Food desert residence and prenatal diet quality

Abstract

Introduction

Suboptimal prenatal nutrition is associated with adverse pregnancy outcomes, yet the contribution of the lived food environment on prenatal diet quality remains incompletely understood. This study evaluated differences in prenatal macro‐ and micronutrient intake, glycemic and lipid intake patterns, and dietary quality indices among pregnant individuals living in food desert (FD) versus non‐FD areas.

Methods

This secondary analysis utilized data from an urban maternal‐child cohort enrolled between 2014 and 2024. Eligible participants were aged 18–40 years, had a singleton pregnancy at 8–14 weeks’ gestation at enrollment, and self‐identified as Black/African American. Dietary intake was assessed using the Block‐Bodnar food frequency questionnaires (FFQs) administered at 8–14 or 24–30 weeks’ gestation, and responses were utilized to calculate dietary intake measures. FD exposure status was assigned by geocoding participants’ primary residential address to census tract–level US Department of Agriculture FD classifications, with FD defined as a census tract meeting criterion for (1) low income AND (2) low access (at least 500 people or 33% of residents living farther than 1 mile from the nearest supermarket or large grocery store in an urban area or 10 miles in rural areas). Descriptive statistics were used to compare dietary intake by FD exposure. Primary outcomes of macronutrient (protein, carbohydrates, fat) and micronutrient (vitamin and mineral) intake were analyzed as continuous variables. Secondary outcomes included dietary glycemic indices, lipid indices, and five dietary scores including Healthy Eating Index 2015 (HEI 2015), Alternate Healthy Eating Index for Pregnancy (AHEI‐P), Dietary Approach to Stop Hypertension (DASH), Dietary Inflammatory Index (DII), and Mediterranean Diet (MED) score. Multivariable linear regression models were used to evaluate associations between FD residence and dietary outcomes, adjusting for maternal age, education level, insurance status, first‐trimester body mass index (BMI), parity, and total energy (caloric) intake.

Results

Of 543 pregnant participants, 242 (45%) resided in FDs. Median BMI (kg/m²) was similar between FD and non‐FD groups (28.0 vs. 27.0; p = 0.32). In both bivariate and adjusted analyses, FD residence was not associated with significant differences in macro‐ or micronutrient intake, or glycemic or lipid dietary patterns. Adjusted analysis of diet quality food scores revealed a significant association of FD residence and higher pro‐inflammatory DII scores (β = 0.51; 95% CI, 0.27–0.98; p = 0.04). Across both geographic groups, median intake of protein and several key micronutrients (iron, folic acid, and vitamin D) fell below the recommended intake in pregnancy, even when accounting for supplement intake. Dietary patterns in both groups were characterized by high intake of total sugar and saturated fats, and low‐to‐moderate adherence to dietary guidelines.

Conclusion

In this urban Black/African American cohort, prenatal dietary quality was suboptimal regardless of FD residence. Pregnant individuals in both FD and non‐FD areas demonstrated insufficient intake of protein and critical micronutrients, as well as dietary indices associated with adverse glycemic indicators, lipid profiles, and inflammation. These findings underscore the need for comprehensive, pregnancy‐centered nutrition strategies that address structural, behavioral, and clinical determinants of diet beyond geographic food access alone. Future research should explore nuanced determinants of diet quality among diverse populations and test innovative public health interventions that extend beyond traditional FD paradigms.

1. INTRODUCTION

Food environment and food access are increasingly recognized as important social drivers of health (SDOH) that shape dietary behaviors. Food insecurity, defined as the limited or uncertain availability to acquire nutritionally adequate and safe foods, reflects systemic constraints related to the four pillars of insecurity: availability, access, utilization, or stability of food resources [1]. Temporal US population trends from 2011 to 2018 indicated rising rates of metabolic syndrome, increasing from 16%–18% to 21%–25% in individuals of child‐bearing age and from 32%–36% up to 37%–41% in women [2, 3]. Simultaneously, rates of adverse cardiometabolic pregnancy outcomes continue to rise, including higher rates of pregestational and gestational diabetes (GDM) by 27%–30% and doubling of chronic hypertension and hypertensive disorders of pregnancy (HDP) [4, 5, 6]. Poor prenatal nutrition has been associated with adverse maternal cardiometabolic outcomes [7]. Thus, there is a need to better understand the underlying contributors to poor prenatal nutritional status and associated pregnancy‐related risks.

A balanced prenatal diet comprising optimal quantities of macronutrients (protein, carbohydrates, and fat) and micronutrients (vitamins and minerals) supplies energy and building blocks for physiologic processes of pregnancy [8, 9]. Inadequate or excessive prenatal dietary intake is associated with pregnancy complications, including fetal growth restriction, macrosomia, preterm birth, HDP, GDM, and long‐term metabolic risk in the mother and child [10, 11, 12, 13, 14, 15, 16]. Vitamins and minerals are essential for cellular and developmental functions, including fertilization, embryogenesis [17], placental development, red blood cell turnover, fetal organogenesis, bone mineralization, and brain development [18]; deficiencies are associated with anemia, miscarriage, preterm birth, low birthweight, and congenital anomalies such as open neural tube defects [19, 20, 21, 22].

Dietary recommendations for pregnant individuals were newly included in the 2020–2025 US Department of Agriculture (USDA) Dietary Guidelines for Americans [23, 24, 25, 26, 27]. Despite these recommendations, poor adherence to dietary guidelines is well‐documented globally [28]. Food environment and food accessibility impact adherence to dietary recommendations [29, 30]. The study of food deserts (FDs), geographic areas with limited access to affordable and nutritious food, facilitates objective evaluation of food access and availability. In pregnancy, limited FD studies suggest that residence in an FD is associated with increased odds of preeclampsia, preterm labor [31], poor periconceptional dietary quality, and higher percent adiposity in the second trimester [32, 33]. However, these studies were conducted in regions outside of the Southeastern United States in populations lacking racial and ethnic diversity, with individuals who identify as non‐Hispanic Black representing 1.3%–22% of cohorts. Studies outside of the United States have demonstrated associations between food insecurity and low‐birth‐weight infants, GDM, and maternal anemia, although FD definitions are not comparable [34, 35, 36]. It is well established that racial disparities exist within US healthcare and food access; national adjusted prevalence reports demonstrate that individuals who identify as non‐Hispanic Black are disproportionately affected by food insecurity compared to non‐Hispanic White individuals (20% vs. 11%, respectively) [37]. With an estimated 19 million people, or 6% of the US population, living in FDs [38], and approximately 20% of census tracts within the state of Georgia classified as FDs compared to 8%–10% of census tracts across the United States [39], we sought to evaluate the association of FD residence on prenatal dietary intake in a geographically and systemically vulnerable pregnant population.

Our primary objective was to investigate differences in macro‐ and micronutrient intake between pregnant individuals living in FD versus non‐FD census tracts within an urban cohort of Black/African American pregnant participants. We hypothesized that macro‐ and micronutrient intake would be lower in those living in FDs. Secondary objectives included assessment of insufficient dietary intake relative to the US Recommended Dietary Allowances (RDAs), comparison of glycemic and lipid dietary patterns, and evaluation of prenatal diet quality using multiple validated dietary indices.

2. MATERIALS AND METHODS

2.1. Study design and population

This study was a secondary analysis of data from an urban maternal‐child cohort study in the state of Georgia designed to investigate the role of the microbiome in local and systemic inflammation and its association with preterm birth [40, 41]. Participants were recruited in‐person during prenatal appointments at two urban academic healthcare centers between 2014 and 2024. The cohort included pregnant individuals who were English‐speaking, born in the United States, and self‐identified as Black or African American. Participants provided written informed consent. Institutional review board (IRB) approval was obtained.

Inclusion criteria for the parent cohort were: ages 18–40 years, with a confirmed singleton intrauterine pregnancy at 8–14 weeks’ gestation, self‐identification as Black or African American, born in the United States, and without any chronic medical conditions (such as diabetes mellitus, hypertension, hyperlipidemia). For inclusion in this secondary analysis, participants were required to complete at least one food frequency questionnaire (FFQ) during pregnancy.

2.2. FD definition (exposure)

Participants’ primary residential address at the time of dietary assessment was geocoded to a latitude and longitude assignment using the ArcGIS method within the tidygeocoder R package [42]. Geocoded addresses were then matched to census tract boundary shapefiles (based on the 2010 US Census Bureau). The matched census tract for a given geocoded address was used to assign FD status according to the USDA Food Access Research Atlas [38]. In this definition, a census tract is classified as an FD if it meets both of the following conditions: (1) low income: if either the poverty rate is ≥20% or the median family income is ≤80% of the statewide or metropolitan area median family income and (2) low access to food retailers: if at least 500 people or at least 33% of residents live farther than 1 mile from the nearest supermarket, supercenter, or large grocery store in an urban area or 10 miles in rural areas.

2.3. Dietary assessment

Dietary intake was assessed using a Block‐Bodnar FFQ, which was administered at 8–14 or 24–30 weeks’ gestation, each capturing dietary intake of the preceding 3 months. These two collection points were determined by the parent study design to accommodate participant scheduling. The Block‐Bodnar FFQ is a dietary assessment that includes a detailed, quantitative intake of dietary choices, portions, and frequencies of usual eating habits over the past three months and has been validated for use in pregnancy [43]. Macronutrient intake (protein, carbohydrate, fat), micronutrient intake (vitamins and minerals), glycemic indices (glycemic index, glycemic load, total sugars, total fiber, soluble fiber, and simple sugar intake), and lipid indices (saturated fat, monounsaturated fat, polyunsaturated fat, cholesterol, omega‐6, and omega‐3 fatty acid intake) were calculated from FFQ responses using NutritionQuest Data‐on‐Demand online validated computational tools [44]. Five dietary scores including Healthy Eating Index 2015 (HEI 2015), Alternate Healthy Eating Index for Pregnancy (AHEI‐P), Dietary Approach to Stop Hypertension (DASH), Dietary Inflammatory Index (DII), and Mediterranean Diet (MED) score were calculated from FFQ responses using Dietaryindex, a validated R‐package that standardizes the calculation of a range of dietary indices of high interest in health research [45]; detailed scoring algorithms for these indices have been described previously and were implemented using validated methods [46].

2.4. Dietary outcomes

Primary outcomes consisted of quantitative measures of daily macro‐ and micronutrient intake analyzed as continuous variables. Macro‐ and micronutrient intake was then dichotomized into sufficient and insufficient intakes based on US Recommended Dietary Allowances (RDAs) in pregnancy [24].

Secondary outcomes included glycemic indices, lipid indices, and five dietary quality scores. Glycemic indices, including glycemic index, glycemic load, total sugar, and total fiber, reflect upon the quality of carbohydrate intake, which potentially impacts glucose metabolism during pregnancy. Lipid indices, including total fat, saturated fat, monounsaturated fat, and polyunsaturated fat, capture dietary fat composition relevant to inflammation and cardiometabolic risk. The five dietary quality scores (HEI 2015, AHEI‐P, DASH, DII, and MED) provide complementary assessments of overall diet quality, guideline adherence, and inflammatory potential, each emphasizing distinct aspects of nutritional intake that have been associated with maternal cardiometabolic health and pregnancy outcomes [47, 48, 49, 50, 51, 52, 53, 54, 55] (Table S3). All dietary outcomes were derived from FFQ data (as described above) and evaluated in relation to FD residence, defined at the census tract level.

2.5. Statistical analysis

Distributions of dietary intake variables were assessed for normality. Differences in the primary outcomes (macro‐ and micronutrient intake) were initially analyzed using Wilcoxon rank‐sum tests for continuous variables and Chi‐square tests for categorical variables as appropriate. These comparisons were repeated to evaluate both dietary intake alone and subsequently total dietary intake including supplement intake. To further evaluate associations between FD residence and dietary outcomes, multivariable linear regression models were constructed for selected macro‐ and micronutrient and dietary indices. Multivariable Poisson regression modeling was performed to investigate the association between FD exposure and insufficient nutrient intake (dichotomized based on RDA). Models were adjusted for maternal age, highest level of education, insurance, early pregnancy body mass index (BMI) (measured at 8–14 weeks’ gestation), parity, and total energy intake. These covariates were selected based on biological plausibility, prior literature linking them to dietary intake during pregnancy and data from our preliminary univariate analyses. Regression results are presented as adjusted β‐coefficients and 95% confidence intervals (CIs; linear regression) or incidence risk ratios (Poisson regression). Statistical analyses were conducted using Stata Statistical Software, version 19.0 [56] with statistical significance defined as a two‐sided p‐value < 0.05. Finally, we presumed that 25% of the population in a non‐FD would be affected by low macro/micronutrient intake compared to 50% in an FD. To test this difference with an alpha = 0.05 and power 0.8, we would need n = 58 subjects in FD and non‐FD groups.

3. RESULTS

Of the 710 pregnancies enrolled in the parent study, 543 had available FFQ data for inclusion in this secondary analysis (Figure 1). Of these, 242 (45%) lived in a census tract designated as an FD. The median BMI (kg/m²) was similar between FD and non‐FD groups (28 vs. 27; p = 0.32); all participants identified as non‐Hispanic Black/African American (Table 1). The majority of participants (n = 495, 91%) completed the FFQ at the 8–14 weeks’ gestation study visit, with the remaining completing the FFQ at 24–30 weeks’ gestation (n = 48, 9%). The same proportions of FD and non‐FD residence were noted in participants who completed the FFQ at each gestational time point (n = 214, 44% FD/8–14 weeks’ gestation and n = 28, 48% FD/24–30 weeks’ gestation, p = 0.55).

FIGURE 1.

Participant flow diagram. FFQ, food frequency questionnaire.

TABLE 1.

Participant demographics by food desert residence.

Demographics	Non–food desert N = 301 (%)	Food desert N = 242 (%)	p value
Age, years, median, IQR [Q1–Q3]	26, 8 [22–30]	24, 8 [21–29]	0.01
BMI a , mg/kg2, median, IQR [Q1–Q3]	27.1 [23.3–33.7]	28.3 [23.8–34.5]	0.32
Gravidity, median, IQR [Q1–Q3]	3 [2–4]	3 [2–4]	0.94
Parity, median, IQR [Q1–Q3]	1 [0–2]	1 [0–10]	0.86
Race			N/A
Black/African American	301 (100)	242 (100)
Ethnicity			N/A
Hispanic	0 (0)	0 (0)
Non‐Hispanic	301 (100)	242 (100)
Education			0.04
Less than high school	39 (13.0)	44 (18.2)
High school diploma or GED	115 (38.2)	105 (43.4)
Some college or more	147 (48.8)	93 (38.4)
Insurance			0.01
Medicaid, low income	101 (33.6)	89 (36.8)
Medicaid, right‐from‐start b	132 (43.9)	127 (52.5)
Private	69 (22.6)	26 (10.7)

Abbreviations: BMI, body mass index; IQR, interquartile range.

^a First trimester.

^b Healthcare coverage provided to pregnant individuals, infants, and children under age 19.

3.1. Macronutrient and micronutrient intake

In bivariate analyses, there were no significant differences in the median macronutrient or micronutrient intake between FD and non‐FD groups (Table 2). However, dietary intake across both groups revealed suboptimal nutritional intake when compared to RDAs in pregnancy. Regarding macronutrient intake, median protein intake (FD group: 62.7 g [interquartile range, IQR, 40.5–95.8], non‐FD group: 62.7 g [IQR 41.6–102], p = 0.72) were both below the protein RDA (71 g), while median intake of total fat (FD group: 74.7 g [IQR 47.0–114], non‐FD group: 69.8 g [IQR 46.4–114], p = 0.91) and carbohydrates (FD group: 223 g [IQR 145–351], non‐FD group: 219 g [IQR 140–350], p = 0.67) exceeded RDA thresholds in both groups (fat: 65 g and carbohydrates:175 g). Similarly, median intakes of key micronutrients (vitamins and minerals) were not significantly different between FD and non‐FD groups (Table 2).

TABLE 2.

Unadjusted prenatal dietary macronutrient and micronutrient intake by food desert residence.

Median	Non–food desert	IQR [Q1–Q3]	Food desert	IQR [Q1–Q3]	p value	RDA
Macronutrients
Total Caloric Intake, kcal	1741	1700 [1103–2803]	1801	1631 [1156–2788]	0.76	2000
Protein, g	62.65	60.28 [41.63–101.9]	62.66	55.29 [40.46–95.75]	0.72	71
Total fat, g	69.78	67.61 [46.43–114.0]	74.69	67.27 [47.02–114.3]	0.91	65
Carbohydrates, g	219.2	209.7 [139.7–349.5]	223.4	205.7 [145.6–351.3]	0.67	175
Micronutrients
Calcium, mg	734.4	714.5 [455.9–1170]	697.4	679.4 [455.3–1135]	0.61	1000
Phosphorous, mg	1109	1081 [704.8–1786]	1074	1017 [687.3–1705]	0.76	700
Iron, mg	12.68	12.37 [8.28–20.65]	13.39	11.88 [7.73–19.61]	0.95	27
Sodium, mg	2862	2777 [1938–4715]	2870	2551 [1896–4447]	0.87	2300
Potassium, mg	2420	2158 [1508–3666]	2382	2059 [1478–3537]	0.66	2900
Vitamin A, mcg RAE	758.7	850.5 [474.6–1325]	764.7	805.0 [458.2–1263]	0.63	770
Thiamine, mg	1.40	1.34 [0.94–2.27]	1.49	1.34 [0.88–2.21]	0.92	1.4
Riboflavin, mg	1.73	1.77 [1.11–2.88]	1.7	1.61 [1.01–2.67]	0.89	1.4
Niacin, mg	18.19	17.57 [11.21–28.78]	18.6	16.04 [11.33–27.37]	0.96	18
Vitamin C, mg	121.2	146.3 [66.94–213.3]	115.5	147.2 [66.71–213.9]	0.92	85
Folic acid, mcg	139.3	162.3 [76.75–239.0]	143.1	158.6 [80.24–238.8]	0.87	600
Zinc, mg	9.45	9.35 [5.86–15.21]	9.10	8.00 [5.81–13.81]	0.65	11
Vitamin B6, mg	1.73	1.60 [1.13–2.73]	1.725	1.74 [1.07–2.81]	0.80	1.9
Magnesium, mg	260.3	236.0 [160.6–396.6]	246.6	218.1 [162.7–380.8]	0.77	350
Vitamin B12, mg	4.12	4.69 [2.42–7.11]	3.81	3.95 [2.50–6.46]	0.61	2.6
Vitamin D, IU	114.0	145.4 [56.79–202.2]	100.2	126.1 [57.19–183.3]	0.52	600
Vitamin E, mg AT	7.25	7.38 [4.67–12.05]	6.89	7.23 [4.75–11.88]	0.65	15
Vitamin K, mcg	171.2	250.6 [85.54–336.1]	145.7	227.4 [76.9–304.3]	0.25	90
Copper, mg	1.19	1.13 [0.771–1.90]	1.14	1.04 [0.779–1.82]	0.72	1
Selenium, mcg	80.55	82.04 [52.29–134.3]	81.74	77.50 [54.10–131.6]	0.94	60

Abbreviations: AT, alpha‐tocopherol; IQR, interquartile range; RAE, retinol activity equivalents; RDA, Recommended Daily Allowance.

When categorizing macro‐ and micronutrient intake as sufficient or insufficient based on the RDAs in pregnancy, no significant differences in the proportion of participants with insufficient intake were found between FD and non‐FD groups. Participants from both geographic groups demonstrated insufficient daily intake below the RDA for several essential micronutrients, including calcium (64% in FD and 62% in non‐FD, p = 0.99), iron (38% in FD and 37% in non‐FD, p = 0.99), folic acid (47% in FD and 45% in non‐FD, p = 0.98), and vitamin D (90% in FD and 87% in non‐FD, p = 0.98), even after accounting for supplement intake (Table 3). Adjusted Poisson regression analysis of the proportion of FD groups with insufficient intake similarly demonstrated no association with FD residence (Table S1).

TABLE 3.

Unadjusted insufficient prenatal nutrient intake by food desert residence.

	Insufficient prenatal intake, n (%)
	Dietary			Dietary + supplements
	Non–food desert	Food desert	p value	Non–food desert	Food desert	p value
Macronutrient
Protein, g	174 (57.8)	141 (58.3)	0.92	–	–	–
Micronutrients
Calcium, mg	200 (66.4)	171 (70.7)	0.97	187 (62.1)	154 (63.6)	0.99
Copper, mg	117 (38.9)	101 (41.7)	0.98	57 (18.9)	45 (18.6)	0.99
Folic acid, mcg	283 (94.0)	231 (95.5)	0.99	134 (44.5)	114 (47.1)	0.98
Iron, mg	254 (84.4)	204 (84.3)	0.99	112 (37.2)	91 (37.6)	0.99
Magnesium, mg	209 (69.4)	173 (71.5)	0.99	198 (65.8)	170 (70.3)	0.97
Niacin, mg	149 (49.5)	114 (47.1)	0.98	68 (22.6)	51 (21.1)	0.98
Phosphorous, mg	75 (24.9)	62 (25.6)	0.99	75 (24.9)	62 (25.6)	0.99
Potassium, mg	183 (60.8)	155 (64.0)	0.98	183 (60.8)	155 (64.1)	0.98
Riboflavin, mg	116 (38.5)	90 (37.2)	0.99	56 (18.6)	42 (17.4)	0.98
Selenium, mcg	96 (31.9)	76 (31.4)	0.99	49 (16.3)	31 (12.8)	0.95
Thiamine, mg	149 (49.5)	115 (47.5)	0.98	63 (20.9)	51 (21.1)	0.99
Vitamin A, mcg RAE	152 (50.5)	122 (50.4)	0.99	71 (23.6)	55 (22.7)	0.99
Vitamin B12, mg	83 (27.6)	69 (28.5)	0.99	42 (14.0)	30 (12.4)	0.98
Vitamin C, mg	96 (31.9)	81 (33.5)	0.98	51 (16.9)	38 (15.7)	0.98
Vitamin D, IU	296 (98.3)	235 (97.1)	0.99	262 (87.0)	218 (90.1)	0.98
Vitamin E, mg AT	256 (85.0)	203 (83.9)	0.99	191 (63.4)	157 (64.9)	0.99
Vitamin K, mcg	81 (26.9)	72 (29.8)	0.97	81 (26.9)	72 (29.8)	0.97
Zinc, mg	180 (59.8)	161 (62.4)	0.98	79 (26.3)	63 (26.0)	0.99

In multivariable linear regression models adjusted for maternal age, education, insurance, BMI, parity, and FD residence, no significant differences were observed in total caloric energy intake between FD and non‐FD residence groups (β = 77.5; 95% CI, −256 to 411; p = 0.65). In multivariable linear regression model adjusted for maternal age, education, insurance, early pregnancy BMI, parity, and total energy intake, FD residence was not significantly associated with macronutrient intake, including protein (β = −1.59; 95% CI, 0.53–2.12; p = 0.40), total fat (β = −2.28; 95% CI, −4.93 to 0.37; p = 0.09), or carbohydrate (β = 5.94; 95% CI, −2.38 to 14.3; p = 0.16). In adjusted micronutrient analysis, FD residence was associated with lower calcium intake (β = −48.9; 95% CI, −95.4 to −2.35; p = 0.04), otherwise no significant associations were observed between FD residence and other micronutrients, including iron (β = −0.21; 95% CI, −0.86 to 0.43; p = 0.52), folic acid (β = −6.96; 95% CI, −26.3 to 12.3; p = 0.48), or vitamin D (β = −9.18; 95% CI, −26.8 to 8.44; p = 0.31) (Table 4).

TABLE 4.

Adjusted dietary intake outcomes by food desert (FD) residence.

Dietary outcome	β coefficient	95% CI*	p value
Macronutrients
Total Caloric Intake, kcal	77.5	−256 to 411	0.65
Protein, g	−1.59	0.53–2.12	0.40
Total fat, g	−2.28	−4.93 to 0.37	0.09
Carbohydrates, g	5.94	−2.38 to 14.3	0.16
Micronutrients
Calcium, mg	−48.9	−95.4 to −2.35	0.04
Phosphorous, mg	−29.2	−74.1 to 15.7	0.20
Iron, mg	−0.211	−0.86 to 0.43	0.52
Sodium, mg	−20.1	−161 to 121	0.78
Potassium, mg	−24.6	−145 to 96.6	0.69
Vitamin A, mcg RAE||	−8.43	−88.6 to 71.7	0.84
Thiamine, mg	0.03	−0.06 to 0.11	0.51
Riboflavin, mg	−0.05	−0.14 to 0.04	0.29
Niacin, mg	0.54	−0.70 to 1.79	0.39
Vitamin C, mg	10.9	−4.32 to 26.2	0.16
Folic acid, mcg	−6.96	−26.3 to 12.4	0.48
Zinc, mg	−0.85	−1.90 to 0.19	0.11
Vitamin B6, mg	0.028	−0.11 to 0.17	0.69
Magnesium, mg	4.71	−13.6 to 23.0	0.62
Vitamin B12, mg	−0.23	−0.75 to 0.30	0.40
Vitamin D, IU	−9.18	−26.8 to 8.44	0.31
Vitamin E, mg AT¶	0.052	−0.52 to 0.63	0.86
Vitamin K, mcg	−0.95	−48.4 to 29.3	0.63
Copper, mg	0.038	−0.064 to 0.14	0.46
Selenium, mcg	−1.95	−7.90 to 4.00	0.52
Glycemic Indices
Glycemic Index (glucose), average daily	0.048	−0.097 to 0.69	0.88
Glycemic Load (glucose), average daily	1.21	−3.03 to 5.46	0.58
Total sugar, g	5.21	−4.22 to 14.6	0.28
Dietary fiber, g	0.25	−1.08 to 1.59	0.71
Soluble fiber, g	−0.003	−0.42 to 0.42	0.99
Fructose, g	3.46	−0.003 to 6.93	0.05
Glucose, g	3.65	0.36 to 6.96	0.03
Lactose, g	−1.37	−2.87 to 0.14	0.08
Maltose, g	0.87	−0.15 to 0.32	0.46
Sucrose, g	−0.65	−4.5 = 3.29	0.75
Lipid indices
Saturated fat, g	−1.43	−2.58 to −0.28	0.02
Monounsaturated fat, g	−0.84	−2.01 to 0.32	0.16
Polyunsaturated fat, g	0.21	−0.48 to 0.90	0.55
Cholesterol, mg	−11.9	−40.3 to 16.5	0.41
Omega‐6 fatty acids, g	0.027	−0.59 to 0.64	0.93
Omega‐3 fatty acids, g	−0.011	−0.12 to 0.99	0.84
Dietary score
Healthy Eating Index 2015 (HEI 2015)	−2.00	−5.25 to 1.25	0.23
Alternate Health Eating Index‐for Pregnancy (AHEI‐P)	−1.57	−4.29 to 1.16	0.26
Dietary Approach to Stop Hypertension (DASH) score	−1.01	−2.29 to 0.28	0.13
Mediterranean Diet (MED) score	−0.29	−0.62 to 0.39	0.08
Dietary Inflammatory Index (DII)	0.51	0.027 to 0.98	0.04

Note: Multivariable linear regression analysis of prenatal dietary intake outcomes by FD residence, adjusted for maternal age, parity, BMI, the highest level of education and insurance, and total caloric energy intake.

Abbreviations: AT, alpha‐tocopherol; BMI, body mass index; CI, confidence interval; RAE, retinol activity equivalents.

3.2. Glycemic and lipid indices

Bivariate comparisons demonstrated no significant difference in glycemic and lipid dietary indices between participants residing in FD and those in non‐FD areas (Table S2). Notably, both geographic groups demonstrated total sugar intake that exceeded the RDA in pregnancy and high levels of saturated fat intake.

In linear regression models adjusted for maternal age, education, insurance, early pregnancy BMI, parity, and total energy intake, FD residence was not associated with glycemic index (β = 0.05; 95% CI, −0.60 to 0.70; p = 0.88), glycemic load (β = 1.21; 95% CI, −3.03 to 5.46; p = 0.58), total sugar intake (β = 5.21; 95% CI, −4.21 to 14.6; p = 0.28), or total fiber intake (β = 0.25; 95% CI, −1.08 to 1.59; p = 0.71). FD residence was associated with higher glucose intake (β = 3.65; 95% CI, 0.36–6.96; p = 0.03) and lower saturated fat intake (β = −1.43; 95% CI, 02.58–−0.28; p = 0.02); no significant associations were observed for other glycemic or lipid indices (Table 4).

3.3. Dietary quality indices

Bivariate analyses of five dietary quality indices (HEI 2015, AHEI‐P, DASH, DII, and MED) demonstrated no significant differences between FD and non‐FD groups. AHEI‐P scores showed low‐moderate adherence to pregnancy dietary guidelines for those residing in both FD and non‐FD geographic areas (41.5 and 40.2, p = 0.65, respectively). DII scores for both geographic groups were in the pro‐inflammatory range, further reflecting poor dietary quality irrespective of geographic food access (Table 5).

TABLE 5.

Dietary scores by food desert residence.

Median dietary score	Non–food desert	IQR [Q1–Q3]	Food desert	IQR [Q1–Q3]	p value	Reference range
Healthy Eating Index 2015 (HEI 2015)	59.5	14.0 [51.5–65.6]	58.4	14.0 [50.0–64.5]	0.13	0–100
Alternate Health Eating Index‐for Pregnancy (AHEI‐P)	41.5	30.8 [28.1–58.9]	40.2	26.4 [29.2–55.7]	0.65	0–90
Dietary Approach to Stop Hypertension (DASH) score	24	8 [20–28]	23	7 [20–27]	0.22	8–40
Mediterranean Diet (MED) score	4	4 [2–6]	3	3 [2–5]	0.15	0–9
Dietary Inflammatory Index (DII)	0.988	5.85 [−2.29–3.56]	1.544	5.68 [−1.81–3.87]	0.34	Pro‐inflammatory: >0 Neutral: 0 Anti‐inflammatory: <0

In multivariable linear regression models adjusted for demographic and clinical covariates as previously described, FD residence was significantly associated with higher mean DII scores (β = 0.51; 95% CI, 0.27 to 0.98; p = 0.04), indicating a more pro‐inflammatory diet. Analysis of other dietary scores demonstrated no significantly association with HEI 2015 (β = −2.00; 95% CI, −5.25 to 1.25; p = 0.23), AHEI‐P (β = −1.57; 95% CI, −4.30 to 1.16; p = 0.26), DASH (β = −1.01, 95% CI; −2.29 to 0.28, p = 0.13), or MED scores (β = −0.29; 95% CI, −0.62 to 0.39; p = 0.08) (Table 4).

4. DISCUSSION

In this secondary analysis of a prospective cohort of pregnant individuals who identified as Black or African American, we found that prenatal macro‐ and micronutrient intakes, as well as dietary glycemic and lipid indices, were similarly suboptimal among participants, regardless of whether they resided in an FD or non‐FD census tract. Pregnant individuals in both FD and non‐FD areas had inadequate intake of protein and key micronutrients such as iron, folic acid, and vitamin D, as well as high sugar intake, high fat intake, and poor adherence to pregnancy dietary guidelines. The presence of suboptimal nutrient intake and inflammatory dietary patterns across both FD and non‐FD settings suggests that structural, cultural, and behavioral factors beyond geographic access likely contribute to prenatal nutrition disparities. Multivariable regression analysis adjusting for confounders and covariates demonstrated similar findings, with the exception of findings that FD residence was significantly associated with lower calcium intake, higher glucose intake, and 0.51‐point higher pro‐inflammatory diet score compared to non‐FD residence.

Pro‐inflammatory diets with higher DII scores typically contain highly processed foods, refined grains, and lack in fiber, fruits, and vegetables. While a reference range of DII scores has not been established, reports of minimum and maximum scores range from −8.87 to +7.98, with 0 being neutral, higher scores (positive, >0) indicating pro‐inflammatory diets, and lower scores (negative, <0) indicating anti‐inflammatory dietary patterns [54]. Higher inflammation in the diet is likely a potential mediator between FD residence and adverse pregnancy outcomes, as inflammatory DII scores have previously been associated with inflammatory obstetric complications such as preterm birth and preeclampsia [50], and these clinical outcomes have previously been associated with FD residence [31].

Our dietary findings contrast with existing literature that has reported lower diet quality among pregnant individuals living in FDs [32]. Notably, our study included a larger proportion of participants residing in FDs (45%, compared to 25%) and was not limited to nulliparous participants. Our study specifically evaluated a self‐identified Black/African American population, focusing on a racially homogenous population historically underrepresented in nutrition research. Wood et al. [33] evaluated diet quality in FDs but the study population only contained <2% of Black/African American participants. While Tipton et al. [31] included a diverse population, diet quality was not assessed. Furthermore, Venkatesh et al. [32] evaluated Healthy Eating Index‐2010 scores, whereas we specifically evaluated individual macro‐ and micronutrients in addition to five validated dietary indices, each of which provides a unique perspective on the nutritional status of a population.

A key strength of this study lies in its ability to address a critical gap in knowledge about geographic and access factors that affect prenatal nutrition. Public health data on nutrition and food access in pregnancy are lacking and understanding nuances of nutritional intake patterns is crucial to understanding risk factors for pregnancy. Use of the Block‐Bodnar FFQ, which is a widely validated dietary assessment tool in pregnancy, allowed for objective and standardized evaluation of dietary intake. Addressing specific macro‐ and micronutrient intake in pregnancy allowed for insight into the clinical significance of nutritional status by population.

Several limitations should be acknowledged. This analysis did not include evaluation of clinical obstetric outcomes but rather focused on the effects of living in an FD on prenatal diet quality. We propose diet quality as a potential mediator between food environment and obstetric outcomes. Future directions will need to further assess associations of diet quality, food environment factors, and perinatal and neonatal outcomes. Further, dietary intake was assessed using an FFQ; while the FFQ is an objective assessment of nutrition intake, it is subject to recall bias and was administered at discrete gestational windows, which may not capture temporal changes in diet across pregnancy. Nutrition research poses challenges due to multifactorial influences of local and systemic factors such as food access, food regulation, and cultural practices. Sociodemographic and behavioral factors such as educational attainment, income, cultural dietary practices, and food literacy may influence dietary intake, but were not fully captured in this analysis. Although multivariable analyses adjusted for several key covariates, residual confounding is possible due to the absence of detailed individual‐level behavioral data on household food insecurity, and participation in nutrition assistance programs such as the Supplemental Nutrition Assistance Program (SNAP) or the Women, Infants, and Children (WIC). Additionally, geographic designation of FD status by census tract may not reflect the diversity of food environments experienced at the individual level. For instance, although the FD census tracts included in our sample had significantly more housing units that utilized SNAP benefits compared to the included non‐FD census tracts (586 vs. 362, p < 0.01), SNAP benefit utilization was not collected at the individual level. This gap limits our ability to assess the direct impact of nutrition assistance programs on dietary quality.

Our study sample exclusively included urban and suburban pregnant individuals who identified as Black or African American. While this offers a focused understanding of nutrition in a historically marginalized group and contributes to the growing body of evidence addressing racial disparities in pregnant populations, findings may not be generalizable to rural populations or individuals from other racial or ethnic backgrounds. Expanding this research to include more diverse geographic and demographic groups will be critical to fully understand the relationship between food environments and prenatal nutrition on a national scale.

Identifying modifiable social and environmental factors is important for developing targeted interventions to improve pregnancy health and outcomes. Existing programs such as SNAP and WIC may address economic barriers to stability and utilization, but gaps in availability and access persist. Other food environment factors that remain unaccounted for are urban food swamps [57] (geographic areas with high saturation of high‐calorie fast food and junk food) and stress on the neighborhood level as well as the individual level. These factors highlight the complexity of food insecurity beyond food availability. Based on our findings that highlight a nutritionally depleted pregnant population, we propose solutions rooted in public policy that uplift community programs at the neighborhood level. Utilizing pregnancy as a unique window of opportunity for programs that have previously shown success in reducing food insecurity, such as produce prescription programs and culturally tailored nutrition counseling, could have the power to change a community's health by utilizing food as medicine.

5. CONCLUSION

In this urban cohort of pregnant individuals identifying as Black or African American, prenatal intake of several essential macro‐ and micronutrients was suboptimal, regardless of residence in an FD. High glucose intake and more pro‐inflammatory diet patterns were significantly associated with FD residence after adjusting for confounders and covariates. Participants residing in both FD and non‐FD census tracts demonstrated dietary patterns characterized by high intake of total sugar, saturated fats, and low‐moderate adherence to pregnancy dietary guidelines. While geographic food access is an important component of the food environment, these findings highlight that geographic food access alone may not fully capture the factors influencing prenatal dietary quality in this population. Future research should explore additional structural and behavioral factors influencing dietary patterns in pregnancy among diverse demographic groups and support the development of innovative public health strategies that go beyond FD classifications to address the nutritional needs of diverse maternal populations.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

This study was conducted in accordance with institutional and international standards for research integrity. Approval from the Emory University Institutional Review Board (IRB) was obtained (IRB # 68441) and written informed consent was obtained for each participant.

Supporting information

Supporting Information

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.