression examined associations of MUFsg with blood and/or urinary metals, adjusted for maternal age, income, pre-pregnancy BMI, ethnicity by nativity and parity, within and between trimesters. We dummy coded covariates including income, parity, and ethnicity by nativity for regression analyses. Additionally, participants with missing data were designated to a “missing” category for these dummy coded covariates. Covariates were selected a prior based on previously established associations between fluoride and metal exposures/ metabolism [18, 30,31,32]. One participant with an extreme and atypical value of MUFsg = 7.99 during trimester three was removed for all analyses that included third trimester urine. A logarithm base 10 transformation was applied to blood and urine metals to satisfy linear regression assumptions. We conducted sensitivity analyses examining covariate-adjusted associations of MUFsg with blood and/or urinary metals among participants who reported fasting for at least 8 h during trimester 3. A False Discovery Rate (FDR) correction accounted for multiple comparisons for associations between MUFsg and sociodemographic variables for statistically significant Kruskal–Wallis tests, as well as for 60 tests of associations between MUFsg and urinary metals and 16 tests of associations between MUFsg and blood metals within and between trimesters. The criterion for statistical significance was a two-tailed p-value or q-value of 0.05, depending on the analysis.