Diabetes complications and depression as predictors of health service costs☆

Article Outline

• Abstract
• 1. Introduction
• 2. Methods
  • 2.1. Setting
  • 2.2. Sample selection
  • 2.3. Survey methods
  • 2.4. Survey measures
  • 2.5. Severity of illness measures
  • 2.6. Utilization and cost measures
  • 2.7. Data analysis
• 3. Results
• 4. Discussion
• References
• Copyright

Abstract 

Objective

The aim of this study was to assess the relative contributions of diabetes complications, depression and comorbid medical disorders to health service costs in adults with diabetes.

Methods

A total of 4398 adult health plan members with diabetes completed a mailed survey. Depression was assessed using the nine-item PHQ. Health service costs, diabetes complications, glycohemoglobin levels and comorbid medical conditions were assessed using computerized health plan records.

Results

Total health service costs were approximately 70% higher for individuals with major depression than for those without any depressive disorder (US$5361 over 6 months vs. US$3120, P<.001); this difference was consistent across all categories of health service costs. Diabetes complications were the strongest predictor of total costs (US$6845 for those with three or more complications vs. US$1719 for those with none), but depression remained strongly associated with increased costs at all levels of diabetes severity.

Conclusions

Among people with diabetes, depression is associated with 50-75% increases in health service costs. This proportional difference is similar to that in general population samples, but the absolute dollar difference is much greater. The effect of depression on health service use is undoubtedly complex and not limited to unexplained physical symptoms among the worried well.

Keywords: Diabetes, Depression, Cost, Complications, Utilization

Back to Article Outline

1. Introduction 

Early literature on health care utilization associated with depression focused on somatization, including unexplained somatic symptoms or exaggerated health concerns [1]. According to this view, many frequent users of medical care are actually the worried well expressing psychological distress through somatic symptoms. Effective treatment of underlying depression or anxiety might eliminate unexplained somatic symptoms and the need for unnecessary medical services.

Subsequent research work suggest a more complex relationship between depression, somatic symptoms, chronic medical illness and use of medical services. Rather than being worried well, those with depressive or anxiety disorders are actually more likely to suffer from chronic medical conditions [2], [3]. Among those with chronic illness, depression is associated with poorer clinical and functional outcomes [4], [5]. While depression may increase perception and reporting of somatic symptoms, depression may also be a consequence of the persistent pain [6] or the functional limitations caused by chronic illness [7], [8]. The association between depression and increased use of general medical services appears to be as strong among those with chronic illness as among those without significant medical conditions [9].

People with diabetes are an important population in which to study the effects of chronic medical illness and depression on use of health care. Total societal costs of diabetes are estimated to exceed US$130 billion annually [10], and the prevalence of Type 2 diabetes is rapidly increasing [11]. Furthermore, diabetes is a prototypical chronic illness with well-defined indicators of severity and progression, allowing an opportunity to examine the interaction of depression and severity of medical illness in predicting health care utilization. Three previous reports have examined the impact of depression on costs of care for diabetes in American national samples. Using Medicare claims, Finkelstein et al. [12] found that treatment of depression was associated with greater use of inpatient and outpatient medical services. Himelhoch et al. [13] found that a diagnosis of depression was associated with greater use of acute care services (emergency department and inpatient care) among Medicare beneficiaries with diabetes and other chronic medical conditions. Using Medical Expenditure Panel Survey data, Egede et al. [14] found that self-reported history of depression was associated with higher total health service costs among respondents with diabetes. In a sample of 367 health maintenance organization members with diabetes, Ciechanowski et al. [15] found that higher Hopkins Symptom Checklist depression scores were associated with significantly higher health service costs.

In this report, we used data from a large population-based sample of people treated for diabetes to examine the association between depression, severity of diabetes, severity of comorbid medical conditions and health service costs. These data add to previous research in three ways. First, depression was assessed using a structured diagnostic measure rather than a symptom scale. Second, detailed accounting records allowed complete capture of health service costs from the insurer perspective. Third, computerized medical records allowed objective examination of diabetes severity and diabetes complications as well as severity of comorbid medical conditions.

Back to Article Outline

2. Methods 

2.1. Setting 

Group Health Cooperative (GHC) is a mixed-model prepaid health plan serving approximately 500,000 members in Washington State. Most GHC members are enrolled via employer-purchased contracts, but approximately 20,000 are enrolled via risk-sharing contracts with Medicare and another 20,000 via risk-sharing contracts with Medicaid or other subsidized low-income programs. GHC enrollment is demographically similar to the area population. Study participants were selected from nine group-model primary care clinics. All study procedures were approved by the institutional review boards of the GHC and the University of Washington.

Adult primary care is provided by staff family practitioners and internists, with each full-time provider responsible for a defined panel of approximately 2300 members. Primary care physicians provide most of the diabetes care and the bulk of antidepressant treatment. Specialty mental health care is provided by staff providers (psychiatrists, psychologists, nurse practitioners and masters-level psychotherapists), with staffing levels similar to those of other group-model health plans [16].

2.2. Sample selection 

Potential participants were sampled from the GHC diabetes registry, a database including all adults meeting the following criteria: filled prescription for insulin or an oral hypoglycemic agent, two fasting plasma glucose levels ≥126 mg/dl in a 12-month period, two random plasma glucose levels ≥200 mg/dl in a 12-month period, two outpatient diagnoses of diabetes or any inpatient diagnosis of diabetes.

2.3. Survey methods 

Potential participants were contacted initially by mail with a package including an invitation letter, a survey booklet and a US$3 gift card. The invitation letter and survey booklet included all elements of informed consent (study purpose, risks, benefits, voluntary nature of participation) as well as a separate specific consent for review of computerized medical records. Those who did not respond but did not refuse were mailed a second survey 4 weeks later and a third (including telephone reminder and an offer to complete the survey by phone) 6 months later.

2.4. Survey measures 

The mail survey included questions regarding sociodemographic characteristics (age, sex, race, educational attainment, marital status, employment), diabetes characteristics (age at onset, initial treatment prescribed, type and duration of treatment), symptoms of depression (see below) and other measures not described here. Participants were classified as having Type 1 diabetes if age at diagnosis was at that younger than 30 years, insulin was the first treatment prescribed and insulin was reported as a current treatment.

Depression was assessed using the nine-item PHQ [17], a self-report measure of depression symptoms based on the American Psychiatric Association's DSM-IV [18] criteria for diagnosis of depressive episodes. Validation studies have shown excellent agreement between the self-report PHQ and a clinician structured interview in samples of general medical outpatients [17], [19] and medical inpatients [20]. Following the standard method, each item was scored as positive if endorsed as “More than half the time” or “Nearly all the time” and a diagnosis of major depression required a positive response to one of the two core symptoms (depressed mood or loss of interest) and a total of five positive symptoms. Criteria for subthreshold depression required at least one core symptom (depressed mood or loss of interest) and a total of two to four positive symptoms.

2.5. Severity of illness measures 

Computerized records of visit diagnoses, hospital discharge diagnoses and laboratory results for the 12 months prior to the survey were used to identify seven specific diabetes complications: retinopathy, nephropathy, neuropathy, cerebrovascular disease, cardiovascular disease, peripheral vascular disease and metabolic complications (ketoacidosis). Computerized pharmacy records were used to compute the R×Risk score [21], a measure of overall medical comorbidity and predicted health service costs. In previous research, the R×Risk score has shown excellent ability to predict health service utilization [21], [22]. The numerical score represents predicted health service costs for the next 6 months. For this study, antidepressant and hypoglycemic medications were not included in the calculation of the R×Risk score. This modified R×Risk score should be considered a measure of overall medical morbidity other than depression or diabetes. Computerized laboratory data were used to identify the most recent HbA_1c level during the last 12 months.

2.6. Utilization and cost measures 

Computerized cost accounting records were used to examine all health service costs during the 6 months following the survey. GHC's cost accounting system assigns budget-based costs (rather than charges) to every unit of health service (visit, hospital day, laboratory test, prescription) provided at GHC facilities. Services purchased from outside facilities or providers are listed at the cost paid by GHC. Primary analyses divided costs into three categories: (a) mental health treatment costs (all outpatient and inpatient specialty mental health care, all antidepressant drugs and all other outpatient visits with a mental health diagnosis); (b) diabetes treatment costs (all visits and hospitalizations associated with any diagnosis of diabetes, all prescriptions for insulin or oral hypoglycemic drugs and all laboratory tests for glucose, HbA_1c or microalbuminuria); and (c) other costs (the remainder). For any visit or service associated with diagnoses of both depression and diabetes, cost was equally divided between the two categories. Visits or hospitalizations for potential diabetes complications (e.g., renal failure, vascular disease) were only included in the diabetes category if associated with a diabetes diagnosis.

Additional analyses examined outpatient visit rates in specific clinical categories. ICD-9 visit diagnoses were classified a priori according to the following scheme (details available upon request):

2.7. Data analysis 

For most cost categories, primary analyses compared mean costs across the three depression groups (no depression, minor depression, major depression) using standard analysis of variance (ANOVA) with untransformed cost as the response variable. The hypothesis of significant heterogeneity across the three groups was tested using an F statistic with a two-sided significance threshold of 5%. Distributions for inpatient costs were highly skewed and included a nonsignificant number of zero observations, so primary comparisons were based on Kruskal–Wallis nonparametric ANOVA. Additional secondary analyses used standard ANOVA to compare log-transformed costs [log of (cost+US$1)]. Results were consistent for all three methods (standard ANOVA, log-transformed ANOVA, Kruskal–Wallis), but we report on the nonparametric analysis for inpatient costs as this test was generally the most conservative.

Because participants did differ modestly from nonparticipants in a variety of clinical and demographic characteristics (see below), logistic regression was used to calculate response probabilities and sampling weights based on 14 individual characteristics measured by administrative data for the prior 12 months: age, sex, most recent HbA_1c value, any treatment with insulin, any use of oral hypoglycemic medicines, any use of specialty mental health care, any depression diagnosis in primary care or specialty care, any prescription for an antidepressant medication, any hospitalization and overall medical comorbidity as measured by the R×Risk score [21]. Analyses using these weights to adjust for nonresponse yielded nearly identical results, so unweighted analyses are reported here.

Back to Article Outline

3. Results 

Surveys were mailed to 9064 potential participants, but 1222 participants were later found to be ineligible because of death (n=128), withdrawal from their health plan or moving out of the area (n=444), erroneous diagnosis of diabetes (n=267), medical illness severe enough to preclude completion of the questionnaire (n=202), cognitive impairment severe enough to preclude participation (n=80), language barriers or other communication difficulties (n=99) or other reasons (n=2). Of the remaining 7842 eligible participants, 4832 (61.7%) returned surveys with complete depression data and 4463 gave permission for use of computerized records (including cost and utilization data). An additional 65 participants withdrew from their health plan within 6 months, leaving 4398 eligible for the analyses presented here. After exclusion of those denying permission to use their computerized records, survey participants were compared with nonparticipants using available computerized data. Compared with those participating, nonparticipants had a lower mean age, higher probability of medical specialty care use, higher probability of comorbid heart disease, higher mean HbA_1c levels and higher probability of having no HbA_1c test in the last year. Participants and nonparticipants did not differ in the probability of mental health diagnosis or treatment in the last year.

The clinical and demographic characteristics of the sample are summarized in Table 1. As reported previously, participants with current depression were younger, more often female, less often married and more likely to use insulin. Depression was also associated with higher HbA_1c levels and more diabetes complications. While significance tests indicated statistically significant heterogeneity in race, R×Risk score and educational attainment, none of these characteristics showed a consistent relationship to level of depression.

Table 1. Demographic and clinical characteristics according to depression status

	No depression (n=3510)	Subthreshold depression (n=371)	Major depression (n=517)	Test statistics
Age [mean (S.D.)]	63.9 (13.0)	64.3 (13.6)	59.1 (13.4)	F=31.7, df=2, P<.001
Women (%)	47	50	58	χ2=21.7, df=2, P<.001
Caucasian (%)	79	73	76	χ2=6.7, df=2, P=.03
Married (%)	64	61	55	χ2=18.9, df=2, P<.001
>12 years of education (%)	75	68	71	χ2=12.3, df=2, P=.002
R×Risk score [mean (S.D.)]	3392 (2955)	3750 (3020)	3492 (3125)	F=9.3, df=2, P<.001
HbA1c [mean (S.D.)]	7.70 (1.51)	7.94 (1.59)	8.09 (1.59)	F=16.7, df=2, P<.001
Diabetes treatment				χ2=55, df=4, P<.001
Diet only (%)	27	23	18
Oral agent only (%)	46	43	40
Any insulin (%)	27	34	42
No. of diabetes complications				χ2=39, df=6, P<.001
0 (%)	33	26	25
1 (%)	32	29	30
2 (%)	18	23	20
≥3 (%)	17	22	25

Table 2 displays specific cost categories according to depression status. As expected, mental health treatment costs were strongly related to current depressive disorder. For all other categories (diabetes treatment, other medical costs and total health service costs), costs for those with major depression were approximately 70% greater than for those with no depressive disorder. These relationships were highly statistically significant using alternative analytic methods such as Kruskal–Wallis nonparametric tests or ANOVA for log-transformed costs (P<.001 for all comparisons). In comparing those who have major depression with those who have none, increased costs attributable to mental health treatment (approximately US$275) accounted for less than 15% of the increase in total costs (approximately US$2240).

Table 2. Categories of health service costs according to depression status

Values represent means (S.D.s).

Other potential predictors of total health service costs are illustrated in Fig. 1. While total cost was significantly related to increasing age, increasing R×Risk score and number of diabetes complications, this relationship was strongest for diabetes complications. Cost was only modestly and inversely related to the most recent HbA_1c level. Secondary analyses examined the association between total health service costs and individual complications. Each category of complications was strongly and significantly associated with increased cost (data not shown).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1. 

  Predictors of total health service costs over 6 months for the sample of 4398 health plan members with diabetes.

The joint effects of depression and diabetes complications on total cost are shown in Table 3. Increasing severity of depression (i.e., moving from left to right in Table 3) was consistently associated with increasing health service costs at all levels of diabetes complications. Increasing number of diabetes complications (i.e., moving from top to bottom in Table 3) was consistently associated with increasing health service costs at all levels of depression. ANOVA was used to examine the independent contributions of depression and diabetes complications to prediction of heath care costs after adjusting for age, sex, R×Risk score and most recent HbA_1c level. Both number of complications (F=47, P<.001) and depression (F=9.4, P<.001) remained strong predictors of total health service costs.

Table 3. Total health service costs according to depression status and number of diabetes complications

No. of diabetes complications	No depression	Subthreshold depression	Major depression	Total
0	1152 (26)	95 (2)	132 (3)	1379 (31)
	US$1719 (13)	US$1576 (1)	US$2331 (2)	US$1768 (31)
1	1126 (25)	108 (2)	153 (3)	1387 (32)
	US$2539 (19)	US$2934 (2)	US$3770 (4)	US$2705 (25)
2	651 (15)	87 (2)	103 (2)	841 (19)
	US$3281 (14)	US$3868 (2)	US$5547 (4)	US$3619 (20)
≥3	581 (13)	81 (2)	129 (3)	791 (18)
	US$6845 (26)	US$7545 (4)	US$10200 (9)	US$7464 (39)
Total	3510 (80)	371 (8)	517 (12)	4398 (100)
	US$3120 (72)	US$3812 (9)	US$5361 (18)	US$3442 (100)

Values for each category represent number of participants (percentage of all participants) and mean health service costs (percentage of costs for the total sample accounted for by those in this cell).

Table 3 also illustrates how depression and diabetes complications predict costs at the population level. For example, the 517 patients with current major depression contribute to approximately 12% of the sample and to 18% of total costs. The 791 patients with three or more diabetes complications contribute to 18% of the sample and to approximately 39% of total costs.

Additional analyses attempted to identify specific clinical presentations responsible for increased costs in those with depressive disorders. As shown in Table 4, depression was associated with significantly higher visit rates for acute illness, chronic illness, ill-defined conditions and mental health care. Dermatological and vision/hearing visits showed no association with depression, and depression was associated with slightly lower visit rates for preventive care.

Table 4. Categories of outpatient visits according to depression status

Values represent means (S.D.s).

Back to Article Outline

4. Discussion 

In a population-based sample of people with diabetes, we found that having major depression was associated with an approximately 70% increase in overall health service costs compared with not having any depressive disorder. Mental health services accounted for less than 15% of this increase. While health service costs were strongly related to diabetes complications and overall medical illness severity, adjustment for these factors had no significant impact on the relationship between depression and costs. Patients with current major depression constituted 12% of this sample but accounted for 18% of total health service costs. Our central findings are broadly consistent with a large body of research on the association between depression and use of health services [9], [23], [24], [25] and a smaller body of research specific to diabetes [12], [14], [15].

Consistent with previous research by Finkelstein et al. [12], Himelhoch et al. [13] and Egede et al. [14], we find that depression comorbid with diabetes was associated with significant increases in health service costs. We assessed depression using a standardized and well-validated measure of DSM-IV criteria [17], [19] in contrast to previous research using self-report of depression treatment [14] or physician diagnosis of depression from billing data [12], [13]. Furthermore, our data allow a more detailed examination of the joint effects of depression and medical illness severity. Depressed patients had higher rates of diabetes complications and comorbid medical illness, the association between depression and increased cost remained significant after adjusting for these differences. As shown in Table 3, depression was associated with an approximately 50% increase in health service cost at every level of diabetes complication.

The strong association between depression and use of health services is clearly not limited to the worried well — those with anxiety or depressive disorders and no significant medical disorder. Previous research in samples of general medical outpatients typically find 50-75% higher costs in those with current depression. We find a similar effect of depression among people with diabetes, including those with the most severe illness. While the proportional increase is similar, the absolute dollar impact is much greater in those with greater medical comorbidity. In earlier samples of unselected primary care patients, 6-month health care costs were approximately US$1000 higher in those with current depression. Among people with diabetes, we observe a mean difference of approximately US$2000, but this difference ranges from approximately US$500 among those with no complications to more than US$3000 among those with three or more diabetes complications.

The excess utilization associated with depression was not limited to specific services (e.g., excess diagnostic testing) or clinical presentations (e.g., unexplained somatic symptoms). The clearest exception to the pattern of generalized increase was preventive services, where depression was associated with lower visit rates. We have previously reported that depression was associated with decreased adherence to hypoglycemic, antihypertensive and lipid-lowering medications [26]. This pattern is consistent with research showing that depressed medical patients may use more overall services [9], [24] but still receive less effective care [26], [27], [28].

Depending on how one views the association between depression and diabetes complications, our methods may underestimate the impact of depression on health service costs. We have previously reported that depression is associated with a greater risk of diabetes complications as well as several important health risk factors such as tobacco use and obesity [29]. Previous cross-sectional and longitudinal studies have demonstrated an association between depression and development of microvascular and macrovascular complications [30], [31], [32]. Depression may directly increase the risk of diabetes complications by its effect on self-care and health behaviors [28], [33], [34] or by its adverse effect on the hypothalamic-pituitary-adrenal axis, the autonomic nervous system or inflammatory response. If some of the effects of depression on health care use are mediated by these mechanisms, then controlling for diabetes complications may underestimate the true impact of depression on service use [35].

Several limitations should be considered in interpreting these data. First, this population-based sample reflected the full range of diabetes severity and included only a small number with more severe illness or Type 1 diabetes. Our results might not generalize to specialty settings with higher rates of Type 1 diabetes or more severe illness. Second, our data were collected from a single group-model managed care organization and findings might not generalize to other health care systems. We have no reason to believe, however, that the effect of depression on health care utilization would be smaller in health care systems where service use is less tightly managed. Third, there may remain unmeasured differences in medical comorbidity between the depressed and nondepressed groups. Our measures of diabetes severity and medical comorbidity cannot completely capture clinical differences that might be associated with increased costs. Fourth, a significant number of those eligible did not respond to the survey and nonrespondents were younger and appeared to have poorer physical and mental health. Propensity score adjustments for nonresponse had no meaningful effect on our findings; we must acknowledge that results might not generalize to the nonrespondents. Finally, patients with successfully treated depression would be included in our nondepressed group. Consequently, mental health treatment costs in this group were small but greater than 0.

Our findings certainly permit different possible explanations for the association between depression and health service use in people with diabetes. A strong and consistent association between depression and greater use of health services does not establish a causal relationship. Higher costs among depressed patients with diabetes may reflect unmeasured differences in severity of diabetes or other comorbid chronic conditions. More severe medical morbidity might produce greater pain and disability, leading to more depression [8]. In this case, one might not expect that effective depression treatment would reduce use of general medical care. Alternatively, depression may interfere with effective self-care (dietary control, medication adherence, physical activity) or increase somatic distress and health-related anxiety, leading to greater use of health services [33]. Our previous research indicate that depression in people with diabetes is associated with both greater sensitivity to physical symptoms [33], [36] and poorer adherence to medication and self-care recommendations [26], [33]. If these mechanisms predominate, then identification and treatment of depression could reduce the use of unnecessary medical services. In all likelihood, these relationships are bidirectional. Only random assignment of depression treatment will determine how much of the association between depression and increased service use is truly causal. Given the multiple predictors and high variability of health service costs, estimating the effects of improved depression treatment on overall costs will require large experimental studies of robust depression interventions.

For health care providers, these findings suggest the need for increased attention to depression in patients with diabetes who are frequent users of medical care. For health care systems, these findings point to the population with coexisting diabetes and depression as an important target for quality improvement and disease management.

Back to Article Outline

References 

1. Simon G. Psychiatric disorder and functional somatic symptoms as predictors of health care use. Psychiatr Med. 1992;10(3):49–60
   • View In Article
   • MEDLINE
2. Katon W. Clinical and health services relationships between major depression, depressive symptoms, and general medical illness. Biol Psychiatry. 2003;54:216–226
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (124 KB)
   • CrossRef
3. Rouchell A, Pounds R, Tierney J. Depression. In:  Wise M,  Rundell J editor. Textbook of consultation–liaison psychiatry. 2nd ed.. Washington (DC): American Psychiatric Publishing, Inc.; 2002;
   • View In Article
4. Wells K, Rogers W, Burnam M, Camp P. Course of depression in patients with hypertension, myocardial infarction, or insulin-dependent diabetes. Am J Psychiatry. 1993;150:632–638
   • View In Article
   • MEDLINE
5. Koike A, Unutzer J, Wells K. Improving the care for depression in patients with comorbid medical illness. Am J Psychiatry. 2002;159:1738–1745
   • View In Article
   • MEDLINE
   • CrossRef
6. Gureje O, Simon G, Von Korff M. A cross-national study of the course of persistent pain in primary care. Pain. 2001;92:195–200
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (69 KB)
   • CrossRef
7. Cavanaugh S, Clark D, Gibbons R. Diagnosing depression in the hospitalized medically ill. Psychosomatics. 1983;24:809–815
   • View In Article
   • Abstract
   • Full-Text PDF (1231 KB)
   • CrossRef
8. Prince M, Harwood R, Thomas A, Mann A. A prospective population-based cohort study of the effects of disablement and social milieu on the onset and maintenance of late-life depression. The Gospel Oak Project VII. Psychol Med. 1998;28:337–350
   • View In Article
   • MEDLINE
   • CrossRef
9. Simon G, VonKorff M, Barlow W. Health care costs of primary care patients with recognized depression. Arch Gen Psychiatry. 1995;52:850–856
   • View In Article
   • MEDLINE
10. Zhang P, Engelgau M, Norris S, Gregg E, Narayan V. Application of economic analysis to diabetes and diabetes care. Ann Intern Med. 2004;140:972–977
    • View In Article
11. Engelgau M, Geiss L, Saaddine J, et al. The evolving diabetes burden in the United States. Ann Intern Med. 2004;140:945–950
    • View In Article
12. Finkelstein E, Bray J, Chen H, et al. Prevalence and costs of major depression among elderly claimants with diabetes. Diabetes Care. 2003;26:415–420
    • View In Article
    • MEDLINE
    • CrossRef
13. Himelhoch S, Weller W, Wu A, Anderson G, Cooper L. Chronic medical illness, depression, and use of acute medical services among Medicare beneficiaries. Med Care. 2004;42:512–521
    • View In Article
    • MEDLINE
    • CrossRef
14. Egede L, Zheng D, Simpson K. Comorbid depression is associated with increased health care use and expenditures in individuals with diabetes. Diabetes Care. 2002;25:464–470
    • View In Article
    • MEDLINE
    • CrossRef
15. Ciechanowski P, Katon W, Russo J. Depression and diabetes: impact of depressive symptoms on adherence, function, and costs. Arch Intern Med. 2000;160:3278–3285
    • View In Article
    • MEDLINE
    • CrossRef
16. Dial T, Bergsten C, Haviland M, Pincus H. Psychiatrist and nonphysician mental health provider staffing levels in health maintenance organizations. Am J Psychiatry. 1998;155:405–408
    • View In Article
    • MEDLINE
17. Spitzer R, Kroenke K, Williams J. Validation and utility of a self-report version of PRIME-MD: the PHQ Primary Care Study. JAMA. 1999;282:1737–1744
    • View In Article
    • MEDLINE
    • CrossRef
18. American Psychiatric Association . Diagnostic and statistical manual of mental disorders. 4th ed.. Washington (DC): American Psychiatric Association; 1994;
    • View In Article
19. Kroenke K, Spitzer R, Williams J. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16:606–613
    • View In Article
    • MEDLINE
    • CrossRef
20. Diez-Quevedo C, Rangil T, Sanchez-Planell L, Kroenke K, Spitzer R. Validation and utility of the patient health questionnaire in diagnosing mental disorders in 1003 general hospital Spanish inpatients. Psychosom Med. 2001;63:679–686
    • View In Article
    • MEDLINE
21. Fishman P, Goodman M, Hornbrook M, Meenan R, Bachman D, Okeeffe Rosetti M. Risk adjustment using automated ambulatory pharmacy data: the R×Risk model. Med Care. 2003;41:84–99
    • View In Article
    • MEDLINE
    • CrossRef
22. Clark D, VonKorff M, Saunders K, Baluch WM, Simon GE. A chronic disease score with empirically derived weights. Med Care. 1995;33:783–795
    • View In Article
    • MEDLINE
    • CrossRef
23. Simon G, Ormel J, VonKorff M, Barlow W. Health care costs associated with depressive and anxiety disorders in primary care. Am J Psychiatry. 1995;152:352–357
    • View In Article
    • MEDLINE
24. Unützer J, Patrick DL, Simon G, et al. Depressive symptoms and the cost of health services in HMO patients aged 65 and over: a four-year prospective study. JAMA. 1997;277:1618–1623
    • View In Article
    • MEDLINE
25. Katon W, Lin E, Russo J, Unutzer J. Increased medical costs of a population-based sample of depressed elderly patients. Arch Gen Psychiatry. 2003;60:897–903
    • View In Article
    • MEDLINE
    • CrossRef
26. Lin E, Katon W, VonKorff M, et al. Relationship of depression and diabetes self-care, medication adherence and preventive care. Diabetes Care. 2004;27:2154–2160
    • View In Article
    • MEDLINE
    • CrossRef
27. Druss B, Bradford D, Rosenheck R, Radford M, Krumholz H. Mental disorders and the use of cardiovascular procedures after myocardial infarction. JAMA. 2000;283:506–511
    • View In Article
    • MEDLINE
    • CrossRef
28. DiMatteo M, Lepper H, Croghan T. Depression is a risk factor for noncompliance with medical treatment: meta-analysis of the effects of anxiety and depression on patient adherence. Arch Intern Med. 2000;160:2101–2107
    • View In Article
    • MEDLINE
    • CrossRef
29. Katon W, von Korff M, Ciechanowski P, et al. Behavioral and clinical factors associated with depression among individuals with diabetes. Diabetes Care. 2004;27:914–920
    • View In Article
    • MEDLINE
    • CrossRef
30. de Groot M, Anderson R, Freedland K, Clouse R, Lustman P. Association of depression and diabetes complications: a meta-analysis. Psychosom Med. 2001;63:619–630
    • View In Article
    • MEDLINE
31. Black S. Increased health burden associated with comorbid depression in older diabetic Mexican Americans. Results from the Hispanic Established Population for the Epidemiologic Study of the Elderly Survey. Diabetes Care. 1999;22:56–64
    • View In Article
    • MEDLINE
    • CrossRef
32. Black S, Markides K, Ray L. Depression predicts increased incidence of adverse health outcomes in older Mexican Americans with type 2 diabetes. Diabetes Care. 2003;26:2822–2828
    • View In Article
    • MEDLINE
    • CrossRef
33. Ciechanowski P, Katon W, Russo J, Hirsch I. The relationship of depressive symptoms to symptom reporting, self-care, and glucose control in diabetes. Gen Hosp Psychiatry. 2003;25:246–252
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (617 KB)
    • CrossRef
34. Wang P, Bohn R, Knight E, Glynn R, Mogun H, Avorn J. Noncompliance with antihypertensive medications: the impact of depressive symptoms and psychosocial factors. J Gen Intern Med. 2002;17:504–511
    • View In Article
    • MEDLINE
    • CrossRef
35. Weinberg C. Toward a clear definition of confounding. Am J Epidemiol. 1993;137:1–8
    • View In Article
    • MEDLINE
36. Ludman E, Katon W, Russo J, et al. Depression and diabetes symptom burden. Gen Hosp Psychiatry. 2004;26:430–436
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (142 KB)
    • CrossRef