Nonadherence to Depression Treatment Guidelines Among Veterans With Diabetes Mellitus

December 1, 2006
Laura E. Jones, PhD
Laura E. Jones, PhD

,
Carolyn Turvey, PhD
Carolyn Turvey, PhD

,
James C. Torner, PhD
James C. Torner, PhD

,
Caroline Carney Doebbeling, MD, MSc
Caroline Carney Doebbeling, MD, MSc

Volume 12, Issue 12

Objective: To assess the adequacy of antidepressant dosage andduration among veterans with and without diabetes mellitus (DM),as well as provider-level and patient-level predictors of depressioncare quality, based on Veterans Health Administration (VHA) evidence-based clinical practice guidelines.

Study Design: Retrospective (1997-2005) cohort study ofadministrative, clinical, and pharmacy data from a midwesternVHA facility.

Methods: The sample included 2332 subjects (773 with DM)who had a new episode of depression, received antidepressanttherapy, and had neither schizophrenia nor bipolar disorder. Antidepressantdosage and duration were evaluated in the acute andcontinuation phases. Dosage was adequate if the treatment dosagemet the minimum therapeutic dosage specified in VHA guidelines.Treatment duration was adequate if the medication possession ratiowas at least 80%. Multivariate logistic regression analysis was usedto calculate odds ratios (ORs), adjusted for demographic, clinical,and healthcare utilization characteristics.

Results: Most subjects received an adequate dosage during theacute (88%) and continuation (58%) phases. Subjects with DMwere 1.51-fold more likely to receive adequate dosage during theacute phase but were similarly likely (OR, 1.15) to receive adequatedosage during the continuation phase. Few subjects (<10%)received adequate treatment duration. Diabetes mellitus was notassociated with less adequate duration during the acute phase (OR,1.14). Few factors were identified as significant predictors of bothantidepressant dosage and duration.

Conclusions: Diabetes mellitus did not adversely affect depressioncare quality. Adequate antidepressant dosages were prescribed,but treatment duration fell short of guideline recommendations.Strategies to more effectively manage depression treatment areneeded.

(Am J Manag Care. 2006;12:701-710)

Depression is a serious debilitating illness withhigh prevalence among the general populationand among Veterans Health Administration(VHA) patients.1,2 Depression disproportionately affectsthose with chronic medical comorbidity, such as diabetesmellitus (DM).3 Unfortunately, depression recognitionand treatment are poor. Only 46% to 51% of patientswith depression are detected in primary care, and onlyhalf of those receive treatment.4

Major national organizations have endorsed clinicalpractice guidelines (CPGs) outlining minimum effectiveantidepressant dosage and duration,5,6 yet undertreatmentof depression is common, although wide variation(11%-90%) is reported.7-18 However, depression care adequacymay be lower than is reported. For example, priorstudies9,11,15,19-21 have based duration adequacy on medicationrefills (which are problematic given the variationin days' supply) or on medication possession ratios(MPRs) less than the standard 80%. Dosage adequacyhas not always been based on guideline-recommendeddosages.13,17 Inadequate depression treatment is clinicallysignificant for several reasons, including poor qualityof life, economic burden, increased healthcare utilization,and poor medical and psychiatric outcomes.22-25

Patients with DM may be at greater risk for lower-qualitydepression care because of competing clinicaldemands (eg, a focus on DM aspects of care), or theymay conversely receive better depression care becauseof increased provider contact and more opportunitiesto receive optimal treatment.4,26,27 We are aware ofonly 2 studies14,15 that provide information regardingdepression care quality among the population withDM. The findings of the first study,14 using computerizedpharmacy records, suggested that 31% receivedadequate antidepressant dosage. The second study15reported that 46% of older persons with DM receivedadequate depression care (based on &#8805;4 antidepressantprescriptions at adequate dosage or &#8805;8psychotherapy visits) but that they were 77% morelikely to receive adequate depression care than adultswithout DM.

We are unaware of any studies that have assessed thelongitudinal nature of depression care quality amongpersons with DM. This is an important issue from a clinicalperspective and from a policy perspective because itmay highlight provider awareness of treatment guidelinesand factors that may contribute to adequate orinadequate depression care among a high-risk population.The objectives of this research were to assess (1)the quality of antidepressant dosage and the durationduring the acute and continuation phases of treatmentand (2) the predictors of depression care quality amongdepressed veterans. We hypothesized that veterans withDM were less likely to receive adequate depression carecompared with those without DM.

METHODS

Study Data

The data included 8.33 years (from January 1, 1997,to April 30, 2005) of administrative, clinical, and pharmacydata from a midwestern Veterans Affairs medicalcenter. The data comprised inpatient encounters andoutpatient encounters (primary care, specialty medicine,and mental health). Pharmacy data included thename and National Drug Code of outpatient drugs, filldate, quantity, dosage, and days' supply.

Study Population

International Classification of

Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)

Subjects with an diagnosis of DM or depression were initially selectedfor inclusion. Further inclusion criteria required that subjects(1) were diagnosed as having a new episode ofdepression, (2) were followed up for at least 180 daysbefore the diagnosis and for the 264-day treatment duration,and (3) received antidepressant therapy within 84days following the depression diagnosis (Figure). Subjectswith comorbid schizophrenia or bipolar disorder wereexcluded because of the focus on unipolar depression.

Cases.

ICD-9-CM

The case population included depressed subjectswith DM. Identification of DM was based on validatedcriteria for VHA data, which specify indication of atleast 2 codes for DM (250.xx) in inpatient oroutpatient data during a 24-month period or receipt of aDM prescription medication.28 All cases were diagnosed ashaving DM before follow-up for depression or treatment.

Control Subjects.

ICD-9-CM

ICD-9-CM

Controls included all subjects atthis facility with an diagnosis of depressionwho received antidepressant therapy and who had noindication of DM, based on codes, DM medications,glycosylated hemoglobin testing results, orimpaired glucose tolerance.

Identification of a New Episode of Depression

ICD-9-CM

ICD-9-CM

Depression was identified via a single codefor major depressive disorder (MDD) (code 296.2 or296.3), dysthymia (code 300.4), or depressive disordernot otherwise specified (code 311). A single code was used for identification because depression isundercoded in administrative and clinical data.7,17,18 Wedid not specifically focus on MDD because many patientswith depression seek care solely in primary care,and depression not otherwise specified is the most commondiagnosis in primary care.9,29 The follow-up periodfor identification of a new episode of depression beganon or after June 30, 1997, to allow for a 180-day pretreatmentperiod. Standardized diagnostic interviewdata to confirm the diagnoses were unavailable.

ICD-9-CM

ICD-9-CM

ICD-9-CM

Prescription data were used to identify new episodesof depression, defined by an diagnosis ofdepression that was not preceded by a 180-day period(an accepted minimum standard in the literature) withindication of antidepressant use or an depressiondiagnosis.7,9,13,30 Symptoms of depression may haveexisted, but there was neither a record of diagnosis nortreatment. If pretreatment-phase requirements were notmet for the first-recorded depression diagnosis,then subsequent periods were analyzed to determinethe next earliest start of a new episode ofdepression. Otherwise, subjects were excluded if thesecriteria were not satisfied.

Depression Treatment Phases

The acute phase of treatment was the 84-day periodfollowing the depression diagnosis. The continuationphase was the 180-day period thereafter.

Guideline-Concordant Treatment of Depression

The 2000 VHA CPGs for depression were used toevaluate depression care quality.6 We focused on antidepressantdosage and duration during the 84-day acuteand 180-day continuation phases. Other forms of therapy(eg, psychotherapy) were not addressed because theyare not included in VHA CPGs.

Antidepressant Duration.

We first analyzed thequantity and days' supply of antidepressants that weredispensed. The end date was calculated by adding thedays' supply to the fill date. If a prescription was refilledbefore the end date, an oversupply of medication wasnoted, which was added to the days' supply of subsequentprescriptions. This made it possible to develop aperson-time calendar to indicate on which days a subjectwas exposed to antidepressant medication. Amongthe 23% of patients who were exposed to multiple antidepressantson the same day (eg, because of drugswitching or augmentation), the prescription with thelongest days' supply was counted. The MPR, defined asthe sum of the days' supply divided by the number ofdays in the treatment phase, was used to assess durationadequacy. The literature suggests that an 80% cutoff isreasonable.20,21 Duration adequacy was defined as adichotomy, with an MPR of at least 80% representing anadequate duration and an MPR of less than 80% representinginadequate duration.

Antidepressant Dosage.

Dosage adequacy refers tothe mean treatment-phase dosage and whether thetreatment dosage met minimum therapeutic dosagesspecified in VHA guidelines.6 First, the mean dailydosage was calculated by multiplying the drug strength(in milligrams) by the quantity of drug dispensed dividedby the days' supply and then converted to fluoxetinehydrochloride equivalents (Table 1). If multiple prescriptionsfor the same agent were filled on the sameday, the dosages were combined if the days' supplywas the same for each prescription. Otherwise, theantidepressant with the highest fluoxetine-equivalentdaily dosage was chosen. Second, the mean treatment-phase dosage was calculated by summing themean daily doses and dividing by the number of prescriptiondays in the period. An adequate treatmentdosage was achieved if the mean fluoxetine-equivalentdosage was at least 20 mg/d for younger subjects (<65years) or at least 10 mg/d for older subjects (&#8805;65years). Otherwise, the treatment dosage was consideredinadequate.

General Considerations.

Subjects who did notreceive antidepressant therapy during the continuationphase were considered to have received inadequatedosage and duration given that VHA guidelines explicitlystate that pharmacologic treatment should be continuedif provided during the acute phase. All subjects werefollowed up during the entire acute and continuationphases of treatment, as stated in the inclusion criteria.

Potential Confounders

Diagnostic and Statistical

Manual of Mental Disorders, Fourth Edition

ICD-9-CM

Demographic, clinical, and healthcare utilizationcharacteristics were examined as possible correlates ofdepression care quality. These characteristics werechosen a priori based on information in the literature.1,3,8,9,14-18 Demographic characteristics included age,sex, race/ethnicity, marital status, educational attainment,and service-connected disability percentage. Agewas included as a dichotomous variable (&#8805;65 vs <65years) in multivariate analyses because antidepressantdosage differs for these age cutoffs. Service-connecteddisability percentage is used to determine the level ofVHA access to healthcare and the amount of copaymentfor prescription medications. Medical comorbidity wasbased on a count of 29 conditions included in thecomorbidity index by Elixhauser et al31 and other conditionsprevalent among the VHA population2 using themethod by Klabunde et al.32 Psychiatric comorbiditywas based on a count of 14 conditions.Prior indication of depression was based on whethercodes for depression were present before thediagnosis of a new episode of depression. A categoricalvariable was created to indicate the type of depression.Healthcare utilization included the location of thediagnosis and the number of outpatient visits to primarycare and mental health during each treatment phase.Clinic stop codes, which identify the type of ambulatoryencounter in the VHA, were used to determine providerspecialty given that specific provider codes wereunavailable in these data.

Statistical Analysis

t

Demographic, clinical, and healthcare utilizationcharacteristics were compared using &#967;2 tests for categoricalvariables and tests for continuous variables.Multivariate logistic regression models were developedto calculate adjusted odds ratios (OR) and 95% confidenceintervals (CIs) to determine if DM status predicteddepression care quality during the acute andcontinuation phases of treatment.33 Univariate logisticregression analysis results are available from the author.All variables included in the multivariate models wereidentified a priori based on salient variables noted inthe literature.1,3,8,9,14-18 A multivariate analysis was notconducted for continuation-phase duration adequacybecause so few patients received adequate treatment.All analyses were conducted using SAS 9.1 (SASInstitute Inc, Cary, NC). Two-tailed tests were used todetermine statistical significance, with a set at .05.The institutional review boards at Roudebush VeteransAffairs Medical Center, Indiana University, and the Universityof Iowa approved this study.

RESULTS

Sample Description

A total of 2332 subjects with depression (773 [33%]with DM) met the study inclusion criteria. Subjects werecommonly excluded because of insufficient follow-up ornonreceipt of antidepressant therapy (Figure).

The subjects with DM were more likely to be male,married, older, and less educated and have more medicalcomorbidity but less psychiatric comorbidity thanthe subjects without DM. They were also less likely tohave been diagnosed as having depression in a mentalhealth clinic. Subjects with DM had less healthcare utilizationduring the acute phase than subjects withoutDM but had more visits during the continuation phase,although the differences were not clinically significant.Serotonergic agents were most commonly prescribed(Table 2).

Duration of Antidepressant Therapy

P

P

Few subjects (<10%) received adequate antidepressanttreatment duration in the acute or continuationphase. Overall, the mean MPR during the 264-day treatmentperiod for all subjects was 27%, corresponding topossession of antidepressant medication for only 71days. The presence of DM did not affect the mean treatmentduration in the acute phase (> .05) but was statisticallyassociated with an increased treatmentduration (albeit a nonclinically significant association)in the continuation phase (= .02) (Table 3). The overalllow MPR in the continuation phase is attributed tothe 35% of subjects who did not receive antidepressanttherapy at any time during this 180-day period and, bydefinition, were considered to have an MPR of 0%. Evenamong the 1525 subjects who received antidepressanttherapy during the continuation phase, only 2% had anMPR of at least 80%. Adequacy of treatment durationwas not associated with the presence of DM in theacute phase (OR, 1.14; 95% CI, 0.81-1.61) (Table 4).Multivariate analysis identified increased odds forreceipt of adequate treatment duration among men,unmarried persons, and those with at least a high schooleducation, increased medical and psychiatric comorbidities,and higher healthcare utilization rates.

Antidepressant Dosage

P

Most subjects received an antidepressant dosagethat was consistent with guideline-recommended minimumtherapeutic dosages during the acute phase (88%)and the continuationphase (58%), regardlessof DM status. Thirty-fivepercent of subjects didnot continue to receiveantidepressant therapyduring the continuationphase of treatmentand, by definition, wereconsidered to have receivedan inadequatedosage (Table 3). However,among the 1525subjects who receivedantidepressant therapyduring the continuationphase, adequacy rates inthe continuation phasewere similar to those inthe acute phase (89% vs88%), and subjects withDM were still significantlymore likely tohave received a therapeuticdosage (92% vs88%, = .01). In multivariateanalyses, subjects with DM were statisticallymore likely to have received an adequate dosage duringthe acute phase (OR, 1.51; 95% CI, 1.10-2.08) butnot during the continuation phase (OR, 1.15; 95% CI,0.94-1.41) compared with subjects without DM (Table 3).In the acute phase, older persons, those with at leasta high school education, and veterans with prior indicationof depression were more likely to receive anadequate dosage. In contrast, medical comorbidity,depression not otherwise specified, and diagnosis in amental health clinic were negatively associated withdosage adequacy. Continuation-phase analyses indicatedthat increased medical comorbidity and healthcareutilization were associated with higher odds of dosageadequacy but that nonwhite race/ethnicity was associatedwith lower odds of dosage adequacy.

DISCUSSION

The following 2 major findings result from thisresearch: (1) most (>85%) depressed veterans receivedadequate antidepressant dosage, yet many (>90%) didnot receive the minimum guideline-recommended treatmentduration, and (2) the presence of DM did notresult in less adequate depression care. The VHA guidelinesexplicitly state that a minimum of 4 to 9 months oftreatment should be provided after acute-phase treatment.In this sample, however, the mean treatmentduration was only 2 to 3 months following depressiondiagnosis. This research uniquely contributes to the literatureby describing the longitudinal nature of depressiontreatment, the use of evidence-based depressiontreatment guidelines in day-to-day practice amongpatients with DM, and the finding that treatment is poor,regardless of DM status.

Our results are similar and dissimilar to resultsreported previously. The literature reports wide variation(11%-90%) regarding adequacy of dosage and duration.7-18 Differences in patient (eg, demographics),provider (eg, guideline awareness), and organizational(eg, resource availability) characteristics in this studymay partially explain the variation. The observed 88%rate of dosage adequacy in this study is comparable tothe 80% to 91% adequacy rates reported in at least 3prior VHA studies9,10,12 but was significantly higher thanthe 29% adequacy rate reported among another VHAsample.7 The observed less than 10% adequacy of treatmentduration in our study is lower than previouslyreported rates (11%-85%)9,16,18 and may be related to themore stringent criteria we used to categorize treatmentadequacy (ie, an MPR of &#8805;80%), which we believed tobetter represent treatment guidelines.

Important methodological differences among ourstudy and others bear mention. First, our longitudinalstudy assessed treatment adequacy during both theacute and continuation phases. A previous study9assessed treatment quality during a single treatmentphase and may have identified subjects with depressionas early as 19 months before antidepressants were prescribed.In contrast, subjects in our study receivedantidepressants within 12 weeks of the depression diagnosis,with most (80%) within 1 week of the depressiondiagnosis. Second, inclusion of patients with any unipolardepression diagnosis in this study, rather than inclusionof patients with only MDD, may also account fordiscrepancies. Treatment initiation and maintenancerates may be higher in studies that focus solely on MDDbecause of disease severity. Certain patient characteristicsmay be associated with the severity of the depression,which may also bias the results if they affectdosage or duration. The VHA CPGs for depression arefor major depression. An assumption of this work is thatdosing and duration are based on the physician's decisionto pharmacologically treat the depression, irrespectiveof the particular depression diagnosis.

Our findings support prior work that DM is not arisk factor for inadequate depression treatment comparedwith controls.15 In fact, persons with DM had14% to 51% higher odds for receipt of adequate treatmentin our study. Contrary to our hypothesis, competingclinical demands may not result in lower-qualitydepression care among the population with DM.Subjects with DM may be more likely to receive adequatedepression care because of their healthcare-seekingbehavior. Prior research shows that depressedpersons with DM use more ambulatory care than nondepressedpersons with DM.27,34 Increased healthcareutilization was a significant predictor of several of theoutcome measures in this study. Increased providercontact corresponds to more opportunities to promotepatient compliance with treatment, while allowingproviders the opportunity to adjust treatment.Adequate follow-up care for depression was the mostsignificant predictor of antidepressant treatment durationin another study.35 In addition, a higher burden ofcomorbidity may result in providers taking a proactiveapproach in treating depression among patients withDM, which may correspond to a greater likelihood foradequate depression care.

These results suggest that factors other than DM areresponsible for inadequate treatment of depression,although few predictors were uniformly identified.Factors associated with adequate depression care in thisstudy (eg, older age and MDD) mirror those reported inother VHA depression studies9,16 and add credibility tothe reported results. Younger adults were less likely toreceive an adequate dosage, but this may be partiallyexplained by the fact that VHA guidelines6 recommendlower starting dosages for older patients. This likelyexplains why age was associated with dosage but notwith duration. The finding that increased medical andpsychiatric comorbidities were associated with lowerodds of dosage adequacy but with higher odds of durationadequacy is unclear, although we speculate thatproviders may be hesitant to initially prescribe higherdosages because of concern for adverse effects or druginteractions. Treatment duration was likely higher forthose with more comorbidity because of increasedprovider contact, resulting in more opportunities to prescribemedications or to educate patients regardingtreatment. We did not address other salient predictors ofdepression care quality, including patient beliefs,provider knowledge, and organizational resources.Inadequate depression treatment does not necessarilyreflect deficiencies in clinician provision of treatmentbut may also reflect patient rejection or underappreciationof the need for long-term treatment. Priorwork suggests that it is difficult to separate patient andprovider adherence, highlighting the importance of promotingpatient and provider partnerships.36

Study Limitations

ICD-9-CM

The results of this study should be interpreted withseveral limitations in mind. First, restricting depressioncare profiling to a single VHA facility may limit generalizabilityto other VHA facilities and to the general public.The veteran population is a predominately oldermale population with significant medical and psychiatriccomorbidities, further limiting generalizability tothe general public. The methods for identification of anew episode of depression may have resulted in anunderestimation of the true treatment rates if the specificityof the diagnosis is low, although treatment rateswere not significantly associated with the number ofdepression diagnoses (data not shown) or thetype of depression. Furthermore, missing information(eg, race/ethnicity), misclassification of known variables,and unmeasured characteristics (eg, access tonon-VHA care) may have affected the observed findings.The use of the MPR to assess treatment durationmay provide spurious results, although recent worksuggests that other metrics available to assess durationadequacy provide results similar to thoseobtained using the MPR.21 Finally, these data do notindicate the clinical necessity of care and treatmentfor depression. Clinical practice guidelines werederived to represent what is best for most patients andshould not be construed as a standard of medical carefor all patients.

Study Strengths

We believe that this is the first study among the veteranpopulation with DM to describe adherence todepression guidelines. The use of rigorous and validatedmethods helped to ensure the reliability and accuracy ofthe reported findings and allowed the results to beextrapolated to other VHA populations. The inclusioncriteria were less stringent, and all pharmacologic therapieswere evaluated. The agreement between the VHA'selectronic medical record database and medical recordreview with respect to treatment of depression is high,supporting the use of VHA administrative, clinical, andpharmacy data to assess depression care quality.7 To ourknowledge, our study is the first to conduct a longitudinalassessment of treatment adequacy during both theacute and continuation phases.

Clinical Implications

ICD-9-CM

The low rates of adequate depression care in thisstudy, despite recognition by providers and treatment(typically within 7 days of the diagnosis), suggestan opportunity for substantial practice improvement.These findings should motivate physicians' heightened awareness of CPG treatment specifications.Because adequate depression care follow-up has beenshown to be the most effective predictor of treatmentduration,35 clinicians and quality managers in the VHAmay wish to implement changes to the current organizationalstructure to ensure that depressed veterans havetimely follow-up care, that they have access to the mostappropriate resources (eg, mental health), and thatstrategies are in place to continually monitor treatmentquality and outcomes. Although this study was not poweredto specifically examine the clinical effects of guideline-concordant depression treatment, the clinicalimplications of inadequate treatment are profound.Inadequate treatment may result in poor patient outcomes(eg, psychiatric hospitalization) and may negativelyaffect the healthcare system as a result of costs.25Because current dosing regimens for selective serotoninreuptake inhibitors make adequate treatment dosage lessconfusing for providers, issues contributing to adherenceand treatment duration are paramount for furtherresearch. Finally, these data elucidate that deficienciesin depression treatment exist but do not pinpoint specificdriving factors for inadequate treatment duration.Further work is required to elucidate why depressiontreatment is not congruent with evidence-basedpractices and to formulate strategies to help aligncurrent practice with evidence-based practices.

Acknowledgments

We thank Ada Yeung and the Roudebush Veterans Affairs Medical CenterHealth Services Research and Development Service Center of Excellencefor administrative support and data access.

From the Roudebush Veterans Affairs Medical Center Health Services Research andDevelopment Service (VAMC HSR&D) Center of Excellence on Implementing Evidence-Based Practice (LEJ), Department of Internal Medicine, Indiana University School ofMedicine (CCD), and Regenstrief Institute (CCD), Indianapolis; and Department ofEpidemiology, College of Public Health (LEJ, JCT, CCD), and Department of Psychiatry,College of Medicine (CT), University of Iowa, and Iowa City VAMC HSR&D Center forResearch in the Implementation of Innovative Strategies in Practice (CT), Iowa City.

This research was presented at the 29th Annual Meeting of the Society of GeneralInternal Medicine; April 27, 2006; Los Angeles, Calif.

Address correspondence to: Laura E. Jones, PhD, Roudebush VAMC HSR&D Center ofExcellence on Implementing Evidence-Based Practice, Roudebush VAMC HSR&D 11H,1481 W 10th St, Indianapolis, IN 46202. E-mail: lj2@iupui.edu.

JAMA.

1. Kessler RC, Berglund P, Demler O, et al. The epidemiology of major depressivedisorder: results from the National Comorbidity Survey Replication (NCS-R). 2003;289:3095-3105.

J Ambul Care Manage.

2. Selim AJ, Fincke G, Ren XS, et al. Comorbidity assessments based on patientreport: results from the Veterans Health Study. 2004;27:281-295.

Diabetes Care.

3. Anderson RJ, Freedland KE, Clouse RE, Lustman PJ. The prevalence of comorbiddepression in adults with diabetes: a meta-analysis. 2001;24:1069-1078.

J Gen Intern Med.

4. Goldman LS, Nielsen NH, Champion HC. Awareness, diagnosis, and treatmentof depression. 1999;14:569-580.

Am J Psychiatry.

VHA/DoD Clinical

5. American Psychiatric Association. Practice guideline for the treatment of patientswith major depressive disorder (revision). 2000;157(suppl):1-45.6. Management of Major Depressive Disorder Working Group.

Practice Guideline for Management of Major Depressive Disorder in Adults.

Washington, DC: Dept of Veterans Affairs; 2000.

Jt Comm J Qual Saf.

7. Kramer TL, Owen RR, Cannon D, et al. How well do automated performancemeasures assess guideline implementation for new-onset depression in the VeteransHealth Administration? 2003;29:479-489.

Psychiatr Serv.

8. Weilburg JB, O'Leary KM, Meigs JB, Hennen J, Stafford RS. Evaluation of theadequacy of outpatient antidepressant treatment. 2003;54:1233-1239.

Med Care.

9. Charbonneau A, Rosen AK, Ash AS, et al. Measuring the quality of depressioncare in a large integrated health system. 2003;41:669-680.

Gen Hosp

Psychiatry.

10. Dobscha SK, Gerrity MS, Corson K, Bahr A, Cuilwik NM. Measuring adherenceto depression treatment guidelines in a VA primary care clinic. 2003;25:230-237.

Med Care.

11. Katon W, von Korff M, Lin E, Bush T, Ormel J. Adequacy and duration of antidepressanttreatment in primary care. 1992;30:67-76.

J Behav

Health Serv Res.

12. Chen RS, Rosenheck R. Using a computerized patient database to evaluateguideline adherence and measure patterns of care for major depression. 2001;28:466-474.

lin Ther.

13. McCombs JS, Shi L, Stimmel GL, Croghan TW. A retrospective analysis of therevocation of prior authorization restrictions and the use of antidepressant medicationsfor treating major depressive disorder. C2002;24:1938-1959.

Med Care.

14. Katon WJ, Simon G, Russo J, et al. Quality of depression care in a population-basedsample of patients with diabetes and major depression. 2004;42:1222-1229.

J Am Geriatr Soc.

15. Harman JS, Edlund MJ, Fortney JC, Kallas H. The influence of comorbid chronicmedical conditions on the adequacy of depression care for older Americans.2005;53:2178-2183.

Med Care.

16. Busch SH, Leslie D, Rosenheck R. Measuring quality of pharmacotherapy fordepression in a national health care system. 2004;42:532-542.

Jt Comm J Qual Improv.

17. Kerr EA, McGlynn EA, Van Vorst KA, Wickstrom SL. Measuring antidepressantprescribing practice in a health care system using administrative data: implicationsfor quality measurement and improvement. 2000;26:203-216.

Arch Gen Psychiatry.

18. Young AS, Klap R, Sherbourne CD, Wells KB. The quality of care for depressiveand anxiety disorders in the United States. 2001;58:55-61.

Dis

Manag Health Outcomes.

19. White T, Vanderplas A, Ory C, Dezii C, Chang E. Economic impact of patientadherence with antidepressant therapy within a managed care organization. 2003;11:817-822.

Am J Manag Care.

20. Sikka R, Xia F, Aubert RE. Estimating medication persistency using administrativeclaims data. 2005;11:449-457.

Med Care.

21. Cantrell CR, Eaddy MT, Shah MB, Regan TS, Sokol MC. Methods for evaluatingpatient adherence to antidepressant therapy: a real-world comparison of adherenceand economic outcomes. 2006;44:300-303.

Am J Manag Care.

22. Sood N, Treglia M, Obenchain RL, Dulisse B, Melfi CA, Croghan TW.Determinants of antidepressant treatment outcome. 2000;6:1327-1336.

Arch

Intern Med.

23. Katon W, Cantrell CR, Sokol MC, Chiao E, Gdovin JM. Impact of antidepressantdrug adherence on comorbid medication use and resource utilization. 2005;165:2497-2503.

Am J Manag Care.

24. Weilburg JB, Stafford RS, O'Leary KM, Meigs JB, Finkelstein SN. Costs of antidepressantmedications associated with inadequate treatment. 2004;10:357-365.

Med Care.

25. Charbonneau A, Rosen AK, Owen RR, et al. Monitoring depression care: insearch of an accurate quality indicator. 2004;42:522-531.

Gen Hosp Psychiatry.

26. Klinkman MS. Competing demands in psychosocial care: a model for the identificationand treatment of depressive disorders in primary care. 1997;19:98-111.

Manag Care Interface.

27. Kalsekar ID, Madhavan SM, Amonkar MM, Scott V, Douglas SM, Makela E.The effect of depression on health care utilization and costs in patients with type 2diabetes. 2006;19:39-46.

Diabetes Care.

28. Miller DR, Safford MM, Pogach LM.Who has diabetes? best estimates of diabetesprevalence in the Department of Veterans Affairs based on computerized patient data.2004;27(suppl 2):B10-B21.

Med Care.

29. Croghan TW, Melfi CA, Dobrez DG, Kniesner TJ. Effect of mental health specialtycare on antidepressant length of therapy. 1999;37(suppl Lilly):AS20-AS23.

Med Care.

30. Melfi CA, Croghan TW. Use of claims data for research on treatment and outcomesof depression care. 1999;37(suppl Lilly):AS77-AS80.

Med Care.

31. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for usewith administrative data. 1998;36:8-27.

J Clin Epidemiol.

32. Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidityindex using physician claims data. 2000;53:1258-1267.

Applied Logistic Regression.

33. Hosmer DW, Lemeshow S. 2nd ed. New York, NY:John Wiley & Sons Inc; 2000.

Diabetes

Care.

34. Egede LE, Zheng D, Simpson K. Comorbid depression is associated withincreased health care use and expenditures in individuals with diabetes. 2002;25:464-470.

Gen Hosp

Psychiatry.

35. Jones LE, Turvey C, Carney Doebbeling C. Inadequate follow-up care for depressionand its impact on antidepressant treatment duration among veterans withand without diabetes mellitus in the Veterans Health Administration. In Press.

J Clin Psychiatry.

36. McCombs JS, Nichol MB, Stimmel GL, Sclar DA, Beasley CM Jr, Gross LS. Thecost of antidepressant drug therapy failure: a study of antidepressant use patterns ina Medicaid population. 1990;51(suppl):60-71.