Maternal Psychosocial Stress and Children’s ADHD Diagnosis: A Prospective Birth Cohort Study

Abstract

Objective:

Examine the association of mothers’ psychosocial stressors before and during pregnancy with their children’s diagnosis of ADHD.

Methods:

This study included 2,140 mother-child pairs who had at least one postnatal pediatric visit at [text removed for blind review] between 2003 and 2015. Child ADHD was determined via ICD-9 codes documented in electronic medical records. Latent factors of maternal stress and social support, and measures of the physical home environment and psychosocial adversities were constructed using exploratory factor analysis. The association between the latent factors and child ADHD diagnosis was examined using multiple logistic regression, controlling for known risk factors for ADHD.

Results:

Children were 1.45 (95% CI: 1.06, 1.99) and 3.03 (95% CI: 2.19, 4.20) times more likely to receive an ADHD diagnosis if their mother experienced a major stressful event during pregnancy or reported a high level of perceived stress, respectively. The number of family adversities increases the risk of ADHD diagnosis (2^nd quartile: OR=1.90 CI (1.31, 2.77); 3^rd quartile: OR=1.96 CI(1.34, 2.88); 4^th quartile: OR=2.89 CI (2.01, 4.16), compared to 1^st quartile.

Conclusion:

In this prospective, predominantly urban, low-income, minority birth cohort, mothers’ psychosocial stress before and during pregnancy appears to be an independent risk factor for the development of ADHD in their children.

Introduction

Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental psychiatric disorder in which deficits in the age appropriate development of self-regulatory processes are characterized by inattention, hyperactivity, or impulsiveness[1]. ADHD is among the most common childhood psychiatric disorders, with prevalence estimates among school-aged children diagnosed by a health care provider ranging from 7–12%[2–5]. Moreover, 66% to 85% of ADHD children will carry their disorder into adolescence and adulthood[4]. The disorder causes impairment in social and academic functioning and is associated with a range of poor outcomes. Prescription ADHD medication and behavior management can provide relatively effective treatment and in the developed world medication use has increased with the prevalence of ADHD[5]. However, ADHD treatments do not reduce the rising incidence or cure ADHD. Medications can have side effects and in many parts of the world are quite costly. The most recent estimation of the annual cost of ADHD to society, including health care utilization, medication utilization, education costs, crime costs, and unemployment costs, in the US alone is $14,500 per child ($42.5 billion in total)[6]. For all these reasons there is an urgent to need to identify important and modifiable early life risk factors that can inform strategies for primary prevention of ADHD.

The causes of ADHD are not completely understood. The increasing prevalence of ADHD[5] may be due to modifiable factors within children’s environments. Although genetic susceptibility likely plays a role as ADHD tends to run in families, environmental factors that disrupt neurodevelopment and self-regulation have also been implicated [7–9]. ADHD has been associated with early life adversity [10]^-[11], including mothers’ psychosocial stress [11, 12], but there is a particular lack of well-designed and adequately powered prospective birth cohort studies to examine the impact of these preventable exposures on the risk of ADHD diagnosis. The temporality of the relationships between early life risk factors and ADHD remains largely unknown, as most ADHD studies employ case-control designs. In such research, mother’s retrospective recall of stress exposure makes it difficult to differentiate the contribution of various aspects of psychosocial stress during pregnancy vs. postnatally. These limitations in the research on precursors of ADHD can be better addressed by capturing the complex, multi-faceted nature of psychosocial stress in prospective research that also examines a wide range of other maternal prenatal factors.

Another threat to the validity of prior research for informing primary prevention is the common practice of operationalizing ADHD as maternal or teacher report of symptoms or behaviors. This is problematic given the known influence of mothers’ psychological states (e.g. anxiety and depression) on their reports of children’s problems [13], as well as the possibility that social and cultural perspectives influence adults’ thresholds for children’s behavior problems. Further, scores on a behavior problem scale do not reflect all necessary diagnostic criteria, such as impairment in functioning across diverse life contexts [14], with the result that differences in findings by reporter and metric are well documented throughout the ADHD literature [15–17]. Integrating electronic medical records (EMR) into a prospective birth cohort study of child outcomes offers an opportunity to integrate maternal prenatal exposures, family demographic data, and mothers’ clinical health and birth outcome measures with subsequent clinical diagnosis of children’s ADHD.

In the present study, we utilized the data from [text removed for blind review] a prospective cohort of predominantly low-income minority participants in urban [text removed for blind review] in the U.S. We sought to examine prospective relationships between maternal and family stressors during the prenatal period, in the context of a wide range of other prenatal factors, and development of ADHD as determined by the ICD-9 codes in the EMR. We hypothesize that children born into family environments characterized by high maternal psychosocial stress before and during pregnancy are more likely to develop ADHD in childhood, even taking into account other known risk factors of ADHD.

Methods

Participants and Data Collection Procedures

As illustrated in the flow chart (supplemental figure 1), this prospective observational study analyzed data from 2,140 mother-child [text removed for blind review] born from 1998 through 2015 [text removed for blind review] with at least one postnatal general follow up pediatric visit. Eligible mothers were those who delivered a singleton live birth. For every preterm (<37 week) and/or low birth weight (<2500 g) infant, two term (≥37 week) and normal birth weight (>2500 g) infants and their mothers were enrolled. The exclusion criteria included pregnancies resulting from in vitro fertilization, multiple-gestation pregnancies, deliveries induced by maternal trauma, or newborns with substantial birth defects. The median age of children by the end of the follow-up period (September 2015) was 9.1 years (the first and third quartiles were 6.5 and 13.0 years, respectively.). The authors’ institutional review boards approved [text removed for blind review].

Recruitment occurred 24 to 72 hours postpartum and informed consent was obtained. Using a standardized questionnaire, an interviewer obtained mothers’ reports on family demographics, health behaviors during pregnancy, level of stress, stressful life events, social support, the physical home environment, and other information. Children who received their medical care at [text removed for blind review] postnatally were followed longitudinally and children’s well-child and specialty medical visits to the [text removed for blind review] were documented in the electronic medical records (EMR) beginning in 2003. For this analysis, physician diagnoses based on the International Classification of Diseases, Ninth Revision (ICD-9) for each postnatal visit were obtained from children’s EMR from 2003 through 2015.

Measures

The main exposure of maternal psychosocial stress was reported by mothers shortly after birth and the outcome of ADHD diagnosis was made when the children were between the ages of 3 and 16 years. During this postpartum interview, mothers rated their psychosocial stress on three scales (Supplemental table 1): daily general life stress during pregnancy (range 0–2); feeling stressed and overwhelmed in the past month on the 4-item Perceived Stress Scale [18, 19]; item scores 0–4); presence/absence of 5 major stressful events [20] and when they occurred (pre-pregnancy, 1^st, 2^nd or 3^rd trimester). Mothers also reported on other adverse exposures during pregnancy, including the presence of mice and cockroaches in home (vermin), amount of public assistance received, and employment level during pregnancy, which along with the stress measures were used to create a Composite Adversities Index (Supplemental table 1). They also reported on father involvement, social support, marital status, whether the pregnancy was planned, alcohol and illicit drug consumption and other health behaviors before and during pregnancy, as well as demographic characteristics, given that these have been linked to ADHD [4] (Table 2).

Table 2.

Bivariate (unadjusted) analysis of the relationship between maternal prenatal exposures and characteristics and children’s risk of ADHD diagnosis^ǂ

	Score	Neurotypical, No(%)	ADHD, No(%)	P-value	Specialist, No(%)	P-value	General, No(%)	P-value
		1780	360		252		108
General life stress scale	0	638 (35.84)	94 (26.11)	<0.001	61 (24.21)	0.001	33 (30.56)	0.002
	1–2	838 (47.08)	174 (48.33)		132 (52.38)		42 (38.89)
	3–4	304 (17.08)	92 (25.56)		59 (23.41)		33 (30.56)
Perceived stress scale	0–7	732 (41.12)	62 (17.22)	<0.001	41 (16.27)	<0.001	21 (19.44)	<0.001
	8–16	1048 (58.88)	298 (82.78)		211 (83.73)		87 (80.56)
Major stressful events scale	0	1415 (79.49)	246 (68.33)	<0.001	181 (71.83)	0.006	65 (60.19)	<0.001
	1–5	365 (20.51)	114 (31.67)		71 (28.17)		43 (39.81)
Family support scale	0–5	406 (22.81)	113 (31.39)	0.001	85 (33.73)	<0.001	28 (25.93)	0.730
	6–8	777 (43.65)	154 (42.78)		110 (43.65)		44 (40.74)
	9	597 (33.54)	93 (25.83)		57 (22.62)		36 (33.33)
Vermin in home index	0	1334 (74.94)	256 (71.11)	0.314	175 (69.44)	0.160	81 (75.00)	0.894
	1	294 (16.52)	68 (18.89)		49 (19.44)		19 (17.59)
	2	152 ( 8.54)	36 (10.00)		28 (11.11)		8 ( 7.41)
Employment scale	0	686 (38.54)	153 (42.50)	0.007	107 (42.46)	0.009	46 (42.59)	0.419
	1–3	421 (23.65)	102 (28.33)		74 (29.37)		28 (25.93)
	4	673 (37.81)	105 (29.17)		71 (28.17)		34 (31.48)
Public assistance index	0	276 (15.51)	54 (15.00)	0.921	36 (14.29)	0.873	18 (16.67)	0.871
	1	836 (46.97)	167 (46.39)		119 (47.22)		48 (44.44)
	2–6	668 (37.53)	139 (38.61)		97 (38.49)		42 (38.89)
Composite adversity index, quartile	Q1	537 (30.17)	54 (15.00)	<0.001	33 (13.10)	<0.001	21 (19.44)	0.025
	Q2	459 (25.79)	93 (25.83)		66 (26.19)		27 (25.00)
	Q3	386 (21.69)	88 (24.44)		64 (25.40)		24 (22.22)
	Q4	398 (22.36)	125 (34.72)		89 (35.32)		36 (33.33)
Composite adversity index, mean(SD)		14.3(5.7)	16.6(5.3)	<0.001	16.8(5.1)	<0.001	16.0(5.5)	0.003

Note:

^ǂChi-square test and t-test were calculated comparing ADHD, Specialist diagnosed ADHD, General diagnosed ADHD to Neurotypical

Medical record abstraction documented medical conditions, including maternal hypertensive disorder, pre-gestational/gestational diabetes, and intrauterine infection during pregnancy, as well as child sex, caesarean or vaginal delivery, gestational age in weeks, and birth weight. To examine the role of blood lead level, a suspected risk factor for ADHD[21], EMR data on blood lead levels, collected at birth and in subsequent medical visits, was analyzed for the subsample of infants and children with these data (72%). The average lead level across all available measurements was calculated. Mothers’ pre-pregnancy BMI was calculated from height and weight measurements in their EMR. Measures are scored in the direction of the title, so higher scores for stress, social support and employment indicate more of each.

Identification of Neurotypical Children and Children with ADHD

ICD-9 codes 314.0–314.9 in children’s EMRs identified those with ADHD, whether diagnosed by a specialist or general physician. The transition to the DSM-V diagnostic criteria in 2013 is not expected to impact determination of ADHD, as criteria for children less than 17 years of age do not substantially differ [22]. Specialists included developmental behavioral pediatricians, pediatric psychologists, pediatric neurologists, and child psychiatrists; general physicians included pediatricians and family medicine physicians. Children without any ICD-9 diagnoses for ADHD, Autism Spectrum Disorders, Intellectual Disability, and Developmental Disorders were assumed to have typical neurological development and were characterized as ‘neurotypical.’ Diagnostic codes used for Autism Spectrum Disorders, Intellectual Disability, and Developmental Disorders were published previously[23].

Statistical Analyses

Exploratory factor analysis (EFA) combined relevant questionnaire items to identify the most informative scales (Cronbach alpha ≥ 0.7) for measuring psychosocial stress and the home environment (Supplemental table 1). Scores on the scales and indices were converted into binary (low vs. high) or categorical (low, medium, high) variables according to their ranges and distributions. Each reference group represents those without significant stress or adversity (a score of 0, or for the Perceived Stress Scale a score below 8). Scores on the composite adversities index range from 0 to 34, truncated as 0–31 in the analyses to eliminate outliers. Chi-square and t-tests compared background characteristics between neurotypical and ADHD cases (Table 2). The significance threshold was set at p≤0.05.

Bivariate regression analyses compared neurotypical children to those diagnosed with ADHD on measures of prenatal stress (Table 3) and other maternal, child and family characteristics (Supplemental Table 4), and also compared specialist with general pediatrician-diagnosed cases across measures to assess the sensitivity of findings to differences across these measures.

Table 3.

Multiple logistic regression analysis of the relationship between maternal prenatal exposures and characteristics and children’s risk of ADHD diagnosis

Index		Model 1				Model 2
		Odds Ratio	95% CI		P-value	Odds Ratio	95% CI		P-value
General life stress scale	0	1				1
	1–2	1.44	1.08	1.93	0.014	1.35	1.00	1.82	0.054
	3–4	2.03	1.43	2.90	<0.001	1.42	0.95	2.11	0.089
Perceived stress scale	0–7	1				1
	8–16	3.02	2.23	4.08	<0.001	3.03	2.19	4.20	<0.001
Major stressful events scale	0	1				1
	1–5	1.74	1.32	2.29	<0.001	1.45	1.06	1.99	0.019
Family support scale	0–5	1				1
	6–8	0.79	0.59	1.07	0.124	0.88	0.65	1.20	0.430
	9	0.65	0.47	0.90	0.010	1.06	0.74	1.51	0.763
Vermin in home index	0	1				1
	1	1.18	0.86	1.61	0.305	1.22	0.89	1.69	0.221
	2	1.20	0.79	1.83	0.381	1.09	0.71	1.67	0.703
Employment scale	0	1				1
	1–3	1.12	0.83	1.51	0.462	1.07	0.78	1.46	0.688
	4	0.81	0.60	1.09	0.158	0.82	0.61	1.11	0.206
Public assistance index	0	1				1
	1	1.11	0.77	1.61	0.566	1.15	0.79	1.68	0.466
	2–6	1.14	0.78	1.67	0.504	1.12	0.76	1.66	0.567
Composite adversity index, quartile	Q1	1
	Q2	1.90	1.31	2.77	0.001
	Q3	1.96	1.34	2.88	0.001
	Q4	2.89	2.01	4.16	<0.001
Composite adversity index, linear trend		1.07	1.04	1.09	<0.001

Note: Model1 tested individual effect of each index, adjusting for maternal age at delivery, race/ethnicity, marital status, education, substance use, smoking, Intrauterine infection, gender, mode of delivery, gestational age, birth weight; Model2 included all indices in the same model, adjusting all the variables in Model

To examine the independent effects of general life stress, perceived stress, and stressful life events on the risk of ADHD diagnosis, three multivariate logistic regressions were conducted, controlling for characteristics significant in bivariate analyses (Table 5, Model 1). To assess the additive impact of all measures on ADHD risk, a single multiple logistic regression analysis was conducted, adjusting for these same covariates (Table 5, Model 2). Model 1 and Model 2 analyses were repeated for specialist diagnosed cases only to assess the sensitivity of findings to diagnosis type (Supplemental Table 6).

Within the subsample of children with blood lead level data, the multiple logistic regression analyses were repeated, adding lead levels to model 1 and model 2 (Supplemental Table 7). Lastly, multiple logistic regression of the composite adversity index (treated as continuous variable) on ADHD, adjusted for all covariates, estimated the average probability of ADHD at each level of this index (Figure 2) and the distribution of this was assessed across boys and girls (Supplemental figure 3).

Results

Supplemental table 1 presents the results of exploratory factor analyses characterizing stress and other psychosocial adversities; general life stress during pregnancy, perceived stress, major stressful events, family support, employment, and public assistance. All latent factors with the exception of the vermin in home and public assistance indices demonstrated acceptable to good internal consistency reliability (alpha > 0.7) and are referred to as scales. Figure 2 shows considerable individual variations in composite adversities index (scores from 0 to 34); but the frequency distribution patterns are similar between boys and girls (supplemental figure 3).

Maternal and child background characteristics comparing neurotypical children to those with ADHD, shown in Table 2, suggests that children with ADHD were more likely to be born to mothers who were unmarried (p=0.003), had not advanced past high school (p=0.008), had ever smoked during pregnancy (p<0.001), had ever reported substance use (p=0.008), or experienced an intrauterine infection during pregnancy (p=0.004). Children with ADHD were themselves more likely to be male (p<0.001), delivered via cesarean section (p=0.024), born pre-term (p<0.001), and of low birth weight (p<0.001). Within the subsample who had blood lead level data, those with elevated blood lead levels were more likely to have ADHD (p=0.005).

The majority of the 360 ADHD diagnoses were made by a specialist (n=252). Those diagnosed by a specialist versus a general pediatrician differed only by race/ethnicity (Supplemental table 4). The median age of first diagnosis was 6 years. Table 3 reports the distribution of explanatory factors and results of three uncontrolled bivariate regression analyses. The first, comparing neurotypical and ADHD cases, shows significantly more maternal general life stress, perceived stress, and major stressful events, lower family support, and lower maternal employment among children with ADHD. The same comparisons hold whether children are diagnosed by specialists or general pediatricians for mothers’ general life stress, perceived stress, and major stressful events.

Table 5 shows the individual impacts of each psychosocial factor on ADHD in nine separate multiple logistic regressions, adjusting for maternal age at delivery, race, marital status, education, substance use, smoking, intrauterine infection, gender, mode of delivery, gestational age, and birth weight. In these controlled bivariate analyses, the odds of ADHD diagnosis increase as mothers’ general life stress increases (score 1–2 vs. score 0: OR=1.44 CI: 1.08, 1.93, p=0.014; score 3–4 vs. score 0: OR=2.03 CI: 1.43, 2.90, p<0.001), and are higher with more perceived stress (score 8–16 vs. score 0–7: OR=3.02 CI: 2.23, 4.08, p<0.001) and major stressful events (score 1–5 vs. score 0: OR=1.74 CI: 1.32, 2.29, p<0.001), and reduced with strong family support (score 9 vs. score 0–5: OR=0.65 CI: 0.47, 0.90, p=0.010). Model one also suggests the composite adversity index is positively associated with the risk of ADHD diagnosis (compared to 1^st quartile OR=1.0, 2^nd quartile OR=1.90 CI: 1.31, 2.77; 3^rd quartile OR=1.96 CI: 1.34, 2.88; 4^th quartile OR=2.89 CI: 2.01, 4.16). Model two examines all maternal psychosocial stress measures in one multiple logistic regression model, adjusted for covariates. Here, perceived stress (OR=3.03 CI: 2.19, 4.20, p<0.001) and major stressful events (OR=1.45 CI: 1.06, 1.99, p=0.019) are significantly associated with an ADHD diagnosis, adjusting for other potential contributors at birth.

The average probabilities of ADHD diagnoses for boys and girls at each level of the composite adversities index are shown in Figure 2. For both boys and girls, probabilities of ADHD diagnoses increase as the level of psychosocial adversity increases from 0 to 31, and boys consistently have a higher probability of diagnosis at each level of the summarized adversity index.

Discussion

This prospective birth cohort study is the first of its kind to examine the independent and combined risks of multiple dimensions of maternal psychosocial stress, support, and maternal and family factors during pregnancy on physician diagnosed ADHD, among low-income, urban minority children in [text removed for blind review]. This study is also distinguished by its demonstration of the increased probability of ADHD diagnosis with children’s increasing early life exposures to overall psychosocial adversity, independent of pre-, peri-, and postnatal factors previously associated with ADHD, as well as less commonly investigated factors such as maternal intrauterine infections.

Prior studies have associated maternal and teacher reports of ADHD symptoms with maternal stress during pregnancy [10] [24] [25], but this is the first to link differential aspects of psychosocial stress before and during pregnancy, assessed at birth, with a later clinical diagnosis of ADHD in a prospective fashion. Moreover, a child’s probability of ADHD diagnosis appears to increase consistently as the level of early life psychosocial adversity increases. This gradient reflects the relationship between children’s adversities and probability of psychiatric diagnosis first described by Rutter and colleagues[26], which was subsequently observed for the likelihood of ADHD diagnosis among white males [7]; and the severity of offspring ADHD symptoms [27].

Understanding the causal elements of the relationship involving adversity, stress, support and ADHD requires further exploration, but research suggests that stressful contexts and adversities, including poverty, discrimination, substandard housing, crowded households, and family turmoil can alter physiological responses to stress over time, across diverse populations [28] [29]. A review by Grizenko et al (2008) posits that a mother’s stress exposure in pregnancy impacts her child’s neurobehavioral development through activation of the maternal hypothalamic-pituitary-adrenal (HPA) axis and the subsequent programming of the fetal HPA axis and related physiology [27]. An alternative or complementary hypothesis is that a high level of stress during pregnancy is likely to persist into the early life of the child, contributing to a home environment that lacks regular routines and is often disrupted such that the infant’s neural systems do not develop the regulatory mechanisms needed for effective behavioral control. Future research should examine the mechanisms by which factors within children’s post-natal social environments increase their susceptibility to neurodevelopmental disabilities over and above that conferred by their prenatal intrauterine and social exposures, and whether this varies across various international contexts.

Lastly, in this prospective study we further confirmed the role of previously identified risk factors for ADHD, such as male gender, maternal smoking, prematurity, low birth weight, cesarean delivery, and elevated blood lead levels that are associated with known disruptions to early neurodevelopment [8]. In addition, we also identified intrauterine infection during pregnancy as a risk factor for ADHD, strengthening the case that ADHD has fetal origins. Those factors such as maternal education that were no longer significant when accounting for multiple early life adversities suggest that these risks may actually operate through their effects on the significant factors included in this study.

The issue of gender difference deserves further investigation. Our finding showed that at a given level of maternal stress, boys had consistently higher risk of ADHD compared to girls (Fig 3). While the significantly higher risk of ADHD among boys is well established, the causes for the gender difference are unknown. It is possible that boys are more susceptible to maternal stress in utero, or there are other unknown risk factors specifically affecting boys. These questions warrant future investigation.

Our study has several limitations. The observed effect size is fairly small in magnitude and may attenuate further if additional confounders were accounted for. Although we examined distinct characterizations of psychosocial stress based on maternal questionnaire interview, no biological measures of stress are included and we were unable to assess whether the timing of maternal psychosocial stress influenced children’s ADHD diagnosis. It is also possible that additional unmeasured mechanisms such as maternal genes influence the degree to which pregnant women experience and perceive stress. In addition, our analyses did not control for child hereditary factors and possible genetic effect modification, and we were unable to account for maternal diagnosis of ADHD due to the limited number of diagnoses documented in EMR. Parenting quality is also known to be associated with poor self-regulatory behaviors that are at the core of ADHD. If mothers in stressful circumstances during pregnancy bring their children into stressful home environments this may directly contribute to the development of offspring ADHD through direct biologic effects of stress and by interfering with positive parenting practices, which may have overestimated the magnitude of the prenatal stress exposures. Future studies should account for these factors, their possible interactions, and whether the timing of maternal psychosocial stress affects the occurrence of children’s ADHD diagnoses. Finally, caution is needed to generalize our findings to other populations with different characteristics and it is plausible that different relationships will be observed in substantially different populations.

Our findings, if further confirmed, have important clinical and public health implications, as they suggest the possibility for primary prevention of ADHD. Our research situates maternal psychosocial stress during pregnancy as a potentially important but modifiable early life contributor to the origins of ADHD. Given that the level of stress experienced by these predominantly low-income and minority women is likely to be higher than that experienced by women with more economic resources, it is particularly noteworthy that there is a gradient in the effects of stressful adversities on the likelihood of ADHD diagnosis; these findings suggest the possibility for a prospective, perinatal equivalent to findings from the study of adverse childhood experiences study, if further confirmed. Moreover, stress is a modifiable risk factor and women’s lives before and during pregnancy may present a window of opportunity during which stressors could be minimized and support provided to promote the health of their children generally and to potentially reduce ADHD specifically. Standard practices to treat ADHD symptoms with pharmaceutical drugs do not prevent its development, may not be available across all global contexts, and involve persistent and adverse side-effects[5]. Lastly, the burden and prevalence of ADHD may warrant incorporating valid and reliable measures of psychosocial stress into existing screening tools for maternal depression and anxiety, as recommended by the American College of Obstetricians and Gynecologists (ACOG) [30].

In summary, our findings suggest that children born to mothers who report more major life events before and during pregnancy, as well as high levels of pregnancy stress, have an elevated risk for ADHD diagnosis even after adjusting for other well established risk factors of ADHD. A directly proportional gradient between the level of psychosocial adversity experienced during pregnancy and the likelihood of children’s ADHD was also observed. Screening for maternal psychosocial stress and strengthening women’s psychosocial resources prior to conception and during pregnancy may represent windows of opportunity for reducing the development of ADHD in the offspring. While our study focused on maternal stress, the methodologies we outlined may be useful for assessing other prenatal risk factors for ADHD.

Figure 1.

The average probability of ADHD diagnosis for each level of Composite adversity index by gender based on MLR estimation

Note: MLR Model tested Composite adversity index, adjusting for maternal age at delivery, race/ethnicity, marital status, education, substance use, smoking, intrauterine infection, mode of delivery, gestational age, birth weight

Table 1.

Maternal, child and family characteristics for typically developing children and children with ADHD

Variable	Total, No(%)	Neurotypical, No(%)†	ADHD, No(%)	P-valueǂ
Total	2140 (100)	1780 (83.18)	360 (16.82)
Maternal characteristics
Maternal age at delivery, y				0.082
≤20	307 (14.35)	242 (13.60)	65 (18.06)
20–30	995 (46.50)	838 (47.08)	157 (43.61)
≥30	838 (39.16)	700 (39.33)	138 (38.33)
Race/ethnicity				0.776
Black	1354 (63.27)	1121 (62.98)	233 (64.72)
White	175 ( 8.18)	149 ( 8.37)	26 ( 7.22)
Hispanic	462 (21.59)	383 (21.52)	79 (21.94)
Others	149 ( 6.96)	127 ( 7.13)	22 ( 6.11)
Marital Status				0.003
Married	677 (31.64)	588 (33.03)	89 (24.72)
Not married	1433 (66.96)	1172 (65.84)	261 (72.50)
Missing	30 (1.40)	20 (1.13)	10 (2.78)
Parity				0.489
Nulliparous	909 (42.48)	762 (42.81)	147 (40.83)
Multiparous	1231 (57.52)	1018 (57.19)	213 (59.17)
Education				0.014
High school and lower	1805 (84.35)	1488 (83.60)	317 (88.06)
College degree and higher	310 (14.49)	273 (15.34)	37 (10.28)
Missing	25 (1.16)	19 (1.06)	6 (1.66)
Substance use				0.008
No	1687 (78.83)	1424 (80.00)	263 (73.06)
Yes	446 (20.84)	353 (19.83)	93 (25.83)
Missing	7 (0.33)	3 (0.17)	4 (1.11)
Smoking				<0.001
Never	1756 (82.06)	1487 (83.54)	269 (74.72)
Ever	384 (17.94)	293 (16.46)	91 (25.28)
Drinking				0.422
Never	1506 (70.37)	1259 (70.73)	247 (68.61)
Ever	634 (29.63)	521 (29.27)	113 (31.39)
Planned Pregnancy				0.125
No	1043 (48.74)	858 (48.20)	185 (51.39)
Yes	1068 (49.91)	905 (50.84)	163 (45.28)
Missing	29 (1.35)	17 (0.96)	12 (3.33)
Pregestational/gestational diabetes				0.758
No	1872 (87.48)	1557 (87.47)	315 (87.50)
Yes	221 (10.33)	182 (10.22)	39 (10.83)
Missing	47 (2.31)	41 (2.31)	6 (1.67)
Hypertensive disorder				0.758
No	1811 (84.63)	1513 (85.00)	298 (82.78)
Yes	297 (13.88)	246 (13.82)	51 (14.17)
Missing	32 (1.50)	21 (1.18)	11 (3.05)
Intrauterine infection during pregnancy				0.004
No	1822 (85.14)	1532 (86.07)	290 (80.56)
Yes	310 (14.49)	240 (13.48)	70 (19.44)
Missing	8 (0.37)	8 (0.45)	0 (0.00)
Prepregnancy BMI, mean (SD)	26.4(6.5)	26.3(6.5)	26.8(6.5)	0.210
Child characteristics
Gender				<0.001
Female	1110 (51.87)	1010 (56.74)	100 (27.78)
Male	1022 (47.76)	766 (43.03)	256 (71.11)
Missing	8 (0.37)	4 (0.23)	4 (1.11)
Mode of delivery				0.014
Cesarean	727 (33.97)	586 (32.92)	141 (39.17)
Vaginal	1394 (65.14)	1182 (66.40)	212 (58.89)
Missing	19 (0.89)	12 (0.67)	7 (1.94)
Gestational age				<0.001
≥37 weeks	1585 (74.07)	1347 (75.67)	238 (66.11)
34–37 weeks	335 (15.65)	273 (15.34)	62 (17.22)
<34 weeks	212 ( 9.91)	156 ( 8.76)	56 (15.56)
Missing	8 (0.37)	4 (0.23)	4 (1.11)
Lead level				0.005
<5 ug/dl	1446 (67.57)	1189 (66.80)	257 (71.39)
≥5 ug/dl	100 ( 4.67)	71 ( 3.99)	29 ( 8.06)
Missing	594 (27.76)	520 (29.21)	74 (20.55)
Birth weight (g), mean(SD)	2961.8(761.6)	2987.8(731.0)	2831.9(888.)	<0.001

Note:

^†Neurotypical is defined as without ID, DD, ASD, and ADHD diagnosis

^ǂChi-square test or t-test between Neurotypical and ADHD.