Lower Dietary Intake of Magnesium is Associated with More Callous-Unemotional Traits in Children

Abstract

Background:

Although researchers increasingly recognize the role of nutrition in mental health, little research has examined specific micronutrient intake in relation to antisocial behavior and callous-unemotional (CU) traits in children. Vitamin B₆ and magnesium are involved in neurochemical processes implicated in modulating antisocial behavior and CU traits. The current study examined dietary intakes of magnesium and vitamin B₆ in relation to antisocial behavior and CU traits.

Method:

We enrolled 11–12 year old children (n = 446, mean age = 11.9 years) participating in the Healthy Brains and Behavior Study. Magnesium and vitamin B₆ dietary intake were assessed with three 24-hour dietary recall interviews in children. CU traits and antisocial behavior were assessed by caregiver-reported questionnaires. We controlled for age, sex, race, total energy intake, body mass index, social adversity, ADHD or learning disability diagnosis, and internalizing behavior in all regression analyses.

Results:

Children with lower magnesium intake had higher levels of CU traits, controlling for covariates (β = −0.18, B = −0.0066, SE = 0.0027, p < 0.05). Vitamin B₆ intake was not significantly associated with CU traits (β = 0.061, B = 0.19, SE = 0.20, p > 0.05). Neither magnesium (β = 0.014, B = 0.0020, SE = 0.0093, p > 0.05) nor vitamin B₆ (β = 0.025, B = 0.33, SE = 0.70, p > 0.05) were significantly associated with antisocial behavior.

Conclusions:

Findings suggest that low dietary intake of magnesium may play a role in the etiology of CU traits but not general antisocial behavior. More studies are needed to determine if magnesium supplementation or diets higher in magnesium could improve CU traits in children.

Introduction

Inadequate micronutrient intake has been identified as a risk factor for antisocial behavior in children [1] and poor dietary quality in children has been found to predict antisocial behavior independent of genetic and environmental influences [2]. However, limited research has examined specific micronutrients in relation to child antisocial behavior, which includes aggression, rule-breaking, and other oppositional behaviors. This is despite the fact that childhood antisocial behavior is an important public health concern. Conduct problems that arise in childhood tend to span across later developmental stages and develop into more severe patterns of behavior than later-onset conduct problems [3]. This can result in lifelong burden on both the healthcare and criminal justice systems, making identification of childhood risk factors for antisocial behavior an important public health concern [4]. However, understanding child antisocial behavior is complicated by heterogeneity in behavioral sub-types of children with conduct problems [5]. One sub-group of children with antisocial behavior problems is children with high levels of callous-unemotional (CU) traits. CU traits include a lack of guilt, lack of empathy, callousness, and lack of emotion. Children with high levels of CU traits represent an important subgroup of antisocial youth [6] and children with CU traits often display more severe and chronic patterns of antisocial behavior than antisocial children without CU traits [7]. CU traits have a high predictive validity for antisocial outcomes, including developing disruptive behavior disorders in childhood [8]. Childhood CU traits predict juvenile delinquency [9], as well as future criminal involvement [10,11]. This makes identifying risk factors for CU traits especially important to prevent concurrent and long-term adverse developmental outcomes.

To our knowledge, no research has examined the relationship between child nutrition and CU traits. The purpose of the current study is to examine the relationship between child antisocial behavior, CU traits, and dietary intake of vitamin B₆ and magnesium. Vitamin B₆ and magnesium are important candidate micronutrients because they are involved in neurochemical processes that have been implicated in modulating antisocial behavior and CU traits.

Magnesium, Vitamin B₆ and Mental Health

Magnesium (Mg) and vitamin B₆ are both essential dietary micronutrients that play a role in healthy neurological function [12,13]. Magnesium and vitamin B₆ are closely related, with vitamin B₆ affecting the uptake of magnesium [14]. Common dietary sources of vitamin B₆ include fortified cereals, meat, fish, potatoes, and non-citrus fruits [15] and common sources of magnesium include green leafy vegetables, nuts, fruits, and whole grains [16]. Magnesium, an essential trace mineral, acts as a catalyst in over 300 enzymatic reactions in the body. Magnesium modulates activity in the hypothalamic pituitary adrenal (HPA) axis and blocks calcium channels of N-methyl-D-aspartate (NMDA) receptors [13]. NMDA receptors and HPA axis activity are thought to be involved in the etiology of CU traits. Amygdala NMDA receptors play an important role in activating neural changes involved in fear learning and extinction [17]. Deficits in fear conditioning and lower amygdala activity in response to others’ distress are thought to characterize antisocial children with high levels of CU traits [18]. Additionally, low cortisol—the end product of the HPA axis—is thought to most strongly characterize antisocial children with high levels of CU traits [19]. This suggests that magnesium may play a more important role in the development of CU traits than antisocial behavior generally.

Pyridoxal 5′-phosphate (PLP), the active form of vitamin B₆, is involved in metabolism of neurotransmitters, including serotonin and dopamine [20]. Deficits in serotonin and heightened dopamine activity are thought to contribute to impulsive and aggressive behavior [21,22]. It has been hypothesized that disruptions in serotonin and dopamine may be more strongly associated with impulsivity and aggression generally, rather than CU or psychopathic traits specifically [23]. Therefore, vitamin B₆ may contribute to the development of antisocial behavior generally, rather than CU traits specifically.

There has been limited research examining magnesium in relation to attention-deficit/hyperactivity disorder (ADHD), which often co-occurs with antisocial behavior problems [24]. A meta-analysis of twelve studies found that serum and hair magnesium concentrations were significantly lower in children with ADHD than controls [25]. Treatment studies have demonstrated beneficial effects of magnesium alone or combined magnesium and vitamin B₆ supplementation on hyperactivity, aggression, oppositional behavior, and attention in children with ADHD [26–29], though most of these studies did not involve treatment randomization or control groups [26, 27, 29], and all had small sample sizes (n = 18 to 75). In another trial that did not involve a control group, symptoms of hyperactivity/impulsivity decreased in children after combined polyunsaturated fatty acid, magnesium, and zinc supplementation [30]. Another small trial without a control group demonstrated significant improvement in parent-reported aggressive behavior after supplementation with a multivitamin that included vitamin B₆ [31]. These studies provide initial suggestive evidence that magnesium and vitamin B₆ may be associated with symptoms of ADHD that overlap with antisocial behavior, though their methodological limitations preclude firm conclusions.

Less research has examined relationships between magnesium, vitamin B₆, and antisocial behavior among children without ADHD. In a sample of adolescents in Australia, lower dietary intake of magnesium was associated with higher levels of antisocial behavior after taking into account potential confounders, including body mass index (BMI) and total energy intake [32]. However, results have been mixed. In a study of 51 Dutch adult psychiatric patients, blood magnesium and vitamin B₆ concentrations were not associated with aggression [33]. Results of this study, however, should be interpreted with caution, as blood draws were performed after participants ate breakfast, and the sample size was small (n = 51). This very limited body of research leaves the relationship between vitamin B₆, magnesium, and antisocial behavior in children uncertain. In addition, this research was not conducted among children living in the United States. This is an important limitation, as consumption of processed food is common in the US, with one study finding that 78% of children aged 9 and older in the US consumed inadequate quantities of fruit and vegetables [34]. As an additional limitation, existing research has not differentiated CU traits from antisocial behavior more generally.

Current Study

The goal of the current study is to address these limitations by examining the relationship between dietary intakes of magnesium and vitamin B₆ and antisocial behavior and CU traits in a sample of US children. We hypothesize that lower dietary intake magnesium will be associated with higher levels of CU traits and that lower dietary intake of vitamin B₆ will be associated with higher levels of antisocial behavior.

Methods

Participants

Data for this study come from the Healthy Brains and Behavior Study [35]. The sample consisted of 11 and 12-year old boys and girls living in Philadelphia County, PA or suburbs of Philadelphia. Within the study area, fliers soliciting enrollment were placed in recreation centers, libraries, health clinics, and other community centers. Targeted mailings were also sent to parents of 11 to 12 year old children living in the geographic catchment area. Youth with diagnosed psychosis, intellectual disability, or a pervasive developmental disorder were excluded. More information about subject recruitment and exclusionary criteria can be found in Liu et al. (2013). The original sample consisted of 454 participants. Of this original group, 8 participants were later deemed ineligible or withdrew, resulting in a final sample of 446. The sample was 49.4% female, 11.9% white, and 80.3% African American. The mean age of the sample was 11.9 years (SD = 0.59). All children were accompanied to the laboratory with a caregiver, who also completed questionnaires about the child’s behavior, demographics, and living circumstances. Written informed consent and assent were obtained from both parents and children, and the study protocols were approved by both the Institutional Review Board of the University of Pennsylvania and the Philadelphia Department of Health.

Dietary Intake

Three 24-hour dietary recall interviews (2 weekdays and 1 weekend day) were collected from each study participant by a registered dietician or dietetic technician experienced in conducting 24-hour dietary recalls in children. All recalls and data points were reviewed by a registered dietician. The 24-hour dietary recall interview provides a detailed summary of all foods and beverages consumed by participants during a complete 24-hour period (from midnight-to-midnight) for the day preceding the interview. Information is obtained on the time of each eating occasion, the type of meal (breakfast, lunch, supper, snack), and the location of the meal (home, school, other), as well as what and how much was consumed.

Recalls were completed by telephone. Telephone dietary recalls have been shown to have similar accuracy to in-person dietary recall interviews in children [36, 37]. The child completed the recall with the parent present. If the child had questions about food preparation or other questions when completing the recall, this information was provided by the caregiver. Parents commonly assisted with details about cooking methods and foods consumed (e.g., brand names, types of milks and cheeses, types of meats, etc.). Families were provided with booklets to help visualize portion sizes. We also used household measurements and visual aids if needed. Families were asked whether the recall day was a typical day. If it was not a typical day, the day was not used and we called again on another day. If a participant was not able to recall a meal, snack, or food amount, the day was not used and we called again another day. We also flagged recalls for potential issues with accuracy/honesty. Three recalls were flagged for possible issues with accuracy/honesty. We re-ran analyses without these recalls and results were substantively unchanged.

The information collected during the 24-hour recall interview was entered into the Nutrition Data System for Research (NDSR 2012, University of Minnesota). This software includes an up-to-date comprehensive food database. NDSR has built-in, standard prompts for guiding the interviewer in obtaining detailed information from the study participant about the foods they ate. Dieticians completed a comprehensive training and certification program with the Nutrition Coordinating Center (NDSR), which included dietary recall quality assurance training. NDSR uses a multipass interview process for the dietary recall, to ensure completeness and improve recall. Each diet record was reviewed in NDSR (prior to generating an output file) to check the accuracy of the food item selected and quantity consumed. Output files from the NDSR include average daily intake estimates for nutrients, including vitamin B₆ (mg), magnesium (mg), and total energy intake (kcal). An extra check on the final data exported from NDSR was also completed to check for outliers.

Behavioral Measures

Child Antisocial Behavior and Internalizing Behavior

Parent-reported antisocial behavior was assessed using the externalizing behavior scale of the Child Behavior Checklist (CBCL) [38]. The CBCL externalizing scales consist of 35 items that measure rule-breaking and aggression. Items are rated by the parent on a 3-point Likert scale (0–2), with higher scores indicating higher levels of externalizing behavior. The Cronbach’s alpha of the CBCL externalizing scale in the current sample was 0.98. Given the comorbidity of externalizing and internalizing behavior, we controlled for CBCL internalizing behavior scores. The CBCL internalizing scale consists of 32 items that measure anxious/depressed symptoms, withdrawn/depressed symptoms, and somatic complaints. The Cronbach’s alpha of the CBCL internalizing scale in the current sample was 0.98.

CU Traits

The Antisocial Personality Screening Device (APSD) was used to assess parent-reported CU traits [39]. Parents rated each item on a 3-point scale (0–2) with higher scores reflecting higher levels of CU traits. The callous-unemotional sub-scale was used in the current analyses (6 items, Cronbach’s alpha = 0.65).

Body Mass Index

Body mass index (BMI) was calculated as kilograms/meters² and was derived from measurements in the laboratory of the child’s height and weight on the day of the study visit.

Covariates

We measured the following covariates: race (0 = white, 1 = non-white), age, sex (0 = male, 1 = female), parent-reported child learning disability or ADHD diagnosis (0 = no learning disability, 1 = ADHD or learning disability present; including reading or writing problems, presence of an individualized education program for special needs or learning disability, speech or hearing problem, learning delay, or dyslexia), total energy intake in kcal (based on dietary recall, described above), CBCL internalizing behavior (described above) and BMI. We also controlled for social adversity using a previously used social adversity index [40]. Social adversity items were coded as either present (1) or absent (0) for each participant, and each participant’s score reflected the sum of the social adversity indicators. Items included: mother’s low education, mother’s teenage pregnancy, problems with living accommodations, not living with both biological parents, early maternal separation between 6 months and 2.5 years of age, living in government housing, mother’s physical illness, mother’s mental illness, and mother ever arrested. Scores ranged from 0–6, with higher scores indicating higher social adversity. More information about the scale construction and descriptive statistics for individual items can be found in Portnoy et al. (2020).

Statistical Analyses

We first calculated descriptive statistics for study variables and performed Pearson’s correlations to examine bivariate relationships between continuous study variables. Independent samples t-tests (or chi-square tests for categorical variables) were used to test for sex differences in study variables. Next, we conducted OLS regression analyses with CU traits and externalizing behavior as the dependent variables and intakes of magnesium and vitamin B₆ as the independent variables. We controlled for race, age, sex, total energy intake, BMI, ADHD or learning disability diagnosis, social adversity, and internalizing behavior in regression analyses. In addition, in the regression model with CU traits as the dependent variable, we controlled for externalizing behavior. In the model with externalizing behavior as the dependent variable, we controlled for CU traits. We did this to isolate the unique effects of magnesium and B₆ on CU traits versus externalizing behavior.

Missing data on study variables ranged from 0% (child sex and age) to 2.5% (CBCL internalizing and externalizing behavior). Analyses were performed using the 415 children with complete data. As an additional sensitivity analysis, we repeated the OLS regression analyses using multiple imputation to handle missing data. Twenty datasets were created with missing values replaced with imputed values using fully conditional specification. Data were analyzed across the imputed datasets to generate pooled parameter estimates and standard errors.

As an additional supplemental analysis, we examined whether magnesium or vitamin B₆ interacted with sex to predict CU traits and externalizing behavior. We added the interactions of sex with magnesium and vitamin B₆ to the regression models.

As a final supplemental analysis, given the high correlation between magnesium and vitamin B₆, we conducted separate regression analyses predicting CU traits and externalizing behavior with only magnesium or vitamin B₆ included as an independent variable (in addition to the covariates). Similarly, given the high correlations between CU traits and externalizing behavior, in the models predicting CU traits, we did not control for externalizing behavior. In the models predicting externalizing behavior, we did not control for CU traits. We controlled for all other covariates in these analyses.

Methods

Sample Characteristics

The mean daily magnesium intake in this sample was 204 mg (SD = 63; Table 1). This slightly exceeded the estimated average requirement (EAR) of 200 mg/day for males and females ages 9–13 years [16]. Dietary intake of magnesium below this recommended level occurred in 53.2% of the sample. The mean daily vitamin B₆ intake in this sample was 1.70 mg (SD = 0.71). This exceeded the EAR for vitamin B₆ of 0.8 mg/day for males and females ages 9–13 years [15]. Only 2.1% of the sample consumed less than the EAR of Vitamin B₆.

Table 1.

Descriptive Statistics (n = 415)

	Full Sample		Female (n = 207)	Male (n = 208)	t /χ2
	Mean/% (SD)	Min-max	Mean (SD)	Mean (SD)
CU Traits	3.82 (2.27)	0–11	3.79 (2.32)	3.85 (2.22)	t = 0.26
Externalizing	9.76 (9.11)	0–45	8.84 (9.50)	10.7 (8.6)	t = 2.05
Magnesium (mg)	204 (63)	75.2–484	204 (61)	204 (66)	t =0.056
Vitamin B6 (mg)	1.70 (.71)	0.44–6.39	1.67 (0.72)	1.74 (0.70)	t =0.96
Age (years)	11.9 (0.59)	10.8–13.1	11.9 (0.57)	11.9 (0.62)	t =0.12
BMI	21.9 (5.8)	13.4–48.8	22.7 (6.2)	21.0 (5.1)	t = −2.95**
ADHD or Learning Disability (% Present)	10.4%		4.3%	16.3%	χ2 = 16.1***
Internalizing	7.51 (7.08)	0–46	7.52 (7.24)	10.7 (8.6)	t = 4.09***
Social Adversity	2.14 (1.40)	0–6	2.14 (1.33)	2.15 (1.47)	t =0.038
Total energy intake (kcal)	1880 (543)	793–4375	1859 (535)	1902 (552)	t =0.80
Female	49.9%
Non-white	87.5%		88.9%	86.1%	χ2 = 0.76

Notes. CU = callous-unemotional. BMI = body mass index. CU traits were measured using the Antisocial Process Screening Devise. Externalizing and internalizing behavior were measured using the Child Behavior Checklist.

^*p < 0.05;

^**p < 0.01,

^***p < 0.001

Bivariate Associations Among Magnesium, Vitamin B₆, CU Traits, and Antisocial Behavior

CU traits were positively associated with externalizing behavior (r = 0.41, p < 0.001), internalizing behavior (r = 0.19, p < 0.001), and social adversity (r = 0.19, p < 0.001; Table 2). CU traits were inversely associated with magnesium intake (r = −0.11, p < 0.05), but were not associated with vitamin B₆ (r = 0.03, p > 0 .05). Externalizing behavior was not associated with magnesium (r = 0.01, p > 0.05) or vitamin B₆ (r = 0.08, p > 0.05). Magnesium intake was positively associated with vitamin B₆ (r = 0.67, p < 0.001) and total energy intake (r = 0.74, p < 0.001) and negatively associated with BMI (r = −0.13, p < 0.05). Children with higher levels of social adversity had lower magnesium intake (r = −0.11, p < 0.05). Vitamin B₆ intake was not significantly associated with any behavioral outcomes or social adversity (p > 0.05).

Table 2.

Bivariate Correlations (n = 415)

	1	2	3	4	5	6	7	8	9
1. CU Traits	--
2. Externalizing	0.41***	--
3. Magnesium	−0.11*	0.01	--
4. Vitamin B6	0.03	0.08	0.67***	--
5. Age	−0.03	−0.01	−0.02	0.02	--
6. BMI	0.10†	0.05	−0.13*	−0.11*	0.10*	--
7. Internalizing	0.19***	0.55***	−0.06	−0.02	0.03	0.12*	--
8. Adversity	0.19***	0.24***	−0.11*	0.02	−0.02	0.05	0.23***	--
9. Total energy	0.003	0.08	0.74***	0.64***	0.04	−0.20***	−0.03	0.01	--

Notes. CU = callous-unemotional. BMI = body mass index. CU traits were measured using the Antisocial Process Screening Device. Externalizing and internalizing behavior were measured using the Child Behavior Checklist.

^*p < 0.05;

^**p < 0.01,

^***p < 0.001;

^†p < 0.10

Dietary Intake of Magnesium and Vitamin B₆ and CU Traits and Antisocial Behavior

In regression analyses, children with lower dietary intake magnesium had higher levels of CU traits, controlling for covariates (β = −0.18, B = −0.0066, SE = 0.0027, p < 0 .05; Table 2). Dietary intake of vitamin B₆ was not significantly associated with CU traits (β = 0.061, B = 0.19, SE = 0.20, p > 0.05). Neither magnesium (β = 0.014, B = 0.0020, SE = 0.0093, p > 0.05) nor vitamin B₆ (β = 0.025, B = 0.33, SE = 0.70, p > 0.05) were significantly associated with externalizing behavior.

Sensitivity Analyses

To confirm robustness of results, we repeated the regression analyses using multiple imputation to handle missing data. Results were substantively unchanged. Magnesium continued to be associated with CU traits, net of covariates (B = −0.0069, SE = 0.0026, p = 0.0087), while vitamin B₆ was not associated with CU traits (B = 0.19, SE = 0.20, p = 0.33). As in the original models, neither magnesium (B = 0.0029, SE = 0.0092, p = 0.75) nor vitamin B₆ (B = 0.29, SE = 0.69, p = 0.67) were associated with externalizing behavior.

Interactions between magnesium and sex (β = 0.063, B = 0.0052, SE = 0.015, p = 0.73) and vitamin B₆ and sex (β = −0.086, B = −0.81, SE = 1.34, p = 0.55) were non-significant in association with externalizing behavior. Similarly, they were not associated with CU traits (magnesium x sex (β = −0.18, B = −0.0037, SE = 0.0044, p = 0.39), and vitamin B₆ x sex (β = 0.17, B = 0.39, SE = 0.39, p = 0.31).

All results were substantively unchanged when we performed regression models separately for vitamin B₆ and magnesium without including CU traits or externalizing behavior as a covariate in the models predicting the other outcome. Magnesium continued to be associated with CU traits (β = −0.16, B = −0.0059, SE = 0.0026, p = 0.024) but not externalizing behavior (β = −0.023, B = −0.0033, SE = 0.0090, p = 0.72). Vitamin B₆ was not associated with CU traits (β = 0.0089, B = 0.028, SE = 0.20, p = 0.86) or externalizing behavior (β = 0.033, B = 0.42, SE = 0.68, p = 0.53).

Discussion

The goal of this study was to examine relationships between dietary intake of magnesium and vitamin B₆ and CU traits and antisocial behavior. We predicted that lower dietary intake of magnesium would be associated with higher levels of CU traits, while lower dietary intake of vitamin B₆ would be associated with higher levels of antisocial behavior. Support for our hypotheses was mixed. Consistent with our initial hypotheses, lower dietary intake of magnesium was associated with higher levels of CU traits, but not antisocial behavior. However, dietary intake of vitamin B₆ was not associated with antisocial behavior or CU traits. To our knowledge, this is the first study to examine the relationship between CU traits and magnesium and vitamin B₆ intake. Findings are suggestive that inadequate magnesium intake may be involved in the etiology of CU traits in children and could have implications for treatment.

Magnesium and CU Traits

Many Americans have insufficient dietary intake of magnesium due to Western diets low in magnesium and high consumption of processed foods [41]. While moderate-to-severe magnesium deficiency and chronic latent deficiency have been associated with seizures, cardiac arrhythmia, osteoporosis, and hypertension [14], inadequate magnesium intake is also thought to be implicated in mental health disorders including depression [42,43] and anxiety [44]. The current study extended these findings by demonstrating an association between magnesium intake and CU traits. Although we cannot determine why dietary intake of magnesium was associated with CU traits, but not antisocial behavior, there are several possibilities. CU traits and antisocial behavior are distinct constructs that may have unique etiologies and CU traits have a stronger biological basis than antisocial behavior without CU traits [45,46]. Although diet is an environmental factor, its hypothesized mechanism of action on behavior is via neurological mechanisms. Therefore, magnesium consumption may play a more important role in the development of CU traits than antisocial behavior without CU traits, where etiology may be more environmentally driven [18]. In addition, magnesium is involved in modulating activity of the HPA axis and NMDA receptors [13], both of which are thought to be implicated in deficits in fear conditioning and emotional responses observed in children with CU traits [19,47].

Vitamin B₆ and CU Traits

In our study, we did not find a relationship between vitamin B₆ and antisocial behavior or CU traits. It is unclear why vitamin B₆ intake was not associated with antisocial behavior or CU traits. Dietary vitamin B₆ intake has been associated with other mental health outcomes, including depression in adults [48]. In addition, preliminary evidence demonstrates beneficial effects of combined magnesium and vitamin B₆ supplementation on hyperactivity and aggression in children with ADHD [26,27]. One possible explanation for the null effects in the current study was the relatively small number of children who had insufficient vitamin B₆ intake. Only 2.1% of the children in this sample consumed less than the EAR of vitamin B_6, and the average dietary intake exceeded the EAR. It is possible that an insufficient number of children had low dietary intake of vitamin B₆ to detect adverse behavioral effects. In addition, we did not measure biomarkers of vitamin B₆. The absorption of both magnesium and vitamin B₆ is dependent on many factors, including hormones, other micronutrients, and overall meal composition [41,49]. Future research is needed that examines serum pyridoxyl-5’-phosphate (PLP) concentration in relation to CU traits and antisocial behavior.

Potential Implications

These findings have potential implications for the treatment of CU traits in children. Traditional family-based interventions for conduct problems are less effective for children with CU traits than children with conduct problems without CU traits [50]. Therefore, novel treatment strategies for children with CU traits are needed. Findings from this study are suggestive that improvement in dietary intake (or possibly supplementation) to increase magnesium consumption could help to improve CU traits in children. A prior study demonstrated the potential beneficial effects of nutritional supplementation on CU traits. In a double-blind randomized, placebo-controlled trial, supplementation with omega-3 fatty acids resulted in reductions in CU traits six months post-treatment [51]. Given the challenge in treating conduct problems in children with CU traits, nutritional supplementation may be a promising alternative that does not involve the commonly recognized side effects of psychotropic medications. Randomized controlled trials are needed to determine the potential effectiveness of magnesium supplementation in children with CU traits. However, it should be noted that 53.2% of the children in the current sample had dietary intake of magnesium below the EAR and that children with higher levels of social adversity had lower dietary intake of magnesium. Increasing consumption of common sources of magnesium, including green leafy vegetables, nuts, fruits, and whole grains, which provide several important nutrients beyond magnesium, and improving the diet as a whole may be the best approach, if possible. However, many of these foods may not be easily available in low-income areas, or foods of choice among some families. From a health standpoint, this suggests that magnesium supplementation should be investigated further as potentially beneficial for children living in difficult social conditions, even if they do not have high levels of CU traits.

Contributions, Limitations, and Future Directions

There were several limitations to the current study. Analyses were cross-sectional, which precludes the determination of whether dietary intake preceded the development of behavioral outcomes. Relatedly, it was not possible to determine causation in the current study. It is unknown whether low magnesium intake increased CU traits or whether children with CU traits were more likely to refuse or choose not to eat foods with high magnesium content. An additional limitation was the lack of biomarker data on magnesium and vitamin B₆. Finally, while this study examined the relationship between magnesium and vitamin B₆ and antisocial behavior for the first time in a US sample, the sample was largely African American, and findings may not generalizable to other groups. Nonetheless, we should note that African Americans tend to have lower dietary intake of magnesium than non-African Americans, making this an especially relevant sample [52, 53].

These limitations should be viewed in light of several important strengths of the current study. To our knowledge, this was the first study to examine the relationship between CU traits and dietary intake of magnesium and vitamin B₆. This study also utilized a larger sample than most prior studies of magnesium and B₆ and related outcomes. Dietary data were collected using three 24-hour dietary recalls, providing a more stable measure of intake than a single day. Finally, we controlled for numerous covariates, including total energy intake, BMI, internalizing behavior, ADHD or learning disability diagnosis, and social adversity.

Conclusions

Researchers increasingly recognize the role of nutrition in mental health, though the role of nutrition in antisocial behavior remains understudied. Findings from the current study provide an important first step in examining the relationship between key micronutrients and antisocial behavior. Future research involving collection of biomarkers is needed to build on the current findings. This will provide a foundation for future randomized controlled trials aimed at reducing behavior problems in children through non-invasive nutritional interventions.

Table 3.

OLS Regression Models Predicting CU Traits and Externalizing Behavior (n = 415)

IV	Dep: CU Traits			Dep: Externalizing
	B (SE)	β	p	B (SE)	β	p
Magnesium	−0.0066 (0.0027)	−0.18	0.014	0.0020 (0.0093)	0.014	0.83
Vitamin B6	0.19 (0.20)	0.061	0.34	0.33 (0.70)	0.025	0.64
Total energy	0.0004 (0.0003)	0.10	0.17	0.0010 (0.0010)	0.059	0.33
BMI	0.025 (0.018)	0.063	0.18	−0.015 (0.063)	−0.0095	0.81
Sex	0.082 (0.21)	0.018	0.69	−1.52 (0.72)	−0.084	0.034
Race	1.01 (0.32)	0.15	0.0019	−0.051 (1.13)	−0.0019	0.96
Age	−0.16 (0.17)	−0.042	0.34	−0.25 (0.59)	−0.017	0.67
Social adversity	0.081 (0.077)	0.050	0.29	0.51 (0.27)	0.078	0.058
ADHD/learning disability	0.0024 (0.34)	0.0003	0.99	1.31 (1.17)	0.044	0.26
Internalizing	−0.021 (0.017)	−0.066	0.22	0.61 (0.05)	0.47	< 0.0001
Externalizing	0.10 (0.014)	0.40	< 0.001
CU				1.20 (0.16)	0.30	< 0.0001
R 2	0.22			.42

Notes. CU = callous-unemotional. BMI = body mass index. CU traits were measured using the Antisocial Process Screening Device. Externalizing and internalizing behavior were measured using the Child Behavior Checklist. Vitamin B₆ and magnesium were measured in mg. Total energy intake was measured in kcal. Race was coded as 0 for white and 1 for non-white. Sex was coded as 0 for male and 1 for female. Learning disability or ADHD diagnosis was coded as 0 for no diagnosis and 1 for positive diagnosis. IV = independent variable. Dep. = dependent variable.