Maternal Fish Intake in the Year Prior to Conception and Birth Defects, National Birth Defects Prevention Study, 1997–2011

ABSTRACT

Background

Epidemiologic data on the association between maternal fish intake and birth defects are sparse. Our objective was to assess associations between maternal fish intake and 52 different birth defects, most of which have not been assessed previously.

Methods

Using logistic regression, data was analyzed for 29,242 mothers of infants with birth defects and 10,973 mothers of control infants who participated in the National Birth Defects Prevention Study (NBDPS) and delivered between 1997 and 2011. We focused on associations between mothers who reported high fish intake of 2 or more servings of fish per week for the year prior to conception versus those who reported little or no fish intake and 52 birth defects. Each birth defect was analyzed separately.

Results

Among control mothers, 13.9% reported high fish intake and 31.3% reported eating very little or no fish. High fish intake was associated with lower odds of 9 birth defects (isolated heterotaxia, double outlet right ventricle transposition of the great arteries, total anomalous pulmonary venous return, Dandy Walker syndrome, holoprosencephaly, choanal atresia, cleft palate, craniosynostosis, and gastroschisis) with adjusted odds ratios (aORs) ranging from 0.32 to 0.83. An elevated association with high fish intake was observed for one birth defect (tricuspid atresia, aOR = 1.76; 95% CI 1.07, 2.89).

Conclusion

Based on our findings, consumption of 2 or more servings of fish per week may help to prevent certain types of birth defects. This conclusion is similar to the existing FDA (Food and Drug Administration) recommendation that women who are or can become pregnant should consume 2–3 servings of low mercury seafood per week. A large majority of pregnant women in this US (United States) based study did not meet the FDA recommendations for fish intake, suggesting that many of them may have benefited from increasing their consumption of seafood.

1. Introduction

The Food and Drug Administration (FDA) recommends that in order to obtain adequate intake of omega 3 fatty acids to promote healthy brain development in their fetus, pregnant women or women who can become pregnant should “consume between 8 and 12 ounces (2–3 servings) of low mercury seafood per week” (US Food and Drug Administration ²⁰²¹). However, many women in the US consume lower amounts of seafood than recommended (Papanikolaou et al. 2014). This may be due to the increased cost of some types of fish (Love et al. 2022), dislike of fish, a vegetarian diet, or concerns about the effects of mercury or other contaminants.

Based on three studies that observed that maternal intake of fish or seafood rich in omega‐3 polyunsaturated fatty acids (PUFAs) was protective for specific birth defects (Miller et al. 2010; Obermann‐Borst et al. ²⁰¹¹; Michikawa et al. ²⁰¹⁹), our objective was to use data from the National Birth Defects Prevention Study (NBDPS) to conduct a comprehensive study of the associations between maternal fish intake and 52 well‐defined categories of birth defects, most of which have not been assessed in previous studies.

2. Methods

The NBDPS was a population‐based, case–control study that collected data on pregnant women who delivered between 1997 and 2011. Mothers of cases were identified from birth defect surveillance programs at 10 sites (Arkansas, California, Georgia, Iowa, Massachusetts, New York, New Jersey, North Carolina, Texas, and Utah). Controls were live‐born infants without major birth defects, selected from birth certificates or birth hospital logs to be representative of the populations from which the cases were sampled. We included all control pregnancies and all case pregnancies whose fetus or infant had a NBDPS birth defect with at least 100 cases in the database.

Cases include live births, fetal deaths, and pregnancy terminations and were diagnosed by physical examination, imaging, autopsy, or pathology reports from surgical procedures. Cases were reviewed by clinical geneticists affiliated with the NBDPS who classified them as “having isolated defects if there was one major birth defect, two or more major birth defects affecting only one organ system, or one major birth defect with a sequence of related defects.” Cases were classified as having multiple birth defects if there were two or more major unrelated defects in different organ systems (Rasmussen et al. 2003; Reefhuis et al. ²⁰¹⁵). Cases that were strongly suspected or known to have been caused by single‐gene disorders or chromosomal abnormalities were excluded.

Maternal interviews were conducted using a computer‐assisted telephone interview, in English or Spanish, between 6 weeks and 24 months after delivery. The frequency of participation was 68% for controls and 64% for cases.

Participating mothers were asked about their intake of fish and 59 other foods using a shortened version of the food frequency questionnaire (FFQ) developed by Willett and colleagues (Willett et al. 1987). Mothers were asked to report how often they consumed a standard sized serving of each food in the year before they became pregnant. Responses were recorded as the number of servings reported per day, week, or month.

Intake of many different nutrients was estimated in grams per day based on the frequency of maternal intake of all foods containing a particular nutrient and the average grams of the nutrient present in a standard size serving of those foods as estimated by the United States Department of Agriculture (USDA version 27).

2.1. Main Statistical Analysis

Mothers with missing information on fish intake or any of the final covariates were excluded. We also excluded mothers who reported having diabetes prior to becoming pregnant because it is a strong risk factor of most of the NBDPS birth defects (Tinker et al. 2020) and we excluded those where the data indicated they were more likely to have poor diet recall. Exclusions for this reason included mothers whose caloric intake was improbably low (< 500 kcal/day), improbably high (≥ 5000 kcal/day) and those who failed to report their intake for more than two foods. As all cases of hypospadias were males, controls for the analysis of hypospadias were restricted to pregnancies with male infants.

Crude prevalence ratios (CPRs) were used to describe significant associations (p < 0.05) between maternal characteristics and fish intake among control mothers. In subsequent tables, logistic regression was used to estimate three adjusted odds ratios (aORs) and 95% confidence intervals for the association between four levels of fish intake (high [2 or more servings per week], moderate [1 serving per week], low [2–3 servings per month], and referent [no intake or less than one serving per month]) and 20 categories of cardiac birth defects (Table 2) and 32 categories of noncardiac defects (Table 3). The same referent group was used to compute aORs throughout. Tests for linear trends between the four levels of fish intake described above and the odds of each of the 52 birth defects were also assessed.

TABLE 2.

Adjusted ^a associations (aOR) between maternal servings of fish and 22 cardiac birth defects, The National Birth Defects Prevention Study (NBDPS) 1997–2011.

Birth defects	Frequency of servings of fish (referent ≤ 1 serving per month)
	1–3 per month		1 per week		≥ 2 per week
	aOR	95% CI	aOR	95% CI	aOR	95% CI
Heterotaxia with CHD	1.15	0.86, 1.53	0.80	0.55, 1.15	1.25	0.86, 1.81
Truncus arteriosus	1.45	0.92, 2.31	1.08	0.61, 1.90	0.79	0.37, 1.65
Tetralogy of Fallot	0.98	0.83, 1.14	0.97	0.81, 1.16	1.04	0.85, 1.28
D‐Transposition of the GA	0.89	0.74, 1.07	0.84	0.67, 1.05	0.90	0.69, 1.16
DORV‐TGA	1.02	0.69, 1.50	1.51	1.00, 2.27	0.42	0.21, 0.84
DORV‐Other	0.85	0.53, 1.36	0.83	0.48, 1.44	1.03	0.57, 1.86
VSD conoventricular	1.01	0.66, 1.55	0.62	0.35, 1.10	1.08	0.62, 1.87
AVSD	0.86	0.65, 1.13	1.11	0.81, 1.51	1.01	0.69, 1.46
TAPVR	0.70	0.52, 0.93	0.68	0.48, 0.97	0.70	0.46, 1.06
Hypoplastic left heart syndrome	0.96	0.78, 1.17	0.83	0.65, 1.06	1.00	0.76, 1.32
Coarctation aorta	1.02	0.88, 1.20	1.00	0.83, 1.20	0.93	0.74, 1.16
Aortic stenosis	1.04	0.83, 1.30	0.98	0.74, 1.28	0.98	0.70, 1.35
Pulmonary atresia	0.94	0.68, 1.29	1.00	0.70, 1.44	0.88	0.56, 1.37
Pulmonary valve stenosis	0.90	0.78, 1.03	0.93	0.79, 1.09	0.84	0.70, 1.02
Ebstein anomaly	0.95	0.66, 1.37	0.74	0.47, 1.17	0.72	0.42, 1.26
Tricuspid atresia	1.28	0.84, 1.94	1.25	0.77, 2.01	1.76	1.07, 2.89
VSD perimembranous	1.06	0.92, 1.21	1.05	0.90, 1.23	1.11	0.93, 1.33
VSD muscular simple	1.30	0.82, 2.05	1.04	0.60, 1.78	1.62	0.88, 2.97
VSD muscular not simple	1.03	0.83, 1.27	1.13	0.89, 1.43	1.11	0.84, 1.46
ASD secundum	0.99	0.88, 1.11	0.99	0.86, 1.13	0.90	0.77, 1.05
ASD NOS	0.90	0.73, 1.12	1.01	0.79, 1.29	1.14	0.86, 1.49
Single ventricle complex	1.17	0.87, 1.58	1.20	0.85, 1.69	1.11	0.73, 1.67

Abbreviations: aOR = adjusted odds ratio; ASD = atrial septal defect; AVSD = atrioventricular septal defect; CHD = congenital heart disease; CI = confidence interval; DORV‐TGA = double outlet right ventricle with transposition of great arteries; GA = great arteries; NOS = not otherwise specified; TAPVR = total anomalous pulmonary venous return; VSD = ventricular septal defect.

^a Model adjusted for dietary quality index, maternal age, education, race, first birth, smoking, alcohol, kilocalorie per day and folic acid from multivitamins; Bolded results = aOR with p < 0.05 or negative test of trend, p < 0.10. Cell counts provided in Supporting Table 2.

TABLE 3.

Adjusted ^a associations (aOR) between maternal fish intake and noncardiac birth defects, National Birth Defects Prevention Study (NBDPS) 1997–2011.

Birth defects	Frequency of servings of fish
	Low intake 1–3/month		Moderate intake 1/week		High intake ≥ 2 per week
	aOR	95% CI	aOR	95% CI	aOR	95% CI
Amniotic band syndrome	0.98	0.74, 1.31	1.03	0.74, 1.44	1.12	0.77, 1.64
Anencephaly	1.05	0.86, 1.29	0.90	0.70, 1.15	0.93	0.69, 1.24
Spina bifida b	1.04	0.90, 1.20	0.92	0.77, 1.10	1.01	0.83, 1.24
Encephalocele	1.04	0.73, 1.49	1.20	0.81, 1.77	1.04	0.66, 1.66
Hydrocephaly	0.98	0.77, 1.24	1.17	0.89, 1.52	1.04	0.76, 1.43
Dandy‐Walker syndrome	0.74	0.50, 1.10	1.07	0.71, 1.61	0.53	0.30, 0.95
Holoprosencephaly	0.87	0.59, 1.30	0.74	0.46, 1.20	0.48	0.25, 0.93
Cataracts	0.86	0.65, 1.14	1.26	0.94, 1.70	0.96	0.66, 1.39
Anophthalmos	1.23	0.88, 1.72	1.04	0.69, 1.56	0.99	0.62, 1.60
Glaucoma	0.89	0.60, 1.31	1.22	0.81, 1.86	0.82	0.48, 1.39
Anotia/microtia	1.06	0.86, 1.31	1.12	0.89, 1.42	1.01	0.76, 1.33
Choanal atresia	0.63	0.43, 0.94	0.52	0.31, 0.85	0.71	0.41, 1.22
Cleft palate c	0.87	0.76, 1.00	0.91	0.78, 1.06	0.81	0.67, 0.98
Cleft lip w/wo cleft palate c	0.94	0.85, 1.04	0.90	0.80, 1.02	0.96	0.84, 1.11
Esophageal atresia	0.94	0.77, 1.14	1.07	0.86, 1.34	0.95	0.73, 1.24
Intestinal atresia	1.03	0.81, 1.31	1.11	0.85, 1.46	0.90	0.64, 1.25
Duodenal atresia	1.08	0.77, 1.51	1.08	0.74, 1.59	0.85	0.53, 1.38
Anorectal atresia/stenosis	0.84	0.71, 0.99	0.88	0.73, 1.06	0.95	0.76, 1.18
Cloacal exstrophy	1.50	0.88, 2.54	1.61	0.89, 2.91	1.08	0.51, 2.30
Biliary atresia	0.80	0.55, 1.16	0.79	0.51, 1.23	0.86	0.51, 1.43
Hypospadias	0.97	0.85, 1.10	0.86	0.74, 1.00	0.94	0.79, 1.11
Bilateral renal agenesis	1.01	0.69, 1.48	1.20	0.78, 1.84	0.88	0.51, 1.53
Longitudinal limb	1.09	0.86, 1.38	0.91	0.68, 1.22	0.96	0.69, 1.35
Longitudinal preaxial limb	1.02	0.74, 1.41	1.14	0.79, 1.64	1.02	0.65, 1.59
Transverse limb	0.93	0.77, 1.12	0.87	0.70, 1.10	0.89	0.68, 1.16
Craniosynostosis	0.96	0.84, 1.10	0.82	0.70, 0.97	0.83	0.68, 1.00
Diaphragmatic hernia	1.09	0.91, 1.30	1.01	0.82, 1.25	0.90	0.69, 1.16
Omphalocele	1.00	0.78, 1.29	1.00	0.75, 1.33	0.82	0.58, 1.16
Gastroschisis	0.96	0.83, 1.11	0.88	0.73, 1.06	0.75	0.60, 0.95
Sacral agenesis	1.20	0.68, 2.12	0.71	0.32, 1.57	0.93	0.40, 2.15

Note: Referent = < no fish or less than one serving a month.

Abbreviations: aOR = adjusted odds ratio; CI = confidence interval; w/wo = with or without.

^a Adjusted for dietary quality index, maternal age, education, race, first birth, smoking, alcohol, kcalorie per day, folic acid from multivitamins. Bolded results = p < 0.05. Cell counts provided in Supporting Table 3.

^b Cases that have both spina bifida and hydrocephaly are only classified under spina bifida.

^c Two separate categories for oral clefts are shown (cleft palate without cleft lip and cleft lip with or without cleft palate).

Fish intake was interpreted as having a meaningful association with a particular birth defect (elevated or decreased) if high fish intake (2 or more servings a week) was significantly associated with the birth defect (p < 0.05) or a test of trend across the four levels of fish intake and odds of the birth defect was significant (p < 0.10).

2.2. Quality of Maternal Diet

We assessed correlation coefficients (adjusted for energy intake) between fish intake and 14 nutrients and 59 other foods (Supporting Table 1). Because we did not identify any nutrients or foods with high correlations with fish intake (r = > 0.50), to adjust for potential confounding from dietary quality, we used a dietary quality index (DQI‐P) described previously (Bodnar and Siega Riz ²⁰⁰²; Carmichael et al. ²⁰¹²). Previous NBDPS studies observed that a high score on the DQI‐P (indicating a higher quality diet) was associated with decreased odds of many different NBDPS birth defects (Carmichael et al. 2012; Ma et al. ²⁰¹²; Study, Feldkamp et al. ²⁰¹⁴; Carmichael et al. ²⁰¹⁶; Botto et al. ²⁰¹⁶; Carmichael et al. ²⁰¹⁸; Weber et al. ²⁰¹⁸; Moore et al. ²⁰²⁰; Carmichael et al. ²⁰²⁰).

For six different beneficial foods and nutrients, each mother was assigned a score of (0, 1, 2 or 3). Each score represented the quartile (among all study participants) for the mother's intake of (fruits, vegetables, grains [including cereal and bread], food folate, calcium and iron). Higher scores were assigned to higher quartiles. In addition, for two detrimental foods, each mother was assigned a score of (0, 1, 2 or 3). Each of these scores represented the quartile for the mother's intake of (sweets [including ice cream, chocolate, non‐chocolate candy, pie, cake, cookies and sugar soda]) and the proportion of total dietary calories from fat. For these two scores, lower scores were assigned to higher quartiles.

Final DQI‐P scores were calculated by summing the eight scores described above and ranged from 0 to 24. DQI‐P scores were entered into the final model as quartiles of the distribution of DQI‐P scores for all participants.

2.3. Other Covariates

We did not identify the most parsimonious models because it would have entailed publishing separate sets of covariates for each of the 52 birth defects. Beginning with a set of covariates often used in analyses of risk factors for birth defects, we first ensured that none of them were colliders or intermediate variables for our study question: What is the association between fish intake in the year prior to conception and selected birth defects?

High maternal BMI is a moderate risk factor for many NBDPS birth defects (Waller et al. 2007; Gilboa et al. ²⁰¹⁰). However, compared to the other covariates, it has a large percentage of missing values (4.4%) and dropping it from our logistic regression models had a negligible effect on all of the aORs. Including study site decreased the precision of the models, and dropping it also had a negligible effect on the aORs. Therefore, both maternal BMI and study site were dropped from the final models.

In addition to the mothers DQI‐P score, final analyses were adjusted for 9 additional maternal characteristics and behaviors; years of age at delivery (< 18, 18–24, 25–29, 30–34, and ≥ 35), race/ethnicity (non‐Hispanic White, non‐Hispanic Black, Hispanic, and other), educational attainment (less than high school [≤ 11 years], completed high school [12 years], some college [13–15 years], and college graduate [≥ 16 years]), whether the participating case infant or control infant was the mothers first live birth (yes, no), any smoking or any alcohol consumption between 1 month before conception and 3 months after conception (yes, no), use of multivitamins (optimal use [daily intake between 1 month before and 1 month after conception], suboptimal use [less than daily intake between 1 month before and 1 month after conception], inadequate use [no intake or intake beginning later than 1 month after conception]), and number of kilocalories consumed per day.

2.4. Sensitivity Analyses

The goal of the first sensitivity analysis was to determine if restricting to isolated cases and rerunning the aORs in the main analysis would result in a meaningful decrease (toward the null) for any of the associations. This was done to ensure that any associations between maternal fish intake and ‘birth defect x’ were due only to that particular birth defect and did not occur because maternal fish intake was associated with other birth defects that co‐occurred with “birth defect x”.

We interpreted the results of the sensitivity analysis restricted to isolated birth defects as consistent with a meaningful association if any of the following criteria were met: (1) high fish intake (2 or more servings a week) had a significant association (p < 0.05) with cases of the isolated birth defect, (2) the p value for the test of trend between the four levels of fish intake and the isolated birth defect was < 0.10 or (3) the deviation between the aOR for isolated cases and the same aOR in the main analysis was small, that is, less than 10.0%.

A second sensitivity analysis was done to determine if excluding mothers who reported taking supplements containing omega three fatty acids or fish oil changed the results.

This study was approved by the Institutional Review Boards for each study site and the Centers for Disease Control and Prevention. All analyses were completed in SAS 9.4 (SAS Institute Inc.; Cary, NC, USA).

3. Results

A total of 43,846 participating mothers and their infants delivered within the study period (1997–2011). We excluded 1754 (4.0%) of study participants who were missing information on the initial question regarding fish intake. An additional 2060 (4.9%) participants were excluded because they had prepregnancy diabetes, missing data for two or more foods, or missing data for one or more covariates. There remained 29,242 mothers of infants with birth defects and 10,973 mothers of control infants. Among mothers of controls, 13.9% reported high fish intake, 21.2% moderate intake, 33.6% low intake, and 31.3% reported they never ate fish or ate it less than once a month (Table 1).

TABLE 1.

Frequency of fish intake by maternal characteristics, 10,973 control mothers, National Birth Defects Prevention Study, 1997–2011.

Maternal characteristics		N	Frequency of servings of fish (%) a				CPR b 95% CI
			< 1 month	1–3 months	1 week	≥ 2/week
	All controls	10,973	31.3	33.3	21.1	13.9
Previous births	≥ 1	6103	29.2	34.9	21.9	14.1	1.04 (0.95, 1.15)
	None	4006	34.6	31.5	20.3	13.5	1.00 (referent)
Education (years)	0–8	504	29.9	29.0	25.2	18.8	1.20 (0.97, 1.49)
	9–11	1126	41.4	28.4	17.0	15.2	1.01 (0.84, 1.20)
	12	2380	38.6	30.8	17.4	15.2	1.00 (referent)
	13–15	2718	31.1	35.2	20.5	15.1	1.00 (0.87, 1.15)
	≥ 16	3318	22.9	36.9	25.4	17.4	1.13 (1.00, 1.28)
Race/Ethnicity	NH Black	1066	16.2	31.7	26.0	26.1	2.51 (2.22, 2.83) *
	NH White	5993	34.3	35.6	19.7	10.4	1.00 (referent)
	Hispanic	2451	30.9	31.3	23.5	14.4	1.38 (1.23, 1.55) *
	Other	598	30.8	26.1	19.5	23.6	2.26 (1.95, 2.63) *
Age (years)	≥ 35	216	17.4	32.4	27.1	23.1	3.12 (2.09, 4.64) *
	30–34	2602	23.9	36.0	24.6	15.5	2.30 (1.55, 3.41) *
	25–29	2856	30.4	35.2	21.4	13.1	1.95 (1.31, 2.90) *
	18–24	2951	42.8	30.4	16.3	10.6	1.57 (1.05, 2.34) *
	< 18	327	52.5	28.4	12.4	6.7	1.00 (referent)
BMI (kg/m2) c	> 30	1736	33.1	33.3	20.1	13.5	0.96 (0.84, 1.09)
	25 to < 30	2227	31.1	34.5	20.8	13.6	0.96 (0.86, 1.09)
	18.5 to < 25	5223	30.4	34.1	21.5	14.1	1.00 (referent)
	< 18.5	516	36.0	31.4	19.4	13.2	0.94 (0.75, 1.17)
Use of vitamins d	Optimal	2777	25.5	34.5	24.7	15.3	1.11 (1.00, 1.24)
	Sup‐optimal	2610	33.7	34.4	19.2	12.7	0.92 (0.82, 1.04)
	Inadequate	4723	33.4	32.5	20.4	13.8	1.00 (referent)
Alcohol use e	No	6319	33.3	32.8	19.6	14.3	1.09 (0.98, 1.20)
	Yes	3732	28.0	34.8	24.0	13.2	1.00 (referent)
Smoked e	No	8313	29.4	33.8	22.4	14.4	1.22 (1.07, 1.40) *
	Yes	1766	40.1	32.3	15.9	11.7	1.00 (referent)

^a Percentages may not add to 100% because of rounding.

^b The crude prevalence ratios (CPRs) compare the prevalence of mothers reporting high fish intake (2 or more servings per day) across levels of each maternal characteristic compared to a referent level.

^c BMI = body mass index (weight in kilograms divided by height in meters squared) at conception; NH = Non‐Hispanic.

^d Use of multivitamins (optimal [daily intake between 1 month before and 1 month after conception], suboptimal [less than daily intake between 1 month before and 1 month after conception], inadequate [no intake or intake beginning later than 1 month after conception]).

^e Any alcohol or smoking between 1 month before pregnancy and third month of pregnancy.

^*Significant CPR p < 0.05.

The data in Table 1 show differences in the prevalence of women reporting high fish intake across categories of maternal race/ethnicity and age and smoking status. Compared with white non‐Hispanic mothers, the CPR for high fish intake (2 or more servings per week) was significantly increased (p < 0.05) among mothers who were non‐Hispanic black (CPR = 2.5, 95% CI 2.2, 2.8), Hispanic (CPR = 1.4, 95% CI 1.2, 1.6), or of other races (Asian, Pacific Islander and other) (2.3, 95% CI 2.0, 2.6). Compared with mothers who were < 18 years of age, the CPR for high fish intake was significantly increased (p < 0.05) among mothers who were 18–24 (CPR = 1.6, 95% CI 1.1, 2.3), 25–29 (CPR = 2.0, 95% CI 1.3, 2.9), 30–34 (CPR = 2.3, 95% CI 1.6, 3.4) and ≥ 35 (CPR = 3.1, 95% CI 2.1, 4.6). A smaller (yet significant) increase in high fish intake was present among women who were nonsmokers compared to smokers (CPR = 1.2, 95% CI 1.1, 4.0) (Table 1).

3.1. Protective Associations

The strongest results were observed for high fish intake (2 or more servings per week), which was significantly (p < 0.05) associated with a protective aOR for 5 of the 52 birth defects (double outlet right ventricle transposition of the great arteries (TGA), Dandy Walker syndrome, holoprosencephaly, cleft palate, and gastroschisis) as shown in Tables 2 and 3. All of the tests for linear trends between the 4 levels of fish intake (low ➔ high) and the odds of these 5 defects were negative in direction, with p values of 0.42, 0.19, 0.02, 0.04, and 0.02, respectively (Supporting Table 4).

A high intake of fish was also associated with protective aORs for isolated heterotaxia with CHD (congenital heart disease), TAPVR (total anomalous pulmonary venous return), pulmonary valve stenosis, choanal atresia, and craniosynostosis. The aORs for these five birth defects all had p values ≥ 0.05 and negative tests of trend, with p values of 0.02, 0.03, 0.10, 0.04, and 0.01, respectively. Thus, except for pulmonary valve stenosis, based on a significant test for trend, each of these birth defects met the study criteria for a protective association with fish intake. Among the 42 remaining birth defects, none of the tests for trend in Supporting Table 4 suggested a significantly protective effect.

Overall, there were nine birth defects that met the criteria for a meaningful protective association with fish intake, with aORs for high fish intake ranging from 0.37 to 0.83. (Cell counts for Tables 2 and 3 are available in Supporting Tables 2 and 3.)

3.2. Elevated Associations

In Table 2 a single significantly elevated association (p < 0.05) was observed between high fish intake and tricuspid atresia (aOR = 1.76; 95% CI 1.07, 2.89), with a positive test for trend (p = 0.04). (Supporting Table 4). No other positive tests of trend were observed with p values < 0.10.

3.3. Isolated Birth Defects

The results in Table 4 address the question of whether the meaningful associations (negative or positive) for high fish intake and the 10 birth defects discussed above remain when restricted to isolated birth defects. For each of these birth defects, Table 4 shows the aOR for high fish intake from the main analysis, the aOR for high fish intake from the analysis restricted to isolated birth defects, and the percent change between the two aORs.

TABLE 4.

Comparison of adjusted ORs ^a for high fish intake from results restricted to isolated cases (Supporting Tables 6 and 8) with results from main analysis (Tables 2 and 3) for 10 selected birth defects ^b , NBDPS 1997–2011.

Birth defects	Isolated cases		Main results
	aOR	95% CI	aOR	95% CI	%Change c
Heterotaxia with CHD	0.32	0.07, 1.44	1.25	0.86, 1.81	+74.4
DORV‐TGA	0.41	0.19, 0.86*	0.42	0.21, 0.84*	+2.4
TAPVR	0.74	0.48, 1.14	0.70	0.46, 1.06	−5.7
Tricuspid atresia	1.76	1.07, 2.89*	1.83	1.08, 3.12*	+3.8
Dandy Walker syndrome	0.57	0.28, 1.14	0.53	0.30, 0.95*	−7.5
Holoprosencephaly	0.35	0.15, 0.80*	0.48	0.25, 0.93*	+27.1
Choanal atresia	0.95	0.48, 1.90	0.71	0.41, 1.22	−33.8
Cleft palate	0.83	0.67, 1.02	0.81	0.67, 0.98*	−2.5
Craniosynostosis	0.83	0.68, 1.02	0.83	0.68, 1.00	0.0
Gastroschisis	0.72	0.57, 0.92*	0.75	0.60, 0.95*	+4.0

Abbreviations: aOR = adjusted odds ratio; CI = confidence interval; DORV‐TGA = double outlet right ventricle with transposition of the great vessels; NBDPS = National Birth Defects Prevention Study; TAPVR = total anomalous pulmonary venous return.

^*Significant result with p < 0.05.

^a All odds ratios are adjusted for dietary quality index, maternal age, education, race, first birth (y/n), smoking (y/n), alcohol (y/n), kcalorie per day and folic acid from multivitamins.

^b Nine birth defects are included because they had meaningful associations with fish intake in the main analyses, that is, aOR for high fish intake (p < 0.05) or a test for trend across the 4 levels of fish intake with a p value < 0.10 (Tables 2 and 3). Heterotaxia is included because a protective effect was present for isolated cases that was not present in the main analysis.

^c Change in aOR = [(aOR for main results−aOR for isolated cases)/aOR from main results] × 100. Results in which the % change in the aOR was negative and > 10.0% are in bold.

For 4 of the birth defects in Table 4, the aOR for isolated cases moved away from the null and remained significantly protective (p < 0.05). For heterotaxia, a nonsignificant protective aOR with a significant test for trend (p value = 0.02 (negative slope)) was present that was not observed in the main analysis. For craniosynostosis, the aORs for isolated cases remained the same as those for non‐isolated cases in the main analysis. And, for the remaining 4 birth defects the aORs for isolated cases moved toward the null, with percent changes in the aORs of −5.7% for TAPVR (total anomalous pulmonary venous return), −7.5% for Dandy Walker syndrome, −33.8% for choanal atresia, and −2.5% for cleft palate. (Table 4). The association between high fish intake and isolated choanal atresia deviated from the association in the main analysis by more than 10% and the test of trend between the four levels of fish intake and the odds of isolated cases of choanal atresia was not significant (p = 0.27) (not shown). Thus, with the exception of choanal atresia, the criteria for a protective association between fish intake and isolated cases of each of 9 birth defects in Table 4 was satisfied. However, an a posteriori analysis revealed that in the main analysis, none of the cases of choanal atresia overlapped with the 8 other birth defects for which we observed a protective association with fish intake in the main analysis.

The complete analysis of isolated birth defects for all 52 birth defects and all four levels of fish intake is provided in Supporting Table 5 (cardiac birth defects) and Supporting Table 6 (noncardiac birth defects). Cell counts for Supporting Tables 5 and 6 are provided in Supporting Tables 7 and 8.

3.4. Supplements of Fish Oil or Omega Three Fatty Acids

Participants who reported taking supplements of fish oil or omega three fatty acids any time in the 3 months before conception or during pregnancy comprised 0.2% (82/40,215) of the study participants. When we repeated the results for Tables 2 and 3 after deleting women who took these supplements, we observed that all changes in the aORs were < 10% and the birth defects for which fish intake was protective remained the same (not shown).

4. Discussion

To our knowledge, this is the first study of the association between maternal fish intake and birth defects to assess a wide range of different categories of birth defects. The large sample size of the NBDPS allowed us to make assessments for different levels of maternal fish intake and 52 birth defects. The strongest findings were for high maternal intake of fish (2 or more servings per week), which was significantly protective for 9 of the 52 birth defects, confirming the hypothesis of protective effects between maternal fish intake and some birth defects.

Under the null hypothesis that high fish intake is not significantly associated (p < 0.10) with any of the 52 birth defects in this study, we would expect that high fish intake would have a protective association (aOR < 1.0) for approximately 2.6 birth defects (52*0.10*0.5) and would be a risk factor (aOR > 1.0) for approximately 2.6 birth defects aORs (52*0.10*0.5). Thus, it is unlikely that all nine of the protective associations between fish intake and specific birth defects can be explained by chance. However, the single positive association observed between high fish intake and an increased risk of tricuspid atresia is consistent with chance.

Fish and shellfish are major dietary sources of essential nutrients, particularly the long‐chain PUFAs, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) (de Oliveira Otto et al. ²⁰¹³). In animal and experimental studies, EPA and DHA incorporate into cellular membranes, altering membrane fluidity and receptor function in ways that modulate pro‐inflammatory signaling and stabilize cardiac ion channels. They also interact directly with membrane proteins, such as ion channels and G‐protein–coupled receptors, contributing to anti‐arrhythmic and anti‐inflammatory effects. Furthermore, omega‐3 PUFAs regulate gene expression by activating nuclear receptors and transcription factors involved in lipid metabolism, energy balance, and inflammatory pathways (Mozaffarian et al. 2011). Altogether, these mechanisms suggest that omega‐3 PUFAs support vascular health.

Additionally, in nonpregnant adults, many prospective cohort studies and clinical trials have shown that low dietary consumption of omega‐3 PUFAs is associated with conditions indicating poor vascular health, such as high blood pressure, high serum LDL cholesterol and ventricular arrhythmias (Saini and Keum 2018). And, studies of associations between maternal conditions that effect vascular health (prepregnancy diabetes (Tinker et al. 2020), maternal obesity (Waller et al. 2007; Gilboa et al. ²⁰¹⁰) and maternal hypertension (Fisher et al. 2017)) and NBDPS birth defects suggest that poor maternal vascular health may affect the risk of many types of birth defects.

Thus, improvements in vascular health among pregnant women who eat fish may wholly or partially explain the inverse association between fish intake and specific birth defects that was observed in this analysis. However, it is also possible that other dietary components present in fish or residual confounding can explain some or all of the protective effects.

Our findings are also consistent with findings from a case control study conducted in the Netherlands that assessed maternal diet using a food frequency questionnaire with 200 items (Obermann‐Borst et al. ²⁰¹¹). Comparing the highest to the lowest tertiles of dietary intake for the one‐carbon‐rich dietary pattern (characterized by a high intake of fish), they observed a protective association with congenital heart defects (in aggregate) (aOR = 0.32; 95% CI 0.18, 0.59). In comparison, the current study observed that high fish intake was associated with a protective association for 3 of 22 specific categories of heart defects.

Similarly, a cohort study of pregnant Japanese women assessed the association between maternal intake of many different types of seafood and 74 cases of gastrointestinal (GI) tract atresia, including esophageal, intestinal, duodenal, and anorectal atresias (Michikawa et al. 2019). Comparing the highest to the lowest tertiles of intake for (1) all types of seafood and (2) six specific species of seafood rich in omega‐3 (PUFAs), they observed protective associations of (aOR = 0.50; 95% CI 0.30, 1.0) and (aOR = 0.50; 95% CI 0.20, 1.3), respectively. In contrast, the current study observed null associations between fish intake and each of these atresias of the GI tract, despite having higher sample sizes for each of them.

Based on an earlier and smaller version of the NBDPS data, an analysis of the epidemiology of holoprosencephaly observed a protective aOR for higher maternal fish consumption, similar to what we observed in the current study (Miller et al. 2010). And, simultaneous with the submission of this manuscript, Williford et al. submitted a manuscript in which they used machine learning and the NBDPS database to assess the epidemiology of TAPVR, observing a crude protective association between maternal fish intake and TAPVR (Williford et al. 2026).

This analysis has some important limitations. Differential recall bias, in which mothers of cases and controls have different probabilities of recalling exposures, is often raised as a concern in case control studies. However, as there has been limited research in this area, it is doubtful that mothers are aware that fish intake may have a protective association with birth defects and case mothers are probably not more (or less) likely to report fish intake than control mothers.

Although the NBDPS questionnaire asked about the frequency of intake of 60 different foods, there was only one question on fish and no question regarding shellfish. The absence of questions on different species of seafood most likely resulted in non‐differential misclassification of seafood intake (occurring among cases and controls with a similar frequency). It follows that the current study may have underestimated the magnitude of the association between high maternal intake of fish or seafood and certain birth defects.

The types of seafood that the NBDPS study participants consumed can be estimated based on data from The National Fisheries Institute. These data do not depend on interviews of consumers, which can be affected by poor recognition and recall of types of seafood. They published lists of the types of seafood most commonly sold in the US beginning in 2005 (National Fisheries Institute ²⁰⁰⁵–2011). The most commonly reported types of seafood remained similar beginning in the first year of their data (2005) through the last year of the NBDPS data collection (2011). In order of highest to lowest pounds per capita, the 8 most common types of seafood sold were shrimp, canned tuna, salmon, pollock, tilapia, catfish, crab, and cod.

An estimate of the extent to which our study participants may have consumed “high mercury fish” is present in an analysis of NHANES dietary interviews (1999–2006) conducted during a time period similar to this study. Among 3692 US women of child bearing age, the NHANES study elicited information on consumption of three categories of fish (swordfish, mackerel and shark) that the USDA classifies as having the highest levels of mercury compared with other seafood (US Food and Drug Administration ²⁰²¹). Consumption of these three types of high mercury fish in the 30 days prior to the interview was reported by 1.2%, 1.5% and 0.2% of the participants, respectively (Razzaghi and Tinker 2014). To the best of our knowledge, there is no data from population‐based studies on the frequency of consumption of 4 other categories of fish that are also classified by the FDA as “highest levels of mercury” (marlin, orange roughy, tilefish from the Gulf of Mexico and bigeye tuna).

With the exception of albacore canned tuna, all of the categories of seafood that were reported as commonly eaten in the US by the National Fisheries Institute are classified by the FDA as “best choices” because they have the lowest levels of mercury compared to other seafoods (US Food and Drug Administration ²⁰²¹). Albacore tuna, comprises approximately 30% of canned tuna (Everyday Health ²⁰²⁵), and is classified by the FDA as a “good choice” because it has levels of mercury intermediate between “best choices” and the “highest mercury levels” (US Food and Drug Administration ²⁰²¹).

Thus, we conclude that this population‐based study primarily captured the intake of fish with relatively low levels of mercury, which comprise the great majority of all fish eaten by women in the US during the study period.

The results of this study are consistent with the existing HHS recommendation that women who are or can become pregnant should consume 2–3 servings per week of “best choices” seafood—the category with the lowest levels of mercury. This recommendation was made because DHA is an important nutrient for the developing fetal brain and the relatively low levels of mercury present in species of fish classified by the FDA as “best choices” have not been associated with adverse developmental outcomes (Starling et al. 2015).

These results suggest that if women of reproductive age consume 2 or more servings of fish per week it may help to protect against the occurrence of 9 well defined birth defects, 8 of which have not previously been linked to maternal fish intake. As the majority of US women of child bearing age consume considerably less than 2 servings of fish per week (Papanikolaou et al. 2014), the possibility of protection against birth defects provides an additional reason for many women to increase their consumption of low mercury seafood prior to becoming pregnant.

Author Contributions

All of the authors on this manuscript have made substantial contributions to the analysis and interpretation of data and/or been involved in drafting the manuscript or revising it critically for important intellectual content. In addition, each author has given final approval of the version to be published.

Funding

This project was supported through NIH/NICHHD R03 # R03HD110803, Centers for Disease Control and Prevention (CDC) cooperative agreements under PA #96043, PA #02081, FOA #DD09‐001, FOA #DD13‐003, NOFO #DD18‐001, NOFO #DD23‐001 and #U01DD001309–01‐00 to the Centers for Birth Defects Research and Prevention participating in the National Birth Defects Prevention Study (NBDPS) and/or the Birth Defects Study To Evaluate Pregnancy exposureS (BD‐STEPS).

Coding of drug information in the National Birth Defects Prevention Study used the Slone Drug Dictionary under license from the Slone Epidemiology Center of Boston University.

Disclosure

The findings and conclusions in this report also do not represent the views of the State Health Departments and State Birth Defects Registries that participated in this study.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Table 1 Spearman correlations between frequency of intake of fish and DQI, foods¹ and nutrients,² among control mothers, NBDPS 1997–2011.

Supporting Table 2. Cell counts for associations between levels of fish intake and cardiac defects shown in Table 2, NBDPS 1997–2011.

Supporting Table 3. Cell counts for associations between levels of fish intake and non‐cardiac defects shown in Table 3, NBDPS 1997–2011

Supporting Table 4. Tests for linear trend between low to high levels of fish intake and odds of cardiac and non‐cardiac birth defects, National Birth Defects Prevention Study (NBDPS) 1997–2011.

Supporting Table 5. Adjusted^a odds ratios (aOR) between maternal servings of fish and 22 isolated cardiac birth defects, The National Birth Defects Prevention Study (NBDPS) 1997–2011.

Supporting Table 6. Cell counts for associations between levels of fish intake and isolated cardiac defects shown in Table 5, NBDPS 1997–2011.

Data Availability Statement

The data that support the findings of this study are available by contacting one of the Principal Investigators from the ten US states that participated in the National Birth Defects Prevention (NBDPS) Study. They can assist interested persons in developing a proposal for use of the data and submitting the proposal to the data sharing committee of the NBDPS for approval.