Mental Health in Early Childhood and Changes in Cardiometabolic Dysregulation by Pre-Adolescence

Abstract

Objective:

Poor mental health in childhood is associated with a greater risk of cardiometabolic disease in adulthood, but less is known about when these associations begin to emerge. This study tests whether poor mental health (indexed by emotional and behavioral problems) in early childhood predicts increases in cardiometabolic dysregulation over 4 years of follow-up.

Methods:

Data are from 4,327 participants in the Generation R Study. Problem behaviors were reported by mothers using the Child Behavior Checklist at age 6 years. Repeated measurements of 6 cardiometabolic parameters were collected at ages 6 and 10 years: high-density lipoprotein cholesterol (HDL-C), non-HDL-C, systolic and diastolic blood pressure, C-reactive protein, and body mass index. Standardized measures were used to create continuous cardiometabolic dysregulation scores at ages 6 and 10 years. Change in dysregulation was defined as the difference in dysregulation scores over time. Cross-sectional and prospective associations were tested using linear regression, sequentially adjusting for relevant confounders. Additional analyses examined whether prospective relationships were robust to adjustment for baseline levels of dysregulation.

Results:

There was no association between child problem behaviors and cardiometabolic dysregulation at age 6 years. However, higher levels of problem behaviors predicted increases in cardiometabolic dysregulation (ß=0.12, 95% Confidence Interval (CI)=0.00, 0.23) from age 6 to 10 years.

Conclusions:

Worse child mental health may be associated with increases in cardiometabolic dysregulation by pre-adolescence. To our knowledge, this is the first study to demonstrate that adverse physiologic effects of psychological distress identified in adult populations may be observed as early as childhood.

INTRODUCTION

A growing body of research indicates that children’s social environments and emotional experiences can influence their risk of acquiring chronic disease in adulthood.^1,2 While numerous studies find early adversity is associated with a greater likelihood of developing heart disease,^1,2 considerably less work has examined the longitudinal impact of children’s psychological well-being on cardiometabolic outcomes. Evidence suggests that psychological distress (e.g., anger, anxiety, and depression) may contribute to poorer cardiometabolic health among adults,^3-6 but research on early distress and cardiometabolic functioning prior to adulthood is limited.⁷ While some studies have examined associations of psychological distress with individual cardiometabolic-related biomarkers among youth (e.g., obesity,⁸ inflammation,⁹ arterial thickness^10,11), no studies to our knowledge have prospectively examined psychological distress in relation to a composite measure of cardiometabolic dysregulation in either childhood or adolescence. Determining how early in life children’s mental health begins to influence their cardiometabolic health will allow researchers to identify new developmental windows for chronic disease prevention.^12,13 Therefore, the goal of this study is to test whether poor mental health at age 6 years predicts increases in cardiometabolic dysregulation from ages 6 to 10 years.

To date, only two studies have prospectively examined the impact of children’s psychological well-being on composite measures of cardiometabolic functioning in adulthood.^14,15 In one study, emotional distress assessed by psychologists at age 7 years predicted a 17%-31% greater risk of developing cardiovascular disease at midlife.¹⁴ A study of 6,714 British children similarly found that those who exhibited psychological distress at ages 7, 11, or 16 years had higher levels of a composite measure of cardiometabolic risk at age 45 years.¹⁵ While investigators accounted for childhood health conditions and body mass index in both studies, they were unable to assess when biological alterations in relevant parameters might begin to manifest because measures of cardiometabolic function in childhood were unavailable.

In this study, we use repeated measures of cardiometabolic functioning from the ongoing Generation R birth cohort in the Netherlands to investigate whether associations between child mental health and subsequent levels of cardiometabolic dysregulation are evident in childhood. Cardiometabolic risk factor clustering in the first decades of life may lead to higher risk of cardiometabolic diseases in adulthood.^16-18 Following the American Academy of Pediatrics’ recommendation to assess risk factor clustering by capturing multiple domains of physiologic function,¹⁹ we examined direct measures of cholesterol, blood pressure, inflammation, and adiposity collected when children were aged 6 and 10 years. Child mental health was indexed by problem behaviors reported by mothers when children were 6 years old. Analyses controlled for sociodemographic factors as well as relevant maternal correlates of child health (e.g., adiposity prior to pregnancy, smoking) and child health factors (e.g., low birthweight, asthma diagnosis) identified in previous literature.¹⁵ Informed by research in adults,¹⁵ we hypothesized that poor mental health would predict increases in cardiometabolic dysregulation over time, even at this relatively young age.

METHODS

Sample

Participants are from the Generation R Study, an on-going population-based, prenatally recruited birth cohort from Rotterdam, The Netherlands.^20,21 Between April 2002 and January 2006, 9,778 pregnant women were recruited into the study (61% participation rate), and their children have been followed from fetal life through childhood to assess early determinants of health and development.²² Data on participating children were collected through blood draws and physical examinations completed during clinic visits at ages 6 and 10 years, as well as through questionnaires administered periodically to mothers. Generation R was approved by the Medical Ethical Committee of the Erasmus Medical Center, Rotterdam, and written informed parental consent was obtained for all participants.

At age 6 years, parents of 8,305 children provided consent for their child’s continued participation in the study. Among those who consented, 6,690 attended the clinical visit. The total eligible sample for the present study was comprised of 6,532 children who did not have an acute infection at the 6- or 10-year clinic (indicated by a C-reactive protein [CRP] level >10 mg/L). A flowchart depicting the sample composition over the study period is provided in Figure 1. Eligible participants were excluded from our study if they were missing mental health data at baseline or the majority of our six cardiometabolic measures at either age 6 or 10 years, yielding a final analytic sample of 4,327 children.

Figure 1:

Flow chart of Generation R participants in the final analytic sample.

Compared to those who were excluded due to missing data, study participants were slightly younger at baseline, more likely to be Dutch and socioeconomically advantaged, and less likely to have a mother with a history of depression (see Table S1, Supplemental Digital Content). See the Online Appendix for more information on the distribution of missing data in our sample.

Measures

Cardiometabolic Functioning.

Two different composite cardiometabolic measures were used to examine both broad changes in risk factor clustering¹⁹ and incremental changes in physiologic function. We created ordinal scores indicating clustered dysregulation on multiple cardiometabolic parameters at ages 6 and 10 years using direct assessments of non-fasting high-density lipoprotein cholesterol [HDL] (mmol/L), non-fasting non-high-density lipoprotein cholesterol [nHDL] (total cholesterol – HDL; mmol/L), systolic blood pressure [SBP] (mm Hg), diastolic blood pressure [DBP] (mm Hg), CRP (mg/L), and body mass index [BMI] (kg/m²). We did not examine low-density lipoprotein cholesterol or triglyceride levels since measurements can be inaccurate when obtained from non-fasting blood samples.²³ Following research on physiologic dysregulation,¹¹ continuous values of individual parameters were first dichotomized to indicate whether a child’s measurement was in the unhealthiest quintile of the sample distribution (i.e., ≥80^th percentile for nHDL, SBP, DBP, CRP, and BMI; ≤20^th percentile for HDL), then summed to create a count score ranging from 0 to 6, with higher values indicating more dysregulated parameters. Children who were dysregulated on ≤1 parameter were classified as being in optimal (versus sub-optimal) health. We then classified children into 4 categories based on the change in clustered cardiometabolic dysregulation observed from ages 6 to 10 years: (1) maintained optimal health; (2) health improved; (3) health declined; and (4) maintained sub-optimal health.

To capture more information on the precise degree of physiologic change children experienced over time, we also constructed continuous cardiometabolic dysregulation scores.²⁴ Following prior pediatric research,^19,24 we standardized values of the six parameters described above and created cardiometabolic dysregulation sum scores at ages 6 and 10, respectively: −1*z(HDL) + z(nHDL) + z(SBP) + z(DBP) + z(CRP) + z(BMI). Change in dysregulation was defined as the difference in scores between ages 6 and 10 (mean=−0.06, SE=0.05, range=−10.1, 15.4), with positive values indicating increases in dysregulation over time. Data for each parameter was collected on-site following standard study protocols (see Online Appendix).

Child Mental Health.

Children’s problem behaviors were assessed through maternal report at age 6 years using the 99-item validated Dutch version of the Child Behavior Checklist for early childhood (CBCL/1.5-5).^25-28 On a 3-point Likert scale from 0 (“not true”) to 2 (“very true or often true”), mothers indicated whether their child displayed a range of behaviors in the prior 6 months. Items were summed to generate a problem behavior score with higher values indicating poorer mental health (α=0.94). Standardized scores were used as the primary exposure variable in all analyses, however a binary measure was also created to identify children in the borderline clinical range of problem behaviors (≥84^th percentile), following criteria defined in the CBCL user manual.²⁹ In secondary analyses, we also separately examined changes in cardiometabolic risk scores in relation to standardized values of each of the 7 subscales of the CBCL (i.e., aggressive behavior, anxious/depressed, attention problems, emotionally reactive, sleep problems, somatic complaints, and withdrawn behavior). Scores were weighted to account for missingness if ≤25% of individual items were incomplete, otherwise the overall measure was classified as missing.

Covariates.

Information on covariates was collected via maternal questionnaires administered during pregnancy and when the participating child was 3 years old. Sociodemographic covariates included the child’s age at baseline (months), sex (male/female), ethnicity defined by parents’ country of birth (Dutch, other European, Caribbean, Middle Eastern, African, other), maternal education (lower vocational/higher vocational and university), and monthly family income reported at age 3 (<€2,000, €2,000-€3,999, ≥€4,000). Additional covariates considered for inclusion were maternal age at the child’s birth (years), maternal self-reported BMI prior to pregnancy (overweight or obese/healthy), maternal and paternal self-reported cardiometabolic conditions (diabetes, high blood pressure, or high cholesterol; yes/no), and maternal self-reported smoking status during pregnancy (yes/no). Newborn and child-level covariates identified from prior literature as predictive of future cardiometabolic health included being born with a low birthweight (<2,500 g), being born preterm (<37 week gestation), and having a history of asthma.^15,30 Retrospective assessments of children’s history of physical or sexual abuse (yes/no) obtained via a semi-structured interview with mothers when children were 10 years old were also considered.

Maternal mental health was examined as a predictor of study non-response. Past week depressive symptoms were assessed at mid-pregnancy using the 6-item depression subscale of the Brief Symptom Inventory (BSI).³¹ Total scores ranging from 0 to 4.0 were created following the BSI manual.³²

Statistical Analyses

Missing Data.

To minimize concerns about selection bias, we used multiple imputation¹⁵ and inverse probability weighting¹⁴ to account for missing data. In the total eligible sample (n=6,532), missing child problem behavior data at age 6 years, cardiometabolic parameters at ages 6 and 10 years, and covariate data were multiply imputed in 85 datasets using chained equations.¹⁶ Imputed data were then used to generate weights reflecting participants’ inverse probability of meeting the study inclusion criteria. All regression analyses were weighted, and effect estimates were obtained only among participants in the final analytic sample (n=4,327).

Primary Analyses.

Bivariate associations between borderline clinical levels of child problem behaviors and study covariates were evaluated using χ² tests. Differences in mean standardized levels of problem behaviors were then examined across our 4-category measure of change in clustered cardiometabolic dysregulation using ANOVA. Post-hoc pairwise comparisons of group means were assessed with a Tukey-Kramer adjustment.³³

Associations with continuous levels of cardiometabolic dysregulation were examined using sequentially adjusted multivariable linear regression models. Cross-sectional analyses were carried out at baseline to determine if problem behaviors were associated with children’s health at 6 years of age, followed by prospective analyses with cardiometabolic dysregulation change scores to examine longitudinal associations. Covariate selection and model building procedures were theory driven, following prior research.¹⁵ Model 1 was unadjusted, while Model 2 controlled for sociodemographic factors and maternal characteristics (sex, ethnicity, monthly family income, maternal age at birth, maternal overweight or obesity, and maternal smoking status during pregnancy). Since family income was strongly correlated with maternal education, income was used as the primary socioeconomic status control variable in our models due to its greater variability in the sample. Model 3 additionally controlled for newborn and child-level correlates of cardiometabolic health (birthweight, gestational age at birth, and history of asthma).

We also conducted a series of sensitivity analyses to assess the robustness of our findings. First, to address potential concerns about composite measures of cardiometabolic risk being highly influenced by adiposity,³⁴ we examined associations with risk scores that excluded BMI. Next, we evaluated the role of additional covariates including parental cardiometabolic history and participants’ history of physical or sexual abuse to determine their effect on primary associations of interest. Lastly, we examined whether longitudinal associations with cardiometabolic dysregulation change scores were robust to further adjustment for cardiometabolic dysregulation at age 6 years to determine if children’s health at baseline influenced our prospective findings.

Secondary Analyses.

Differences in the strength of associations by sex and ethnicity were tested via interaction terms and stratification. We also examined relationships between each CBCL subscale and cardiometabolic risk scores in separate models using the same procedures described above to identify whether specific behavioral domains were contributing to our findings. Finally, we tested associations between problem behaviors and change scores in standardized levels of each individual cardiometabolic parameter in six different fully adjusted models to assess whether overall associations were driven by specific cardiometabolic factors.

Since some children (6.8%) had siblings in the study, all regression models used a clustered sandwich estimator to calculate standard errors that accounted for potential non-independence among study participants.³⁵ All analyses were conducted using Stata MP v.15.1, and statistical significance was defined as a 2-tailed p-value ≤0.05.

RESULTS

Sample Description.

On average, participants were 6.1 years old (SD=0.4) at baseline and 9.8 years old (SD=0.3) at follow-up. Approximately half of participants were female, 63.8% were Dutch, and 59.7% had a mother with a higher vocational or university level education (Table 1). Overall, 6.7% of children experienced borderline clinical levels of problem behaviors. Compared to those with fewer problem behaviors, children with borderline clinical levels were more likely to be male, non-Dutch, socioeconomically disadvantaged, and to have a mother who smoked during pregnancy or was overweight/obese prior to having a child (Table 1).

Table 1:

Descriptive statistics of Generation R sample, by child problem behaviors.^a

	Total Sample n (%)	Child Problem Behaviorsb		p-valuec
		Low	Borderline Clinical
		n (%)	n (%)
Total Sample	4,327 (100.0)	4,035 (93.3)	292 (6.8)	--
Child Sociodemographic Factors
Sex
Female	2,166 (50.1)	2,053 (94.8)	113 (5.2)	<0.001
Male	2,161 (49.9)	1,982 (91.7)	179 (8.3)
Ethnicity
Dutch	2,761 (63.9)	2,629 (95.2)	132 (4.8)	<0.001
Other European	343 (7.9)	326 (95.0)	17 (5.0)
Caribbean	386 (8.9)	342 (88.6)	44 (11.4)
Middle Eastern	411 (9.5)	355 (86.4)	56 (13.6)
African	181 (4.2)	160 (88.4)	21 (11.6)
Other	242 (5.6)	220 (90.9)	22 (9.1)
Maternal Education
Lower Vocational	1,742 (40.3)	1,568 (90.0)	174 (10.0)	<0.001
Higher Vocational and University	2,579 (59.7)	2,461 (95.4)	118 (4.6)
Monthly Family Income at 6y
Less than €2,000	1,499 (35.7)	762 (87.5)	109 (12.5)	<0.001
€2,000-€3,999	1,828 (43.5)	1,710 (93.5)	118 (6.5)
More than €4,000	871 (20.8)	1,446 (96.5)	53 (3.5)
Maternal Characteristics
Maternal Age at Child’s Birth
35 Years or Older	1,058 (24.5)	995 (94.1)	63 (5.9)	0.24
Younger than 35 Years	3,269 (75.6)	3,040 (93.0)	229 (7.0)
Pre-Pregnancy BMI
Overweight or Obese (i.e., ≥25)	817 (24.9)	747 (91.4)	70 (8.6)	0.003
Healthy (i.e., <25)	2,466 (75.1)	2,326 (94.3)	140 (5.7)
Smoked During Pregnancy
Yes	526 (13.6)	472 (89.7)	54 (10.3)	<0.001
No	3,334 (86.4)	3,137 (94.1)	197 (5.9)
Child Health Factors
Birthweight
Less than 2,500 g	237 (5.5)	221 (93.3)	16 (6.7)	0.99
2,500 g or More	4,084 (94.5)	3,809 (93.3)	275 (6.7)
Gestational Age at Birth
< 37 Weeks	257 (6.0)	238 (92.6)	19 (7.4)	0.67
≥ 37 Weeks	4,043 (94.0)	3,772 (93.3)	271 (6.7)
Asthma
Yes	250 (7.0)	230 (92.0)	20 (8.0)	0.21
No	3,304 (93.0)	3,105 (94.0)	199 (6.0)

^an’s may vary due to missing data.

^bProportions of children with low and borderline clinical problem behaviors presented as row percentages.

^cCalculated using χ² test

Child Mental Health and Cardiometabolic Dysregulation.

Roughly one third of the sample (32.4%) maintained optimal cardiometabolic health from early to middle childhood. In contrast, 32.1% experienced health improvements over time, 28.4% experienced declines, and only 7.1% experienced persistent sub-optimal health. Figure 2 illustrates differences in mean standardized problem behavior scores by child health profiles (F=3.6, p=0.014). Pairwise comparisons found two differences that nearly reached significance. Children whose health declined had a 0.14-standard deviation higher problem behavior score compared to both those who maintained optimal health (t=2.5; p=0.061) and those whose health improved (t=2.5; p=0.057). Although children in persistently sub-optimal health had the highest levels of problem behaviors (mean=0.09, SD=1.0), no significant differences were noted in comparisons with other health profiles (all p>0.05).

Figure 2:

Mean standardized problem behavior scores by categories of change in cardiometabolic health from age 6 to 10 years (N=2,120).^a

^a Mean values estimated among participants with complete cardiometabolic risk and child problem behavior data.

Cross-sectional analyses at baseline found no evidence of an association between problem behaviors and cardiometabolic dysregulation in unadjusted (ß=0.03, 95% CI=−0.08, 0.14) or fully adjusted models (ß=−0.07, 95% CI=−0.18, 0.03). Results from prospective analyses are provided in Table 2. After controlling for all study covariates, each standard deviation higher problem behavior score predicted a 0.12-unit increase in cardiometabolic dysregulation over time (95% CI=0.00, 0.23). Analyses excluding BMI from risk scores yielded nearly identical results (ß_{fully adjusted}=0.11, 95% CI=−0.01, 0.22). Findings were also unchanged after further adjustment for parental cardiometabolic history and participants’ history of physical or sexual abuse (Table S2, Supplemental Digital Content). Fully adjusted associations were slightly attenuated but largely robust to further adjustment for baseline levels of dysregulation (ß=0.09, 95% CI=−0.01, 0.20).

Table 2:

Adjusted associations between standardized problem behaviors at age 6 years and within-individual change in cardiometabolic dysregulation from age 6 to 10 years (N=4,327). ^a,b

	Change in Cardiometabolic Dysregulation c
	Model 1		Model 2		Model 3
	ß (95% CI)	p-value	ß (95% CI)	p-value	ß (95% CI)	p-value
Problem Behaviors, Per 1-SD	0.18 (0.06, 0.30)	0.003	0.11 (0.00, 0.21)	0.043	0.12 (0.00, 0.23)	0.048
Sociodemographic Factors
Age, Months			−0.04 (−0.06, −0.03)	<0.001	−0.05 (−0.06, −0.03)	<0.001
Female			0.09 (−0.10, 0.27)	0.36	0.01 (−0.19, 0.21)	0.90
Ethnicity
Dutch			Reference	--	Reference	--
Other European			0.13 (−0.23, 0.49)	0.47	0.13 (−0.24, 0.51)	0.48
Caribbean			0.50 (0.13, 0.86)	0.008	0.42 (0.04, 0.80)	0.032
Middle Eastern			0.63 (0.27, 1.05)	0.001	0.66 (0.26, 1.06)	0.001
African			0.63 (0.10, 1.16)	0.020	0.66 (0.11, 1.21)	0.019
Other			0.13 (−0.29, 0.54)	0.55	0.11 (−0.32, 0.54)	0.62
Monthly Family Income at 6y
More than €4,000			Reference	--	Reference	--
€2,000-€3,999			0.31 (0.09, 0.52)	0.005	0.32 (0.10, 0.54)	0.004
Less than €2,000			0.37 (0.05, 0.69)	0.023	0.41 (0.08, 0.73)	0.014
Maternal Characteristics
Age at Child’s Birth, Years			0.01 (−0.01, 0.03)	0.28	0.02 (−0.01, 0.04)	0.14
Overweight or Obese Pre-Pregnancy			0.57 (0.32, 0.83)	<0.001	0.56 (0.30, 0.82)	<0.001
Smoked During Pregnancy			0.39 (0.08, 0.70)	0.012	0.36 (0.04, 0.68)	0.027
Child Characteristics
Low Birthweight					0.77 (0.19, 1.34)	0.009
Preterm Birth					−0.26 (−0.79, 0.28)	0.35
Asthma					−0.37 (−0.85 0.12)	0.14

^aEffect estimates were calculated using linear regression and p-values were calculated using t-tests.

^bModel 1 Adjusted R² = 0.0040; Model 2 adjusted R² = 0.032; Model 3 adjusted R² = 0.038

^cChange in dysregulation was defined as the difference in the sum of six Z-scored cardiometabolic parameters measured at ages 6 years and 10 years.

Differences in Associations by Sex and Ethnicity.

After adding interaction terms to fully adjusted models, there was no evidence of significant differences by either sex (p_interaction=0.68) or ethnicity (all p_interaction>0.10). Fully adjusted stratified models indicated similar findings for boys and girls, but somewhat different results were noted by ethnicity. Specifically, the most substantial associations were observed among African (ß=0.39, 95% CI=0.01, 0.78) and Middle Eastern children (ß=0.30, 95% CI=−0.05, 0.66), while associations among children of other ethnic groups were somewhat weaker (e.g. Dutch: ß=0.06, 95% CI=−0.08, 0.20).

Analyses by behavioral subscales found significant associations between nearly all behaviors and increases in cardiometabolic dysregulation (Table 3). Estimates were largely attenuated after adjusting for study covariates, but associations across domains generally mirrored our overall results (e.g., ß_{anxious/depressed}=0.10, 95% CI=−0.01, 0.22; ß_{sleep problems}=0.10, 95% CI=0.00, 0.21). When considering fully adjusted associations by individual cardiometabolic parameter, we found that problem behaviors were associated with small increases in dysregulation across all parameters, but confidence intervals were overlapping with zero (Table 4). The most substantial associations were noted with increases in CRP and SBP.

Table 3:

Adjusted associations between standardized problem behavior subscale scores at age 6 years and within-individual change in cardiometabolic dysregulation from age 6 to 10 years (N=4,327). ^a,b

	Change in Cardiometabolic Dysregulation
	Model 1		Model 2		Model 3
	ß (95% CI)	p-value	ß (95% CI)	p-value	ß (95% CI)	p-value
Aggressive Behavior	0.13 (0.03, 0.24)	0.014	0.08 (−0.02, 0.19)	0.12	0.09 (−0.02, 0.19)	0.099
Anxious/Depressed	0.14 (0.02, 0.25)	0.020	0.10 (−0.02, 0.21)	0.11	0.10 (−0.01, 0.22)	0.08
Attention Problems	0.11 (0.00, 0.21)	0.041	0.06 (−0.04, 0.17)	0.23	0.07 (−0.04, 0.17)	0.22
Emotionally Reactive	0.10 (−0.01, 0.22)	0.077	0.07 (−0.04, 0.19)	0.21	0.08 (−0.03, 0.19)	0.16
Sleep Problems	0.18 (0.07, 0.28)	0.001	0.10 (−0.01, 0.20)	0.062	0.10 (0.00, 0.21)	0.049
Somatic Complaints	0.13 (0.03, 0.23)	0.009	0.06 (−0.04, 0.16)	0.27	0.07 (−0.03, 0.17)	0.20
Withdrawn Behavior	0.09 (−0.02, 0.20)	0.12	0.04 (−0.07, 0.15)	0.44	0.04 (−0.07, 0.15)	0.43

^aEffect estimates were calculated using linear regression and p-values were calculated using t-tests.

^bModel was unadjusted. Model 2 additionally controlled for child’s sex, ethnicity, maternal age at child’s birth, family income at age 6, maternal overweight or obesity pre-pregnancy, and maternal smoking status during pregnancy. Model 3 further controlled for child characteristics at birth (low birthweight, preterm birth) and asthma.

^cChange in dysregulation was defined as the difference in the sum of six Z-scored cardiometabolic parameters measured at ages 6 years and 10 years.

Table 4:

Fully adjusted associations between standardized child problem behaviors at age 6 years and within-individual changes in cardiometabolic parameters from age 6 to 10 years (N=4,327).^a,b

	Change in Standardized Cardiometabolic Parametersc
	∆ nHDL	∆ HDL	∆ SBP	∆ DBP	∆ CRP	∆ BMI
	ß (95% CI)	ß (95% CI)	ß (95% CI)	ß (95% CI)	ß (95% CI)	ß (95% CI)
Problem Behaviors, Per 1-SD	0.01 (−0.02, 0.04)	−0.01 (−0.05, 0.03)	0.03 (−0.01, 0.06)	0.01 (−0.02, 0.05)	0.04 (−0.01, 0.09)	0.01 (−0.01, 0.03)

^aEffect estimates were calculated using linear regression and p-values were calculated using t-tests.

^bModel controlled for child’s sex, ethnicity, maternal age at child’s birth, family income at age 6, asthma, child characteristics at birth (low birthweight, preterm birth), maternal overweight or obesity pre-pregnancy, and maternal smoking status during pregnancy.

^cChange in dysregulation was defined as the difference in standardized (per 1-SD) cardiometabolic parameters measured at ages 6 years and 10 years.

DISCUSSION

This study suggests that associations between child mental health and elevated cardiometabolic risk previously documented among adults may be observed as early as the first decade of life. In a large, population-based cohort of children, we found suggestive evidence that worse mental health was associated with deterioration in cardiometabolic functioning over time. In preliminary analyses, we found that children who experienced increases in clustered dysregulation tended to have worse mental health at age 6. To quantify more precisely the degree of physiologic change children experienced, we also examined associations with continuous cardiometabolic dysregulation scores. Results indicated that higher levels of problem behaviors were not associated with cardiometabolic dysregulation at baseline but were related to increases in dysregulation by age 10 years. Although the alterations in children’s cardiometabolic functioning we identified were incremental, they provide evidence that the early childhood period may be a developmental window during which emotion-related health deteriorative processes are beginning to manifest. In light of prior work linking poor mental health in childhood and adult cardiometabolic risk,^14,36,37 it is possible that these increases in dysregulation may accrue over decades to predispose children to chronic disease in adulthood.

In line with prior findings among adults,¹⁴ we found no evidence of sex differences in observed associations. There were also no differences noted by ethnicity, however stratified analyses indicated that relationships may be stronger among some ethnic minority children, namely those of Middle Eastern and African descent. Research among adults in the United States has found similar evidence of stronger associations between depression and elevated cardiovascular risk among racial/ethnic minority individuals (e.g., African Americans) compared to White individuals.³⁸ Some authors posit that this trend may relate to minorities experiencing higher levels of psychological distress,³⁸ which may be the case in our sample as Middle Eastern and African youth also had the highest prevalence of borderline clinical behavior problems.

Research on the life course impact of psychological distress provides some insight into underlying mechanisms that might explain our findings. Prior studies suggest that poor mental health can exert negative impacts on cardiometabolic outcomes through both behavioral pathways (e.g., the adoption of risk behaviors) and physiological pathways (e.g., by triggering direct biological alterations to stress response systems).^39,40 Since our study was conducted in early childhood – before health behaviors like smoking and unhealthy diets are fully established – a behavioral pathway may be less probable. However, evidence suggests that physical activity in childhood is associated with lower levels of cardiometabolic risk and therefore may be a contributing factor.⁴¹

Altered biological responses to stress may also play a role. Prior work has demonstrated that psychological distress can lead to over-activation of the amygdala and subsequently, the hypothalamic-pituitary-adrenal axis and sympathetic nervous system.⁴² Over time, persistent over-activation can affect cardiovascular, immune, and metabolic function, and ultimately cause biological damage and disease.⁴³ In this study, we observed incremental physiological alterations in multiple systems over time with increased dysregulation noted across measures of cholesterol, blood pressure, inflammation, and adiposity. Although prior evidence has linked elevated BMI with dysregulated levels of other cardiometabolic biomarkers,^39-40 we did not find that our results were disproportionately influenced by BMI. In contrast, sensitivity analyses found the clearest evidence for associations with changes in CRP and SBP, while associations with changes in BMI were less substantial. With respect to specific problem behaviors, associations were largely similar across domains, indicating that different dimensions of psychological distress appear to confer similar risk. Future research should determine whether these alterations might track from childhood through adulthood to gain a stronger understanding of the mechanisms underlying disease development over the life course.

This study has some limitations. Since our findings are based on observational data, causality cannot be conclusively ascertained. However, we accounted for many potential confounders and health correlates. Furthermore, since biological measurements were directly assessed and collected starting in early childhood, we were also able to examine change in dysregulation over time, which mitigated the threat of confounding by baseline health. Lastly, our study was ethnically diverse but disproportionately comprised of socioeconomically advantaged children, all of whom resided in a large, urban center in western Europe, and therefore may not be generalizable to other populations in different geographic regions.

This study also has noteworthy strengths. Psychological distress was assessed using a well-validated measure of child problem behaviors. Additionally, our longitudinal study design minimized potential concerns about reverse causal associations through both the study of change in cardiometabolic dysregulation over time and adjustment for baseline outcome levels. Lastly, this is the first population-based study to our knowledge to examine cardiometabolic alterations related to psychological distress in early childhood. Since clinical levels of both psychiatric diagnoses and cardiometabolic conditions are uncommon in the early childhood period,^44,45 children are unlikely to be using medications that may influence their levels of cardiometabolic risk, thereby addressing an important source of confounding in the adult literature. Furthermore, assessing associations in childhood also provides novel insight into how early the accumulation of cardiometabolic risk may begin.

Our findings may have important implications for pediatric care. In a past policy statement, the American Academy of Pediatrics argued for incorporating mental health competencies more comprehensively into current practice.⁴⁶ Recommended strategies for improving care include screening children for early signs of emotional and behavioral difficulties, connecting families with resources to address their children’s behavioral needs, and greater coordination of care with mental health specialists.⁴⁶ More recent evidence suggests that integrating mental health services into pediatric practice through collaborative care models may be an effective systems-based strategy to reduce the burden of emotional and behavioral problems in childhood.⁴⁷ While these recommendations were initially developed to address undiagnosed and sub-clinical mental health disorders among youth, our findings suggest that these efforts may also provide the additional benefit of ameliorating cardiometabolic dysregulation in childhood, and possibly even the development of chronic disease in adulthood. Researchers should consider examining the longitudinal impact of integrated pediatric care on cardiometabolic health as children transition to adulthood to determine its effectiveness as a potential strategy for primordial prevention.⁴⁸

CONCLUSION

Although substantial research has investigated emotion-related factors in relation to chronic disease risk among adults,⁴⁰ this is the first study to examine whether emotional factors are linked to cardiometabolic changes occurring from early to middle childhood, when lifelong health trajectories may be established.⁴⁹ Using prospective data from a large, population-based birth cohort, we found that poor mental health in early childhood was associated with increases in cardiometabolic dysregulation over a 4-year follow-up period. In light of prior work indicating that the biological effects of early psychological distress develop over the life course,¹⁵ our results indicate that risk accumulation may begin earlier than previously appreciated. Whether integrated models of pediatric care can address children’s emotional and behavioral needs, and also mitigate the potential for subsequent adverse cardiometabolic effects is a critical issue that deserves more attention.

Abbreviations:

BMI

Body Mass Index

CRP

C-reactive Protein

CBCL

Child behavior checklist

DBP

Diastolic Blood Pressure

HDL

High-Density Lipoprotein Cholesterol

nHDL

non-HDL Cholesterol

SBP

Systolic Blood Pressure