Longitudinal Examination of Hair Cortisol Concentrations and Weight Changes in Preschool-Aged Children of Latino Farmworkers

Abstract

Background:

Hair cortisol concentrations may serve as a measure of biologically embedded stress. While the cross-sectional association between hair cortisol and obesity in children has been examined, the data examining this relationship over time are limited.

Methods:

We examined hair cortisol and anthropometrics in 40 children with obesity from Latino families enrolled in a Head Start program serving farmworkers. All participants were enrolled in a clinical trial using parent mentors to encourage healthy lifestyles. We analyzed the proximal 3 cm of hair at the beginning and the end of the trial, a period of about 8 months. Linear mixed models were used to examine if changes in hair cortisol were associated with changes in adiposity.

Results:

Children had a median BMI percentile of 98% and parents with lower education with 25 of 40 having less than high school diploma and high food insecurity (17 of 40, 43%). Among the 40 children with valid data for both time points, the median hair cortisol concentration at baseline was 4.09 pg/mg (interquartile range [IQR] 2.65–8.68) and 6.05 pg/mg (IQR 3.95–9.33) at the end point. Increases in cortisol from baseline to follow-up had a small but significant association with decreased obesity over time. Hair cortisol did not moderate an association between food security and weight.

Conclusion:

In children with chronic stressors and obesity, we found that increases in cortisol over time were associated with decreases in adiposity. Further studies following hair cortisol concentrations over time are needed to understand how this biomarker relates to weight status and stressors. Clinicaltrials.gov ID: NCT03330743.

Introduction

Psychosocial stressors such as child abuse, poverty, and food insecurity are well established risk factors for childhood obesity. This association can be partially explained by interacting biological and behavioral pathways that include the dysregulation of the stress-response mechanism through the hypothalamic-pituitary-adrenal (HPA) axis, increased chronic peripheral inflammation, and structural changes to neurological reward processes that in turn promote risky behavior, including binge eating behaviors.^1,2 Developmentally, children have a rapid maturation of their HPA axis over the 1st year of life that has substantial influence and interaction with parenting. Generally, more responsive and sensitive parenting helps modulate the cortisol response to stressful situations in the preschool aged child, and higher basal cortisol is associated with less responsive caregiving practices.³

Many cross-sectional studies have found an association between elevated hair cortisol concentrations and obesity in children. In a sample of prepubertal girls with obesity matched by age to a sample of healthy weight girls, investigators observed a correlation between hair cortisol and BMI z-scores.⁴ A meta-analysis examining 26 studies in children found a small positive correlation between hair cortisol concentrations and BMI, BMI z-scores, and waist circumference.⁵

Some studies have examined the mechanisms underpinning the relationship between exposure to psychosocial stressors and obesity. A study on preadolescent, low-income Latino children found that hair cortisol acted as an effect modifier in the association between BMI and food insecurity—only those with high hair cortisol had a significant association between BMI and food insecurity.⁶ This suggests that children's weight may only be affected when exposure to a stressor such as food insecurity translates to an experience of chronic biologic stress, as shown by changes in hair cortisol. Adversity and hair glucocorticoids have been examined in limited studies with experiences of abuse associated with higher cortisone in hair and with BMI, although they did not identify that cortisol or cortisone mediates the association between maltreatment and BMI,⁷ suggesting that mechanisms may vary across different types of stressors (abuse vs. food insecurity). The adversity faced by children of Latino farmworkers, some of whom are undocumented, may also include larger threats related to immigration and deportation.⁸ Acculturation and the challenges faced with being a cultural minority are additional potential stressors.

Longitudinal data are limited, and many studies use hair cortisol at one time point. Hair cortisol measured at 6 years of age was associated with higher systolic blood pressure at 10 years of age although the association between hair cortisol and blood pressure was not significant after adjusting for BMI.⁹ Some longitudinal studies suggest that weight status may predict lower cortisol reactivity rather than the reverse. For example, Doom et al found that overweight/obesity at preschool age (mean 4.3 years) was associated with lower morning cortisol and lower cortisol stress reactivity in later childhood (mean 8 years).¹⁰ Hair cortisol measured in mid-childhood (median age of 7.7) followed for 5 years and measured again in adolescence was associated with insulin resistance but no other measures of adiposity, including BMI, waist circumference, laboratory values (Interleukin-6, C-reactive protein [IL-6, CRP]), and blood pressure. Notably, this was in a relatively high-income, all White, and mostly metabolically healthy sample.¹¹

Outside of the study of obesity and chronic stress, studies examining hair cortisol over time show varying results. In children with trauma exposure followed over time, investigators showed that increases in hair cortisol concentration were associated with improvements in behaviors.¹² However, in adults receiving interventions targeting mindfulness or meditation, studies have found decreases in hair cortisol associated with treatment for stress reduction.^13,14

Despite the many cross-sectional studies and a few longitudinal studies, the effect of chronic stress on weight status in people with obesity remains unclear. One recent study of adults who had successfully reduced their weight as part of a clinical trial found no association between changes in hair cortisol and changes in body weight measures.¹⁵ However, they did find some association between an increase in hair cortisol concentrations over the first 12 months and weight variability in the subsequent period (12–18 months), suggesting that increases in physiologic stress may be associated with continued challenges in weight.¹⁵

To our knowledge, no study has examined whether, in children with obesity, changes in hair cortisol are associated with changes in weight. We sought to address that question by examining a cohort of children who participated in a clinical trial for obesity who had hair cortisol measured over time. The overall hypothesis was that decreased biologic stress (cortisol) would be associated with a decrease in weight measures (e.g., BMI). Our analysis plan was designed to address four questions with these data. First, how does this sample with obesity and relatively low income compare to other samples of children using hair cortisol concentration as an indicator of chronic cortisol exposure. We also sought to test the hypothesis that elevated hair cortisol in children is associated with reduced weight loss over time. Therefore, the second question was: does baseline cortisol predict weight change over time? Third, we examined if changes in hair cortisol concentration between baseline and follow-up were associated with changes in weight by BMI variables and waist circumference, as well as systolic and diastolic blood pressure, using mixed model analyses. Fourth, we examined whether hair cortisol acts as a moderator of the association between food security and weight.⁶

Methods

Sample Selection

Participants for this study were recruited from a clinical trial aimed at reducing the body weight of children in obesity. Participants were children primarily from low-income, Latino families with parents working as migrant or seasonal farmworkers in the state of Oregon. All participants included in this study were in a state of obesity (BMI ≥95th percentile) which was calculated using CDC guidelines with objective measures (stadiometer and scale) of weight and height collected as part of the clinical trial. Parents provided consent for the participation of their children in this study. The protocol of this study was approved by the Institutional Review Board at Oregon Health and Science University.

Clinical Trial

Parents participating in this clinical trial were randomized to one of three treatment groups, one control and two active trial groups.¹⁶ The two active trial arms utilized a parent mentor model to deliver a mix of group and individual skill-based instruction focused on healthy behaviors in this age group. Parent mentors met with a group of parents on a monthly basis and then followed up with each individual parent on a weekly basis either in person or through text or phone call to provide support and feedback. All measurements described at baseline were collected at enrollment, including hair samples. Measurements described as collected at the end of the study were gathered after the intervention, on average 8 months, with a range from 6 to 12 months.

Hair Cortisol

In the clinical trial, 188 children participated overall, and 97 children provided no sample at all. We collected 71 hair samples at baseline and 63 at the end of the trial period, a period of 8 months on average between enrollment and the end of the trial. We had 43 case pairs (baseline and end of study) with valid cortisol samples. Only data from these complete pairs were included in this analysis. Hair samples were processed and analyzed as described in Meyer et al¹⁷ with minor modifications. Briefly, samples were weighed and then washed twice with 1.0 mL of isopropanol to remove external contaminants. After overnight drying, hair was ground to a fine powder using a bead mill, and the cortisol was extracted for 18–24 hours into 1.5 mL methanol with slow rotation. Tubes were centrifuged, after which 1.0 mL of the supernatant was transferred to a clean microcentrifuge tube and evaporated using a vacuum evaporator. The dried extract was redissolved in 0.25 mL of assay buffer and then spin-filtered to remove residual particulate matter. Finally, the extracted cortisol was analyzed in duplicate along with standards and quality controls using the Arbor Assays DetectX ELISA Kit. The cortisol content of each sample was corrected for sample weight to yield final results as pg cortisol per mg hair. The intra- and interassay coefficients of variation for this assay are <10%.

Health Outcomes

Children had their height and weight measured as part of the larger clinical trial, at baseline and the end of the study 8 months later.¹⁶ Waist circumference was measured in triplicate using the umbilicus and anterior superior iliac spine as markers. Blood pressure was measured using an appropriately sized automatic sphygmomanometer, also in duplicate. We used height and weight to calculate both the BMI z-score and percentile and the percent distance from the median BMI. We used this last measure as the primary outcome given recent data showing it to be a more robust indicator for children with obesity.¹⁸

Covariates

We measured parental educational attainment and reported family income using categorical measures as proxies of socioeconomic status that may influence coping, self-efficacy, and stress tolerance. Physical activity and sleep have been shown to be associated with the biologic stress response, and we measured these using a structured review of typical week and weekend days using previously tested questionnaires.^19,20 The Brief Acculturation Scale for Hispanics was administered to parents; this scale has a range from 4 to 20 with higher scores indicating greater acculturation.²¹ Food insecurity was measured using the USDA screening tool for families.²² This 6-item short form shows 98% correct classification compared with the 18-item USDA food insecurity measure.²³

Analysis Plan

We examined the distribution of cortisol concentration values. One value exceeded 4000 pg/mg and was considered biologically implausible. In addition, outliers were defined as those values >3 standard deviation from the mean and were excluded from further analyses.

To address our objectives, first, we used descriptive statistics to examine this sample in contrast to others' published work. For the other three objectives, we used a linear mixed model approach. For the second objective, we used hair cortisol concentration at baseline as a predictor (nontransformed) in the first set of models with age, sex assigned at birth, and a time variable included. Randomization group was included in the models as well. The dependent or outcome variables were measures of percent distance from median BMI, BMI, blood pressure, and waist circumference. For the third objective, we included the end of trial hair cortisol concentration variable as an independent fixed variable. The subject was included as a random effect in all models. Finally, we stratified the sample by cortisol level (above and below the median) to examine whether cortisol acts as a modifier of the relationship between food security and weight. SPSS 26.0 was used for all analyses.

Results

Of the 43 children with valid cortisol samples at both time points, we excluded three outliers from the analysis (Fig. 1). Overall, the average age of the 40 children included in the analysis was 4.2 years. Thirty-six (90%) of their parents attained at most a high school diploma (Table 1). All of the participant families included in this analysis identified as Hispanic or Latino. Food insecurity was reported by 17 (43%) of the participant families. The median hair cortisol concentration among the included children was 4.09 pg/mg (interquartile range [IQR] 2.65–8.68) at baseline and 6.05 pg/mg (IQR 3.95–9.33) at the end of the study. Table 1 shows descriptive characteristics of participants by their baseline cortisol concentration above or below the median. No major differences were found in demographics or physiologic parameters, including percent distance from the median BMI (p = 0.48), baseline BMI percentile (p = 0.59), or systolic (p = 0.88) or diastolic (p = 0.85) blood pressure. The median parental perceived stress score at baseline was 8 (IQR 5–10) on a 1–10 point scale and remained unchanged at the end of the study at 7 (IQR 5–8). Acculturation overall was low at 6.9 on a scale from 4 to 20, with no difference by group. Parent hair cortisol at baseline was correlated with child hair cortisol, Spearman's coefficient 0.53, p = 0.006 (n = 26 parents provided hair samples with their children). We also examined whether the 43 who provided samples at both time points differed from participants who provided a sample only at baseline or not at all and found no significant differences by age, gender, parental education, family income, or child BMI (data not shown).

Figure 1.

Flow chart of subjects participating in the overall clinical trial who provided hair samples and were included in the analysis of hair cortisol over time.

Table 1.

Demographics of the Parent-Child Dyad Participants Examined by Baseline Hair Cortisol Levels, Median Split

	All ( n  = 40)	Below the median (1.51–4.01)	Above the median (4.18–94.87)	p
Age, years, mean (SD)	4.2 (0.7)	4.2 (0.4)	4.3 (0.8)	0.75
Sex, female, n (%)	21 (53)	9 (45)	12 (60)	0.34
Acculturation scale, mean (SD)	6.9 (4.5)	6.0 (2.9)	7.8 (5.6)	0.21
Language at home, n (%)				0.07
Spanish or mostly Spanish	37 (93)	20 (100)	17 (85)
English or mostly English	3 (8)	0	3 (15)
Family income, n (%)				0.59
<$20,000	5 (13)	3 (15)	2 (10)
$20,000–50,000	25 (63)	12 (60)	13 (65)
$50,000 or more	4 (10)	1 (5)	3 (15)
Don't know	6 (15)	4 (20)	2 (10)
Parental education, n (%)				0.57
Less than high school	25 (63)	13 (65)	12 (60)
High school	11 (28)	6 (30)	5 (25)
Any college	4 (10)	1 (5)	3 (15)
Food insecure, n (%)	17 (43)	9 (45)	8 (40)	0.75
Percent distance from the median, adjusted, median (IQR)	23.2 (18.7–30.4)	22.5 (17.0–36.7)	23.2 (19.1–28.8)	0.75
BMI percentile, median (IQR)	98.3 (96.7–99.6)	98.3 (96.6–99.7)	98.2 (97.1–99.5)	0.75
Systolic BP, mean (SD)	101 (8)	100 (7)	102 (8)	0.56
Diastolic BP, mean (SD)	59 (6)	59 (6)	58 (6)	0.44
Age parent, mean (SD)	34 (7)	34 (7)	33 (7)	0.71
BMI parent, mean (SD)	32 (7)	33 (8)	31 (7)	0.38
Parental perceived stress at baseline, median (IQR)	8 (5–10)	8 (5–10)	8 (5–10)	0.65
Sleep, total in hours, mean (SD)	10.9 (1.2)	11.2 (1.0)	10.6 (1.4)	0.16
Physical activity, baseline, median (IQR)	5.5 (2.5–7.5)	6.5 (3.8–7.4)	4.0 (2.0–8.3)	0.63
Cortisol at end of intervention, median (IQR)	6.05 (3.95–9.33)	5.93 (3.89–9.20)	6.50 (3.95–9.33)	0.75

BP, blood pressure; IQR, interquartile range; SD, standard deviation.

In the linear mixed models, we did not find a significant association between baseline hair cortisol and change in percent distance from the median, BMI, or waist circumference (Table 2). Higher baseline hair cortisol was significantly associated with higher systolic blood pressure (Table 2).

Table 2.

Linear Mixed-Model Analysis Examining Association Between Baseline Cortisol Concentrations and Clinical Outcomes (Model 1) and End Cortisol Concentrations (Model 2)

Predictors	Model 1		Model 2
	Estimates (95% CI)	p	Estimates (95% CI)	p
Percent distance from median BMI
Intercept	3.6 (−22.1 to 29.3)	0.78	3.7 (−20.0 to 27.5)	0.83
Baseline cortisol concentration	−0.01 (−0.18 to 0.17)	0.94	−0.02 (−0.18 to 0.14)	0.79
Time	−2.1 (−4.6 to 0.5)	0.12	−2.1 (−4.6 to 0.5)	0.12
Sex	−1.6 (−9.2 to 6.1)	0.68	0.6 (−6.6 to 7.8)	0.92
Age	6.0 (0.2 to 11.8)	0.04	6.1 (0.7 to 11.4)	0.03
End cortisol concentration			−0.08 (−0.13 to −0.02)	<0.01
BMI
Intercept	16.4 (12.0 to 20.8)	<0.001	16.4 (12.4 to 20.5)	<0.001
Baseline cortisol concentration	−0.00 (−0.03 to 0.03)	0.91	−0.00 (−0.32 to 0.02)	0.77
Time	−0.4 (−0.9 to 0.1)	0.10	−0.4 (−0.9 to 0.1)	0.10
Sex	−0.5 (−1.8 to 0.8)	0.46	−0.1 (−1.4 to 1.1)	0.84
Age	1.0 (−0.0 to 1.9)	0.06	1.0 (0.0 to 1.9)	0.04
End cortisol concentration			−0.01 (−0.02 to −0.00)	0.01
Waist circumference
Intercept	41.1 (30.7 to 51.5)	<0.001	41.1 (30.5 to 51.7)	<0.001
Baseline cortisol concentration	−0.01 (−0.08 to 0.06)	0.84	−0.01 (−0.08 to 0.06)	0.83
Time	−2.2 (−3.4 to −1.0)	<0.001	−2.2 (−3.4 to −1.0)	<0.001
Sex	−1.1 (−4.2 to 2.0)	0.48	−1.0 (−4.2 to 2.2)	0.53
Age	5.4 (3.0 to 7.7)	<0.001	5.4 (3.0 to 7.7)	<0.001
End cortisol concentration			−0.0 (−0.03 to 0.02)	0.84
Systolic BP
Intercept	93.1 (80.3 to 105.9)	<0.001	93.1 (80.2 to 106.0)	<0.001
Baseline cortisol concentration	0.09 (0.00 to 0.17)	0.05	0.08 (0.00 to 0.17)	0.06
Time	−1.5 (−4.1 to 1.1)	0.25	−1.5 (−4.1 to 1.1)	0.25
Sex	−1.7 (−5.5 to 2.1)	0.38	−1.4 (−5.3 to 2.6)	0.48
Age	2.2 (−0.7 to 5.1)	0.13	2.2 (−0.7 to 5.1)	0.13
End cortisol concentration			−0.01 (−0.04 to 0.02)	0.50
Diastolic BP
Intercept	48.7 (38.1 to 59.2)	<0.001	48.6 (37.9 to 59.4)	<0.001
Baseline cortisol concentration	0.03 (−0.04 to 0.10)	0.37	0.03 (−0.04 to 0.11)	0.37
Time	−2.4 (−5.2 to 0.4)	0.09	−2.4 (−5.2 to 0.4)	0.09
Sex	−0.5 (−3.7 to 2.6)	0.73	−0.6 (−3.9 to 2.7)	0.71
Age	2.8 (0.4 to 5.2)	0.02	2.8 (0.4 to 5.2)	0.03
End cortisol concentration			0.00 (−0.02 to 0.02)	0.86

Bold values indicate statistical significance at the p < 0.05 level.

CI, confidence interval.

We examined the end point hair cortisol concentration while controlling for the baseline cortisol concentration (model 2 in Table 2). We found that increases in end point hair cortisol had a small but significant association with decreases in percent distance from the median BMI, β = −0.08 (95% confidence interval [CI]: −0.13 to −0.02) (Table 2). We had similar findings for BMI but not for waist circumference. We did not find that changes in cortisol were associated with changes in either systolic or diastolic blood pressure (Table 2). We did not observe an effect of randomization group on outcomes.

Overall, food security was not significantly associated with the primary adiposity outcome of percent distance from the median BMI in the models with hair cortisol concentration, food security β = −1.2 (95% CI: −8.2 to 5.9), p = 0.74. We also examined separate models stratified by hair cortisol level, above and below the median, with the model examining subjects below the median (n = 21) showing no significant association between food security status and percent distance from the median BMI (food security β = −6.8, 95% CI: −18.6 to 5.1, p = 0.24), and the model examining subjects above the median cortisol level (n = 19) also showing no significant association (food security β = 3.8, 95% CI: −5.7 to 13.4, p = 0.41).

Discussion

The main finding of our study was a small but significant association between increases in hair cortisol concentration with reduced adiposity in a sample of children with obesity from Latino families with a low income. This study is one of the first to report changes in hair cortisol concentration over time in a group of young children with obesity. Our findings do not support the hypothesis that decreased hair cortisol would be associated with decreased adiposity over time. This hypothesis was based on observations between children with a healthy weight and children with obesity, and clearly further work examining children with obesity over time is warranted to better understand the role of cortisol in children with obesity.

There are several potential explanations for the negative association between increasing hair cortisol concentrations and decreasing adiposity we found in our sample. First, it is possible that greater exposure to cortisol reflects normal adrenal function and coping with stress with higher values over time meaning an increased biologic response to stressors. It is possible that parents who were more engaged in monitoring and controlling their children's activities and eating habits stressed their children more by forcing behavior change in their children, when the children had little or no control over the changes. From prior analysis of the clinical data, we found that parents reported improvements over time in self-efficacy related to physical activity behaviors and increased monitoring of unhealthy foods.¹⁶ An analysis of a similar sample of children from families with low income and high exposure to stress or trauma showed that increases in hair cortisol over time were associated with an improvement in problematic behaviors over time.¹² The common hypothesis being that a change in behavioral trajectory for a child creates stress, even if that change may be perceived as positive. The contrast between these findings and the adult studies showing an association between decreased cortisol and stress reduction training^13,14 may be related to the differences in stress regulation between children and adults in the context of changing habits—children lack the locus of control when parents are trying to enact change for them, whereas parents, acting on their own habits, have more control. We also do not understand how weight loss may cause a stress response or change in the function of the HPA axis.

It is also worth noting that all of the end point hair samples were obtained between 4 and 5 months into the COVID-19 pandemic when substantial uncertainty and external stressors were present for many families. However, for this sample, at least the reported perceived stress did not change significantly in the early months of the pandemic, and was elevated at baseline, likely secondary to the socioeconomic stressors commonly described in this largely immigrant, farmworker population.^24,25

A challenge with interpreting these findings is that while hair cortisol examines exposure to cortisol over time, the information gained on how the HPA axis was functioning is more limited (e.g., cannot differentiate the awakening response from multiple stressors or take into account the slope). Recent data showed that children with a mean age of 9 years from low socioeconomic status (SES) families had a lower awakening response but higher nighttime concentrations, showing that even in younger children there may be a biologic response to chronic stress and highlighting one of the limitations of hair cortisol interpretation.²⁶

Our finding provides an opposing perspective to most results currently reported in this body of literature. Most studies have shown an association between hair cortisol concentration and adiposity assessed by weight parameters and with measures of blood pressure.⁹ A cohort study showed that hair cortisol at 6 years of age was associated with later nonalcoholic fatty liver disease (NAFLD) at 10 years of age.²⁷ Larsen et al used Generation R data and found an association between higher cortisol, cortisone, and several measures of adiposity at 6 years of age.²⁸ However, there are other studies with contradictory findings as well. One larger study (n = 300) did not identify an association between hair cortisol and weight measures, blood pressure, or other metabolic disturbances.²⁹ Previously published data from hair samples of mothers enrolled in Head Start showed that higher maternal hair cortisol was associated with lower BMI z-scores for their children and lower BMI in the mothers as well.³⁰

Our sample described had higher cortisol concentrations than levels observed in some US population-based samples, although important differences in sample selection and methods need to be considered. Two of the studies used mass spectrometry to determine concentration rather than enzyme-linked immunosorbent assay (ELISA) methods.^9,11 For instance, in a large population-based cohort of 2598 children 6 years of age, the lowest quintile of hair cortisol concentration was 0.13–0.74 pg/mg and the highest 2.93–6.76 pg/mg with a significant association between upper quintiles and BMI and systolic blood pressure.⁹ A study of children at mean 7.9 years of age found hair cortisol concentrations at a median 0.98 pg/mg (IQR 0.49–2.44).¹¹ A study examining hair cortisol in a more similar sample of children enrolled in Head Start and using ELISA methods of quantification showed a mean of 32 pg/mg with a range from 0.5 to 157.2, with no median reported.³¹ Reported normal ranges for healthy children from Netherlands showed a mean of 5 pg/mg for 4–5 years old (data seem to be skewed but report mean).³² The finding of potentially elevated hair cortisol concentrations in our sample is expected as prior literature has shown that hair cortisol concentrations are associated with obesity⁴ and lower socioeconomic status.³³ This sample was selected to participate in a clinical trial for families of children with obesity who have a low income. In addition, our sample was largely composed of children of migrant or seasonal farmworkers, with over 90% speaking mostly Spanish at home, and families who have recently immigrated may experience significant acculturative stress.

Finally, we did not find evidence of any effect modification of cortisol on an association between food security and change in BMI. This result does not coincide with that of a recent analysis by Distel et al that showed that hair cortisol concentrations moderated the association between food security and BMI among 52 children 6–10 years of age who were Mexican-American immigrants.⁶ A key characteristic of our sample was that all children had obesity compared with only 29% in the sample used by Distel et al.⁶ Food insecurity was common in our sample, although the limited variation in the adiposity outcome may make examining this association in our sample challenging.

Strengths and Limitations

This study is the first to examine the relationship between temporal changes in hair cortisol and adiposity in preschool-aged children with obesity. We used hair cortisol concentration as a measure for the total exposure to psychosocial stressors over time. While this measure has some advantages compared with salivary cortisol in examining long-term stress exposure, hair cortisol does not provide information about the diurnal cortisol rhythm or the cortisol awakening response. We also had substantial attrition for the hair sampling with only 43 of 188 subjects providing a sample at both time points, and thus, the results may not be representative of the whole sample. An additional limitation is that hair samples from the two time points for a given participant were run in separate assays. It would have been desirable to analyze each participant's samples side-by-side on the same assay plate. However, the samples from each time point were analyzed across multiple plates, which reduce the likelihood of a systematic bias that could have affected the results. This study is also unique by looking at this relationship in a sample of Latino children from families with low income. However, this also limits the capacity to generalize these results to children from other backgrounds or children with a healthy weight. In addition, the relatively small sample size did not allow for mediation analyses by physical activity, diet, or other reported behavioral changes.

Authors' Contributions

B.A.F. conceived of and designed the study, contributed to the writing, and assisted with the analysis. H.O.A. contributed to the writing and revisions of the article. T.P. assisted in the recruitment, enrollment, and data collection of subjects and contributed to writing and revisions of the article. J.S.M. supervised the hair cortisol extraction and analysis and contributed to the writing and revisions of the article.

Ethics and Incentives

This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Institutional Review Board (IRB) of Oregon Health and Science University. Written informed consent was obtained from all parents.

Funding Information

Research reported in this publication was supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under Award Number K23DK109199 to B.A.F.

Author Disclosure Statement

No competing financial interests exist.