C

DALPEa

SDUSD

Aa

gHo

a

w

1

0d

LINICAL RESEARCH STUDY

epression and Incident Diabetic Foot Ulcers:Prospective Cohort Study

isa H. Williams, MD, MS,a Carolyn M. Rutter, PhD,b Wayne J. Katon, MD,c Gayle E. Reiber, PhD, MPH,d

aul Ciechanowski, MD, MPH,c Susan R. Heckbert, MD, PhD,b,e Elizabeth H. B. Lin, MD, MPH,b

vette J. Ludman, PhD,b Malia M. Oliver, BA,b Bessie A. Young, MD, MPH,f Michael Von Korff, ScDb

Department of Medicine, Division of Dermatology, University of Washington, Seattle; bGroup Health Center for Health Studies,eattle, Wash; cDepartment of Psychiatry and Behavioral Sciences, University of Washington, Seattle; dHealth Services Research andevelopment Center of Excellence, VA Puget Sound Health Care System, and Departments of Epidemiology and Health Services,niversity of Washington, Seattle; eCardiovascular Health Research Unit, Department of Epidemiology, University of Washington,eattle; fEpidemiologic Research and Information Center and Primary and Specialty Care, VA Puget Sound Health Care System, and

epartment of Medicine, Division of Nephrology, University of Washington, Seattle.

OMwfHWpcdfRQ9HCop©

E-mail address

002-9343/$ -see foi:10.1016/j.amjm

ABSTRACT

BJECTIVE: To test whether depression is associated with an increased risk of incident diabetic foot ulcers.ETHODS: The Pathways Epidemiologic Study is a population-based prospective cohort study of 4839 patientsith diabetes in 2000-2007. The present analysis included 3474 adults with type 2 diabetes and no prior diabetic

oot ulcers or amputations. Mean follow-up was 4.1 years. Major and minor depression assessed by the Patientealth Questionnaire-9 were the exposures of interest. The outcome of interest was incident diabetic foot ulcers.e computed the hazard ratio and 95% confidence interval (CI) for incident diabetic foot ulcers, comparing

atients with major and minor depression with those without depression and adjusting for sociodemographicharacteristics, medical comorbidity, glycosylated hemoglobin, diabetes duration, insulin use, number ofiabetes complications, body mass index, smoking status, and foot self-care. Sensitivity analyses also adjustedor peripheral neuropathy and peripheral arterial disease as defined by diagnosis codes.ESULTS: Compared with patients without depression, patients with major depression by Patient Healthuestionnaire-9 had a 2-fold increase in the risk of incident diabetic foot ulcers (adjusted hazard ratio 2.00;5% CI, 1.24-3.25). There was no statistically significant association between minor depression by Patientealth Questionnaire-9 and incident diabetic foot ulcers (adjusted hazard ratio 1.37; 95% CI, 0.77-2.44).ONCLUSION: Major depression by Patient Health Questionnaire-9 is associated with a 2-fold higher riskf incident diabetic foot ulcers. Future studies of this association should include better measures oferipheral neuropathy and peripheral arterial disease, which are possible confounders or mediators.

2010 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2010) 123, 748-754

KEYWORDS: Complications; Depression; Diabetes; Foot ulcers

sst2subda

b

mong individuals with diabetes, depression is commonnd associated with multiple adverse outcomes. Depressive

Funding: This research was supported by National Institutes of Healthrant MH-073686. Dr Williams was supported by National Institutes ofealth/American Skin Association grant F32 AR-056380 and a Dermatol-gy Foundation Dermatologist Investigator Research Fellowship.

Conflict of Interest: None of the authors have any conflicts of interestssociated with the work presented in this manuscript.

Authorship: All authors had access to the data and played a role inriting this manuscript.

Reprint requests should be addressed to Lisa H. Williams, MD, MS,25 16th Ave E, mail stop CSB-5, Seattle, WA 98112.

s: lisahw@u.washington.edu

ront matter © 2010 Elsevier Inc. All rights reserved.ed.2010.01.023

ymptoms were associated with a higher risk of developingelf-reported macrovascular and microvascular complica-ions in a large study of older Hispanic Americans with type

diabetes.1 Mild to severe depressive symptoms were as-ociated with the development of retinopathy2 and protein-ria3 in younger African-American adults with type 1 dia-etes. In addition, several studies in patients with type 2iabetes showed that baseline major and minor depressionre associated with an increased risk of mortality.1,4-6

The relationship between depression and incident dia-etic foot ulcers is not well characterized, however. Cohort

tudies of patients with prevalent diabetic foot ulcers have

ssnnitjcia2

M

SGimpWmirU

SThdhpmmhmma9ijorc

alrGbdlru(d

k(

MT

sdptpse

Ic77v

of fo

749Rutter et al Depression and Incident Diabetic Foot Ulcers

hown that depression is associated with larger6 and moreevere7 foot ulcers at presentation and a higher risk ofonhealing and recurrent foot ulcers during follow-up.7 It isot known whether depression is a risk factor for develop-ng a first diabetic foot ulcer. The purpose of this study waso determine whether baseline ma-or and minor depression are asso-iated with an increased risk ofncident diabetic foot ulcers usinglarge cohort of patients with typediabetes.

ATERIALS AND METHODS

ettingroup Health Cooperative (GHC)

s a mixed-model prepaid healthaintenance organization with 30

rimary care clinics in westernashington State. Nine GHC pri-ary care clinics were selected for

nclusion in the study. Institutionaleview boards at GHC and theniversity of Washington approved all study procedures.

tudy Cohort Selectionhe cohort included in the Pathways Epidemiologic Studyas been described in detail.8 Briefly, adults in the GHCiabetes registry were invited to participate if they receivedealthcare between 2000 and 2002 at any of the 9 GHCrimary care clinics. The GHC diabetes registry includesembers having any of the following in the preceding 12onths: a dispensed prescription for insulin or an oral

ypoglycemic agent; 2 fasting plasma glucose levels 126g/dL or greater; 2 random plasma glucose levels 200g/dL or greater; or 2 outpatient diagnoses of diabetes or

ny inpatient diagnosis of diabetes. Surveys were mailed to063 potentially eligible patients; 1222 were found to beneligible (Figure). Of 7841 eligible participants, 4839 sub-ects returned the baseline questionnaire, for a response ratef 61.7% (see Online Appendix A for characteristics ofesponders vs nonresponders). These 4839 participantsomprised the Pathways cohort.

Approximately 5 years after enrollment, between 2005nd 2007, we contacted surviving members of the cohort byetter and a follow-up telephone call to ask for permission toeview their electronic and paper medical records. BothHC and University of Washington institutional reviewoards approved a waiver of consent for participants whoied after enrollment in the Pathways cohort. We excludediving participants who did not consent to medical recordeview (n � 682) and participants with prior diabetic footlcers (n � 159), prior amputations (n � 35), type 1 diabetesn � 211), unknown diabetes type (n � 5), missing covariate

CLINICAL SIGNIF

● Major depressionQuestionnaire-9primary care pat

● Major depressiotwice the risk offoot ulcer over 4

● Clinicians shouldprevalence of dwith diabetes anthe development

ata (n � 253), disenrollment before baseline (n � 11), un-

nown date of foot ulcer (n � 1), and missing charts (n � 8)Figure).

easureshe predictor of interest, depression at baseline, was as-

sessed using the Patient HealthQuestionnaire-9, a self-reportmeasure of depression symptomsbased on criteria for depression inthe Diagnostic and StatisticalManual of Mental Disorders,Fourth Edition.9 By following thestandard method, each item wasconsidered positive if endorsed“more than half the days” in thelast 2 weeks.9 A diagnosis of ma-jor depression required at least 1of the 2 cardinal symptoms (de-pressed mood or loss of interest)and at least 5 of 9 positive symp-toms. Criteria for minor depres-sion required at least 1 cardinalsymptom and 2 to 4 positive

ymptoms. Compared with a structured interview majorepression diagnosis, the sensitivity, specificity, positiveredictive value, and negative predictive value of this Pa-ient Health Questionnaire-9 scoring method in a generalrimary care sample were 88%, 88%, 35%, and 99%, re-pectively,9 and 35%, 97%, 69%, and 87%, respectively, inlderly patients with diabetes.10

Participants were screened for diabetic foot ulcers usingnternational Classification of Diseases, Ninth Revisionodes (707.1x, 707.8-707.9, 681.1x, 681.9x, 682.7x,30.06-730.09, 730.17, 730.19, 730.27, 730.29, and85.4x). Foot ulcer diagnoses were confirmed by chart re-iew and defined as a break in the skin extending through

CE

he Patient Healthsent in 11.5% ofwith diabetes.

associated withloping a diabetics.

ware of the highsion in patientsassociation withot ulcers.

ICAN

by tis preients

n isdeveyear

be aepresd its

Figure Recruitment flow chart.

tclacied

sitos

utfq

ritimtbs

nAFHNNHDPN

PPR

BCFCC

750 The American Journal of Medicine, Vol 123, No 8, August 2010

he dermis to deeper tissue (subcutaneous fat, fascia, mus-le, joint, or bone) in a location distal to the medial andateral malleoli that had not healed in 30 days.11 Ulcersbove the malleoli, spider bites, malignant neoplasms, vas-ulitis, pyoderma gangrenosum, venous stasis ulcers, fungalnfections, and routine cuts and scratches were not consid-red diabetic foot ulcers. Prior amputations were defined byiagnosis and procedure codes.

The baseline mailed survey included questions aboutociodemographic characteristics (age, gender, race/ethnic-ty, education, and marital status), diabetes duration, diabe-es treatment at diagnosis (none, diet, oral hypoglycemicnly, and any insulin), body mass index, current smoking

Table 1 Baseline Characteristics of Participants with Type 2 DQuestionnaire-9*

T

(%)ge, y, mean (SD)emale, n (%) 1igh school education or less, n (%)ot married/living together, n (%) 1on-white, n (%)bA1c, %, mean (SD)iabetes duration, y, mean (SD)rescribed insulin, n (%)o. of diabetes complications†

0 11 12345-7

eripheral neuropathy,‡ n (%)eripheral arterial disease,§ n (%)xRisk score,� US$, n (%)

�13001300-25992600-4399�4400

ody mass index, mean (SD)urrently smoking, n (%)oot self-carehecked feet, No. of days in last week, mean (SD)hecked inside of shoes, No. of days in last week, mean (SD)

HbA1c � glycosylated hemoglobin; PHQ-9 � Patient Health QuestionnColumn percentages may not add up to 100 because of rounding.

*The Patient Health Questionnaire-9 is a self-report measure of depresDiagnostic and Statistical Manual of Mental Disorders, Fourth Edition.

†Retinopathy, cardiovascular disease, stroke, nephropathy, metabolicand procedure codes. Cardiovascular disease and stroke cases were confi

‡Defined by diagnosis codes.§Defined by diagnosis and procedure codes.�RxRisk is a pharmacy-based measure of medical comorbidity. Higher s

annual healthcare costs as a function of age, gender, and the chronic condiabetes medications are excluded from this modified RxRisk score.

tatus, and foot self-care. Foot self-care was ascertained c

sing the 2 core foot self-care items (the number of days inhe past week the participant checked his/her feet and shoes)rom the Summary of Diabetes Self-Care Activitiesuestionnaire.12

Medical record review and automated pharmacy, labo-atory, and diagnosis data provided information on use ofnsulin, glycosylated hemoglobin, and the number of diabe-es complications (cardiovascular disease, nephropathy, ret-nopathy, peripheral arterial disease, stroke, neuropathy, andetabolic) during the 12 months before study entry. Pa-

ients were classified as having type 1 diabetes if onset wasefore 30 years of age, insulin was the first treatment pre-cribed, and they were currently taking insulin. For diabetes

by Depressed Status, as Defined by the Patient Health

ampleMajor Depressionby PHQ-9

Minor Depressionby PHQ-9 No Depression

4 401 (11.5) 290 (8.3) 2783 (80.1)2.6) 59.5 (12.6) 64.6 (13.2) 64.7 (12.3)8.0) 239 (59.6) 144 (50.0) 1284 (46.1)4.5) 104 (25.9) 94 (32.4) 652 (23.4)3.3) 174 (43.4) 103 (35.5) 879 (31.6)9.8) 82 (20.4) 74 (25.5) 533 (19.2).6) 8.1 (1.7) 7.9 (1.6) 7.7 (1.5).2) 8.4 (6.9) 9.9 (9.0) 8.4 (8.3)6.2) 159 (39.7) 87 (30.0) 664 (23.9)

1.8) 107 (26.7) 73 (25.2) 926 (33.3)3.2) 128 (31.9) 89 (30.7) 937 (33.7)9.0) 82 (20.5) 66 (22.8) 513 (18.4).9) 46 (11.5) 29 (10.0) 270 (9.7).4) 31 (7.7) 22 (7.6) 98 (3.5).6) 7 (1.7) 11 (3.8) 39 (1.4)4.2) 80 (20.0) 58 (20.0) 356 (12.8).2) 1 (0.3) 5 (1.7) 35 (1.3)

6.3) 104 (25.9) 79 (27.2) 731 (26.3)4.7) 111 (27.7) 52 (17.9) 696 (25.0)3.6) 67 (16.7) 73 (25.2) 678 (24.4)5.4) 119 (29.7) 86 (29.7) 678 (24.4).1) 35.0 (9.0) 32.5 (7.1) 31.2 (6.7).5) 62 (15.5) 28 (9.7) 204 (7.3)

.7) 4.5 (2.7) 4.6 (2.6) 4.5 (2.7)

.8) 2.0 (2.8) 2.0 (2.7) 2.1 (2.8)

SD � standard deviation.

mptoms based on the criteria for diagnosis of a depressive episode in the

eral neuropathy, and peripheral arterial disease as defined by diagnosischart review.

dicate higher medical comorbidity. The score is an estimate of expectedlasses in which prescription drug fills are observed. Antidepressants and

iabetes

otal S

34764.1 (1667 (4850 (2156 (3689 (17.8 (18.5 (8910 (2

106 (3154 (3661 (1345 (9151 (457 (1

494 (141 (1

914 (2859 (2818 (2883 (231.7 (7294 (8

4.5 (22.1 (2

aire-9;

sion sy

, periphrmed by

cores indition c

omplications, we used the Diabetes Complications Sever-

iu(aFRfiiac

SSIoltmdacweS

mhrwaugtidaliie

ay

SWanTa2ps

n

wc(obacOH

RBspmCmddslsgmbts

oemwp

bwja(adiadaacPufwbnwt

751Rutter et al Depression and Incident Diabetic Foot Ulcers

ty Index,13 modifying cardiovascular disease and strokesing additional chart review data and procedure codesavailable on request) and removing peripheral neuropathynd peripheral arterial disease for separate examination.or medical comorbidity, we used the pharmacy-basedxRisk score, an estimate of annual healthcare costs as a

unction of age, gender, and the chronic condition classesn which prescription fills are observed; higher scoresndicate higher medical comorbidity.14 For this study,ntidepressants and diabetes medications were not in-luded in the RxRisk score.

tatistical AnalysesAS/PC for Windows (Stat software version 9.1.3, SASnstitute Inc, Cary, NC) was used for all analyses. Theverall incidence rate for diabetic foot ulcers was calcu-ated by dividing the number of incident foot ulcers byhe total person-years of risk. We used Cox regression toodel time to incident foot ulcers, censoring at death and

isenrollment. All covariates were measured at baselinend chosen a priori. Age, diabetes duration, foot self-are, glycosylated hemoglobin, and body mass indexere included in the models as continuous variables. We

valuated the proportional hazards assumption usingchoenfeld residuals.

To assess the influence of particular covariates (bodyass index, smoking status, foot self-care, glycosylated

emoglobin, peripheral arterial disease, and peripheral neu-opathy) on the hazard ratio for incident diabetic foot ulcers,e used a series of several proportional hazards models,

dding the covariates sequentially. The first model wasnadjusted. The second model was adjusted for sociodemo-raphic characteristics (age, gender, race/ethnicity, educa-ion, and marital status). In the third model, diabetes sever-ty (diabetes duration, insulin prescription, and number ofiabetes complications, excluding peripheral neuropathynd peripheral arterial disease, which were added separatelyater) was added. The remaining covariates were addedndividually in the following order: body mass index, smok-ng status, foot self-care, glycosylated hemoglobin, periph-ral arterial disease, and peripheral neuropathy.

Few amputation events, 25 major (transtibial and above)nd 20 minor (ankle or below), precluded meaningful anal-sis of this outcome.

ensitivity Analysese examined death as a competing risk in 1 sensitivity

nalysis. We censored individuals at disenrollment butot at death, thus keeping those who died in the risk set.here were no meaningful differences in the estimatedssociation between depression and foot ulcers under thecensoring conditions, indicating that death as a com-

eting risk did not affect the estimated effect of depres-ion on foot ulcer risk.

We did not have standard clinical measures of peripheral

europathy and peripheral arterial disease in our study, but 9

e wanted to examine the effects of adding these possibleonfounders or mediators as defined by diagnostic codeslisted in Online Appendix B). We also assessed the effectf any antidepressant medication use in the year beforeaseline; we did not include antidepressants in our finalnalysis because treatment with antidepressants in primaryare is often simply a marker of depression severity.15 Seenline Appendix C for an analysis using continuous Patientealth Questionnaire-9.

ESULTSaseline characteristics are shown in Table 1. Of 3474

ubjects with type 2 diabetes, 401 (11.5%) had major de-ression by Patient Health Questionnaire-9, 290 (8.3%) hadinor depression, and 2783 (80.1%) had no depression.ompared with those without depression, participants withajor depression were younger, and participants with minor

epression were more often non-white and had longer meaniabetes duration. Compared with those with no depression,ubjects with either major or minor depression were moreikely to be female, single, less educated, and currentlymoking. They also had higher mean body mass index,reater medical comorbidity (higher RxRisk scores), andore severe diabetes (higher mean glycosylated hemoglo-

in, more prescribed insulin, and more diabetes complica-ions). There were no differences in patient-reported footelf-care among the groups.

Mean follow-up was 4.1 years. The unadjusted incidencef diabetic foot ulcers was 8.4/1000 person-years in thentire sample, 16.2/1000 person-years in participants withajor depression, 12.1/1000 person-years in participantsith minor depression, and 7/1000 person-years in partici-ants without depression.

Table 2 shows estimated hazard ratios for incident dia-etic foot ulcers, comparing major and minor depressionith no depression by Patient Health Questionnaire-9. Ma-

or depression, compared with no depression, was associ-ted with more than twice the hazard for incident foot ulcershazard ratio 2.17; 95% confidence interval (CI), 1.34-3.49),fter adjusting for sociodemographic characteristics (gen-er, age, race/ethnicity, education, and marital status), med-cal comorbidity (RxRisk score), and diabetes severity. Thisssociation did not change substantially when potential me-iators such as foot self-care, body mass index, smoking,nd glycosylated hemoglobin were added to the model (haz-rd ratio 2.00, 95% CI, 1.24-3.25). There was no statisti-ally significant association between minor depression byatient Health Questionnaire-9 and incident diabetic footlcers (unadjusted hazard ratio 1.74; 95% CI, 0.99-3.06;ully adjusted hazard ratio 1.37; 95% CI , 0.77-2.44). Thereere no substantial changes in the observed associationetween major depression by Patient Health Question-aire-9 and diabetic foot ulcers in sensitivity analyses wheree added each of the following covariates sequentially to

he full model: peripheral arterial disease (hazard ratio 2.09;

5% CI, 1.29-3.38), peripheral neuropathy (hazard ratio

1b

CWtidjvtbcpducai

aruiabdcthatar

rcdpecsaasra

rtisDitrcs

pctgsmlr

MMcMMMMMM

of day

752 The American Journal of Medicine, Vol 123, No 8, August 2010

.99; 95% CI, 1.22-3.24), and antidepressant use in the yearefore baseline (hazard ratio 2.20; 95% CI, 1.33-3.62).

ONCLUSIONSe found that major depression by Patient Health Ques-

ionnaire-9 was associated with a 2-fold increased risk ofncident diabetic foot ulcers among patients with type 2iabetes over 4 years. This association persisted after ad-ustment for sociodemographic characteristics, diabetes se-erity (diabetes duration, insulin use, number of complica-ions, and glycosylated hemoglobin), medical comorbidity,ody mass index, smoking, and patient-reported foot self-are. This finding is consistent with prior cohort studies ofatients with prevalent diabetic foot ulcers showing thatepression is associated with larger6 and more severe7 footlcers at presentation, as well as delayed healing and in-reased recurrence of foot ulcers.7 To our knowledge, were the first to examine the association of depression andncident diabetic foot ulcers.

Major foot ulcer risk factors, such as peripheral neurop-thy and peripheral arterial disease, could be mediators in aelationship between depression and incident diabetic footlcers. Cross-sectional studies have shown that depressions associated with peripheral neuropathy16-18 and peripheralrterial disease.19-21 In a large, prospective, population-ased study of mostly nondiabetic participants, baselineepressive symptoms are linked to the development of in-ident peripheral arterial disease.22 In addition, a prospec-ive study of veterans, half with diabetes, showed that aistory of treated depression is associated with decreasedrtery patency and recurrent symptoms after revasculariza-ion.23 We are aware of no prospective studies showing anssociation between baseline depression and incident pe-

Table 2 Proportional Hazards Models for Risk of Incident DiabStatus, as Defined by the Patient Health Questionnaire-9

HR (95% CI

No Depressi

odel 1 (unadjusted) 1 (referenceodel 2 (Model 1 plus sociodemographicharacteristics*)

1 (reference

odel 3 (Model 2 plus RxRisk score) 1 (referenceodel 4 (Model 3 plus diabetes severity†) 1 (referenceodel 5 (Model 4 plus body mass index) 1 (referenceodel 6 (Model 5 plus current smoking status) 1 (referenceodel 7 (Model 6 plus foot self-care‡) 1 (referenceodel 8 (Model 7 plus HbA1c) 1 (reference

CI � confidence interval; HR � hazard ratio; HbA1c � glycosylated heAll covariates were measured at baseline.

*Sociodemographic characteristics: age, gender, race/ethnicity, educ†Diabetes severity: diabetes duration, prescribed insulin, and numb

arterial disease, which were added separately in sensitivity analyses.‡Foot self-care: number of days participant checked feet and number

ipheral neuropathy. t

Alternatively, symptomatic peripheral neuropathy or pe-ipheral arterial disease could cause depression and thusonfound the relationship between depression and incidentiabetic foot ulcers. Cross-sectional studies in patients witheripheral arterial disease have shown that decreased lowerxtremity functioning by 6-minute walk test20,21 and in-reased self-reported rest pain24 are associated with depres-ive symptoms, although the directionality of these associ-tions is not clear. A prospective study showed anssociation between baseline diabetic peripheral neuropathyeverity and depressive symptoms 18 months later, and thiselationship was partially mediated by the symptoms of painnd unsteadiness.25

Our ability to detect mediation or confounding by pe-ipheral neuropathy and peripheral arterial disease in sensi-ivity analyses was limited. We did not have standard clin-cal measures of neuropathy or peripheral arterial disease. Inensitivity analyses, we used International Classification ofiseases, Ninth Revision code definitions, which are likely

naccurate compared with standard methods, and found lit-le changes from the results of the main analysis. Theelationship between depression and these diabetic foot ul-er risk factors is likely bidirectional and requires furthertudy.

Other potential mediators of a relationship between de-ression and diabetic foot ulcers include impaired glycemicontrol and poor self-care, but we found little evidence ofhis in our study. Depression has been associated with poorlycemic control in several studies.8,26 In our study, depres-ion was associated with modestly higher glycosylated he-oglobin values at baseline, but adding baseline glycosy-

ated hemoglobin to the model made little difference in theesults (Table 2). Depression has been linked with medica-

ot Ulcers in Participants with Type 2 Diabetes by Depressed

Major Depression by PHQ-9 Minor Depression by PHQ-9

2.35 (1.50-3.68) 1.74 (0.99-3.06)2.80 (1.76-4.45) 1.77 (1.00-3.14)

2.38 (1.49-3.80) 1.69 (0.95-2.98)2.17 (1.34-3.49) 1.43 (0.80-2.54)2.07 (1.28-3.36) 1.40 (0.79-2.49)2.02 (1.24-3.27) 1.40 (0.79-2.49)2.02 (1.24-3.27) 1.39 (0.78-2.46)2.00 (1.24-3.25) 1.37 (0.77-2.44)

in; PHQ-9 � Patient Health Questionnaire-9.

nd marital status.iabetes complications, excluding peripheral neuropathy and peripheral

s participant checked shoes in past week.

etic Fo

)

on

))

))))))

moglob

ation, aer of d

ion nonadherence,27,28 unhealthy diet,27,28 sedentary life-

sswfnctic

idfivQdcs

duhncwertdsesTpatws

authtsnmbam

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

753Rutter et al Depression and Incident Diabetic Foot Ulcers

tyle,27,28 obesity,8 smoking,8 and impaired foot-specificelf-care28 in patients with diabetes. Studies examininghether improvements in self-care can decrease diabetes

oot ulcer rates have had conflicting results.29,30 There waso association between depression and baseline foot-spe-ific self-care in the present study, which is consistent withhe findings of other groups,6,27,31 and adding body massndex, foot self-care, and smoking to the model did nothange the results (Table 2).

Our study has several strengths. To our knowledge, thiss the first study of the association of depression with inci-ent diabetic foot ulcers. Foot ulcer diagnoses were con-rmed by chart review. We did not have structured inter-iew diagnoses of depression, but the Patient Healthuestionnaire-9 is a well-validated self-report measure ofepression.9,10 We also used validated measures of self-are.12 Automated data allowed us to estimate diabeteseverity and medical comorbidity.

There also are limitations of our study. We may not haveetected incident or prior diabetic foot ulcers if the patientsed another healthcare system for care (unlikely in thisealth maintenance organization), did not seek care, or didot receive 1 of the screening diagnosis codes. Patientharacteristics, including depression and other covariates,ere measured at baseline only, so we were unable to

xamine how changes in depression may have affected theisk of incident foot ulcers or fully evaluate mediation byhe covariates. As stated above, we did not have reliableiagnoses of possible important confounders or mediatorsuch as peripheral neuropathy and peripheral arterial dis-ase, which could only be measured with International Clas-ification of Diseases, Ninth Revision codes in our study.he relationship between depression and diabetic foot com-lications is probably bidirectional, but we were not able tossess this directly. Finally, we analyzed less than 50% ofhe patients eligible for the original cohort, and the studyas completed in 1 geographic region of the country, pos-

ibly limiting generalizability.Major depression by Patient Health Questionnaire-9 is

ssociated with a 2-fold increased risk of incident footlcers in patients with type 2 diabetes. Screening for andreatment of depression in patients with diabetes could beelpful in the prevention of foot ulcers, but further studies ofhis relationship and of the effectiveness of depressioncreening and treatment in the prevention of foot ulcers areeeded before specific clinical recommendations can beade. Future studies of this association should include

etter measures of peripheral neuropathy and peripheralrterial disease, which are likely important confounders orediators.

eferences1. Black SA, Markides KS, Ray LA. Depression predicts increased in-

cidence of adverse health outcomes in older Mexican Americans withtype 2 diabetes. Diabetes Care. 2003;26:2822-2828.

2. Roy MS, Roy A, Affouf M. Depression is a risk factor for poorglycemic control and retinopathy in African-Americans with type 1

diabetes. Psychosom Med. 2007;69:537-542.

3. Roy MS, Affouf M, Roy A. Six-year incidence of proteinuria intype 1 diabetic African Americans. Diabetes Care. 2007;30:1807-1812.

4. Katon WJ, Rutter C, Simon G, et al. The association of comorbiddepression with mortality in patients with type 2 diabetes. DiabetesCare. 2005;28:2668-2672.

5. Lin EH, Heckbert SR, Rutter CM, et al. Depression and increasedmortality in diabetes: unexpected causes of death. Ann Fam Med.2009;7:414-421.

6. Ismail K, Winkley K, Stahl D, et al. A cohort study of people withdiabetes and their first foot ulcer: the role of depression on mortality.Diabetes Care. 2007;30:1473-1479.

7. Monami M, Longo R, Desideri CM, et al. The diabetic person beyonda foot ulcer: healing, recurrence, and depressive symptoms. J AmPodiatr Med Assoc. 2008;98:130-136.

8. Katon W, Von Korff M, Ciechanowski P, et al. Behavioral and clinicalfactors associated with depression among individuals with diabetes.Diabetes Care. 2004;27:914-920.

9. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of abrief depression severity measure. J Gen Intern Med. 2001;16:606-613.

0. Lamers F, Jonkers CC, Bosma H, et al. Summed score of the PatientHealth Questionnaire-9 was a reliable and valid method for depressionscreening in chronically ill elderly patients. J Clin Epidemiol. 2008;61:679-687.

1. Reiber GE, Smith DG, Wallace C, et al. Effect of therapeutic footwearon foot reulceration in patients with diabetes: a randomized controlledtrial. JAMA. 2002;287:2552-2558.

2. Toobert DJ, Hampson SE, Glasgow RE. The summary of diabetesself-care activities measure: results from 7 studies and a revised scale.Diabetes Care. 2000;23:943-950.

3. Young BA, Lin E, Von Korff M, et al. Diabetes complications severityindex and risk of mortality, hospitalization, and healthcare utilization.Am J Manag Care. 2008;14:15-23.

4. Fishman PA, Goodman MJ, Hornbrook MC, et al. Risk adjustmentusing automated ambulatory pharmacy data: the RxRisk model. MedCare. 2003;41:84-99.

5. Gilchrist G, Gunn J. Observational studies of depression in primarycare: what do we know? BMC Fam Pract. 2007;8:28.

6. de Groot M, Anderson R, Freedland KE, et al. Association of depres-sion and diabetes complications: a meta-analysis. Psychosom Med.2001;63:619-630.

7. Vileikyte L, Leventhal H, Gonzalez JS, et al. Diabetic peripheralneuropathy and depressive symptoms: the association revisited. Dia-betes Care. 2005;28:2378-2383.

8. Moreira RO, Papelbaum M, Fontenelle LF, et al. Comorbidity ofpsychiatric disorders and symmetric distal polyneuropathy amongtype II diabetic outpatients. Braz J Med Biol Res. 2007;40:269-275.

9. Smolderen KG, Aquarius AE, de Vries J, et al. Depressive symptomsin peripheral arterial disease: a follow-up study on prevalence, stabil-ity, and risk factors. J Affect Disord. 2008;110:27-35.

0. Arseven A, Guralnik JM, O’Brien E, et al. Peripheral arterial diseaseand depressed mood in older men and women. Vasc Med. 2001;6:229-234.

1. McDermott MM, Greenland P, Guralnik JM, et al. Depressivesymptoms and lower extremity functioning in men and women withperipheral arterial disease. J Gen Intern Med. 2003;18:461-467.

2. Wattanakit K, Williams JE, Schreiner PJ, et al. Association of angerproneness, depression and low social support with peripheral arterialdisease: the Atherosclerosis Risk in Communities Study. Vasc Med.2005;10:199-206.

3. Cherr GS, Wang J, Zimmerman PM, Dosluoglu HH. Depression isassociated with worse patency and recurrent leg symptoms afterlower extremity revascularization. J Vasc Surg. 2007;45:744-

750.

2

2

2

2

2

2

3

3

754 The American Journal of Medicine, Vol 123, No 8, August 2010

4. Smolderen KG, Hoeks SE, Pedersen SS, et al. Lower-leg symptoms inperipheral arterial disease are associated with anxiety, depression, andanhedonia. Vasc Med. 2009;14:297-304.

5. Vileikyte L, Peyrot M, Gonzalez JS, et al. Predictors of depressivesymptoms in persons with diabetic peripheral neuropathy: a longitu-dinal study. Diabetologia 2009;52:1265-1273.

6. Lustman PJ, Clouse RE, Ciechanowski PS, et al. Depression-relatedhyperglycemia in type 1 diabetes: a mediational approach. PsychosomMed. 2005;67:195-199.

7. Lin EH, Katon W, Von Korff M, et al. Relationship of depression anddiabetes self-care, medication adherence, and preventive care. Diabe-

8. Gonzalez JS, Safren SA, Delahanty LM, et al. Symptoms of depressionprospectively predict poorer self-care in patients with Type 2 diabetes.Diabet Med. 2008;25:1102-1107.

9. Lincoln NB, Radford KA, Game FL, Jeffcoate WJ. Educationfor secondary prevention of foot ulcers in people with diabetes:a randomised controlled trial. Diabetologia. 2008;51:1954-1961.

0. Malone JM, Snyder M, Anderson G et al. Prevention of amputation bydiabetic education. Am J Surg. 1989;158:520-524.

1. Johnston MV, Pogach L, Rajan M, et al. Personal and treatment factorsassociated with foot self-care among veterans with diabetes. J Rehabil

tes Care. 2004;27:2154-2160. Res Dev. 2006;43:227-238.

nA

FH

R

OHOOMP

S

D

754.e1Rutter et al Depression and Incident Diabetic Foot Ulcers

Online Appendix A Baseline Characteristics of Responders versus Nonresponders

Total Sample Nonresponder Responder P Value

* 6554 2084 4470ge, y �.0001

18-34 179 84 (4.0) 95 (2.1)35-44 441 207 (9.9) 234 (5.2)45-54 1292 566 (27.2) 726 (16.2)55-64 1728 582 (27.9) 1146 (25.9)65-74 1468 340 (16.3) 1128 (25.2)75-84 1204 251 (12.0) 953 (21.4)85-100 242 54 (2.6) 188 (4.2)

emale 3187 1006 (48.3) 2181 (48.8) .70bA1c �.0001

0-6.999 1890 455 (21.8) 1435 (32.1)7-7.999 1742 438 (21.0) 1304 (29.2)8-8.999 1162 343 (16.5) 819 (18.3)9-9.999 620 225 (10.8) 395 (8.8)10� 738 334 (16.0) 404 (9.0)No test in past year 402 289 (13.9) 113 (2.5)

xRisk score in past year �.0001�1300 1635 719 (34.5) 916 (20.5)1300-2599 1680 532 (25.5) 1148 (25.7)2600-4399 1602 422 (20.2) 1180 (26.4)�4399 1637 411 (19.7) 1226 (27.4)

n antidepressants in past year 1672 488 (23.4) 1184 (26.5) .008ospitalized in past year 673 196 (9.4) 477 (10.7) .12n insulin in past year 1872 528 (25.3) 1344 (30.1) �.0001n oral hypoglycemic in past year 3938 1281 (61.4) 2657 (59.4) .12ental health visit in past year 431 149 (7.2) 282 (6.3) .20rimary care visits in past year �.0001

0-2 1675 649 (31.1) 1026 (23.0)3-6 2820 904 (43.4) 1916 (42.9)7-9 1018 255 (12.2) 763 (17.1)10-14 698 188 (9.0) 510 (11.4)15� 343 88 (4.2) 255 (5.7)

pecialist visits in past year �.00010 1044 492 (23.6) 552 (12.3)1-2 2709 863 (41.4) 1846 (41.3)3-5 1718 487 (23.4) 1231 (27.5)6� 1083 242 (11.6) 841 (18.8)

epression ICD-9 code in last year 852 271 (13.0) 581 (13.0) .99

HbA1c � glycosylated hemoglobin; ICD-9 � International Classification of Diseases, Ninth Revision.*Of the 4839 participants who responded to the survey, there were 369 who declined permission for us to access their medical records, leaving 4470

responders for this analysis. We had contact with 1287 nonresponders who declined permission to access their medical records at the time of contact(mail-in refusal n � 50, call in refusal n � 187, refusal at reminder call n � 679, refusal by non-respondent n � 2), leaving 2084 nonresponders for thisanalysis.

1

2

754.e2 The American Journal of Medicine, Vol 123, No 8, August 2010

Online Appendix B International Classification of Diseases, Ninth Revision Diagnostic Codes Used to Define Peripheral Neuropathyand Peripheral Arterial Disease in Sensitivity Analyses

) Peripheral neuropathy was defined by any of the following in the 12 months before study entry, provided there were no otherdiagnoses to explain the neuropathy (Hartsfield CL, Korner EJ, Ellis JL, et al. Painful diabetic peripheral neuropathy in a managedcare setting: patient identification, prevalence estimates, and pharmacy utilization patterns. Popul Health Manag. 2008;11:317-328.)a) one 357.2 diagnosis (polyneuropathy in diabetes),b) two 250.6 diagnoses (diabetes with neurologic manifestations) at least 30 days apart,c) two 337.1 diagnoses (peripheral autonomic neuropathy in disorders classified elsewhere) at least 30 days apart, ord) two 337.9 diagnoses (unspecified disorder of autonomic nervous system) at least 30 days apart.

) Peripheral arterial disease was defined by any of the following in the 12 months before study entry:a) ICD-9 codes: 250.7x (diabetes with peripheral circulatory disorders), 442.3x (aneurysm of artery of lower extremity),

443.81 (peripheral angiopathy in diseases classified elsewhere), 443.9x (peripheral vascular disease, unspecified), 444.22 (lowerextremity arterial embolism or thrombosis), 785.4 (gangrene)

b) CPT codes for lower limb artery procedures: thromboendarterectomy (35302-6, 35351, 35355, 35361, 35363, 35371-2; openangioplasty (35454, 35456, 35459); percutaneous angioplasty (35470, 35473, 35474); open atherectomy (35482, 35483,35485); percutaneous atherectomy (35492, 35493, 35495); artery bypass graft with harvested vein (35521, 35533, 35537,35538, 35539, 35540, 35548, 35549, 35551, 35556, 35558, 35563, 35565, 35566, 35571); artery bypass graft with in-situ vein(35583, 35585, 35587); artery bypass graft with other than vein (35621, 35623, 35637, 35638, 35646, 35647, 35651, 35654,35656, 35661, 35663, 35665, 35666, 35671, 35686, 35700); drug-eluting peripheral vessel stents (00.55)

MMMMMMMM

754.e3Rutter et al Depression and Incident Diabetic Foot Ulcers

Online Appendix C Risk of Incident Foot Ulcers Associated with a One-Point Increase in the Patient Health Questionnaire-9 Score

HR (95% CI) P Value

odel 1 (unadjusted) 1.049 (1.020-1.079) .0009odel 2 (Model 1 plus sociodemographic characteristics) 1.062 (1.031-1.094) �.0001odel 3 (Model 2 plus RxRisk score) 1.050 (1.018-1.082) .002odel 4 (Model 3 plus diabetes severity) 1.040 (1.008-1.073) .01odel 5 (Model 4 plus body mass index) 1.037 (1.004-1.070) .03odel 6 (Model 5 plus current smoking status) 1.035 (1.003-1.069) .03odel 7 (Model 6 plus foot self-care) 1.036 (1.004-1.070) .03odel 8 (Model 7 plus HbA1c) 1.036 (1.003-1.069) .03

CI � confidence interval; HR � hazard ratio; HbA1c � glycosylated hemoglobin.