Socioeconomic Disparities in Insulin Resistance: Results From the Princeton
School District Study
ELIZABETH GOODMAN, MD, STEPHEN R. DANIELS, MD, PHD, AND LAWRENCE M. DOLAN, MD
Objectives: The objectives of this study were to determine whether lower socioeconomic status (SES) is associated with changes
in insulin resistance in adolescents over a 3-year period and explore moderators of this effect. Methods: A total of 1167 healthy
non-Hispanic black and white participants in the Princeton School District Study, a longitudinal study of fifth to 12th graders in
a suburban Midwestern public school district were included in this study. Inclusion criteria were a) physical examination and fasting
morning blood draw at baseline and 3 years later, b) younger than 20 years old at follow up, and c) information available on SES
provided by a parent. The influence of SES on insulin resistance and change in insulin resistance over time was examined using
general linear models adjusting for multiple covariates. Models also assessed if race or baseline weight status changed the
SES–insulin resistance relationship and explored the role of perceived stress. Results: Blacks and lower SES youth had higher body
mass index z score and increased insulin resistance (p ? .001). In multivariable models, lower parent education, but not household
income, was associated with higher baseline insulin resistance (F ? 7.84, p ? .001) and worsening insulin resistance over time (F ?
18.86, p ? .001). Parent education’s effect on change in insulin resistance was more pronounced for obese youth compared with
nonobese (F interaction ? 10.12, p ? .001) even with adjustment for multiple covariates. Perceived stress did not alter these
relationships. Conclusions: Lower parent education appears to be related to increased insulin resistance both cross-sectionally and
over time in black and white adolescents. Worsening insulin resistance is especially problematic for obese adolescents from families
with low parent education. Key words: obesity, insulin resistance, disparities, race, SES.
BMI ? body mass index; CDC ? Centers for Disease Control and
Prevention; CV ? coefficient of variation; HDL-C ? high-density
lipoprotein cholesterol; HOMA ? homeostasis model assessment
score; PSD ? Princeton School District; SES ? socioeconomic
nsulin resistance is strongly associated with obesity and nu-
merous other morbidities (1). There has been increasing con-
cern over the past decade that rising rates of obesity will lead to
a marked increase in the frequency of type 2 diabetes, especially
among non-Hispanic black adolescents. Prevalence of obesity is
higher among non-Hispanic black youth compared with non-
more insulin-resistant (3). In addition to these physiological dif-
ferences between non-Hispanic black and white youth, there are
also social differences that place non-Hispanic black youth at
increased risk. Non-Hispanic black adolescents are more likely to
be from less well educated and lower income families than their
non-Hispanic white counterparts. These differences in social cir-
cumstances can also have deleterious effects on diabetes risk,
because coming from a lower socioeconomic status family has
been associated with both obesity and insulin resistance in ado-
lescence (4–6) and with obesity and type 2 diabetes in adulthood
Adolescence may be a particularly critical developmental
period vis-a `-vis the intersections between obesity, insulin re-
sistance, and social factors. Pubertal changes in insulin secretion
to predispose to obesity in adulthood (11–13), but the relation-
ships between obesity and insulin resistance in adolescence
and the tracking of these factors into adulthood is less clear.
Studies suggest that racial differences in body mass index
(BMI) tracking exist (14) as well as insulin secretion and
clearance during adolescence (15). However, the role of so-
cioeconomic status (SES) in these associations remains unex-
plored. Given the correlation between minority race/ethnicity
and low SES, this is an important gap in the literature.
To our knowledge, no longitudinal study has examined
socioeconomic disparities in insulin resistance and changes in
insulin resistance over time in a diverse cohort of youth. The
purpose of this study was twofold: first to determine if parent
education and household income, two commonly used indi-
cators of SES, influence both baseline insulin resistance and
changes in insulin resistance over time and second, to assess
if race/ethnicity or weight status altered the SES–insulin re-
sistance relationship. In addition, because the stress mediation
hypothesis is one of the leading theories on how social dis-
advantage becomes embodied to create health disparities (16–
18), the role perceived stress may play in any demonstrated
socioeconomic disparities in insulin resistance was explored.
The data for this study were drawn from the Princeton School District
(PSD) Study, a school-based longitudinal study (19,20). The PSD study
follows a cohort of students who were in the fifth to 12th grades (initial ages
12–19 years) in the 2001 to 2002 academic year. There were 4269 eligible
students when we went into the field in September 2001. Of these students,
2501 (58.6%) participated in year 1 of the PSD Study, eight of whom were
known to have diabetes. The cohort is 95% non-Hispanic black and white,
which reflects the demographics of the school district. However, non-His-
panic whites were slightly more likely to participate than non-Hispanic
blacks. This difference, although statistically significant, was small (60.3%
non-Hispanic white versus 58.2% non-Hispanic black, p ? .001). There was
From the Department of Pediatrics, Tufts-New England Medical Center
and Tufts University School of Medicine, Boston, MA (E.G.); Department of
Pediatrics, Denver Children’s Hospital, and the University of Colorado School of
Medicine, Denver, CO (S.R.D.); and the Division of Endocrinology, Cincinnati
Children’s Hospital Medical Center, Cincinnati, OH (L.M.D.).
Address correspondence and reprint requests to Elizabeth Goodman, MD,
Department of Pediatrics, Tufts-New England Medical Center, NEMC Box
351, 750 Washington Street, Boston, MA 02111. E-mail: egoodman@tufts-
Received for publication April 25, 2006; revision received July 24, 2006.
This work was supported by NIH grants HD41527, DK59183, and M01 RR
This work was presented, in part, at the Pediatric Academic Societies
Annual Meeting, April 30, 2006, San Francisco, CA.
61Psychosomatic Medicine 69:61–67 (2007)
Copyright © 2007 by the American Psychosomatic Society
also higher participation among females (62.7% versus 54.9%, p ? .01).
Mean grade among participants was 9.1 ? 1.6 compared with 9.3 ? 1.6
among nonparticipants (p ? nonsignificant). Thus, although there are some
statistically significant differences in participation, the cohort is quite repre-
sentative of the underlying population and brings a great deal of diversity.
Parental consent and student assent was obtained from all participants as previ-
ously described (21). The study was approved by the Institutional Review Boards
at both the local children’s hospital and the participating university.
These analyses are restricted to healthy non-Hispanic black and white
participants, because sample sizes were too small for inclusion of other
racial/ethnic groups. All healthy (nondiabetic) non-Hispanic black and white
participants, hereafter referred to as black and white, who were a) seen at
baseline and 3 years later for a follow-up visit; b) had data on insulin
resistance, lipids, and weight status from both time points; c) had a parent
provide information on parent education at baseline; and d) were less than
240.5 months of age at follow up were included in the analyses. The age
restriction allowed us to assess the influence of change in BMI z score based
on the Centers for Disease Control and Prevention (CDC) 2000 growth charts,
which include information for derivation of BMI z scores only up through age
240.5 months. A total of 1167 PSD study participants fulfilled inclusion
criteria. This is a larger, younger cohort than was included in an earlier
cross-sectional study of social inequalities in multiple cardiovascular risks,
including fasting insulin and glucose (5). The sample for the current analyses
was 46.7% non-Hispanic black and 19.0% were obese at baseline. Mean time
to follow up was 33.3 months with a standard deviation of 2.4 months.
The study visits, which took place in the school setting or local children’s
hospital after a verified overnight minimum 10-hour fast, included a physical
examination and venipuncture. Participants who were in the seventh through
12th grades at baseline (N ? 856) also completed a survey that assessed
SES was reported by a parent or guardian through a questionnaire dis-
tributed in the informed consent process. Completed parental surveys were
brought by students to the baseline visit along with completed consent forms.
If the participant did not have the parental survey, stamped self-addressed
envelopes were sent home for the parent to return the survey by mail. Parent
education for the reporting parent and his or her current spouse/partner was
obtained in categories ranging from never attended school to professional
training beyond a 4-year college or university. The highest level for either
parent was used in analyses. Analysis categories were high school or less,
some college or vocational training after high school, or college graduate or
higher. Household income was reported in nine ordered categories ranging
from less than $5000 to greater than $100,000. The midpoint of the range was
used in analyses. Because 13.5% (N ? 158) were missing data on household
income, multiple imputation was used to impute the missing values in
Date of birth, gender, and parent-identified race/ethnicity were available
from school records.
Body Mass Index
BMI was calculated from measured height and weight according to the
following equation: BMI ? weight (kg)/height (m2). Height and weight were
measured per a standardized protocol (19).
BMI z score and a dichotomous variable representing obesity were used
as adiposity measures. BMI z scores and percentiles were derived from CDC
2000 growth chart standards based on nationally representative data (22).
Obesity was defined as a BMI-for-age greater at or above the 95% based on
the CDC Growth Chart standards.
Stage of Pubertal Development
Pubertal status (prepubertal, pubertal, and postpubertal) was assessed
using plasma estradiol concentration and the presence or absence of menarche
for 2 years in females and plasma-free testosterone concentration and the
stage of axillary hair in males per a validated protocol (20).
Plasma insulin concentration was measured by radioimmunoassay using
an antiinsulin serum raised in guinea pigs, 125I labeled insulin (Linco, St.
Louis, MO) as a standard and a double antibody method to separate bound
from free tracer. The sensitivity is 2 pM with intra- and interassay coefficients
of variation (CVs) of 5% and 8%, respectively. Glucose was measured by an
enzymatic method. Intra and interassay CVs are 1.2% and 1.6%, respectively.
Results from the glucose and insulin assays were used to derive insulin
resistance measured by the homeostasis model assessment scores (HOMA)
model (23). HOMA is calculated as [fasting insulin (mIU) ? fasting glucose
(mM/L)]/22.5. Lipid profiles were performed on the Hitachi 704. National
Cholesterol Education Program performance criteria for accuracy and preci-
sion are followed. Direct measurement of high-density lipoprotein cholesterol
(HDL-C) was made using the HDL C-plus kit from Roche (Boehringer
Mannheim) The intraassay CV is 1.3% and the interassay CV is 2.6%.
Triglycerides were measured using a single reagent system from Roche-
BMD. The intra- and interassay CVs were approximately 4%.
Stress was assessed with the Perceived Stress Scale, a measure of global
stress (24). This 14-item scale, which measures a person’s appraisal of how
stressful his or her life was during the past month, has been shown to be
reliable and valid in adolescents (24). Responses are provided on a 5-point
Likert scale ranging from never to very often and scores can range from 0 to
56. Cronbach alpha in the PSD cohort was 0.65.
Because some variables were not normally distributed, nonparametric
tests were used in bivariate analyses. ?2test was performed for categorical
variables, and Mann-Whitney U tests were used for continuous variables.
Correlations were assessed with Spearman’s rho. Multivariable analyses were
performed using general linear models. Baseline HOMA and triglycerides
were log transformed to improve normality of the distribution before inclu-
sion in the multivariable models. The multivariable analyses occurred in two
phases. Inclusion of variables in the model in both phases was theoretically
driven; automated stepwise selection procedures were not used. The first
phase assessed the influence of parent education and household income on
baseline HOMA adjusting for factors known to be associated with insulin
resistance (age, gender, race, baseline HDL-C, triglycerides, pubertal stage,
and weight status). Both BMI z score and the dichotomous variable repre-
senting obesity were included. The former assessed a linear relationship
between general adiposity and HOMA and the latter assessed a threshold
effect. In phase 2, change in HOMA was regressed on these same factors. In
addition, the model adjusted for change in BMI z score and baseline HOMA.
In both phase 1 and phase 2, two-way and three-way interactions were
explored with a particular interest in interactions among SES measures, race,
and weight status. In addition, we ran a parallel set of analyses that used
fasting insulin as the dependent variable and included adjustment for fasting
glucose. Results were nearly identical as those for models that used HOMA
as the dependent variable. Because HOMA is the widely used estimate of
insulin resistance, we report results of the HOMA models. Last, perceived
stress was added to the regression models to determine if addition of this
variable altered any demonstrated SES–insulin resistance relationships. Pa-
rameter estimates (?) and their standard errors (SE) are reported from the
general linear models.
E. GOODMAN et al.
62Psychosomatic Medicine 69:61–67 (2007)
Table 1 provides a description of the demographic and
clinical characteristics of the sample. Although there were no
racial differences in age, gender, or pubertal status, black–
white differences were present for all other characteristics.
Blacks had lower parent education, lower household income,
were more likely to be obese at both baseline and follow up,
and had higher BMI z score and HOMA at both time points
and were more stressed at baseline than whites. However,
blacks also had better lipid profiles at baseline with lower
triglycerides and higher HDL-C. Similarly, obese youth dif-
fered from nonobese youth in all characteristics except age,
gender, and perceived stress. Baseline obese youth had lower
parent education, lower household income, were more likely
to be postpubertal, were more insulin-resistant at both time
points, and had worse lipid profiles than nonobese youth.
Cumulative incidence of obesity was 7%, whereas 21.2% of
obese youth had become nonobese by follow up. SES mea-
sures were strongly correlated with each other (rho ? 0.57,
p ? .001) and weakly correlated with perceived stress (parent
education rho ? ?0.18, p ? .001; income rho ? ?0.17, p ?
.001). Perceived stress was also weakly correlated with base-
line HOMA (rho ? 0.11, p ? .001) but was not associated
with change in HOMA (rho ? 0.02, p ? .49).
Socioeconomic Status Effects on Baseline
Although household income was not associated with base-
line HOMA, parental education had a strong effect on baseline
insulin resistance (Table 2). Lower parental education was
associated with progressively increased HOMA. Estimated
TABLE 1.Sociodemographic and Clinical Characteristics of the Study Sample
(N ? 1167)
(N ? 542)
(N ? 625)
(N ? 945)
Obese at Baseline
(N ? 222)
N PercentN PercentN Percentpa
High school or less
?High school, ?college
College or higher
Pubertal stage at baseline
Obese at baseline
Obese at follow up
Baseline age (years)
Homeostasis model assessment
Body mass index z score
Baseline high-density lipoprotein
Baseline Triglycerides (mg/dL)
Baseline perceived stress (N ? 854)
SOCIAL INEQUALITIES IN INSULIN RESISTANCE
63Psychosomatic Medicine 69:61–67 (2007)
marginal means in HOMA from the multivariable models for
the parent education categories were 4.35 for high school or
less parent education, 4.06 for more than high school but less
than college, and 3.60 for college graduate or higher. The
regression model, which accounted for 35% of the variance in
baseline HOMA, also demonstrates that females, blacks,
obese subjects, those who are pubertal, or have higher trig-
lycerides are more insulin-resistant. No interactions were
demonstrated. Addition of perceived stress to the model did
not alter these results, and perceived stress was not associated
with HOMA (?stress ? 0.005, SE ? 0.003, p ? .14).
Socioeconomic Status Effects on Change in Insulin
Resistance Over 3 Years
Results of longitudinal analyses to assess predictors of
change in HOMA are presented in Table 3. Because results of
the longitudinal models were nearly identical when log trans-
formed or untransformed HOMA was used, we present results
of the model using the difference in untransformed HOMA,
which is more straightforward. Overall, the regression model
accounted for 36% of the variance in change in HOMA.
Lower parental education, but not lower household income,
was associated with worse insulin resistance (F ? 18.86, p ?
.001). This was especially true for obese youth as evidenced
by the baseline obesity by parent education interaction (F ?
10.12, p ? .001). The effect was most pronounced for those
who came from families in which the highest level of parental
education was high school or less (? interaction ? 3.66, p ?
.001). No interaction was noted between race/ethnicity and
parent education, suggesting that the influence of parent edu-
cation is not different between non-Hispanic black and white
teens. No three-way interactions were noted. Like in the
baseline analyses, addition of perceived stress did not alter
these relationships (?stress ? 0.016, SE ? 0.024, p ? .49).
Figure 1 illustrates the effect of parent education and the
interaction with baseline obesity using data derived from the
multivariable longitudinal model. At all levels of parent edu-
Score From Multivariable General Linear Model Analysesa
Determinants of Baseline Homeostasis Model Assessment
ParameterB Standard Errorp
Body mass index z score
High school or less (E1)
College or higher (E3)
Household income ($)
aModel R2? 0.35 (adjusted R2? 0.35).
bNatural log transformed.
Assessment Score (HOMA) During 3-Year Follow-Up Period in 1167
Princeton School District Study Participants Based on General Linear
Determinants of Change in Homeostasis Model
ParameterB Standard Errorp
Baseline age (years)
Months of follow up
Pubertal stage at baseline
Baseline high-density lipoprotein
Baseline BMI z score
Baseline BMI z score ? HOMA
Change in BMI z score
High school or less (E1)
?High school, ?college (E2)
College or higher (E3)
Obese ? parent education
Obese ? E1
Obese ? E2
Obese ? E3
Household income ($)
aModel R2? 0.38 (adjusted R2? 0.36).
bNatural log transformed.
BMI ? body mass index.
assessment score (HOMA) over 3 years. From multivariable general linear
model adjusting for age, gender, race, months of follow up, baseline pubertal
stage, baseline household income, baseline high-density lipoprotein choles-
terol and triglycerides, baseline body mass index z score, baseline HOMA,
and change in body mass index z score. Closed circles ? nonobese; open
circles ? obese.
Influence of parent education on change in homeostasis model
E. GOODMAN et al.
64Psychosomatic Medicine 69:61–67 (2007)
cation, insulin resistance was higher for obese youth for whom
the slope of the line for parent education is much steeper than
for the nonobese. Over the 3 years, on average, insulin resis-
tance decreased among all nonobese adolescents as evidenced
by the negative estimated change in HOMA for at each level
of parent education. This is likely the result of resolution of
pubertal insulin resistance. The gradient effect is shown by the
increasingly negative magnitude of the estimated change as
parent education increased. Thus, lower education appears to
be associated with less of a decline in insulin resistance than
expected during the late pubertal and early postpubertal years.
In contrast, for obese youth, insulin resistance decreased only
for those from the most highly educated families. For obese
youth from less well educated families, the average change in
insulin resistance was positive reflecting worsening insulin
resistance. The increase among obese youth from parents with
high school or less education was especially marked.
The longitudinal model has a number of other notable
findings. Baseline HOMA was an important determinant of
change in insulin resistance over time. The negative sign of
this parameter estimate reflects the fall in insulin resistance,
which occurs as young people transition through puberty.
However, this decrease is mitigated by increasing adiposity as
demonstrated by the interaction between baseline BMI z score
and baseline HOMA. Baseline BMI z score itself was not
predictive of change in HOMA. This lack of significance may
be because the dichotomous variable representing baseline
obesity captured most of the effect of baseline BMI z score. In
addition, although baseline BMI z score was not significant,
change in BMI z score was strongly associated with change in
HOMA. The positive sign of the parameter estimate indicates
that loss of adiposity was related to decreasing insulin resis-
tance, whereas gain in adiposity was related to worsening
insulin resistance independent of baseline values.
There were some other notable similarities and differences
between the baseline and longitudinal models. Like in the
baseline model, higher baseline triglycerides and black race
were related to worsening HOMA. However, age, female
gender, and puberty were not predictive of change in HOMA.
The direction of the pubertal effect was in the correct direction
(baseline pubertal youth had decreasing insulin resistance as
they aged consistent with resolving pubertal insulin resis-
tance). However, the estimate did not reach statistical signif-
icance. This was also true for the relationship of lower HDL-C
to higher insulin resistance. The parameter estimates did not
reach statistical significance in either baseline (p ? .07) or
longitudinal (p ? .051) models.
This study has two major findings. First, among non-
Hispanic black and white teens, lower parent education, but
not household income, was related to increased insulin resis-
tance both cross-sectionally and over time; these effects were
independent of race/ethnicity. Second, the influence of lower
parent education was especially marked for obese youth. For
those whose parent(s) had a high school degree or less edu-
cation, estimated mean change in HOMA was fourfold higher
for obese youth compared with nonobese youth. For those
whose parent(s) had between a high school and college de-
gree, there was approximately a 1.5-fold difference. These
data suggest that the influence of parent education on insulin
resistance is profound and sustained, especially because lower
parent education has been associated with obesity in adoles-
cence (4,25). In addition, this study provides preliminary
evidence that the influence of parent education on insulin
resistance is not mediated through the conscious appraisal of
the environment as stressful. However, because perceived
stress does not correlate well with physiological stress,
chronic adaptation to a more challenging environment and the
resultant wear and tear on regulatory systems may still play a
role in the development of these disparities (17).
The strong relationship between parent education and in-
sulin resistance provides a striking contrast to the lack of
association between household income and insulin resistance.
Cross-sectional studies of Danish youth have also documented
the lack of association between income and insulin resistance
in multivariable models, including parent education, which
itself, has a robust effect (6). These differential effects of
income and education highlight the need to carefully consider
the mechanisms underlying social inequalities in health. Par-
ent education, which is likely to be a stable factor in the lives
of youth, may exert its effect though influences through psy-
chosocial factors such as the ability to adapt to one’s environ-
ment (26). Those from less well-educated families may have
fewer psychological reserves to cope with the difficulties
inherent in their environments and thus, may experience more
physiological stress as a result of living in a more challenging
setting (17,27). Recently, lower parent education has been
associated with decreased optimism among youth, and this
decreased optimism partially mediated the influence of lower
parent education on perceived stress (28). Because insulin is
sensitive to signals along the hypothalamic–pituitary–adrenal
axis, processive stress-related responses secondary to living in
a lower status environment may explain increased insulin
resistance among youth from less well-educated families
(5,29). In contrast, income’s effect may relate more to material
goods, which may not have direct links to insulin signaling
pathways. Testing these potential mechanisms and others such
as intergenerational effects and the role of relevant health-
related behaviors (i.e., physical activity) is beyond the scope
of the current study.
There are some limitations to this study. First, because this
was a large epidemiological study, insulin resistance was
measured using the HOMA model rather than euglycemic
clamp. However, HOMA, which is widely used in epidemio-
logic studies of insulin resistance among youth, has been
shown to adequately estimate clamp studies (30). Second, we
could not assess whether regional differences exist in the
pattern of social inequalities, as has been suggested by Lawlor
et al. (6), because our data were derived from a single school
district. Balancing these limitations are the strengths of this
study—its nearly equal representation of non-Hispanic black
SOCIAL INEQUALITIES IN INSULIN RESISTANCE
65Psychosomatic Medicine 69:61–67 (2007)
and white youth from a wide range of socioeconomic back-
grounds, its prospective design, parental report of SES, the
careful measurement of multiple physiological parameters,
and use of BMI z score, which accounts for normal growth
and development as a measure of adiposity rather than BMI.
Although insulin resistance is associated with type 2 dia-
betes, metabolic syndrome, and multiple other morbidities,
little attention has been paid to the socioeconomic patterning
of this hormonal risk, especially in adolescence. Cross-sec-
tional studies of adolescents have differed on the direction of
this relationship. In both the United States and Denmark,
lower SES has been associated increased insulin resistance,
whereas higher SES has been associated increased insulin in
Estonia and Portugal (5,6). To our knowledge, this is the first
prospective study to explore whether the influence of socio-
economic status on insulin resistance. This study identifies
insulin as a potential key hormone mediating the development
of health disparities, thereby adding to a growing literature,
which suggests that physiological mechanisms may underlie
social inequalities in health (5,31–33).
Although the socioeconomic patterning of insulin resis-
tance has not been well studied, racial/ethnic differences in
insulin resistance are well described. Non-Hispanic blacks are
generally considered to be more insulin-resistant than non-
Hispanic whites (3,34–37). However, the intersections be-
tween race and SES in regard to insulin regulation have
received little attention. In this study, both non-Hispanic black
race/ethnicity and lower parent education were independently
associated with increased insulin resistance at baseline,
whereas only parent education influenced change in insulin
resistance over time. This highlights the importance of longi-
tudinal studies. These data suggest that, given the same level
of insulin resistance at baseline, black and white adolescents
do not differ in the degree of change over time, whereas those
with lower parent education, especially obese youth with less
educated parent(s), fared worse than youth from better edu-
cated families. These findings have important implications for
understanding risk trajectories among youth. They suggest
that interventions aimed at obese adolescents from poorly
educated families may be the most effective in changing risk
for morbidities associated with sustained or worsening insulin
We thank the students, parents, teachers, administration, and staff of
the Princeton City School district and the PSD study staff.
1. Reaven GM. Pathophysiology of insulin resistance in human disease.
Physiol Rev 1995;75:473–86.
2. Hedley AA, Ogden CL, Johnson CL, Carroll MD, Curtin LR, Flegal KM.
Prevalence of overweight and obesity among US children, adolescents,
and adults, 1999–2002. JAMA 2004;291:2847–50.
3. Gower BA, Nagy TR, Goran MI. Visceral fat, insulin sensitivity, and
lipids in prepubertal children. [erratum appears in Diabetes 2001;50:
477–8]. Diabetes 1999;48:1515–21.
4. Goodman E. The role of socioeconomic status gradients in explaining
differences in US adolescents’ health. Am J Public Health 1999;89:
5. Goodman E, McEwen BS, Huang B, Dolan LM, Adler NE. Social
inequalities in biomarkers of cardiovascular risk in adolescence. Psycho-
som Med 2005;67:9–15.
6. Lawlor DA, Harro M, Wedderkopp N, Andersen LB, Sardinha LB,
Riddoch CJ, Page AS, Anderssen SA, Froberg K, Stansbie D, Davey
Smith G. Association of socioeconomic position with insulin resistance
among children from Denmark, Estonia, and Portugal: cross sectional
study. BMJ 2005;331:183.
7. Haffner SM: Epidemiology of type 2 diabetes: risk factors. Diabetes Care
8. Lawlor DA, Ebrahim S, Davey Smith G. Socioeconomic position in
childhood and adulthood and insulin resistance: cross sectional survey
using data from British women’s heart and health study. BMJ 2002;325:
9. Power C, Manor O, Matthews S. Child to adult socioeconomic conditions
and obesity in a national cohort. Int J Obes Relat Metab Disord 2003;
10. Caprio S, Plewe G, Diamond MP, Simonson DC, Boulware SD, Sherwin
RS, Tamborlane WV. Increased insulin secretion in puberty: a compen-
satory response to reductions in insulin sensitivity. J Pediatr 1989;114:
11. Freedman DS, Khan LK, Serdula MK, Dietz WH, Srinivasan SR, Berenson
GS. The relation of childhood BMI to adult adiposity: the Bogalusa Heart
Study. Pediatrics 2005;115:22–7.
12. Speiser PW, Rudolf MC, Anhalt H, Camacho-Hubner C, Chiarelli F,
Eliakim A, Freemark M, Gruters A, Hershkovitz E, Iughetti L, Krude H,
Latzer Y, Lustig RH, Pescovitz OH, Pinhas-Hamiel O, Rogol AD,
Shalitin S, Sultan C, Stein D, Vardi P, Werther GA, Zadik Z, Zuckerman-
Levin N, Hochberg Z. Childhood obesity. J Clin Endocrinol Metab
13. Steinberger J, Moran A, Hong CP, Jacobs DR Jr, Sinaiko AR. Adiposity
in childhood predicts obesity and insulin resistance in young adulthood.
J Pediatr 2001;138:469–73.
14. Freedman DS, Khan LK, Serdula MK, Dietz WH, Srinivasan SR, Berenson
GS. Racial differences in the tracking of childhood BMI to adulthood. Obes
15. Jiang X, Srinivasan SR, Radhakrishnamurthy B, Dalferes ER, Berenson
GS. Racial (black–white) differences in insulin secretion and clearance in
adolescents: the Bogalusa Heart Study. Pediatrics 1996;97:357–60.
16. Adler NE, Ostrove JM. Socioeconomic status and health: what we know
and what we don’t. Ann NY Acad Sci 1999;896:3–15.
17. McEwen BS. From molecules to mind: stress, individual differences, and
the social environment. Ann NY Acad Sci 2001;935:42–9.
18. Wilkinson RG. Health, hierarchy, and social anxiety. Ann NY Acad Sci
19. Goodman E, Adler NE, Daniels SR, Morrison JA, Slap GB, Dolan LM.
Impact of objective and subjective social status on obesity in a biracial
cohort of adolescents. Obes Res 2003;11:1018–26.
20. Dolan LM, Bean J, D’Alessio D, Cohen RM, Morrison JA, Goodman E,
Daniels SR. Frequency of abnormal carbohydrate metabolism and dia-
betes in a population-based screening of adolescents. J Pediatr 2005;146:
21. Cline A, Schafer-Kalkhoff T, Strickland E, Hamann T. Recruitment
strategies for the Princeton (Ohio) city school district epidemiologic
study. J Sch Health 2005;57:189–91.
22. CDC Growth Charts: United States, vol 2001. National Center for Health
23. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF,
Turner RC. Homeostasis model assessment: insulin resistance and beta-
cell function from fasting plasma glucose and insulin concentrations in
man. Diabetologia 1985;28:412–9.
24. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived
stress. J Health Soc Behav 1983;24:385–96.
25. Goodman E, Slap GB, Huang B. The public health impact of socioeco-
nomic status on adolescent depression and obesity. Am J Public Health
26. Lynch JW, Smith GD, Kaplan GA, House JS. Income inequality and
mortality: importance to health of individual income, psychosocial envi-
ronment, or material conditions. BMJ 2000;320:1200–4.
27. Gallo LC, Matthews KA. Understanding the association between socio-
economic status and physical health: do negative emotions play a role?
Psychol Bull 2003;129:10–51.
28. Finkelstein D, Kubzansky LD, Goodman E. Socioeconomic differences
in adolescent stress: the role of psychological resources. J Adolesc Health
E. GOODMAN et al.
66 Psychosomatic Medicine 69:61–67 (2007)
29. GoodmanE,McEwenBS,DolanLM,Schafer-KalkhoffT,AdlerNE.Social Download full-text
disadvantage and adolescent stress. J Adolesc Health 2005;37:484–92.
30. Gungor N, Saad R, Janosky J, Arslanian S. Validation of surrogate
estimates of insulin sensitivity and insulin secretion in children and
adolescents. J Pediatr 2004;144:47–55.
31. Williams RB. Lower socioeconomic status and increased mortality: early
childhood roots and the potential for successful interventions. JAMA
32. McEwen BS, Seeman T. Protective and damaging effects of mediators of
stress. Elaborating and testing the concepts of allostasis and allostatic
load. Ann NY Acad Sci 1999;896:30–47.
33. Brunner EJ, Marmot MG, Nanchahal K, Shipley MJ, Stansfeld SA,
Juneja M, Alberti KG. Social inequality in coronary risk: central obesity
and the metabolic syndrome. Evidence from the Whitehall II study.
34. Haffner SM, D’Agostino R, Saad MF, Rewers M, Mykkanen L, Selby J,
Howard G, Savage PJ, Hamman RF, Wagenknecht LE, Bergman RN.
Increased insulin resistance and insulin secretion in nondiabetic African-
Americans and Hispanics compared with non-Hispanic whites. The In-
sulin Resistance Atherosclerosis Study. Diabetes 1996;45:742–8.
35. Tershakovec AM, Kuppler KM, Zemel BS, Katz L, Weinzimer S,
Harty MP, Stallings VA. Body composition and metabolic factors in
obese children and adolescents. Int J Obes Relat Metab Disord 2003;
36. Ku CY, Gower BA, Hunter GR, Goran MI. Racial differences in insulin
secretion and sensitivity in prepubertal children: role of physical fitness
and physical activity. Obes Res 2000;8:506–15.
37. Arslanian S, Suprasongsin C, Janosky JE. Insulin secretion and sensitiv-
ity in black versus white prepubertal healthy children. J Clin Endocrinol
SOCIAL INEQUALITIES IN INSULIN RESISTANCE
67 Psychosomatic Medicine 69:61–67 (2007)