**3. Results**

We present study participant demographic factors overall, and by birth weight (Table 1) and Mediterranean diet adherence (Table 2). Of the 298 women included in our sample, 36% were Black, 28% were white, 32% were Latina or Hispanic, and 4% were of other race/ethnicities. Approximately one-quarter of women had a BMI of 30 or greater, and about 75% of obese women were either African American or Hispanic. Over half of our sample (55%) had a high school diploma. Among infants in our study, 6% were born prior to 37 weeks gestation and 5.7% weighed 2500 g or less at birth. Approximately 20% of mothers in our study reported smoking during pregnancy. The median (IQR) of cadmium concentration in blood was 0.24 (0.34) ug/g of blood weight, and MDS scores were normally distributed and ranged from 0 to 9 and approximately half of the women in our study had a score at or below 4. Women who did not deliver a low birth weight infant were more likely to be white, non-smokers, and have a college degree.

#### *3.1. Cadmium Exposure and Birth Weight*

Results of the associations between cadmium and birth weight are summarized in Table 3. After adjustment for prepregnancy BMI, smoking during pregnancy, gestational age, and sex of the infant, we observed that the women in the highest quartile of cadmium exposure during the prenatal period had infants whose birth weights were 210 g lower (β = −210.4; 95% CI: −332.0, −88.8; *p* = 0.008) than those in the lower three quartiles. Further including other co-occurring metals, lead and arsenic, that also have been linked to lower birth weights did not alter these associations. We explored removing gestational age from these models, as it may be on the causal pathway. This somewhat attenuated the association, however, it remained statistically significant (β = −161.6; 95% CI: −311.3, −11.9; *p* = 0.04). Further adjusting the co-occurrence of other metals, including lead or arsenic, did not materially alter these findings (data not shown).


**Table 1.** Sociodemographic characteristics of study participants by birth weight.

– No missing data in this category.


**Table 2.** Sociodemographic characteristics of study participants by Mediterranean diet adherence.

> – No missing data in this category.

We also regressed gestational age at birth on cadmium exposure, controlling for the same covariates (Table 4). We found that preterm birth is marginally associated with prenatal cadmium exposure (β = −0.11; 95% CI: 0.01, 0.72; *p* = 0.04) (Table 2). We included birth weight as a confounder in our main analysis and explored the effect of removing it in a supplemental analysis. When birth weight was removed from the model, the association between elevated prenatal cadmium exposure and preterm birth was no longer statistically significant (β = 1.2; 95% CI: 0.37, 3.31; *p* = 0.74). Again, further adjustment for the co-occurring metals, lead and arsenic, did not alter these findings, suggesting that growth restriction, rather than preterm birth, may be the major contributor to these birth outcomes.


**Table 3.** Regression coefficients and 95% confidence intervals for the association/relationship between, cadmium exposure and weight.

a Adjusted for smoking during pregnancy, prepregnancy BMI, gestational age and sex of the infant. b Adjusted for smoking during pregnancy, prepregnancy BMI, gestational age and sex of the infant, among mothers with a Mediterranean diet score at or below 4. c Adjusted for smoking during pregnancy, prepregnancy BMI, gestational age and sex of the infant, among mothers with a Mediterranean diet score above 4. d Adjusted for smoking during pregnancy, prepregnancy BMI, and sex of the infant.

**Table 4.** Odds ratio and 95% confidence intervals of the association between elevated prenatal cadmium exposure and preterm birth.


a Adjusted for smoking during pregnancy, prepregnancy BMI, birth weight and sex of the infant. b Adjusted for smoking during pregnancy, prepregnancy BMI, birth weight and sex of the infant, among mothers with a Mediterranean diet score at or below 4. c Adjusted for smoking during pregnancy, prepregnancy BMI, birth weight and sex of the infant, among mothers with a Mediterranean diet score above 4. d Adjusted for smoking during pregnancy, prepregnancy BMI, and sex of the infant.

#### *3.2. Stratification by Mediterranean Diet Adherence*

To determine whether these associations were modified by adherence to a Mediterranean diet pattern, we first dichotomized the MDS below the median (a score of 4 of 9) among the *n* = 185 of 310 pregnan<sup>t</sup> women who also completed the food frequency questionnaire. We examined cadmium-birth outcome associations among low and high adherers to the Mediterranean diet. We found no evidence for effect measure modification by Mediterranean diet adherence in the association between prenatal cadmium exposure and either gestational age or birth weight. Among women who reported high adherence to a Mediterranean diet pattern during pregnancy, the magnitude of the association between prenatal cadmium exposure and birth weight (β = −126.46; 95%CI: −453.14, 200.22; *p* = 0.44) was indistinguishable from the β = −210.38 observed among all participants. Similarly, the association between prenatal cadmium exposure and birth weight among women with low Mediterranean adherence was not statistically significant (β = −64.76; 95% CI: −359.90, 230.37; *p* = 0.66) (Table 3). However, these risk estimates lacked precision as confidence intervals were wide. We also observed no evidence for effect measure modification by maternal Mediterranean diet adherence on the association between prenatal cadmium exposure and preterm birth (high adherence: β: 0.01; 95% CI: 0, 1.59; *p* = 0.20; low adherence: β: 0.07; 95% CI: 0.0008, 1.31; *p* = 0.14) (Table 4). As expected from stratified analyses, including the interaction term of the Mediterranean diet adherence score and cadmium exposure did not alter these findings. The p-values for the interaction terms of cadmium and MDS in the overall birthweight or preterm birth models were not significant (*p* > 0.15). Defining "high Mediterranean adherence" with a more stringent cut-off of MDS of 5, 6 or 7 did alter these findings (data not shown).

#### *3.3. Exploratory Analyses: Apgar Scores and PI*

In our exploratory analyses we found no association between elevated prenatal cadmium exposure and Apgar scores (β = −0.009; 95% CI: −0.13, 0.12; *p* = 0.89) or PI (β = −0.03; 95% CI: −0.11, 0.06; *p* = 0.51). Results available in Supplemental Table S3.
