ARTICLE IN PRESS
Using Hair Cortisol to Examine the Role of Stress in
Children’s Health Inequalities at 3 Years
Hannah E. Bryson, BA; Fiona Mensah, PhD, MSc;
Sharon Goldfeld, FRACP, FAFPHM, PhD; Anna M.H. Price, PhD
From the Murdoch Children’s Research Institute (HE Bryson, F Mensah, S Goldfeld, and AMH Price); Centre for Community Child Health
(HE Bryson, S Goldfeld, and AMH Price); Clinical Epidemiology and Biostatistics Unit (F Mensah), The Royal Children’s Hospital; and
Department of Paediatrics (HE Bryson, F Mensah, S Goldfeld, and AMH Price), The University of Melbourne, Parkville, Victoria, Australia
The authors have no conflicts of interest to disclose.
Address correspondence to Hannah Bryson, Murdoch Children’s Research Institute, Flemington Rd, Parkville VIC 3052, Australia
(e-mail: hannah.bryson@mcri.edu.au).
Received for publication January 2, 2019; accepted May 11, 2019.
TAGEDPABSTRACT
OBJECTIVE: Children exposed to early adversity (eg, financial
hardship, family violence, parent mental health difficulties)
are at greater risk of poor health outcomes. Physiological stress
is one mechanism thought to explain this pathway. We investi-
gated associations between adversity and young children’s
health and whether child stress (measured using hair cortisol)
mediated these associations.
METHODS: This was a cross-sectional study of 3-year-old chil-
dren whose mothers were recruited during pregnancy, through
the right@home trial, for their experience of adversity. Using
total counts of 9 sociodemographic and 9 psychosocial indica-
tors of adversity, regression models examined relationships
among adversity risk counts, child hair cortisol (potential
mediator), and 5 health outcomes: externalizing and internaliz-
ing problems, physical and socioemotional wellbeing, and
overweight/obesity.
RESULTS: Hair cortisol data were available for 297 out of 500
(59%) participating children. When examined separately, soci-
odemographic adversity risk was associated with higher exter-
nalizing problems, and psychosocial adversity risk was
TAGEDPWHAT’S NEW
We investigated whether physiological stress (hair cor-
tisol) is a mechanism that explains (mediates) the asso-
ciation between adversity and health in 3-year-old
children. We found that greater adversity risk was
associated with poorer health but no evidence that
stress mediated these effects.
TAGEDPHEALTH
CONDITIONS SUCH as poor mental health and
overweight/obesity are common in young children.1,2 Chil-
dren who are exposed to social adversity (eg, raised by
parents experiencing unemployment, financial hardship,
poor mental health) are at greater risk of these conditions,
creating inequalities in children’s health.1−4 Total adversity
counts such as the adverse childhood experiences frame-
work and cumulative risk models described by Evans et al5
highlight the increasingly detrimental effect that multiple
ACADEMIC PEDIATRICS
Copyright © 2019 by Academic Pediatric Association
associated with higher externalizing problems and poorer
physical/socioemotional wellbeing. When examined together
in a single model, psychosocial (but not sociodemographic)
adversity was associated with higher externalizing problems
(unstandardized mean difference [b], 0.53; P = .002) and
poorer physical wellbeing (b, 1.19; P = .009); higher hair corti-
sol was associated with higher externalizing problems (b, 0.76;
P = .02). There was no evidence that stress (hair cortisol) medi-
ated associations between adversity and health.
CONCLUSIONS: In 3-year-old children, we found no evidence
that physiological stress (hair cortisol) mediated associations
between adversity risk and children’s health. Hair cortisol may
be limited as a single measure of stress, or physiological stress
may not be a mechanism for explaining the effects of adversity
on these young children’s health.
TAGEDPKEYWORDS: hair cortisol; adversity; child; stress; health
inequalities
ACADEMIC PEDIATRICS 2019;XXX:1−10
adverse exposures can have on children’s outcomes.5,6
Although this association is well established, the biological
mechanisms that operate on the pathways from adversity to
poor health are less understood. The physiological stress
response is one potential mechanism thought to play a piv-
otal role via its effects on inflammation, metabolic function,
structural changes to the central nervous system, and accel-
erated cellular aging.5−7 Although evidence supporting the
role of physiological stress on the pathways from adversity
to poor health has been reported in studies of trauma and via
experimental and animal models,6,7 evidence from commu-
nity cohorts is limited.
The hypothalamic-pituitary-adrenal (HPA) axis and its
production of the hormone cortisol has been a focus of exist-
ing research. Traditionally, research examining the link
among adversity, physiological stress, and health has used
cortisol measured in saliva as a biomarker of physiological
stress. These studies typically report associations among
Volume 000
1 XX 2019
 ARTICLE IN PRESS
TAGEDEN2 BRYSON ET AL
greater adversity, higher cortisol, and poorer health; yet,
studies also report the inverse or no association.8−12 Differ-
ing outcomes may result from time- and age-dependent
effects.7,13,14 In young children or when adversity is recent,
adversity and poorer health are often associated with ele-
vated cortisol (hypercortisolism),9,13,15,16 but this may
become blunted over time, resulting in diminished cortisol
(hypocortisolism).9,13,14 The nature of the adversity may
also determine the cortisol response. Although chronic or
extreme trauma may lead to hypocortisolism,8,15 broader
social adversities have been associated with hypercortiso-
lism.12,15−17 Inconsistencies may also arise from the use of
salivary cortisol, which measures a single time point rather
than longer term stress.18 Measuring cortisol concentrations
in hair is a promising, novel method for measuring longer
term activity of the physiological stress response, as it cap-
tures the accumulation of cortisol produced and incorpo-
rated into the growing hair over time.18,19
To date, few studies have used hair cortisol to examine
the physiological stress pathway and determine whether it
mediates (explains) the association between adversity and
children’s health. Ursache et al20 found no evidence that
stress mediated associations between socioeconomic fac-
tors and internalizing symptoms but noted the small sam-
ple size (N = 26). White et al21 examined internalizing
and externalizing symptoms in a cohort of 537 children
ages 3 to 16 years with or without a history of maltreat-
ment. They found that children with a history of maltreat-
ment had higher externalizing symptoms and that lower
hair cortisol in the maltreated group partially mediated
this difference. In addition, White et al21 found age- and
time-dependent effects, where chronicity of maltreatment
was associated with lower cortisol and effects were spe-
cific to children older than 9.7 years. However, hair corti-
sol research more broadly suggests that social adversities
are associated with higher hair cortisol, particularly in
young children or when adversity is recent.17,19,22,23
Given this evidence, it can be hypothesized that higher
hair cortisol is associated with poorer health in young
children and may mediate the association between social
adversity and poor health outcomes.
The current study aimed to investigate this proposed
mediation pathway by concurrently examining adversity,
physiological stress (measured using hair cortisol), and
young children’s health outcomes at 3 years. Due to
the multifinality of adversity (affecting multiple health
outcomes),3,4 this study investigated 3 domains of child
health—mental health, wellbeing, and overweight/
obesity—that are commonly associated with both adversity
and stress. The aims were to determine 1) the association
between composite measures of social adversity risk and
children’s health outcomes, 2) the association between
children’s physiological stress and health, and 3) whether
physiological stress mediated any associations between
social adversity risk and health. We hypothesized that
greater social adversity risk and higher physiological stress
(reflected in higher hair cortisol) would be associated with
poorer child health outcomes and that higher stress would
mediate the association between adversity and health.
ACADEMIC PEDIATRICS
TAGEDH1METHODSTAGEDEN
TAGEDH2DESIGN AND SETTINGTAGEDEN
We conducted a cross-sectional study nested within
the right@home randomized controlled trial (RCT) of
nurse home visiting. The right@home RCT was com-
prised of a large cohort of women recruited from a
community-setting for their experience of adversity
during pregnancy, as well as their subsequent children.
Detailed study methods for the RCT are described in
the published protocol.24
TAGEDH2PARTICIPANTSTAGEDEN
Children included in the current study were those whose
mothers were enrolled in the right@home trial, completed a
3-year follow-up assessment, and had a hair sample col-
lected for cortisol analysis. The trial recruited pregnant
women attending antenatal clinics between April 30, 2013,
and August 29, 2014, from 10 public maternity hospitals in
metropolitan and regional areas of Victoria and Tasmania,
Australia. Inclusion criteria were women 1) whose expected
due dates were before October 1, 2014; 2) who were less
than 37 weeks into their pregnancy at recruitment; 3) who
had sufficient English proficiency to complete assessments;
4) whose home addresses were within the travel boundaries
of the study; and 5) who self-reported 2 or more of 10 ante-
natal risk factors. These antenatal risk factors included
young pregnancy; not living with another adult; no support
in pregnancy; poorer health; a long-term illness, health prob-
lem, or disability that limited daily activities; currently
smokes; coping difficulties; low education; no person in the
household currently earning an income; and never having
had a job before. The study excluded women who 1) were
enrolled in an existing Tasmanian nurse home visiting pro-
gram, 2) did not comprehend the recruitment invitation (eg,
had an intellectual disability such that they were unable to
consent to participation or had insufficient English to com-
plete assessments), 3) had no way to be contacted, and 4)
experienced a critical event that excluded their participation
(eg, termination of pregnancy, stillbirth, participant or child
death).
T ROCEDURETAGEDEN
AGEDH2P
Eligible women were identified in antenatal clinics and
invited into the RCT. Women who chose to enroll pro-
vided informed consent and completed a comprehensive
baseline home-based assessment. The initial RCT consent
included follow-up to child age 2 years, at which time
consent was obtained for follow-up to 5 years. Measures
in the current study were collected at the 3-year home-
based assessment (conducted between child ages 36 and
40 months), except items regarding mothers’ age and
Aboriginal or Torres Strait Islander status, which were
collected at the baseline assessment.
T EASURESTAGEDEN
AGEDH2M
Exposures were study-designed composite measures
of sociodemographic and psychosocial adversity risk.
 ARTICLE IN PRESS
TAGEDENACADEMIC PEDIATRICS CHILD ADVERSITY, HAIR CORTISOL, AND HEALTH OUTCOMES 3

Table 1. Adversity Indicators Used to Derive Total Counts of Sociodemographic and Psychosocial Adversity Risk

Measure Indicator of Adversity

Sociodemographic Adversity Risk
Age at pregnancy Age in years calculated from year of birth
Categorized as young pregnancy (<23 y, adverse) vs not (≥23 y, not adverse)
Highest level of education Highest level of education attained
“Did not complete high school or any further training” (adverse) vs “completed high school and/or
any further training” (not adverse)
Marital status Marital status
“Single, not living with partner, separated, or divorced” (adverse) vs “living with partner or
married” (not adverse)
Household income Main source of household income
“Benefit, pension, or no income” (adverse) vs “paid employment” (not adverse)
Employment Currently employed
“No” (adverse) vs “yes” (not adverse)
Housing tenure Type of housing tenure
“Public rental, paying board, or living rent free“ (adverse) vs “fully owned, being purchased, or
private rental” (not adverse)
Aboriginal or Torres Strait Islander Aboriginal or Torres Strait Islander
“Yes” (adverse) vs “no” (not adverse)
Housing problems Self-report of housing problems, including overcrowding, lead, mold, rodents, being threatened
with eviction
“Yes” (adverse) vs “no” (not adverse)
Financial hardship Six-item measure from the Longitudinal Study of Australian Children;36 “yes” or “no” report that
any of the following happened in the past 12 months due to shortage of money: adults or
children went without meals, they were unable to heat or cool their home, they pawned or sold
something, or they sought assistance from a welfare or community organization
“Two or more hardships” (adverse) vs “one or none” (not adverse)
Psychosocial Adversity Risk
Current smoker Single-item self-report of being a current smoker
“Yes” (adverse) vs “no” (not adverse)
Family violence in the past year Self-report that partner or any other family member pushed, punched, kicked, hit, or threatened
mother in the past year and self-report of feeling threatened in home in the past year; from the
Family Psychosocial Screening Questionnaire37
“Yes” to either (adverse) vs “no” to both (not adverse)
Live in a safe place Single-item self-report of feeling safe in the home (“Do you feel that you live in a safe place?”);
from the Family Psychosocial Screening Questionnaire37
“No” (adverse) vs “yes” (not adverse)
Ever had a drug problem Single-item self-report of ever having had a drug problem; from the Family Psychosocial
Screening Questionnaire37
“Yes” (adverse) vs “no” (not adverse)
Alcohol problem in the past year Single-item self-report of having had a drinking problem in the past year; from the Family
Psychosocial Screening Questionnaire37
“Yes” (adverse) vs “no” (not adverse)
Emotional abuse Emotional abuse subscale of the Composite Abuse Scales; 11-item measure rated on a 6-point
scale (“never” to “daily”) assessing emotionally abusive partner behavior, where higher scores
indicate greater emotional abuse38
Score of 3 or more indicating the presence of emotional abuse (adverse) vs less than 3
(not adverse)38
Mental health Depression, Anxiety, and Stress Scales;39 21-item measure rated on a 4-point scale (“not at all”
to “most of the time”) assessing the negative emotional states of depression, anxiety, and
tension/stress; each subscale had 7 items each, where higher scores indicate greater
symptoms, and thus more adverse circumstances
Top 15% scores (adverse) vs bottom 85% scores (not adverse) based on population
representative data40

Child hair cortisol concentration, as a measure of physi-
ological stress, was an exposure and potential mediator.
Outcomes were child mental health, wellbeing, and
overweight/obesity.
TAGEDPSOCIODEMOGRAPHIC AND PSYCHOSOCIAL ADVERSITY RISKTAGEDEN
Sociodemographic and psychosocial adversity risks were
measured as total unweighted risk counts of maternal-
reported indicators of adversity (Table 1). Sociodemographic
(eg, education, marital status, income) and psychosocial (eg,
family violence, drug and alcohol problems, mental health)
adversity risk counts were defined by the presence versus
absence of 9 indicators of adversity with totals ranging from
0 to 9 (Cronbach’s alpha: sociodemographic a = 0.54, psy-
chosocial a = 0.63). Items were selected for each index on
face validity and in consideration of the cumulative risk liter-
ature.4,5,12,16 Sociodemographic and psychosocial adversity
risks were independently examined, as evidence suggests
they have specific and collective impacts on children’s health
outcomes.4,25
 ARTICLE IN PRESS
TAGEDEN4 BRYSON ET AL
TAGEDPHAIR CORTISOL CONCENTRATIONSTAGEDEN
Hair samples were collected and analyzed according to
international standard practice.18 Hair was collected from
1 to 4 portions cut from the posterior vertex region of the
scalp. When a child’s hair was too short at the posterior
vertex, hair samples were collected from the temporal or
parietal regions of the head. The first 3 cm of hair cut clos-
est to the scalp were analyzed by the Australian laboratory
Stratech Scientific. Samples were weighed and mechani-
cally crushed, and methanol was used for extraction. Sam-
ples were then dried and reconstituted in phosphate-
buffered saline for analysis. Cortisol was analyzed in
duplicate using a commercially available enzyme-linked
immunosorbent assay (Salimetrics; Carlsbad, CA) accord-
ing to the manufacturer’s instructions (limit of detection,
0.007 mg/dL). Intra- and inter-assay coefficients of vari-
ability were 4.3% and 4.9%, respectively, indicating high
precision. Values are expressed as a single continuous
concentration (pg/mg) of cortisol, transformed to the natu-
ral log due to positive skewness.
T HILD
AGEDPC MENTAL HEALTHAGEDNTE
We used the Strengths and Difficulties Questionnaire
(SDQ), a 25-item parent report measure assessing a child’s
mental health and behavior, with each item rated on a 3-
point scale (“not true,” “somewhat true,” or “certainly
true”).26 Two constructs were used to examine child mental
health: 1) externalizing problems, comprised of conduct and
hyperactivity/inattention subscales; and 2) internalizing
problems, comprised of emotional and peer relationship
problems subscales. Higher scores indicate greater prob-
lems, with possible scores ranging from 0 to 20.
T HILD
AGEDPC WELLBEINGTAGEDEN
We also used the Pediatric Quality of Life Inventory
(PedsQL), a 21-item parent report measure assessing a
child’s general, physical, and socioemotional functioning,
with each item rated on a 5-point scale (“never,” “almost
never,” “sometimes,” “often,” or “always”).27 Two sub-
scales (physical functioning and socioemotional function-
ing) were used to examine child wellbeing. Scores were
reversed for analysis so that higher scores reflected poorer
wellbeing, with possible scores ranging from 0 to 100.
T HILD
AGEDPC OVERWEIGHTTAGEDEN
Researchers collected 2 measurements for height using an
Invicta Height Measure stadiometer (Invicta Education;
Bicester, Oxfordshire, UK) and 2 for weight using Tanita
HD-315 digital scales (Tanita; Arlington Heights, IL). These
measurements were then averaged to calculate age- and gen-
der-specific body mass index (weight/height2) z-scores, cat-
egorized into international classifications of weight status.28
Weight classifications of “overweight/obese” were com-
pared to “normal weight/underweight.” Underweight and
normal weight were combined due to low numbers of under-
weight children (n = 14; 5%).
ACADEMIC PEDIATRICS
TAGEDH2STATISTICAL ANALYSISTAGEDEN
Preliminary analyses examined the bivariate correlation
coefficients between each exposure and outcome. Frac-
tional polynomial fitted plots showed no evidence of non-
linear associations between hair cortisol and each
adversity risk count and health outcome; thus, linear asso-
ciations were examined in all analyses. To address the
aims of determining the association between composite
measures of social adversity risk and children’s health
comes and the association between children’s physiologi-
cal stress and health, associations between each exposure
(sociodemographic adversity risk, psychosocial adversity
risk, and hair cortisol) and each outcome were separately
examined using adjusted linear (continuous outcomes of
externalizing and internalizing problems and physical and
socioemotional wellbeing) and logistic (categorical out-
come of overweight/obesity) regression analyses. To
address the aim to determine whether physiological stress
mediated any associations between social adversity risk
and health, we used adjusted linear and logistic regression
analyses in which models including both sociodemo-
graphic and psychosocial adversity risks were examined
without hair cortisol (Model 1) and then with hair cortisol
(Model 2) to determine whether the inclusion of hair cor-
tisol attenuated associations between adversity and out-
comes, as evidence of mediation by hair cortisol. To
formally check this, unadjusted univariable mediation
tests (Sobel-Goodman mediation test for continuous out-
comes and mediation test for categorical outcomes29,30)
were undertaken in which sociodemographic and psycho-
social adversity were separately examined with each
health outcome for indirect effects mediated by hair corti-
sol. Regression coefficients are presented as adjusted
unstandardized mean differences (b, linear regression for
continuous outcomes) and odds ratios (logistic regression
for categorical outcome) with 95% confidence intervals
(CIs) and P values.
Adjusted analyses accounted for child age and sex, as
well as the season of hair sample collection as a proxy for
sun exposure, in line with previously identified potential
confounders of hair cortisol.17 The site on the head from
which the hair sample was taken was included to account
for variation in hair growth rates across different regions of
the scalp,18 and RCT randomization status was included to
address potential confounding due to the study design.
Regression models were repeated using multiple impu-
tation methods to account for missing data. The longitudi-
nal nature of this cohort meant that hair cortisol data were
available for 319 children at 2 years. Missing data were
imputed for 393 children who had hair cortisol data at
either 2 or 3 years. Results considering the 283 children
with complete data were compared to those for the 393
children with imputed data. The imputation model
included all indicators of adversity, health outcomes, con-
founders, and hair cortisol at 3 years and hair cortisol at
2 years as an auxiliary variable; 50 datasets were imputed
by chained equations. Results for children with complete
data are presented in Table 3, and results for the imputed
 ARTICLE IN PRESS
TAGEDENACADEMIC PEDIATRICS
data are presented in the Supplementary Tables. Data
were analyzed using Stata 14.1 (StataCorp; College Sta-
tion, TX).
TAGEDH2ETHICS APPROVALTAGEDEN
This study was approved by the Human Research
Ethics Committee of The Royal Children’s Hospital
(HREC 32296), Australia.
TAGEDH1RESULTSTAGEDEN
The RCT enrolled 722 pregnant women, of whom 558
consented to the extended follow-up at child age 3 to
5 years (Figure). Out of the 558 women, 495 (89%) com-
pleted the 3-year follow-up assessment for 500 children,
including 5 women who had twins. Hair samples were
collected from 297 of those 500 children (59%). Reasons
for child hair samples not being collected are shown in
Figure. Cortisol concentrations were analyzed for all 297
children (mean age, 3.1 years; 117 [39%] male). Table 2
shows participant characteristics, and Supplementary
Table 1 shows observed proportions of adversity indica-
tors used to derive the adversity risk counts. Adversity
indicators were comparable at 3 years for children with
and without hair cortisol data (Supplementary Table 1);
differences in baseline adversity indicators between the
original enrolled cohort and those with hair cortisol at
3 years suggest those lost to follow-up had higher rates of
adversity indicators (Supplementary Table 2). Supple-
mentary Table 3 shows bivariate correlations between all
exposures and outcomes.
Figure. Participant flow.
CHILD ADVERSITY, HAIR CORTISOL, AND HEALTH OUTCOMES 5
Models examining associations between each exposure
and outcome (individual factors; Table 3) for the first aim
of determining the association between composite meas-
ures of social adversity risk and children’s health out-
comes showed that greater sociodemographic adversity
risk was associated with higher externalizing problems
(b = 0.41; 95% CI, 0.15−0.66), reflecting a mean differ-
ence in externalizing scores of 0.41 for each 1-point
increase in sociodemographic adversity risk. Greater psy-
chosocial adversity risk was associated with higher exter-
nalizing problems (b = 0.58; 95% CI, 0.29−0.87) and
poorer physical (b = 0.96; 95% CI, 0.18−1.73) and socio-
emotional (b = 1.18; 95% CI, 0.25−2.11) wellbeing. With
regard to our second aim of determining the association
between children’s physiological stress and health, our
results showed that higher child hair cortisol was associ-
ated with higher externalizing problems (b = 0.69; 95%
CI, 0.09−1.28) but was not associated with other health
outcomes.
With regard to our aim of determining whether physio-
logical stress mediated any associations between social
adversity risk and health, models examining sociodemo-
graphic and psychosocial adversity risk together (Model
1, Table 3) showed that, when both adversity risk totals
were included in the same model, associations between
sociodemographic adversity risk and outcomes attenu-
ated, but psychosocial adversity risk was still associated
with externalizing problems and physical wellbeing.
When hair cortisol was added to the model (Model 2,
Table 3), all estimates between adversity risk and health
outcomes remained similar, indicating that hair cortisol
did not mediate any associations. In Model 2, the
 ARTICLE IN PRESS
TAGEDEN6 BRYSON ET AL
Table 2. Participant Characteristics
Characteristic
Mother
Age at pregnancy (y), mean (SD) [range]
Intervention status, n (%)
Usual care
Program group
Child
Age (y), mean (SD) [range]
Gender, n (%)
Male
Female
Child hair sample
Hair cortisol (pg/mg), mean (SD) [range]
Season of hair sample collection, n (%)
Summer
Autumn
Winter
Spring
Location of hair sample collection, n (%)
Posterior vertex
Temporal
Parietal
Adversity risk counts, mean (SD) [range]
Socioeconomic adversity risk
Psychosocial adversity risk
Child health outcomes
SDQ Externalizing problems, mean (SD) [range]
SDQ Internalizing problems, mean (SD) [range]
PedsQL Physical functioning, reverse, mean (SD) [range]
PedsQL Socioemotional functioning, reverse, mean (SD) [range]
BMI weight status, n (%)
Normal weight
Overweight/obese
SD indicates standard deviation; SDQ, Strengths and Difficulties Questionnaire; PedsQL, Pediatric Quality of Life Inventory; and BMI,
body mass index.
*N ranges from 242 to 297 due to missing data.
association between higher hair cortisol and greater
externalizing problems remained (b = 0.76; 95% CI,
0.12−1.39). Formal tests for mediation (Table 4) showed
no evidence that any associations between either of the
adversity risk counts and health outcomes were mediated
by hair cortisol.
Compared to the complete cases, results for the
imputed analyses (Supplementary Table 4) showed simi-
lar results, aside from strengthened associations between
sociodemographic adversity risk and externalizing prob-
lems and between psychosocial adversity risk and inter-
nalizing problems and poorer socioemotional wellbeing.
TAGEDH1DISCUSSIONTAGEDEN
In this cohort of 3-year-old children whose mothers
were recruited for their experience of adversity, greater
adversity risk was associated with higher externalizing
problems and poorer physical wellbeing. Higher hair cor-
tisol was associated with higher externalizing problems;
however, there was no evidence that hair cortisol was
associated with any other health outcomes including inter-
nalizing problems, physical or socioemotional wellbeing,
or overweight/obesity. Also contrary to our hypothesis,
there was no evidence that hair cortisol, as a measure of
ACADEMIC PEDIATRICS
Value (N = 297)*
28.2 (6.1) [15.1−46.2]
144 (48.5)
153 (51.5)
3.1 (0.1) [3.0−3.4]
117 (39.4)
180 (60.6)
8.5 (7.8) [1.1−45.5]
59 (19.9)
77 (25.9)
76 (25.6)
85 (28.6)
239 (80.5)
12 (4.0)
46 (15.5)
2.6 (1.8) [0−7]
1.6 (1.7) [0−7]
8.1 (3.9) [0−19]
3.8 (2.6) [0−14]
10.1 (10.7) [0−68.8]
15.3 (12.2) [0−52.5]
212 (77.1)
63 (22.9)
physiological stress, mediated any associations between
adversity and young children’s health outcomes.
The associations identified between adversity and
poorer child mental health and wellbeing at 3 years rein-
force the well-established associations between cumula-
tive adversity risk and children’s health,1,3−5 further
highlighting the young age at which they emerge.2,3 How-
ever, we found no evidence that physiological stress
mediated the associations between adversity and young
children’s health outcomes. These findings add to a small
field of research; indeed, only 2 previous studies (Ursache
et al20 and White et al21) have used hair cortisol to exam-
ine this potentially mediating role of stress, with inconsis-
tent results. The lack of evidence of mediation by stress
appears to be due to an absence of associations between
adversity and child hair cortisol. This finding may be due
to the young age of this cohort; for example, White et al21
identified an age band from 9.7 to 16 years in which mal-
treatment was associated with lower hair cortisol but
found no association in younger children ages 3 to
7.9 years.21 Young children may not experience the
adversity risks measured in the current study as stressful;
thus, the expected physiological stress response may not
be evoked.15 That said, the associations identified
between psychosocial adversity and health suggest such
 TAGEDENACADEMIC PEDIATRICS
Table 3. Regression Analyses of Association Among Social Adversity Risk, Child Hair Cortisol, and Child Health Outcomes

Individual Factors† Model 1‡ Model 2‡

Coef* 95% CI P Value Coef* 95% CI P Value Coef* 95% CI P Value
SDQ externalizing problems
Sociodemographic adversity risk 0.41 0.15, 0.66 .002 0.13 −0.18, 0.43 .42 0.09 −0.21, 0.40 .55
Psychosocial adversity risk 0.58 0.29, 0.87 <.001 0.49 0.16, 0.82 .004 0.53 0.20, 0.86 .002
Hair cortisol (pg/mg) 0.69 0.09, 1.28 .02 0.76 0.12, 1.39 .02

ARTICLE IN PRESS
SDQ internalizing problems
Sociodemographic adversity risk 0.10 −0.08, 0.27 .28 −0.0004 −0.22, 0.22 1.00 −0.003 −0.22, 0.22 .98
Psychosocial adversity risk 0.17 −0.04, 0.37 .11 0.18 −0.06, 0.41 .14 0.17 −0.06, 0.41 .15
Hair cortisol (pg/mg) 0.06 −0.33, 0.46 .76 −0.08 −0.54, 0.38 .74
PedsQL physical wellbeing (reverse)
Sociodemographic adversity risk −0.10 −0.81, 0.61 .78 −0.80 −1.61, 0.02 .06 −0.75 −1.57, 0.06 .07
Psychosocial adversity risk 0.96 0.18, 1.73 .02 1.24 0.35, 2.12 .006 1.19 0.30, 2.07 .009
Hair cortisol (pg/mg) −0.94 −2.57, 0.68 .25 −1.09 −2.80, 0.61 .21

CHILD ADVERSITY, HAIR CORTISOL, AND HEALTH OUTCOMES

PedsQL socioemotional wellbeing
(reverse)
Sociodemographic adversity risk 0.61 −0.20, 1.41 .14 0.14 −0.85, 1.13 .77 0.17 −0.82, 1.17 .73
Psychosocial adversity risk 1.18 0.25, 2.11 .01 1.08 0.003, 2.15 .05 1.05 −0.03, 2.13 .06
Hair cortisol (pg/mg) −0.31 −2.17, 1.54 .74 −0.67 −2.75, 1.41 .52
Overweight/obese BMI z-score status
Sociodemographic adversity risk 1.10 0.93, 1.31 .27 1.14 0.92, 1.41 .23 1.14 0.92, 1.42 .22
Psychosocial adversity risk 0.93 0.76, 1.14 .48 0.90 0.71, 1.13 .36 0.89 0.70, 1.13 .33
Hair cortisol (pg/mg) 0.76 0.51, 1.12 .16 0.79 0.51, 1.23 .30
CI indicates confidence interval; SDQ, Strengths and Difficulties Questionnaire; PedsQL, Pediatric Quality of Life Inventory; and BMI, body mass index.
*Coef is the unstandardized adjusted mean difference (b, linear regression) in outcome (SDQ externalizing problems, SDQ internalizing problems, PedsQL physical wellbeing, PedsQL socioemotional
wellbeing) for a 1-point change in exposure, except for overweight/obesity status, which is the odds ratio (logistic regression). All are adjusted for child age, gender, season of assessment, site of hair collec-
tion, and randomized controlled trial randomization status.
†N ranges from 223 to 294.
‡N ranges from 220 to 238.

7
 ARTICLE IN PRESS
TAGEDEN8 BRYSON ET AL
Table 4. Indirect Effects of Individual Adversity Risk Counts on Health Outcomes Mediated by Hair Cortisol
Coef*
Sociodemographic adversity risk
SDQ externalizing problems 0.01
SDQ internalizing problems 0.001
PedsQL physical wellbeing (reverse) −0.01
PedsQL socioemotional wellbeing (reverse) −0.01
Overweight/obese BMI z-score status −0.0004
Psychosocial adversity risk
SDQ externalizing problems −0.01
SDQ internalizing problems 0.001
PedsQL physical wellbeing (reverse) 0.01
PedsQL socioemotional wellbeing (reverse) 0.01
Overweight/obese BMI z-score status 0.01
CI indicates confidence interval; SDQ, Strengths and Difficulties Questionnaire; PedsQL, Pediatric Quality of Life Inventory; and BMI,
body mass index.
*Unadjusted mediation coefficient.
adversities are having a detrimental impact on these
young children. Moreover, inconsistent hair cortisol find-
ings have been reported for both younger and older chil-
dren.17 Given that the physiological stress response is
comprised of multiple interrelated processes across the
nervous, endocrine, and immune systems,5−7 the potential
of using hair cortisol as a single measure of young child-
ren’s physiological stress may be limited.
Studies using hair cortisol to examine children’s physi-
ological stress and health (without mediation) have also
provided inconsistent findings. We drew on evidence sug-
gesting that adversity is associated with higher physiolog-
ical stress to hypothesize that higher hair cortisol would
be associated with poorer health; however, existing stud-
ies report either no association20,22,31 or a negative associ-
ation between hair cortisol and externalizing or
internalizing problems in school-aged children.21 The cur-
rent study reports the only positive association between
child hair cortisol and externalizing problems. This may
be specific to the young age of the cohort; for example,
the current findings align with a meta-analysis that exam-
ined basal salivary cortisol and externalizing problems
and found a negative association among school-aged chil-
dren but a positive moderate association for children
younger than 5 years.9 This positive association specific
to younger children is thought to reflect a common precur-
sor, such as early adversity, associated with both stress
and the development of externalizing problems.9
Although we found that adversity and physiological stress
(hair cortisol) were each associated with externalizing
problems, we found no evidence that physiological stress
mediated effects of adversity on young children’s exter-
nalizing problems.
Although some studies have found associations
between higher child hair cortisol and higher body mass
index/obese weight status,23,32,33 those examining primar-
ily normal-weight cohorts have not.22,34 We found no evi-
dence of association between child hair cortisol and
overweight/obese weight status. This may be due to the
ACADEMIC PEDIATRICS
Indirect Effect Mediated by Hair Cortisol N = 223−288
95% CI P Value
−0.03, 0.04 .75
−0.01, 0.01 .93
−0.06, 0.05 .80
−0.07, 0.06 .86
−0.02, 0.02 .97
−0.05, 0.03 .68
−0.01, 0.01 .93
−0.06, 0.08 .81
−0.07, 0.08 .88
−0.01, 0.03 .54
small number of children in the obese weight range
(n = 14; 5%), which limited our ability to detect such asso-
ciations. Taken with the existing research, the current
findings provide little support for the proposed role of
physiological stress in the emergence of young children’s
health conditions.
A strength of this study is the large cohort of children
whose mothers were recruited for their experience of
adversity during pregnancy. By using a cohort enriched
for risks, indicators of adversity were more prevalent and
diverse than in previous hair cortisol studies, which have
primarily used population cohorts unselected for adver-
sity. This design meant we could collectively examine
sociodemographic and psychosocial adversity risks, hair
cortisol, and multiple domains of health for children at a
young age when health inequalities are beginning to
emerge. Previous studies tend to examine stand-alone
associations between child hair cortisol and health in lim-
ited sample sizes (eg, N < 100) of school-aged chil-
dren20,22,33 or among larger, more homogeneous cohorts
comprised of relatively advantaged participants. For
example, 3 studies with sample sizes greater than 1000
are drawn from the Generation R cohort, in which over
80% of mothers are married or living with a partner and
54% have completed a university education.23,31,32
A limitation of the current study is the reliance on
maternal report for both adversity exposures and child-
ren’s health outcomes. The use of brief self-reported
measures to identify adversity indicators may have over-
simplified women’s experiences; yet, these were necessar-
ily maternal report to capture the multifaceted nature of
social adversity in this large cohort of young children.
Maternal reporter bias may also partly explain the associ-
ations between psychosocial adversity and children’s
health; for example, mothers reporting poorer mental
health may be more likely to report poorer child out-
comes.35 Despite possible reporter bias, maternal mental
health is an important determinant of child outcomes,25
and mothers provide a valuable source of information
 ARTICLE IN PRESS
TAGEDENACADEMIC PEDIATRICS
about their young children.35 Reliance on parental report
is not specific to the current study and is a limitation of
the broader child hair cortisol literature.20−22,31 Continued
follow-up of this cohort through the ongoing right@home
trial, as children turn 4 and 5 years old, will allow the cur-
rent findings to be extended to longitudinal analyses,
including objective health measures at older ages. Such
longitudinal analyses may also determine whether a role
of physiological stress, measured via hair cortisol, is evi-
dent at older ages or emerges over time; however, the cur-
rent findings fill an important gap in the literature by
focusing on an early circumscribed age to comprehen-
sively examine these cross-sectional associations.
Finally, despite being a large cohort with diverse
adversity risks, there was some evidence those lost to
follow-up from the original cohort had more adversity
risk at baseline than those participating at 3 years and
who had a child hair sample collected. Such differential
attrition may have reduced the variability in adversity
risks and health outcomes examined, potentially con-
tributing to the lack of associations for internalizing
and overweight health outcomes. However, our findings
align with previous literature, suggesting that inequal-
ities in internalizing and overweight outcomes emerge
at older ages.2,3 Complete case analyses may also have
had insufficient power to detect independent effects
of co-occurring sociodemographic and psychosocial
adversity risks. Multiple imputation analyses identified
independent associations between sociodemographic
adversity and externalizing problems that were not evi-
dent in complete cases. However, even with imputation,
hair cortisol did not show evidence of any mediating
effect, supporting that the lack of mediation by stress
was a robust finding.
TAGEDH1CONCLUSIONSTAGEDEN
In this study of 3-year-old children, we demonstrated
associations between adversity and poorer health out-
comes and found new evidence that higher physiologi-
cal stress, measured using hair cortisol, was associated
with greater externalizing symptoms. However, findings
did not support our hypothesis that physiological stress
mediates the effects of adversity on young children’s
health. These unexpected findings may be because a
simple measure of hair cortisol is limited for assessing
the complex interrelated processes involved in physio-
logical stress responses. Alternatively, physiological
stress may not be a mechanism through which adversity
affects children’s health at this young age. These find-
ings highlight the need to better understand the biologi-
cal mechanisms that operate on the pathways from
adversity to poor health. The role of stress as a mecha-
nism explaining the effects of adversity on health is a
well-held premise; however, when examined at the
community level and using hair cortisol as the marker
of stress, the fragmented findings provide limited evi-
dence that this mechanism explains the effects of adver-
sity on health in young children.
CHILD ADVERSITY, HAIR CORTISOL, AND HEALTH OUTCOMES 9
TAGEDH1ACKNOWLEDGMENTSTAGEDEN
We thank all families; the researchers, nurses, and social care practi-
tioners working on the right@home trial; the antenatal clinic staff at par-
ticipating hospitals who helped facilitate the research; and the Expert
Reference Group for their guidance in designing the trial.
Financial disclosure: The right@home sustained nurse home visiting
trial is a research collaboration among the Australian Research Alliance
for Children and Youth; the Translational Research and Social Innova-
tion Group at Western Sydney University; and the Centre for Community
Child Health, which is a department of The Royal Children’s Hospital
and a research group of Murdoch Children’s Research Institute (MCRI).
right@home is funded by the Victorian Department of Education and
Training, the Tasmanian Department of Health and Human Services, the
Ian Potter Foundation, Sabemo Trust, Sidney Myer Fund, the Vincent
Fairfax Family Foundation, and the Australian National Health and Med-
ical Research Council (NHMRC, Project Grant 1079418). Research at
the MCRI is supported by the Victorian Government’s Operational Infra-
structure Support Program. H.E.B. is supported by an MCRI Research
Group Scholarship and an Australian Government Research Training
Program Scholarship, S.G. is supported by NHMRC Career Develop-
ment Fellowship 1082922, and F.M. is supported by NHMRC Career
Development Fellowship 1111160. The funding sources had no involve-
ment in the study design; collection, analysis, and interpretation of
the data; writing of this article; or the decision to submit the article for
publication.
TAGEDH1SUPPLEMENTARY DATATAGEDEN
Supplementary data related to this article can be found
online at https://doi.org/10.1016/j.acap.2019.05.008.
TAGEDH1REFERENCESTAGEDEN
1. Lawrence D, Hafekost J, Johnson SE, et al. Key findings from the
second Australian Child and Adolescent Survey of Mental Health
and Wellbeing. Aust N Z J Psychiatry. 2015;50:876–886.
2. Rougeaux E, Hope S, Law C, et al. Have health inequalities changed
during childhood in the New Labour generation? Findings from the
UK Millennium Cohort Study. BMJ Open. 2017;7:e012868.
3. Nicholson JM, Lucas N, Berthelsen D, et al. Socioeconomic inequal-
ity profiles in physical and developmental health from 0-7 years:
Australian National Study. J Epidemiol Community Health.
2012;66:81–87.
4. Goldfeld S, O’Connor M, Chong S, et al. The impact of multidimen-
sional disadvantage over childhood on developmental outcomes in
Australia. Int J Epidemiol. 2018;47:1485–1496.
5. Evans GW, Li D, Whipple SS. Cumulative risk and child develop-
ment. Psychol Bull. 2013;139:1342–1396.
6. Shonkoff JP, Garner AS, et al. The lifelong effects of early child-
hood adversity and toxic stress. Pediatrics. 2012;129:e232–e246.
7. McEwen BS, Gianaros PJ. Central role of the brain in stress and
adaptation: links to socioeconomic status, health, and disease. Ann N
Y Acad Sci. 2010;1186:190–222.
8. Bernard K, Frost A, Bennett CB, et al. Maltreatment and diurnal
cortisol regulation: a meta-analysis. Psychoneuroendocrinology.
2017;78:57–67.
9. Alink LR, van Ijzendoorn MH, Bakermans-Kranenburg MJ, et al.
Cortisol and externalizing behavior in children and adolescents:
mixed meta-analytic evidence for the inverse relation of basal corti-
sol and cortisol reactivity with externalizing behavior. Dev Psycho-
biol. 2008;50:427–450.
10. Incollingo Rodriguez AC, Epel ES, White ML, et al. Hypothalamic-
pituitary-adrenal axis dysregulation and cortisol activity in obesity:
a systematic review. Psychoneuroendocrinology. 2015;62:301–318.
11. Saridjan NS, Velders FP, Jaddoe VW, et al. The longitudinal
association of the diurnal cortisol rhythm with internalizing and
externalizing problems in pre-schoolers. The Generation R Study.
Psychoneuroendocrinology. 2014;50:118–129.
 ARTICLE IN PRESS
TAGEDEN10 BRYSON ET AL
12. Evans GW, Kim P. Childhood poverty and health: cumulative risk
exposure and stress dysregulation. Psychol Sci. 2007;18:953–957.
13. Miller GE, Chen E, Zhou ES. If it goes up, must it come down?
Chronic stress and the hypothalamic-pituitary-adrenocortical axis in
humans. Psychol Bull. 2007;133:25–45.
14. Trickett PK, Noll JG, Susman EJ, et al. Attenuation of cortisol
across development for victims of sexual abuse. Dev Psychopathol.
2010;22:165–175.
15. Gunnar MR, Quevedo K. The neurobiology of stress and develop-
ment. Annu Rev Psychol. 2007;58:145–173.
16. Blair C, Raver CC, Granger D, et al. Allostasis and allostatic load in
the context of poverty in early childhood. Dev Psychopathol.
2011;23:845–857.
17. Gray NA, Dhana A, Van Der Vyver L, et al. Determinants of hair
cortisol concentration in children: a systematic review. Psychoneur-
oendocrinology. 2018;87:204–214.
18. Stalder T, Kirschbaum C. Analysis of cortisol in hair—state of the
art and future directions. Brain Behav Immun. 2012;26:1019–1029.
19. Stalder T, Steudte-Schmiedgen S, Alexander N, et al. Stress-related
and basic determinants of hair cortisol in humans: a meta-analysis.
Psychoneuroendocrinology. 2017;77:261–274.
20. Ursache A, Merz EC, Melvin S, et al. Socioeconomic status, hair
cortisol and internalizing symptoms in parents and children. Psycho-
neuroendocrinology. 2017;78:142–150.
21. White LO, Ising M, von Klitzing K, et al. Reduced hair cortisol after
maltreatment mediates externalizing symptoms in middle childhood
and adolescence. J Child Psychol Psychiatry. 2017;58:998–1007.
22. Simmons JG, Badcock PB, Whittle SL, et al. The lifetime experi-
ence of traumatic events is associated with hair cortisol concentra-
tions in community-based children. Psychoneuroendocrinology.
2016;63:276–281.
23. Rippe RC, Noppe G, Windhorst DA, et al. Splitting hair for cortisol?
Associations of socio-economic status, ethnicity, hair color, gender
and other child characteristics with hair cortisol and cortisone. Psy-
choneuroendocrinology. 2016;66:56–64.
24. Goldfeld S, Price A, Bryson H, et al. ‘right@home’: a randomised
controlled trial of sustained nurse home visiting from pregnancy to
child age 2 years, versus usual care, to improve parent care, parent
responsivity and the home learning environment at 2 years. BMJ
Open. 2017;7:e013307.
25. Kiernan KE, Mensah FK. Poverty, maternal depression, family sta-
tus and children’s cognitive and behavioural development in early
childhood: a longitudinal study. J Social Policy. 2009;38:569–588.
ACADEMIC PEDIATRICS
26. Goodman R. Psychometric properties of the Strengths and Difficul-
ties Questionnaire. J Am Acad Child Adolesc Psychiatr.
2001;40:1337–1345.
27. Varni JW, Limbers CA, Burwinkle TM. Parent proxy-report of their
children’s health-related quality of life: an analysis of 13,878 parents’
reliability and validity across age subgroups using the PedsQL 4.0
Generic Core Scales. Health Qual Life Outcomes. 2007;5:2.
28. Cole TJ, Bellizzi MC, Flegal KM, et al. Establishing a standard defi-
nition for child overweight and obesity worldwide: international sur-
vey. BMJ. 2000;320:1240–1243.
29. Mackinnon DP, Dwyer JH. Estimating mediated effects in preven-
tion studies. Eval Rev. 1993;17:144–158.
30. Sobel ME. Asymptotic confidence intervals for indirect effects in
structural equation models. Sociol Methodol. 1982;13:290–312.
31. Windhorst DA, Rippe RC, Mileva-Seitz VR, et al. Mild perinatal
adversities moderate the association between maternal harsh parent-
ing and hair cortisol: evidence for differential susceptibility. Dev
Psychobiol. 2017;59:324–337.
32. Noppe G, van den Akker EL, de Rijke YB, et al. Long-term gluco-
corticoid concentrations as a risk factor for childhood obesity and
adverse body-fat distribution. Int J Obes. 2016;40:1503–1509.
33. Veldhorst MA, Noppe G, Jongejan MH, et al. Increased scalp hair
cortisol concentrations in obese children. J Clin Endocrinol Metab.
2014;99:285–290.
34. Larsen SC, Fahrenkrug J, Olsen NJ, et al. Association between hair
cortisol concentration and adiposity measures among children and
parents from the “Healthy Start” study. PLoS One. 2016;11:
e0163639.
35. Richters JE. Depressed mothers as informants about their children:
a critical review of the evidence for distortion. Psychol Bull.
1992;112:485–499.
36. Australian Institute of Family Studies. The Longitudinal Study of
Australian Children. Available at: http://growingupinaustralia.gov.
au/. Accessed November 30, 2017.
37. Kemper KJ, Kelleher KJ. Family psychosocial screening: instru-
ments and techniques. Ambul Child Health. 1996;4:325–339.
38. Hegarty K. Composite Abuse Scale Manual. Melbourne, Australia:
Department of General Practice, University of Melbourne; 2007.
39. Lovibond SH, Lovibond PF. Manual for the Depression Anxiety Stress
Scales. 2nd ed. Sydney, Australia: Psychology Foundation; 1995.
40. Henry JD, Crawford JR. The short-form version of the Depression
Anxiety Stress Scales (DASS-21): construct validity and normative data
in a large non-clinical sample. Br J Clin Psychol. 2005;44:227–239.