Longitudinal Study of Depressive Symptoms and Health-Related Quality of Life During Pregnancy and...

Post on 04-May-2023

0 views 0 download

Transcript of Longitudinal Study of Depressive Symptoms and Health-Related Quality of Life During Pregnancy and...

Longitudinal Study of DepressiveSymptoms and Progression of InsulinResistance in Youth At-Risk for AdultObesityLAUREN B. SHOMAKER, PHD

1,2

MARIAN TANOFSKY-KRAFF, PHD1,2

ELIZABETH A. STERN, BA1

RACHEL MILLER, MA2

JACLYN M. ZOCCA, BA1

SARA E. FIELD, BA2

SUSAN Z. YANOVSKI, MD3

VAN S. HUBBARD, MD, PHD4

JACK A. YANOVSKI, MD, PHD1

OBJECTIVEdThe purpose of this study was to determine whether having childhood de-pressive symptoms is a risk factor that prospectively predicts impairment in glucose homeostasis.

RESEARCH DESIGN ANDMETHODSdA non–treatment-seeking sample of 115 chil-dren (aged 5–13 years), oversampled for being at risk for adult obesity, was assessed at baselineand again ~6 years later. Children self-reported depressive symptoms using the Children’s De-pression Inventory at baseline. Insulin resistance was assessed at baseline and follow-up with thehomeostasis model assessment of insulin resistance index (HOMA-IR).

RESULTSdChildren’s depressive symptomswere a significant predictor of follow-upHOMA-IR,fasting insulin, and fasting glucose in models accounting for baseline HOMA-IR, insulin, or glucosevalues; sex; race; baseline age; baseline BMI; change in BMI at follow-up; family history of type 2diabetes; and time in the study (P, 0.01).

CONCLUSIONSdIn this study, depressive symptomatology at baseline predicted the pro-gression of insulin resistance during child and adolescent development independent of changesin BMI. Research is needed to determine whether early intervention to decrease elevated de-pressive symptoms in youth ameliorates later development of insulin resistance and lessens therisk of type 2 diabetes.

Worsening insulin resistance, inconcert with impaired insulin se-cretion, is a major physiological

precursor of type 2 diabetes. Insulin re-sistance often develops through excessivegain of body weight or adiposity. Amongyouth, those who are obese (BMI [kg/m2]$95th percentile) have greater insulin re-sistance than their nonobese peers.

Psychological factors also may play arole in the development of insulin re-sistance and consequent type 2 diabetes.In particular, evidence from adult studies

suggests that symptoms of depressionmay be a risk factor for insulin resistanceand type 2 diabetes, independent of BMI(1). Depressive symptoms and perceivedpsychosocial stress are positively associ-ated with insulin resistance in women,even after accounting for body weight(2,3). Furthermore, in adult women andmen, elevated depressive symptoms orthe presence of major depressive disor-ders predict type 2 diabetes onset afteraccounting for measures of body weightor adiposity (4–8). A meta-analysis of

prospective adult studies concluded thatelevated depressive symptoms were asso-ciated with a 37% increased risk of devel-oping type 2 diabetes at a later point intime, and this effect did not seem to besignificantly diminished by confounderssuch as BMI (9). Although themechanismsthat explain depression’s impact on insulinresistance are not well understood, de-pressive symptoms may promote worsen-ing insulin resistance through affectingbehavioral patterns, such as increasedtotal energy or carbohydrate intake andlowered voluntary energy expenditure.Depressive symptoms also may activateunderlying physiological stress pathways,such as distress-induced upregulation ofcounterregulatory hormone systems likethe hypothalamic-pituitary-adrenal axis,sympathetic nervous system activation, al-teration of insulin signaling in the brain,and/or an increase in proinflammatoryfactors (10).

Despite the strong evidence linkingdepressive symptoms and insulin resis-tance or type 2 diabetes in adults, thereare limited data on the relationship be-tween the symptoms of depression andinsulin resistance during childhood andadolescent development. Given the pos-sible role that adult studies suggest fordepression in the development of type 2diabetes, understanding how depressivesymptoms are linked to insulin resistanceduring childhood and adolescence maybe important for identifying early risks forpotentially adverse health outcomes. Ina large community study of youth aged6–18 years, parents’ perceptions of theirchildren’s negative emotionality were as-sociated cross-sectionally with children’sfasting insulin, particularly among boys(11). Among healthy nonoverweight andoverweight adolescents aged 12–17 years,depressive symptoms were positively re-lated to poorer insulin sensitivity cross-sectionally, even after adjusting for bodycomposition (12). Depressive symptomsalso were associated cross-sectionallywith higher fasting insulin among over-weight middle-school children with a

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the 1Section on Growth and Obesity, Program in Developmental Endocrinology and Genetics, EuniceKennedy Shriver National Institute of ChildHealth andHumanDevelopment, National Institutes of Health,Bethesda,Maryland; the 2Department ofMedical and Clinical Psychology, Uniformed Services University ofthe Health Sciences, Bethesda, Maryland; the 3Division of Digestive Diseases and Nutrition, National In-stitute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland;and the 4Division of Nutrition Research Coordination, National Institutes of Health, Bethesda, Maryland.

Corresponding author: Jack A. Yanovski, jy15i@nih.gov.Received 15 June 2011 and accepted 5 August 2011.DOI: 10.2337/dc11-1131. Clinical trial reg. no. NCT00001522, clinicaltrials.gov.© 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited, the use is educational and not for profit, and thework is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

care.diabetesjournals.org DIABETES CARE 1

C a r d i o v a s c u l a r a n d M e t a b o l i c R i s kO R I G I N A L A R T I C L E

Diabetes Care Publish Ahead of Print, published online September 12, 2011

family history of type 2 diabetes (13).However, it remains to be determinedwhether depressive symptoms are pro-spectively related to worsening progres-sion of insulin resistance among youth.

We therefore sought to evaluatewhether children’s depressive symptomspredicted worsening insulin resistanceover time. Understanding the link be-tween depressive symptoms and insulinresistance during the transition fromchildhood to adolescence is important.Normative endocrine changes associatedwith the onset of puberty produce in-creased insulin resistance relative tochildhood. Moreover, entry into adoles-cence marks a peak time for increases inelevated depressive symptoms. Thesemarked biological and psychologicalshifts render the transition from child-hood to adolescence an important devel-opmental window for understanding howsymptoms of depression are associatedwith insulin resistance. On the basis ofadult longitudinal data (9) and cross-sectional studies in youth (12,13), we hy-pothesized that children’s depressivesymptoms at a baseline assessment wouldbe related to greater insulin resistancewhen followed-up in adolescence, inde-pendent of initial BMI and BMI changeover time.

RESEARCH DESIGN ANDMETHODSdParticipants were a non–treatment-seeking sample of children tak-ing part in a longitudinal study designed toinvestigate risk factors for excessive weightgain and adversemetabolic outcomes (clin-ical trial reg. nos. NCT00001522 andNCT00001195, clinicaltrials.gov) be-tween July 1996 and November 2010.By design, the sample was oversampledto include children at increased risk forthe development of adult obesity by virtueof either the child’s overweight status (BMI$85th percentile) or a parental history ofbeing overweight (BMI $25 kg/m2). Par-ticipants were recruited through mailingsto pediatricians, family physicians, andtwo waves of notices to families with ele-mentary school–aged youth in Marylandand the Washington, DC, metropolitanarea. Advertisements requested the partici-pation of children willing to undergo phle-botomy and X rays for studies investigatinghormones and metabolic functioning inchildren. All recruitment materials speci-fied that no treatment would be offered.Approximately 7% of families respondedto each of the school mailings, and sub-jects recruited directly from these mailings

constituted 88% of all subjects studied.Youth were in good general health andwere not taking medications known to af-fect body weight or metabolism for at least2 weeks prior to study entry. Exclusioncriteria included a significant medicalhealth problem, such as renal, hepatic,most endocrinologic (e.g., hyperthyroid-ism, Cushing syndrome, or type 2 diabe-tes), or pulmonary disorders (other thanmild asthma not requiring chronic medi-cation) or major psychiatric illnesses re-quiring treatment. Children providedwritten assent and parents gave writtenconsent to participate. Children were fi-nancially compensated for their participa-tion at baseline ($120) and follow-up($120). All study procedures were ap-proved by the Eunice Kennedy ShriverNational Institute of Child Health and Hu-man Development Institutional ReviewBoard.

AssessmentAt a baseline assessment visit, participantsunderwent a physical examination thatincluded pubertal staging by a pediatricendocrinologist or trained pediatric orfamily nurse practitioner. Parents repor-ted the child’s and the family medical his-tory, including parental height andweightand family history of type 2diabetes. Child-ren’s height and weight measurementswere obtained after an overnight fast. Par-ticipants were clothed but with shoes re-moved. Each child’s height was measuredthree times to the nearest millimeter by astadiometer (Holtain, Crymmych, Wales,U.K.) that was calibrated before eachchild’s measurement. Weight was mea-sured to the nearest 0.1 kg with a calibra-ted digital scale (Scale-Tronix, Wheaton,IL). Participants’ height and weight wereused to compute BMI (kg/m2). BMI z (SD)scores for sex and age were calculated ac-cording to the Centers for Disease Controland Prevention 2000 standards.

Participants completed the 27-itemChildren’s Depression Inventory (CDI) toassess the extent and severity of depressivesymptoms (14). Younger children (aged#8 years) had the questions read to themso that any difficult concepts could be ex-plained. A total raw score, ranging from0 to 54, is derived from the sum of itsitems. A total score that exceeded 12 wasproposed as the cutoff for screening forat-risk for clinical depression (15). We ex-amined the CDI total score both as a contin-uous variable and as a dichotomous variable(lower depressive symptoms [CDI ,13]vs. elevated depressive symptoms [CDI

$13]). The CDI is reliable and well vali-dated (14), and the total score has demon-strated adequate internal reliability instudies oversampled to include youthaged#14 years who were at risk for adultobesity in our laboratory (a = 0.79). Anychild who endorsed active suicidal ideationwas referred for an immediate psychiatryconsultation at the National Institutes ofHealth Clinical Center and was excludedfrom participation in the current study.

Each participant provided fastingblood samples for serum insulin and glu-cose after an overnight fast. Children wereasked at least twice about caloric consump-tion before blood was drawn. A fasted statewas encouraged to be reported truthfullyby providing compensation for the visiteven if subjects reported not having fasted.Youth consumed their habitual diet (in-variably reported on a food-frequencyquestionnaire as$35% energy from carbo-hydrates) during the week before theywerestudied. Glucose was measured using aHitachi 917 analyzer using reagents fromRoche Diagnostics (Indianapolis, IN). In-sulin concentrations were determinedusing a commercially available immuno-chemiluminometric assay purchased fromDiagnostic Product Corporation (LosAngeles, CA) and calibrated against insulinreference preparation 66/304. The insulinassay uses a monoclonal anti-insulin anti-body and was run on an Immulite2000machine (Diagnostic Product Corpora-tion). The cross-reactivity of the insulinassay with proinsulin was ,8% and withC-peptide was ,1%, sensitivity was 2mU/mL, and the mean inter- and intra-assay coefficients of variation were 5.8 and3.6%, respectively. Insulin resistance wasestimated with the homeostasis model as-sessment of insulin resistance (HOMA-IR)index, calculated as follows: (fasting insu-lin [mU/mL]3 fasting glucose [mmol/L])/22.5. AlthoughHOMA-IR as a continuousvariable was considered the primaryoutcome measure in the current study,secondary outcomes included insulinresistance defined dichotomously (absence[HOMA-IR,3.16] vs. presence [HOMA-IR$3.16]), impaired fasting glucose (ab-sence [fasting glucose ,100 mg/dL] vs.presence [fasting glucose $100 mg/dL]),and hyperinsulinemia (fasting insulin$15 mU/L).

At a follow-up appointment intendedto take place 5 years later, participants’height and weight were reassessed to cal-culate BMI, as completed at baseline. Par-ticipants again provided fasting bloodsamples for serum insulin and glucose,

2 DIABETES CARE care.diabetesjournals.org

Depression and insulin resistance in youth

which were used to reassess HOMA-IR.For ~2% of participants, plasma glucose(collected in tubes containing powderedsodium fluoride) was used in place of se-rum glucose when the latter was not avail-able. For participants who did notcomplete a 5-year follow-up, available an-thropometric and phlebotomy data wereused either from a somewhat shorter orlonger follow-up period, whenever suchdata were available.

Statistical analysisAnalyses were conducted using SPSS18.0. Study variables were examined todetermine whether the assumptions ofunivariate and multivariate analyses weremet. The skew and kurtosis were satisfac-tory on all variables, and outliers (,3% ofall data points) were adjusted to fall 1.5times the interquartile range below orabove the 25th or 75th percentile. Thisstrategy was used because it minimizesoutliers’ influence on the characteristicsof the distribution, minimally changesthe distribution overall, and avoids poten-tial bias associated with eliminating outliersaltogether.Missingdata patternswere char-acterized to test baseline differences instudy variables between children whodid and those who did not completea follow-up assessment. Correlations orindependent-sample t tests were used toexamine the bivariate associations amongbaseline demographic and anthropometriccharacteristics, depressive symptoms, and

insulin resistance. Hierarchical multiple re-gressionswere conducted regressing follow-up insulin resistance on baseline depressivesymptoms, including in themodel baselineinsulin resistance, sex, race (non-Hispanicwhite vs. other), first- or second-degreefamily history of type 2 diabetes (presencevs. absence), baseline age (years), baselineBMI (kg/m2), change in BMI from baselineto follow-up, and time in study (years be-tween baseline and follow-up). Baselinepubertal staging was considered to be a co-variate but was removed because it wasnonsignificant in all models (P . 0.65).Parallel analyses also were conducted forthe secondary outcomes of fasting insulinand fasting glucose. ANCOVA was used totest whether similar relationships betweendepressive symptoms and insulin resis-tance, fasting insulin, or glucose were ob-served if depressive symptoms wereconsidered categorically as children withlower depressive symptoms (CDI totalscore ,13) versus elevated symptoms(CDI total score $13). Logistic regres-sions were conducted to test whether de-pressive symptoms predicted greater oddsof categorical outcomes of insulin resis-tance (absence [HOMA-IR ,3.16] vs.presence [HOMA-IR $3.16]), hyperin-sulinemia (fasting insulin $15 mU/L),and impaired fasting glucose (absence[fasting glucose ,100 mg/dL] vs. pres-ence [fasting glucose $100 mg/dL]). Par-allel sets of covariates were used in theseanalyses. Differences and associations

were considered significant when P valueswere # 0.05. All tests were two tailed.

As a result of a large missing datafraction, we also reran the primary anal-yses using multiple imputation with SAS18.0 to handle missing data. Ten imputeddatasets were produced (16). Followingstandard multiple imputation procedures,each dataset was analyzed separately, andthen the effects were combined using theSAS MIANALYZE procedure. Becausethese results did not significantly differfrom the nonimputed findings, resultsobtained in the nonimputed analyses arepresented.

RESULTSdA total of 198 children(51.9% female) aged 8.6 6 1.7 years(range 6–13) completed a baseline assess-ment visit. Three children were excludedbecause of medical illnesses identified attheir baseline visit. Fifty percent of partic-ipants were obese (BMI $95th percen-tile), 16% were overweight (BMI $85thand ,95th percentile), and 34% werenonoverweight (BMI ,85th percentile)but had at least one overweight parent.Demographic, anthropometric, and met-abolic characteristics of the study partic-ipants by depressive symptom status aredescribed in Table 1. Compared withchildren with lower depressive symptoms(n = 160), children with elevated depres-sive symptoms (n = 38) were slightly, butsignificantly, younger (aged 8.1 6 1.5

Table 1dParticipant characteristics at baseline assessment

Variable

Lower depressive symptoms Elevated depressive symptoms

PCDI total score ,13 CDI total score $13

n 160 38Sex (% female) 51.3 55.3 0.66Race (%)* 0.11Non-Hispanic black 30.6 47.4Non-Hispanic white 66.9 52.6Hispanic 2.5 0

Family history of type 2 diabetes (% presence) 58.5 55.9 0.79Baseline BMI (kg/m2) 22.0 6 5.9 (13.2–40.1) 23.7 6 7.8 (13.2–40.1) 0.21Baseline BMI z score 1.4 6 1.1 (21.9 to 3.2) 1.6 6 1.2 (21.9 to 3.4) 0.27Baseline fasting insulin (mU/L) 9.1 6 6.3 (1.0–26.1) 11.6 6 7.9 (2.0–26.1) 0.07Hyperinsulinemia (% presence)† 14.4 31.6 0.01Baseline fasting glucose (mg/dL) 88.1 6 6.7 (70.5–106.5) 89.1 6 6.3 (79.0–103.0) 0.38Impaired fasting glucose (% presence)‡ 3.8 5.3 0.67Baseline HOMA-IR 2.0 6 1.4 (0.2–5.9) 2.6 6 1.8 (0.4–5.9) 0.07Elevated HOMA-IR (% presence)x 16.9 34.2 0.02Data are means 6 SE (range), unless otherwise indicated. n = 176–198. *When race/ethnicity was defined dichotomously as non-Hispanic white vs. other, its as-sociation with depressive symptoms status remained nonsignificant (P = 0.10). †Hyperinsulinemia defined as fasting insulin $15 mU/L. ‡Impaired fasting glucosedefined as fasting glucose $100 mg/dL. xElevated HOMA-IR defined as $3.16.

care.diabetesjournals.org DIABETES CARE 3

Shomaker and Associates

years vs. 8.7 6 1.6 years, P = 0.03), weremore likely to have hyperinsulinemia(31.6 vs. 14.4%, P = 0.01), and weremore likely to have elevated insulin resis-tance (HOMA-IR$3.16; 34.2 vs. 16.9%,P = 0.02) at baseline. Fifty-eight percent ofchildren (n = 115) returned for a follow-upphlebotomy appointment an average of5.8 6 1.3 years (range 3.1–8.3) later.Mean age at follow-upwas 14.66 2.3 years(range 8.9–20.3). One child had confirmedtype 2 diabetes at follow-up. Youth whodid not complete a follow-up appointmenthad greater baseline depressive symptoms(8.46 6.5 vs. 6.86 5.2, P = 0.05), higherbaseline fasting insulin (11.0 6 7.7 vs.8.5 6 5.6, P = 0.01), and higher baselineinsulin resistance (2.56 1.8 vs. 1.96 1.3,P = 0.01) than youth who did return for afollow-up. There were no differences be-tween completers and noncompleters onany other variable.

Table 2 summarizes the analyses ex-amining baseline depressive symptomsas a predictor of follow-up insulin resis-tance, fasting insulin, and fasting glucose.Controlling for all other variables in themodels, children’s baseline depressivesymptoms predicted greater insulin resis-tance and higher fasting insulin and glu-cose at follow-up (P , 0.01). Baselinedepressive symptoms explained 8% ofthe variance in follow-up insulin resis-tance (P , 0.001), 7% of the variance infollow-up fasting insulin (P, 0.001), and4% of the variance in follow-up fasting

glucose (P = 0.019), after accounting forthe other variables in the model. An iden-tical pattern of results was observed whendepressive symptoms were consideredcategorically. Youth with elevated depres-sive symptoms at baseline (n = 17) hadhigher insulin resistance, fasting insulin,and fasting glucose at follow-up thanyouth with lower depressive symptomsat baseline (P # 0.001) (Fig. 1). As anexploratory analysis, we also testedwhether baseline insulin resistance pre-dicted follow-up depressive symptoms,accounting for baseline depressive symp-toms and a parallel set of covariates. In-sulin resistance was not a significantpredictor of follow-up depressive symp-toms (P = 0.99).

At follow-up, 43 (37.4%) childrenmet the criteria for elevated insulin re-sistance, 43 (37.4%) for hyperinsuline-mia, and 4 (3.8%) for impaired fastingglucose. Table 3 presents a summary ofanalyses examining baseline depressivesymptoms as a predictor of elevated insu-lin resistance, hyperinsulinemia, and im-paired fasting glucose. Accounting for allother variables in the model, each one-unit increase in CDI total score at baselinewas associated with a 1.14 greater odds ofelevated insulin resistance at follow-up(95%CI 1.01–1.28, P, 0.05). Depressivesymptoms did not significantly predictfollow-up hyperinsulinemia or impairedfasting glucose. When depressive symp-toms were considered categorically, those

with and without elevated depressivesymptoms at baseline did not significantlydiffer in their odds of elevated insulin re-sistance (P = 0.10), hyperinsulinemia (P =0.52), or impaired fasting glucose (P=0.98),adjusting for all of the same covariates.

CONCLUSIONSdThe current studyprovides evidence that children’s depres-sive symptoms are a prospective risk fac-tor for worsening insulin resistance. Evenwhen accounting for known additionalrisk factors, including family history oftype 2 diabetes, children’s baseline BMI,and changes in children’s BMI over time,depressive symptoms were associatedwith greater insulin resistance ~6 yearslater. Depressive symptoms’ impact on in-sulin resistance was clinically meaningfulsuch that depressive symptoms were as-sociated with a significantly greater likeli-hood of developing clinically elevatedHOMA-IR (defined as $3.16). Of note,children’s degree of insulin resistanceis a significant predictor of type 2 diabetesonset in young adulthood, even after ac-counting for BMI (17).

The current findings are consistentwith previous cross-sectional studiesdemonstrating a link between depressivesymptoms or negative affect and insulinresistance in youth independent of bodycomposition (12,13). Moreover, the pres-ent results are consistent with adult datademonstrating that depressive symptomsare related to greater odds of developing

Table 2dMultiple hierarchical regressions examining children’s baseline depressive symptoms as a predictor of follow-upinsulin resistance, fasting insulin, and fasting glucose

Predictor

Follow-up

HOMA-IR Fasting insulin Fasting glucose

B SE b B SE b B SE b

Sex (female) 0.87 0.31 0.22* 4.27 1.36 0.23* 22.26 1.35 20.15†Race (black or Hispanic) 20.48 0.32 20.12 22.25 1.40 20.12 21.67 1.39 20.11Family history of type 2 diabetes(present) 0.64 0.31 0.16‡ 3.23 1.37 0.17‡ 21.09 1.36 20.07

Baseline age (years) 20.24 0.10 20.20‡ 21.03 0.43 20.19‡ 21.04 0.43 20.23‡Baseline BMI (kg/m2) 0.04 0.04 0.11 0.17 0.18 0.12 20.13 0.12 20.10BMI change 0.09 0.04 0.22* 0.46 0.16 0.24* 20.11 0.16 20.07Time in study (years) 20.27 0.11 20.19‡ 21.28 0.50 20.20* 20.45 0.49 20.09Baseline HOMA-IR 0.52 0.17 0.40* d d d d d dBaseline fasting insulin (mU/L) d d d 0.59 0.17 0.43x d d dBaseline fasting glucose (mg/dL) d d d d d d 0.22 0.10 0.20‡

R2 = 0.37x R2 = 0.43 R2 = 0.11Baseline depressive symptoms 0.10 0.03 0.29x 0.44 0.11 0 0.28x 0.29 0.12 0.23*

R2 = 0.45x R2 = 0.50x R2 = 0.15‡DR2 = 0.08x DR2 = 0.07x DR2 = 0.04‡

n = 98–102. *P # 0.01. †P # 0.10. ‡P # 0.05. xP # 0.001.

4 DIABETES CARE care.diabetesjournals.org

Depression and insulin resistance in youth

type 2 diabetes (9). The mechanisms ex-plaining the relationship between depres-sive symptoms and insulin resistance arenot well understood. Symptoms of depres-sion, including fatigue, lack of energy, oranhedonia (referring to loss of pleasureover activities that one previously found en-joyable), may prompt behavioral decreasesin voluntary energy expenditures, such asexercise, which, in turn, may heighten in-sulin resistance. Consistent with this no-tion, adolescent depressive symptoms areassociatedwith poorer cardiorespiratoryfit-ness (18). Depressive symptoms also havebeen concurrently related to emotional eat-ing patterns (19), which possibly may pro-mote insulin resistance independent ofweight gain. From a neurohumoral frame-work, depressive symptoms are hypothe-sized to promote insulin resistance byupregulating cortisol and enhancing itsdownstream effects, including increasing

the production of theneurotransmitter neu-ropeptide Y (10,20,21).

Strengths of the current investigationinclude the longitudinal nature of data,the examination of depressive symptomsand insulin resistance in a sizeable sampleof children, and the adjustment for im-portant covariates, including measuredBMI and BMI change. Study limitationsinclude the use of surrogate measures forassessing insulin resistance and depres-sion. The measure of insulin resistancewas derived from fasting values, which,although highly related to clamp-derivedmeasures, is not considered as precise anassessment. Likewise, although the CDIis a widely used, reliable, and valid mea-sure of depressive symptoms, it does notprovide a diagnostic assessment of clinicaldepression. Future longitudinal studiesexamining the impact of depressive symp-toms on insulin resistance using criterion

measures are warranted. Another signifi-cant study shortcoming was the very highdegree of attrition that diminished the sam-ple size at follow-up. Although the greaterlikelihood of dropout among youth withgreater baseline depressive symptoms andhigher insulin resistance could be expectedto attenuate the significance of the results,this pattern, as well as the nature of thesample being oversampled to include youthat risk for adult obesity, may limit the gen-eralizability of the findings. In addition, theeffect of depressive symptoms on insulinresistance was small relative to the effect ofanthropometric variables. Youthwithmajordepression or active suicidal ideation wereexcluded from participation. Examinationof the depression-insulin resistance rela-tionship in samples of adolescents withclinically elevated symptomatology mayshed more light on the magnitude of thedepression-insulin relationship.

Figure 1dCompared with children with lower depressive symptoms at baseline (n = 97; CDI total score ,13), youth with elevated depressivesymptoms at baseline (n = 18; CDI total score$13) had greater follow-up: insulin resistance (HOMA-IR: [means6 SE] 2.86 0.2 vs. 4.36 0.4, P,0.001) (A), fasting insulin (13.26 0.8mU/L vs. 19.46 1.7mU/L, P = 0.001) (B), and fasting glucose (85.36 0.8 mg/dL vs. 92.96 1.7 mg/dL, P,0.001) (C), adjusting for the respective baseline values, sex, race, family history of type 2 diabetes, baseline age, baseline BMI, BMI change, and timein study.

Table 3dChildren’s baseline depressive symptoms as a predictor of follow-up elevated insulin resistance, hyperinsulinemia,and impaired fasting glucose

Predictor

Follow-up

Elevated HOMA-IR$3.16

Hyperinsulinemia$15 mU/L

Impaired fastingglucose .100 mg/dL

Sex (female) 2.75 (0.83–9.08) 0.90 (0.29–2.78) 0.56 (0.01–61.65)Race (black or Hispanic) 0.53 (0.13–2.16) 0.83 (0.23–2.99) 0.00 (0.00–0.00)Family history of type 2 diabetes (present) 1.69 (0.57–5.01) 1.86 (0.68–5.14) 41.61 (0.45–3,819.55)Baseline age (years) 0.88 (0.57–1.36) 1.00 (0.66–1.50) 5.10 (0.39–66.33)Baseline BMI (kg/m2) 1.11 (0.95–1.28)* 1.07 (0.93–1.23) 0.83 (0.57–1.21)BMI change 1.17 (1.00–1.37) 1.19 (1.03–1.39)† 0.88 (0.45–1.72)Time in study (years) 0.67 (0.39–1.17) 0.84 (0.50–1.40) 1.80 (0.07–45.28)Baseline HOMA-IR 1.65 (0.83–3.26) d dBaseline fasting insulin (mU/L) d 1.16 (1.00–1.35)† dBaseline fasting glucose (mg/dL) d d 0.88 (0.60–1.28)Baseline depressive symptoms‡ 1.14 (1.01–1.28)† 1.03 (0.92–1.14) 2.23 (0.95–5.19)*Data are odds ratios (95% CIs). n = 98–102. *P # 0.10. †P # 0.05. ‡CDI total score (continuous).

care.diabetesjournals.org DIABETES CARE 5

Shomaker and Associates

Adolescence marks a developmentalperiod notable for a normative increase ininsulin resistance that typically resolvesby the end of puberty (22–25). Yet, youthvulnerable for type 2 diabetes may displaythe largest increases in insulin resistanceand continued progression of worseninginsulin resistance throughout late adoles-cence and possibly into young adulthood.Therefore, investigation of the impact ofchild or adolescent depressive symptomson the progression of insulin resistanceduring an even longer follow-up intervalwould be important to clarify the role ofpediatric depressive symptoms in the de-velopment of type 2 diabetes. An equallyimportant task for future research is toelucidate the mechanisms by which de-pressive symptoms may impact insulinresistance. An understanding of the puta-tive behavioral and/or physiological fac-tors that explain how depression relatesto insulin resistance is crucial to the de-sign of effective interventions.

In the current study, we observed thatdepressive symptoms begin to exert animpact on insulin resistance early in life.Among adults, interventions targeting de-pressive symptoms among adults withtype 2 diabetes and/or major depressionhave been shown to improve indices ofglucose impairment or insulin resistanceeven without altering body weight oradiposity (1). Research is needed to deter-mine whether early interventions to de-crease youths’ depressive symptoms willameliorate worsening insulin resistanceand consequently lessen the risk of devel-oping type 2 diabetes. If treating or pre-venting the onset of major depression inadolescents improves insulin resistance,routine depression screening in primarycare settings, especially among youth at-risk for type 2 diabetes, might have thepotential to delay or possibly prevent type2 diabetes onset in a considerable subset ofindividuals.

AcknowledgmentsdThis research was sup-ported by National Research Service Award1F32HD056762 from the Eunice KennedyShriver National Institute of Child HealthandHuman Development (NICHD) (to L.B.S.)and Intramural Research Program Grant1ZIAHD000641 from the NICHD, with sup-plemental funding from the Division of Nu-trition Research Coordination, the NationalInstitute of Minority Health and Health Dis-parities, and the Office of Behavioral and SocialSciences Research (to J.A.Y.). J.A.Y. is a Com-missioned Officer in the U.S. Public HealthService, Department of Health and Human

Services. The funding organizations played norole in the design and conduct of the study; thecollection, management, analysis, and interpre-tation of data; or the preparation or review of themanuscript.No potential conflicts of interest relevant to

this article were reported.L.B.S. conceived the hypothesis for this ar-

ticle, wrote the first draft of the manuscript,conducted data analysis, participated in theinterpretation of the results, reviewed andedited the manuscript, and approved the finalversion of the manuscript. M.T.-K. conceivedthe hypothesis for this article, conducted dataanalysis, participated in the interpretation ofthe results, reviewed and edited the manu-script, and approved the final version of themanuscript. E.A.S. participated in the interpre-tation of the results, reviewed and edited themanuscript, approved the final version of themanuscript, and collected data. R.M. participatedin the interpretation of the results, reviewed andedited the manuscript, and approved the finalversion of the manuscript. J.M.Z. and S.E.F.conducted data analysis, participated in the in-terpretation of the results, reviewed and editedthe manuscript, and approved the final versionof the manuscript. S.Z.Y. and V.S.H. partici-pated in the interpretation of the results, re-viewed and edited themanuscript, approved thefinal version of the manuscript, and superviseddata collection. J.A.Y. conceived the hypothesisfor this article, participated in the interpretationof the results, reviewed and edited the manu-script, approved the final version of the manu-script, and supervised data collection.The authors thank the families who partic-

ipated in these studies. They also are grateful tothe staff of the National Institutes of HealthClinical Center who assisted in data collection.

References1. Lustman PJ, Penckofer SM, Clouse RE.

Recent advances in understanding de-pression in adults with diabetes. CurrPsychiatry Rep 2008;10:495–502

2. Räikkönen K, Keltikangas-Järvinen L,Adlercreutz H, Hautanen A. Psychosocialstress and the insulin resistance syndrome.Metabolism 1996;45:1533–1538

3. Suarez EC. Sex differences in the relation ofdepressive symptoms, hostility, and angerexpression to indices of glucosemetabolismin nondiabetic adults. Health Psychol 2006;25:484–492

4. Eaton WW, Armenian H, Gallo J, Pratt L,Ford DE. Depression and risk for onset oftype II diabetes: a prospective population-based study. Diabetes Care 1996;19:1097–1102

5. Kawakami N, Takatsuka N, Shimizu H,Ishibashi H. Depressive symptoms and oc-currence of type 2 diabetes among Japanesemen. Diabetes Care 1999;22:1071–1076

6. Everson-Rose SA, Meyer PM, Powell LH,et al. Depressive symptoms, insulin re-sistance, and risk of diabetes in women at

midlife. Diabetes Care 2004;27:2856–2862

7. Arroyo C, Hu FB, Ryan LM, et al. De-pressive symptoms and risk of type 2 di-abetes in women. Diabetes Care 2004;27:129–133

8. Golden SH, Williams JE, Ford DE, et al.;Atherosclerosis Risk inCommunities study.Depressive symptoms and the risk of type 2diabetes: the Atherosclerosis Risk in Com-munities Study. Diabetes Care 2004;27:429–435

9. Knol MJ, Twisk JW, Beekman AT, HeineRJ, Snoek FJ, Pouwer F. Depression as arisk factor for the onset of type 2 diabetesmellitus: ameta-analysis.Diabetologia 2006;49:837–845

10. Golden SH. A review of the evidence for aneuroendocrine link between stress, depres-sion and diabetesmellitus. CurrDiabetes Rev2007;3:252–259

11. Ravaja N, Keltikangas-Järvinen L. Temper-ament and metabolic syndrome precursorsin children: a three-year follow-up. PrevMed 1995;24:518–527

12. Shomaker LB, Tanofsky-Kraff M, Young-Hyman D, et al. Psychological symptomsand insulin sensitivity in adolescents. Pe-diatr Diabetes 2010;11:417–423

13. Jaser SS, Holl MG, Jefferson V, Grey M.Correlates of depressive symptoms inurban youth at risk for type 2 diabetesmellitus. J Sch Health 2009;79:286–292

14. Kovacs M. The Children’s Depression In-ventory. North Tonawanda, NY, Multi-Health Systems, 2003

15. Kazdin AE, Colbus D, Rodgers A. As-sessment of depression and diagnosis ofdepressive disorder among psychiatricallydisturbed children. J Abnorm Child Psy-chol 1986;14:499–515

16. Allison PD. Missing data. In The SAGEHandbook of Quantitative Methods inPsychology. Millsap RE, Maydeu-Olivares A,Eds. Thousand Oaks, CA, Sage Publica-tions, 2009, p. 72–89

17. Morrison JA, Glueck CJ, Horn PS, Wang P.Childhood predictors of adult type 2diabetes at 9- and 26-year follow-ups.Arch Pediatr Adolesc Med 2010;164:53–60

18. Shomaker LB, Tanofsky-Kraff M, ZoccaJM, Field SE, Drinkard B, Yanovski JA.Depressive symptoms and cardiorespira-tory fitness in obese adolescents. J AdolescHealth. In press

19. Epel E, Jimenez S, Brownell K, Stroud L,Stoney C, Niaura R. Are stress eaters atrisk for the metabolic syndrome? Ann N YAcad Sci 2004;1032:208–210

20. Rozanski A, Blumenthal JA, Kaplan J.Impact of psychological factors on thepathogenesis of cardiovascular diseaseand implications for therapy. Circulation1999;99:2192–2217

21. McIntyre RS, Soczynska JK, Konarski JZ,et al. Should depressive syndromes be

6 DIABETES CARE care.diabetesjournals.org

Depression and insulin resistance in youth

reclassified as “Metabolic Syndrome TypeII”? Ann Clin Psychiatry 2007;19:257–264

22. Amiel SA, Sherwin RS, Simonson DC,Lauritano AA, Tamborlane WV. Impairedinsulin action in puberty. A contribut-ing factor to poor glycemic control in

adolescents with diabetes. N Engl J Med1986;315:215–219

23. Caprio S, Plewe G, Diamond MP, et al.Increased insulin secretion in puberty:a compensatory response to reductions ininsulin sensitivity. J Pediatr 1989;114:963–967

24. Moran A, Jacobs DR Jr, Steinberger J,et al. Insulin resistance during puberty:results from clamp studies in 357 chil-dren. Diabetes 1999;48:2039–2044

25. Goran MI, Gower BA. Longitudinal studyon pubertal insulin resistance. Diabetes2001;50:2444–2450

care.diabetesjournals.org DIABETES CARE 7

Shomaker and Associates