Vitamin D is an essential micronutrient derived from sunlight, dietary intake and supplements that supports multiple cellular and metabolic processes promoting health throughout the lifespan. The 2020–2025 U.S. Dietary Guidelines (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020) define vitamin D levels sufficient for bone health as blood 25-hydroxyvitamin D [25(OH)D] concentrations of 20 to < 50 ng/ml. Although pregnant women are generally considered to be at high deficiency risk (Bodnar et al., Reference Bodnar, Simhan, Powers, Frank, Cooperstein and Roberts2007; Lee et al., Reference Lee, Smith, Philipp, Chen, Mathieu and Holick2007), optimal 25(OH)D concentrations for prenatal health remain uncertain (El-Hajj Fuleihan et al., Reference El-Hajj Fuleihan, Bouillon, Clarke, Chakhtoura, Cooper, McClung and Singh2015; Marshall, Mehta, & Petrova, Reference Marshall, Mehta and Petrova2013; Maxmen, Reference Maxmen2011), partly because those with low levels are often underrepresented in clinical research (Kiely, Wagner, & Roth, Reference Kiely, Wagner and Roth2020). This knowledge gap disproportionately affects Black pregnant women since darker skin pigmentation reduces cutaneous vitamin D synthesis (Ames, Grant, & Willett, Reference Ames, Grant and Willett2021; Institute of Medicine, 2011; Webb et al., Reference Webb, Kazantzidis, Kift, Farrar, Wilkinson and Rhodes2018). Furthermore, some evidence suggests that a lack of prenatal multivitamin use is unrelated to vitamin D insufficiency among Black women (Burris, Thomas, Zera, & McElrath, Reference Burris, Thomas, Zera and McElrath2015), raising questions about the utility of commonly used thresholds across diverse populations, and highlighting a need for research focused on the health implications of vitamin D that is relevant for Black families.

The developing fetus depends on maternal vitamin D transported across the placenta (Liu & Hewison, Reference Liu and Hewison2012). Research with animals has shown that prenatal 25(OH)D associates with neuronal differentiation, maturation, neurotransmitter synthesis, and synaptic function in the fetal brain (Cui, Gooch, Petty, McGrath, & Eyles, Reference Cui, Gooch, Petty, McGrath and Eyles2017; Kesby et al., Reference Kesby, Cui, O'Loan, McGrath, Burne and Eyles2010; Wagner & Hollis, Reference Wagner and Hollis2018); neurobiological mechanisms that support behavioral, emotional and social functioning in humans. Indeed, several population-based and registry studies in Europe have reported associations between low concentrations of prenatal 25(OH)D (measured continuously and categorically) and symptoms of attention-deficit hyperactivity disorder (ADHD) among school-age children and adolescents (Daraki et al., Reference Daraki, Roumeliotaki, Koutra, Chalkiadaki, Katrinaki, Kyriklaki and Papavasiliou2018; Morales et al., Reference Morales, Julvez, Torrent, Ballester, Rodriguez-Bernal, Andiarena and Tardon2015; Sucksdorff et al., Reference Sucksdorff, Brown, Chudal, Surcel, Hinkka-Yli-Salomäki, Cheslack-Postava and Sourander2021). Convergent results were reported among 6- to 13-year-olds in a large, nationwide study in the U.S. (Melough et al., Reference Melough, Li, Hamra, Palmore, Sauder, Dunlop and Switkowski2023). Findings have been more mixed in the early childhood period (Francis et al., Reference Francis, Charron, Li, Chen, Mayo, Butler and Rennert2021; Keim, Bodnar, & Klebanoff, Reference Keim, Bodnar and Klebanoff2014; López-Vicente et al., Reference López-Vicente, Sunyer, Lertxundi, González, Rodríguez-Dehli, Espada Sáenz-Torre and Torrent2019; Melough et al., Reference Melough, Li, Hamra, Palmore, Sauder, Dunlop and Switkowski2023; Sammallahti et al., Reference Sammallahti, Holmlund-Suila, Zou, Valkama, Rosendahl, Enlund-Cerullo and Tiemeier2023), highlighting the need for a more specific focus on dimensional measures of behavioral and emotional problems (e.g. overactivity, temper tantrums, fearful behaviors) within discrete developmental windows.

Few studies have examined associations between prenatal 25(OH)D and offspring behavioral and emotional health in Black-identifying mothers and children. In one notable exception, gestational 25(OH)D was inversely related to externalizing behaviors, but unrelated to internalizing problems, in a large sample of young children aged 1.5–5 years (Melough et al., Reference Melough, Li, Hamra, Palmore, Sauder, Dunlop and Switkowski2023). In another that included 66 toddlers identified as Black by their parent, Chawla and colleagues reported counterintuitive findings: the lowest prenatal 25(OH)D quartile was associated with fewer child internalizing problems and greater social competence (Chawla et al., Reference Chawla, Fuemmeler, Benjamin-Neelon, Hoyo, Murphy and Daniels2019). Although the small sample and unequal representation of participants across quartiles were important limitations urging the need for further research, this study is noteworthy for its additional focus on prosocial child behaviors that are consistent with the theorized beneficial effects of prenatal vitamin D. Existing studies have also been limited by a lack of consideration of social-contextual factors such as stress exposure and depressive symptoms during pregnancy, which have demonstrated associations with both prenatal vitamin D status and future child health outcomes (Aghajafari, Letourneau, Mahinpey, Cosic, & Giesbrecht, Reference Aghajafari, Letourneau, Mahinpey, Cosic and Giesbrecht2018; Gutteling et al., Reference Gutteling, De Weerth, Willemsen-Swinkels, Huizink, Mulder, Visser and Buitelaar2005; Jahan et al., Reference Jahan, Went, Sultan, Sapkota, Khurshid, Qureshi and Alfonso2021; Morgan, Channon, Penny, & Waters, Reference Morgan, Channon, Penny and Waters2021; Parker, Brotchie, & Graham, Reference Parker, Brotchie and Graham2017). Adjusting for these potential confounds is critical for clarifying the influence of prenatal vitamin D on child behavioral and emotional development.

Finally, a life-course conceptual framework, recognizing that current health is shaped by earlier exposures to environmental and psychosocial factors (Barker, Reference Barker2004; Ben-Shlomo & Kuh, Reference Ben-Shlomo and Kuh2002), aligns with calls for optimizing the health and development of children well before conception (Black et al., Reference Black, Liu, Hartwig, Villavicencio, Rodriguez-Martinez, Vidaletti and You2022; Vaivada et al., Reference Vaivada, Lassi, Irfan, Salam, Das, Oh and Sharma2022; World Health Organization, 2013). There is robust evidence that nutritional status prior to pregnancy is an important contributor to offspring health (Keenan, Hipwell, Class, & Mbayiwa, Reference Keenan, Hipwell, Class and Mbayiwa2018; Stephenson et al., Reference Stephenson, Heslehurst, Hall, Schoenaker, Hutchinson, Cade and Barker2018). A good example is the use of folic acid supplementation to reduce the risk of neural tube defects in the fetus; the recommended timing of supplementation is prior to conception in order to achieve sufficient levels (Abate et al., Reference Abate, Kumsa, Abebe, Wodaynew, Habtie, Kassa and Merchaw2024). Results from other nutrition-based interventions have shown that preconception, rather than prenatal, supplementation is associated with healthy offspring neurodevelopment (D'Souza et al., Reference D'Souza, Behere, Patni, Deshpande, Bhat, Bhalerao and Yajnik2021; Hynes, Seal, Otahal, Oddy, & Burgess, Reference Hynes, Seal, Otahal, Oddy and Burgess2019; Keats et al., Reference Keats, Das, Salam, Lassi, Imdad, Black and Bhutta2021; Robinson et al., Reference Robinson, Crozier, Miles, Gale, Calder, Cooper and Godfrey2018). In addition, an Australian cohort study showed that higher diet quality prior to pregnancy was associated with a reduced risk of behavior problems (e.g. hyperactivity) in school-aged offspring (Gete, Waller, & Mishra, Reference Gete, Waller and Mishra2021), although shared-method variance from maternal reports may have contributed to these results. Despite a growing number of human studies highlighting the importance of preconception nutrition factors, we are not aware of any longitudinal investigation of the extent to which preconception levels of vitamin D associate with positive and negative child behavioral health, or indeed might enhance beneficial prenatal levels, that could help inform early (i.e. pre-pregnancy) risk assessment and interventions to support children's health.

The current prospective study sought to examine relationships between preconception and prenatal vitamin D and offspring behavioral and emotional functioning among Black-identifying families that could inform the timing of interventions to support child health. Specifically, we hypothesized that prenatal 25(OH)D concentration would be associated with fewer offspring behavioral and emotional problems and more prosocial behaviors (e.g. kindness, helpfulness) after accounting for contextual factors such as prenatal stress and depression. Given emerging evidence for the importance of preconception health, we also examined the independent effects of preconception 25(OH)D on child outcomes. Finally, we explored the extent to which preconception 25(OH)D might strengthen the association between prenatal 25(OH)D and child behavioral and emotional outcomes.

Methods

Sample

The sample included participants from the population-based Pittsburgh Girls Study (PGS), an ongoing longitudinal study of 2450 women originally recruited during childhood (ages 5–8 years) in 1999–2000. PGS recruitment used a random sampling process that targeted 103 238 households in Pittsburgh, PA, with oversampling of families in low-income neighborhoods (Hipwell et al., Reference Hipwell, Loeber, Stouthamer-Loeber, Keenan, White and Kroneman2002; Keenan et al., Reference Keenan, Hipwell, Chung, Stepp, Stouthamer-Loeber, Loeber and McTigue2010). More than half of PGS participants (57%, N = 1402) identified as Black or Black and another race. Annual PGS interviews have been conducted for over 20 years; in assessment waves 17 and 18 when participants were 21–25 years, a venous blood draw was added to the study protocol. Across the duration of the study, participant retention has been high: 87.4% of the original sample was retained at the close of annual assessment wave 20.

In 2018, PGS participants who became pregnant were invited to participate in a study examining the effects of preconception and prenatal stress on offspring neurodevelopment. In this study, enrolled participants completed a venous blood draw and self-report measures during pregnancy and follow-up assessments, including a research lab visit when the child was age 2–4 years. Approval for all study procedures was obtained from the University of Pittsburgh Human Research Protection Office. Participants provided written informed consent prior to data collection.

Among Black-identifying PGS participants, 328 women became pregnant at least once after 2018 and enrolled in the substudy. Of this group, 280 (85.4%) also provided a prenatal venous blood sample. At the time of analysis, prenatal serum 25(OH)D assay results were available for 239 participants. Among this group, 185 participants had one or more age-eligible children (2–4 years) enrolled in the study (N = 199) of whom 157 (78.9%) also had data on child health outcomes (see Fig. 1).

Figure 1.

Consort diagram of the final sample. W, PGS participant; p, pregnancy; C, child. ^a One outlier was excluded from analysis resulting in a final sample of N = 156.

Measures

Vitamin D concentrations

Non-fasting venous blood samples were collected from participants during the preconception and prenatal periods into silicone-coated collection tubes (Monoject 8881301512) and allowed to clot. Tubes were centrifuged at 4 °C at 1500 rpm for 30 min to induce serum separation, aliquoted into plastic vials (Fisherbrand 02-681-338), and immediately stored at −80 °C. Frozen sera were transferred on dry ice to, and subsequently assayed by the Central Ligand Assay Satellite Services Laboratory, a CAP-accredited and College of American Pathologists and Clinical Laboratory Improvement Act (CLIA)-certified lab at the University of Michigan School of Public Health. Serum samples were assayed for total 25(OH)D using an enzyme-linked immunosorbent assay (ELISA) kit according to manufacturer instructions (R&D Systems RDKAP1971). The analytical range of the 25(OH)D assay was 2.81–333 ng/ml and the average intra-assay coefficient of variation (CV) was 7.8% at 5.5 ng/ml, 5.7% at 27.4 ng/ml, and 2.5% at 81.2 ng/ml. Concentration was measured continuously and dichotomized following common dietary guidelines for sufficiency (0 = 20–49.9 ng/ml) v. insufficiency (1 = <20 ng/ml) (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020).

Child behavioral and emotional health was measured using the parent-rated preschool version of the Strengths and Difficulties Questionnaire (SDQ; Goodman, Reference Goodman1997) for children aged 2–4 years. Biological mothers completed SDQ subscales assessing conduct problems, hyperactivity/inattention, emotional symptoms, peer problems, and prosocial behaviors. The five items of each subscale were rated on 3-point Likert scales as not true (0), somewhat true (1) and certainly true (2). In population samples, a three-subscale division of the preschool SDQ is recommended (Goodman, Lamping, & Ploubidis, Reference Goodman, Lamping and Ploubidis2010) consisting of behavioral/externalizing problems (sum of conduct problems and hyperactivity/inattention), emotional/internalizing problems (sum of emotional symptoms and peer problems) and prosocial behaviors (sum of five items, e.g. kindness, helpfulness, sharing). The SDQ preschool version has acceptable psychometric properties, including convergent validity with other measures of child behavioral and emotional problems (Croft, Stride, Maughan, & Rowe, Reference Croft, Stride, Maughan and Rowe2015; Goodman, Reference Goodman1997). In the current sample, internal consistency (Cronbach's alpha) was 0.71 for behavioral problems, 0.81 for emotional problems and 0.82 for prosocial behavior. These three subscales provide indicators of behavioral and emotional health and the absence/presence of behavioral and emotional problems.

Potential covariates

Participants reported on parity and receipt of public assistance (e.g. Special supplemental nutrition program for Women, Infants and Children [WIC], Supplemental Nutrition Assistance Program [SNAP], Temporary Assistance for Needy Families [TANF]). Given the impact of season on vitamin D synthesis, the timing of blood draws was categorized as: winter (December, January, February), spring (March, April, May), summer (June, July, August) and fall (September, October, November). Child sex assigned at birth (1 = male, 2 = female), child date of birth, birthweight, and gestational age at birth were collected from electronic birth record or by maternal report. Maternal and child ages were calculated from research visit dates relative to their respective birth dates, and gestational week of the prenatal blood draw was calculated from date of research visit and child gestational age at birth. Research visit dates were also used to determine the length of interval between the preconception and prenatal assessments.

Factors associated with both maternal 25(OH)D concentration and offspring behavioral and emotional health (Casseb, Kaster, & Rodrigues, Reference Casseb, Kaster and Rodrigues2019; Choy & Raine, Reference Choy and Raine2022; Naugler, Zhang, Henne, Woods, & Hemmelgarn, Reference Naugler, Zhang, Henne, Woods and Hemmelgarn2013) were considered for inclusion as covariates. Preconception maternal BMI, derived from interviewer-measured height and weight, was used as a proxy for the effects of adiposity on vitamin D metabolism (Wortsman, Matsuoka, Chen, Lu, & Holick, Reference Wortsman, Matsuoka, Chen, Lu and Holick2000), and tobacco use (0 = no, 1 = yes) was self-reported using the Nicotine, Alcohol and Drug Use measure (Pandina, Labouvie, & White, Reference Pandina, Labouvie and White1984). Prenatal depression was measured by self-report using the Adult Symptom Inventory (ASRI-4; Gadow, Sprafkin, & Weiss, Reference Gadow, Sprafkin and Weiss2004), which assesses the frequency of DSM-IV symptoms of major depressive disorder plus two related symptoms: low self-esteem and hopelessness. Seven symptoms were rated on 4-point scales (0 = never to 3 = very often), and four symptoms (changes in appetite, sleep, activity, concentration) were scored as absent (0.5) or present (2.5) following standard scoring procedures (Gadow et al., Reference Gadow, Sprafkin and Weiss2004). Items were summed to form a total depression severity score. The ASRI-4 depression scale demonstrates convergent and discriminant validity and has been shown to differentiate between clinical and non-clinical samples (Gadow et al., Reference Gadow, Sprafkin and Weiss2004). Participants also reported on the presence/absence of life stressors (e.g. inadequate housing, financial difficulties, household conflict, exposure to violence in the community) during pregnancy using the 28-item Difficult Life Circumstances Scale (DLC; Barnard, Reference Barnard1994). Items were summed to indicate number of life stressors (score range = 0–28). The DLC has good psychometric properties and in pregnant samples has demonstrated predictive validity with adverse child outcomes (Curry, Campbell, & Christian, Reference Curry, Campbell and Christian1994).

Analytic plan

Statistical analyses were performed with IBM SPSS 28.0 software. Potential covariates were evaluated with a correlation matrix. Study hypotheses were tested in three stepwise linear regression models with SDQ subscale scores (behavior problems, emotional problems, prosocial behavior) as the dependent variables (DVs). Prenatal 25(OH)D was entered as the primary independent variable (IV) in Step 1 after adjusting for covariates. Preconception 25(OH)D was added in Step 2 to examine possible timing effects, and the two-way interaction between preconception and prenatal 25(OH)D was added in Step 3 to examine synergistic effects on child behavioral and emotional health. We aimed to probe significant interactions using regions of significance to explore correspondence with sufficiency guidelines. A variance inflation factor ⩾1.5 was used to indicate collinearity among covariates (Kim, Reference Kim2019).

Results

Sample characteristics

Descriptive statistics are summarized in Table 1. Participants self-identified as Black (94.2%) or Black and another race (5.8%) and had a mean age of 23.4 years (s.d. = 1.2) and 25.2 years (s.d. = 1.4) at the preconception and prenatal visits respectively. Approximately one third of participants were primiparous, 40.4% received public assistance during pregnancy, and 48.1% were in the second trimester at the time of the prenatal blood draw. The mean interval between the preconception and prenatal blood draws was 1.8 years (s.d. = 0.7) years. Participants reported on average, three life stressors (range = 0–11) during pregnancy.

Table 1.

Characteristics of study participants (n = 156)

Note: a Preconception 25(OH)D data were missing for eight participants; b Birthweight was missing for 15 children.

Mean serum 25(OH)D concentrations were 15.5 ng/ml (s.d. = 7.7) before pregnancy and 18.0 ng/ml (s.d. = 9.2) during pregnancy; levels designated as insufficient according to U.S. Department of Agriculture and U.S. Department of Health and Human Services (2020) guidelines. In the preconception period, 72.3% of participants had insufficient 25(OH)D concentrations (<20 ng/ml), and during pregnancy the proportion was 68.6%. An outlier with vitamin D concentrations exceeding eight standard deviations above the mean was excluded to avoid disproportionate influence on analyses (Chambers, Hentges, & Zhao, Reference Chambers, Hentges and Zhao2004). The overall mean increase from preconception to pregnancy was very small (2.1 ng/ml, s.d. = 8.9), but there was considerable variability in the extent and direction of change in 25(OH)D over time (range: −19.2 to 42.7 ng/ml).

Offspring (53% female) completed a research visit on average at 2.8 years (s.d. = 0.5). Mean gestational age at birth was 38.6 weeks (s.d. = 2.4) and 12.8% of infants were born preterm (< 37 weeks). Mean birthweight was 3016 g (s.d. = 570); 11.3% had a birthweight < 2500 g, including two infants with birthweight < 1000 g. Child birthweight and gestational age at birth were highly correlated (r = 0.76, p < 0.001), and as birthweight was missing for 15 participants, we excluded it from further analyses. Maternal reports of child behavioral and emotional problems were similar to other community samples: 25.6% of children scored in the high/very high range for behavioral problems, 28.5% scored in the high/very high range for emotional problems, and 37.8% in the low/very low range for prosocial behavior (http://sdqinfo.org).

Concentrations of 25(OH)D before and during pregnancy were moderately stable (r = 0.48, p < 0.001) even after adjusting for the length of the interval between the two blood draws (partial r = −0.31, p < 0.001). As shown in Table 2, preconception 25(OH)D (continuously measured) was significantly and negatively correlated with prenatal life stressors and child behavior problems. There were no associations between sufficient/insufficient levels in the preconception period and other study variables. Prenatal 25(OH)D was higher in first-time mothers, was negatively associated with depression severity and with offspring behavioral and emotional problems and was positively associated with child prosocial behavior. Prenatal insufficiency showed a similar pattern of associations with child outcomes. Primiparous mothers reported higher levels of behavior problems in their child and fewer prosocial behaviors. Older child age was associated with more positive child behavior (i.e. fewer emotional problems and more prosocial behavior). Additionally, prenatal depression and life stress were positively correlated with child behavioral problems; prenatal depression was also associated with child emotional problems. Finally, offspring behavioral and emotional outcomes showed small to moderate correlations with each other, indicating that they were related but distinct domains of functioning

Table 2.

Bivariate correlations among study variables

Note: ** p < 0.01; * p < 0.05.

Given their associations with maternal 25(OH)D at either timepoint and/or the DVs, primiparity, prenatal depression, life stressors, child sex assigned at birth, and child age were selected for inclusion as covariates in the regression models. Other potential covariates identified in prior research (preconception BMI and smoking status, season, prenatal age, child gestational age at birth and receipt of public assistance) were excluded given their lack of association with either the IVs or DVs in the current study.

Associations between prenatal and preconception 25(OH)D and behavioral and emotional outcomes in early childhood

Results of stepwise linear regression analyses examining prospective associations between prenatal and preconception 25(OH)D and later child behavioral and emotional problems and prosocial behavior are shown in Table 3. Model statistics indicated reliable prediction of behavior problems in each step with modest effect sizes (R ² = 0.16–0.17). Prenatal 25(OH)D was inversely related to child behavior problems (β = −0.19, 95% CI −0.12 to −0.01) after adjusting for primiparity, prenatal depression severity, life stressors, child age, and child sex. Preconception 25(OH)D explained no additional variance in child behavior problems when added in Step 2, and attenuated the positive association between prenatal 25(OH)D and behavior problems to non-significance. These results were unchanged when the two-way interaction between preconception and prenatal 25(OH)D was added to the model in the final step.

Table 3.

Prospective associations between prenatal and preconception 25(OH)D and child behavioral and emotional outcomes

Note: β, standardized coefficient; 95% CI, 95% Confidence interval. Significant findings bolded for emphasis. Model results were unchanged with the addition of the two-way preconception × prenatal 25(OH)D interaction in Step 3.

Model summary: Behavior problems: Step 1: R ² = 0.16, F _(6,149) = 4.74, p < 0.001; Step 2: R ² = 0.17, F _(7,148) = 4.22, p < 0.001; Step 3: R ² = 0.17, F _(8,147) = 3.67, p < 0.001; Emotional problems: Step 1: R ² = 0.09, F _(6,149) = 2.53, p < 0.05; Step 2: R ² = 0.09, F _(7,148) = 2.15, p < 0.05; Step 3: R ² = 0.10, F _(8,147) = 1.93, ns; Prosocial behaviors: Step 1: R ² = 0.14, F _(6,149) = 4.08, p < 0.01; Step 2: R ² = 0.14, F _(7,148) = 3.50, p < 0.01; Step 3: R ² = 0.16, F _(8,147) = 3.44, p = 0.001. Variance Inflation Factors ranged from 1.03 to 1.41.

Step 1 results remained the same in sensitivity analyses that excluded eight participants without preconception 25(OH)D data. Results were also unchanged if the two extremely low birth weight infants were excluded.

Results showed no association between prenatal 25(OH)D in Step 1 and child emotional problems after adjusting for covariates. Similarly, preconception 25(OH)D explained no additional variance in emotional problems when added to the model in Step 2. In addition, results showed no synergistic effects of preconception × prenatal 25(OH)D on later child emotional problems.

When prosocial behavior was examined as the outcome, results revealed a significant positive association between prenatal 25(OH)D concentration and positive child behavior (β = 0.27, 95% CI 0.03–0.12) after accounting for covariates in the adjusted model. This positive association remained in Step 2 (β = 0.28, 95% CI 0.03–0.13), after also accounting for 25(OH)D concentration in the preconception period. There was no unique association between preconception 25(OH)D and early childhood prosocial behavior after accounting for prenatal 25(OH)D and covariates, and no interaction between preconception and prenatal 25(OH)D on child prosocial behavior.