Collaborative Care Management of Late-Life Depression in the Primary Care Setting: A Randomized Controlled Trial | Depressive Disorders | JAMA | JAMA Network
[Skip to Navigation]
Sign In
Figure 1. Flowchart of Participants in the Trial
Image description not available.
Asterisk indicates that most (90%-95%) of the ineligible subjects did not meet screening or research diagnostic criteria for depression. Dagger indicates the cumulative number of participants who were deceased and who were therefore excluded from analysis; analyses included all other patients after multiple imputation of unit-level missing data. SCID indicates structured clinical interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition.
Figure 2. Mean SCL-20 Depression Score
Image description not available.
Scores on the 20 depression items from the Symptom Checklist–90 (SCL-20) ranged from 0 to 4. Error bars indicate SEs.
Table 1. Characteristics of Participating Organizations*
Image description not available.
Table 2. Baseline Patient Characteristics*
Image description not available.
Table 3. Depression Care
Image description not available.
Table 4. Clinical Outcomes*
Image description not available.
Original Contribution
December 11, 2002

Collaborative Care Management of Late-Life Depression in the Primary Care Setting: A Randomized Controlled Trial

Author Affiliations

Author Affiliations: Center for Health Services Research, UCLA Neuropsychiatric Institute, Los Angeles, Calif (Drs Unützer, Tang, and Belin and Ms Oishi); Department of Psychiatry, University of Washington (Dr Katon), and Center for Health Studies, Group Health Cooperative of Puget Sound (Dr Lin), Seattle; Indiana University Center for Aging Research, Regenstrief Institute for Health Care, Indianapolis (Dr Callahan); Center for Health Services Research in Primary Care, Durham Veterans Affairs Medical Center and Duke University Medical Center (Dr Williams), and Department of Medicine, Duke University Medical Center (Dr Harpole), Durham, NC; Division of Research, Kaiser Permanente of Northern California, Oakland (Ms Hunkeler); Desert Medical Group, Palm Springs, Calif (Dr Hoffing); Kaiser Permanente of Southern California, San Diego (Dr Penna); South Texas Veterans Health Care System/University of Texas Health Science Center, San Antonio (Dr Noël); Department of Psychiatry, University of California, San Francisco (Dr Areán); Department of Psychiatry, Dartmouth Medical School, Hanover, NH (Dr Hegel); and John A. Hartford Foundation, New York, NY (Dr Langston).

JAMA. 2002;288(22):2836-2845. doi:10.1001/jama.288.22.2836

Context Few depressed older adults receive effective treatment in primary care settings.

Objective To determine the effectiveness of the Improving Mood–Promoting Access to Collaborative Treatment (IMPACT) collaborative care management program for late-life depression.

Design Randomized controlled trial with recruitment from July 1999 to August 2001.

Setting Eighteen primary care clinics from 8 health care organizations in 5 states.

Participants A total of 1801 patients aged 60 years or older with major depression (17%), dysthymic disorder (30%), or both (53%).

Intervention Patients were randomly assigned to the IMPACT intervention (n = 906) or to usual care (n = 895). Intervention patients had access for up to 12 months to a depression care manager who was supervised by a psychiatrist and a primary care expert and who offered education, care management, and support of antidepressant management by the patient's primary care physician or a brief psychotherapy for depresssion, Problem Solving Treatment in Primary Care.

Main Outcome Measures Assessments at baseline and at 3, 6, and 12 months for depression, depression treatments, satisfaction with care, functional impairment, and quality of life.

Results At 12 months, 45% of intervention patients had a 50% or greater reduction in depressive symptoms from baseline compared with 19% of usual care participants (odds ratio [OR], 3.45; 95% confidence interval [CI], 2.71-4.38; P<.001). Intervention patients also experienced greater rates of depression treatment (OR, 2.98; 95% CI, 2.34-3.79; P<.001), more satisfaction with depression care (OR, 3.38; 95% CI, 2.66-4.30; P<.001), lower depression severity (range, 0-4; between-group difference, −0.4; 95% CI, −0.46 to −0.33; P<.001), less functional impairment (range, 0-10; between-group difference, −0.91; 95% CI, −1.19 to −0.64; P<.001), and greater quality of life (range, 0-10; between-group difference, 0.56; 95% CI, 0.32-0.79; P<.001) than participants assigned to the usual care group.

Conclusion The IMPACT collaborative care model appears to be feasible and significantly more effective than usual care for depression in a wide range of primary care practices.

Major depression and dysthymic disorder affect between 5% and 10% of older adults seen in the primary care setting.1-3 Late-life depression is often chronic or recurrent4-8 and is associated with substantial suffering, functional impairment, and diminished health-related quality of life.9 Depressed, older primary care patients are frequent users of general medical services5,10-12 and may have poor adherence to medical treatments.13 They are also at increased risk of death from suicide14 and medical illnesses.15-17 Although late-life depression can be successfully treated with antidepressant medications or psychotherapy,18-21 few depressed older adults receive adequate trials of such treatments in primary care22-28 or see a mental health specialist.25,29-35 Efforts to improve late-life depression treatment using screening and health care practitioner feedback and education have not resulted in consistent improvements in depression.22,23 A more comprehensive intervention strategy may be needed to improve outcomes for this population.

We enrolled 1801 depressed, older adults from 18 primary care clinics across the United States in a randomized trial of a primary care–based collaborative care model for late-life depression, the Improving Mood–Promoting Access to Collaborative Treatment (IMPACT) program,36 compared with care as usual. The IMPACT intervention includes key components of evidence-based models for chronic illness care37,38: collaboration among primary care practitioners, patients, and specialists on a common definition of the problem, development of a therapeutic alliance, a personalized treatment plan that includes patient preferences, proactive follow-up and outcomes monitoring by a depression care manager, targeted use of specialty consultation, and protocols for stepped care. Intervention patients had access to an IMPACT care manager for up to 12 months. Usual care patients could use any primary care or specialty mental health care services available to them in usual care. After 12 months, all study participants continued with their regular primary care practitioners as usual.

This article examines health outcomes throughout 12 months. We hypothesized that patients assigned to the IMPACT intervention would have higher rates of depression treatment, greater satisfaction with depression care, greater improvements in depression, less health-related functional impairment, and higher quality of life than usual care patients.


The IMPACT study is a multisite randomized controlled trial of a collaborative intervention program for late-life depression in primary care.36,39-41 Study protocols were developed in collaboration with all participating organizations, reviewed by a study advisory board, and approved by the institutional review boards at all sites and the study coordinating center. All participants gave written informed consent.


Seven study sites representing 8 diverse health care organizations with a total of 18 primary care clinics in 5 states participated in the study (Table 1). We estimated that 650 participants each were required in the intervention and control groups to have a 95% chance of detecting as significant (at the 2-sided .05 level) a difference of 0.10 (SD, 0.50) in the mean score of the 20 depression items from the Symptom Checklist–90 (SCL-20)42 depression scores. To compensate for patient attrition, we planned to enroll 875 patients per group. To identify a sample of depressed, older adults who could participate in a quality improvement intervention such as IMPACT under real-world practice conditions, each site used a 2-pronged strategy to recruit study participants from July 1999 to August 2001 (Figure 1).36

The first strategy relied on referrals of depressed older adults from primary care practitioners, other clinic staff, or patients themselves in response to clinic promotions of the program. The second method consisted of systematic depression screening of English-speaking, older adults who used the participating primary care clinics with a 2-item depression screener adapted from the PRIME-MD study. These screens were administered either in person or by telephone.43 Of the 32 908 patients approached for screening, 5246 (16%) either refused to be screened or participated in the initial screening but refused further interviews. A total of 1791 (5%) of the initial screens were incomplete, and 23 233 (71%) of those screened were not eligible because they did not endorse one of the core depression symptoms (95% of those ineligible) or because of logistic reasons, such as lack of transportation or access to a telephone (5% of those ineligible). Of the 2190 patients referred to the study, 308 (14%) refused the initial eligibility screen or further interviews. Fifty-four (3%) had incomplete initial screens, and 202 (9%) were ineligible because they were younger than 60 years or they did not plan to use the participating clinic during the coming 12 months.

The remaining 2638 (8%) of those screened and 1626 (74%) of those referred underwent a 30- to 60-minute structured, computer-assisted interview by trained lay interviewers to determine study eligibility.36 This interview included the structured clinical interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (SCID) to assess whether patients met research diagnostic criteria for major depression or dysthymia.44,45 Inclusion criteria were age 60 years or older, plans to use one of the participating clinics as the main source of general medical care in the coming year, and a diagnosis of current major depression or dysthymic disorder according to the SCID. Approximately 2% (n = 99) of otherwise eligible patients were excluded because of current drinking problems (a score of ≥2 on the CAGE questionnaire)46; 3% (n = 145) were excluded because of a history of bipolar disorder or psychosis36; 2% (n = 85) were excluded because they were in ongoing treatment with a psychiatrist; and approximately 1% (n = 44) were excluded because they met screening criteria for severe cognitive impairment defined by a score of less than 3 on a 6-item cognitive screen.47 Less than 1% were excluded because they were found to be at acute risk for suicide and needed immediate care. Our 2-pronged recruitment method identified 2102 eligible older adults with major depression or dysthymic disorder (approximately 2% of all older adults served at the participating clinics); 1801 (86% of those eligible) enrolled in the study and completed a structured baseline interview.36

After the baseline interview, participants were randomly assigned to the IMPACT intervention or usual care. The random assignment was stratified by recruitment method (screening or referral) and clinic. Within each stratum, participants were assigned according to a random number sequence that was developed using a computer random number generator at the coordinating center. Random assignment information was contained in a set of numbered, sealed envelopes for each stratum that were used sequentially for newly enrolled patients at each clinic.36


The IMPACT intervention has been described in detail elsewhere.36,39,40,48 Intervention participants received a 20-minute educational videotape and a booklet about late-life depression49,50 and were encouraged to have an initial visit with a depression care manager at the primary care clinic. Care managers were nurses or psychologists who were trained for the study as a depression clinical specialist (DCS).36,39,40 During the initial visit, the DCS conducted a clinical and psychosocial history, reviewed the educational materials, and discussed patient preferences for depression treatment (antidepressant medications or psychotherapy). New cases and cases needing treatment plan adjustments were discussed with a supervising team psychiatrist and a liaison primary care physician during a weekly team meeting. The DCS then worked with the patient and his/her regular primary care practitioner to establish a treatment plan according to a recommended treatment algorithm, but patients and their primary care practitioners made the actual treatment choices.36 The IMPACT treatment algorithm suggested an initial choice of an antidepressant medication (usually a selective serotonin reuptake inhibitor) or a course of Problem Solving Treatment in Primary Care (PST-PC), a 6- to 8-session, brief, structured psychotherapy for depression,51-55 delivered by the DCS in the primary care setting. For patients who were already taking antidepressant medications but who were still depressed, the recommendation was to increase the dose or to augment the antidepressant with a trial of PST-PC (for partial responders) or to switch to a different medication or PST-PC (for nonresponders). Patients' regular primary care practitioners were asked to write all antidepressant prescriptions. The DCSs also encouraged patients to schedule pleasant life events and referred them to additional health or social services as clinically indicated.

As care managers, DCSs attempted to follow up patients for up to 12 months, monitoring treatment response with the Patient Health Questionnaire 956 and a Web-based clinical information system.57 During the acute treatment phase, in-person or telephone follow-up contacts were suggested at least every other week. Patients who achieved recovery from depression (≥50% reduction in the Patient Health Questionnaire 9 score and fewer than 3 of 9 symptoms of major depression) were engaged in developing a relapse prevention plan and then followed up monthly by the DCS. Patients who did not respond to initial treatment were discussed with the IMPACT team and a "step 2" treatment plan was developed that could include augmentation of an antidepressant medication, a switch to a different antidepressant, a switch from medications to PST-PC, or vice versa. Team psychiatrists were encouraged to see patients who presented diagnostic challenges or who had persistent depression for in-person consultations in the primary care setting. Patients who did not respond after 10 weeks of step 2 treatment were again reviewed by the team, and additional treatments, such as further medication changes, psychotherapy, hospitalization, or electroconvulsive therapy, were considered.

Data Collection

We used baseline and 3-, 6-, and 12-month follow-up data from all 1801 study participants. Baseline interviews were conducted by trained lay interviewers using structured computerized interviews36 before randomization; thus, the interviewers were blind to study assignment. Blind follow-up interviews were performed at 3, 6, and 12 months by trained interviewers at a telephone survey research group using computer-assisted telephone interviews,36 with survey response rates of 90% at 3 months, 87% at 6 months, and 83% at 12 months (Figure 1).

Baseline interviews assessed sociodemographic characteristics, the severity of depressive symptoms using the SCL-20,42 SCID diagnoses of major depression or dysthymia,44,45 and health-related functional impairment using an index developed from the Sheehan disability scale that incorporates impairments in work, family, and other social functioning.58,59 Respondents rated their overall quality of life in the past month (including physical and mental well-being) on a scale from 0 (about as bad as dying) to 10 (life is perfect) and indicated whether they had been diagnosed as having or had been treated for any of 10 common chronic medical problems in the past 3 years. The Cornell Services Index60 and additional questions about the use of antidepressant medications, counseling, or psychotherapy assessed health services use in the past 3 months.61 Earlier research at one of our study sites found high rates of agreement between self-reported antidepressant use and prescription fill data from a pharmacy database.62,63

Outcomes Examined

Dependent variables in our analyses included self-reported use of antidepressants or psychotherapy, satisfaction with depression care (percentage who answered "excellent" or "very good"), mean SCL-20 depression scores, treatment response (≥50% decrease in SCL-20 score from baseline), complete remission of depression symptoms (SCL-20 score <0.5), major depression as assessed by the SCID, health-related functional impairment, and quality of life. We estimated the costs of providing IMPACT intervention services based on detailed study records of all patient contacts, mean salary and benefit costs of DCSs plus 30% overhead costs, and the cost of supervision and consultation from team psychiatrists and primary care experts.


We conducted bivariate analyses to compare demographic and clinical characteristics of intervention and usual care patients at baseline (Table 2). For each dependent variable, we conducted an intention-to-treat analysis of repeated measures. We fitted mixed-effects regression models for continuous variables or mixed-effects logistic regression models for dichotomous variables using baseline and 3-, 6-, and 12-month follow-up data with regression adjustment for recruitment method (screening or referral) and participating study organizations. In the mixed-effects models, we treated time as a categorical variable and examined the fixed effects for time, intervention condition, and their interactions. We specified the covariance structure within patients using an unstructured model to account for the within-patient correlation over time.64 For predicting depression severity and major depression at follow-up, we performed additional analyses that tested the interaction of intervention status with recruitment method (referral or screening), participating organizations, and depression diagnosis (major depression or dysthymia). Because of multiple comparisons, we used a conservative P value of less than .01 to detect statistically significant differences. All analyses were conducted using SAS software, version 8 (SAS Institute Inc, Cary, NC).

We used an extended hot deck multiple imputation technique that modifies the predictive mean matching method65,66 to impute item-level missing data.67 Rates of item-level missing data were less than 2% for all variables discussed in this article. The results across 5 imputed data sets were combined by averaging, and SEs were adjusted to reflect both within-imputation variability and between-imputation variability.67 Although there were no significant differences in the completion rate of follow-up interviews between the intervention and usual care groups, we found somewhat different predictors of follow-up response in intervention and usual care patients. We used an approximate Bayesian bootstrap multiple imputation method68 to impute unit-level missing data and adjust for these differences. Imputations were conducted separately in the intervention and usual care groups.


The enrolled sample was clinically and sociodemographically diverse (Table 2). The mean age of participants was 71.2 years (SD, 7.5 years), and 65% were women. With 23% of participants from ethnic minority groups (12% African Americans, 8% Latinos, and 3% other ethnic minorities), we had a somewhat greater representation of ethnic minorities than a national sample of older adults.69 Most participants (53%) met diagnostic criteria for major depression and dysthymic disorder, and 71% reported 2 or more prior depressive episodes. The mean SCL-20 depression score42 was 1.7 (SD, 0.6), indicating moderate to severe depression. Six percent of participants reported thoughts of suicide in the past month. One third (35%) showed some evidence of cognitive impairment, and 29% screened positive for panic disorder or posttraumatic stress disorder. Participants reported a mean of 3.2 (SD, 1.7) of 10 common comorbid medical conditions. About half (46%) reported depression treatment (antidepressant medication or psychotherapy) in the past 3 months.63 We found no significant differences in sociodemographic and clinical characteristics between the intervention and control groups.

Intervention Implementation

Most (98%) of the 906 intervention participants completed an initial visit with a DCS. Intervention participants had a mean of 9.15 (SD, 6.17) in-person visits and 6.10 (SD, 5.13) telephone contacts with a DCS, and 11% were seen for a consultation by a team psychiatrist. The majority (80%) had at least 1 trial of an antidepressant, and approximately one third (30%) received PST-PC. The mean number of PST-PC sessions was 6.34 (SD, 4.26).

Process of Care

Intervention patients were significantly more likely to use antidepressants or psychotherapy than usual care participants at all follow-ups (Table 3). Intervention patients reported antidepressant use for 6.6 months (SD, 4.9 months) of the 12-month study period compared with 4.6 months (SD, 5.0 months) in the usual care group (t = 8.12, P<.001). They also reported greater satisfaction with depression care at 3 and 12 months (satisfaction was not assessed at 6 months). Four patients in the usual care group and 5 intervention patients reported psychiatric hospitalizations during the 12-month study period.

Clinical Outcomes

Intervention patients had significantly lower depression severity (measured by SCL-20 depression scores) during all follow-up points, with the difference between intervention and usual care increasing from the 3- to the 12-month follow-up (Figure 2). Intervention patients also had significantly higher rates of treatment response (at least 50% reduction in the SCL-20 score from baseline) and of complete remission of depressive symptoms (SCL-20 score of <0.5)70 than usual care participants (Table 4). We are not aware of any attempted or completed suicides in either group.

We observed significant main effects of participating organizations on depression. For example, the proportion of patients who met criteria for major depression across the participating sites ranged from 57% to 84% at baseline and from 18% to 36% at 6 months. We did not find any significant interaction effects of intervention status with organization (F = 1.61; P = .13), intervention with recruitment method (t = 1.17; P = .24), or intervention with baseline depression diagnosis (major depression or dysthymia, t = −1.12; P = .26). Across all sites, intervention patients had a significantly greater reduction in rates of major depression (from a mean of 71% at baseline to 22% at 6 months) than usual care participants (from 68% to 35%; the SCID to assess major depression was not administered at 3 or 12 months).

Intervention patients also reported less health-related functional impairment (P<.001 at 3 and 12 months, P = .02 at 6 months) and greater overall quality of life in the past month (P<.001 at all follow-ups) than usual care participants (Table 4). We conducted sensitivity analyses using design-based permutation tests. For each outcome variable, we used the imputation version least favorable to the intervention. These conservative analyses gave similar results and are not presented in this article.

Intervention Costs

We estimate the mean costs of providing IMPACT services to be $553 per intervention patient for a 12-month period. These costs include $7 for the educational brochure and videotape, $418 for DCS services, $70 for supervision and in-person consultations with team psychiatrists, and $58 for supervision of DCSs by primary care experts. All visits with DCSs and team psychiatrists were provided free of charge to the patients. Patients and their insurers were responsible for all other health care costs, including antidepressant medications. Information on these costs will be reported in a subsequent article.


Recent studies69,71-73 have reported significant increases in rates of antidepressant use during the past 10 years. Almost half of our patients reported depression treatment during the 3 months before the study and more than half of our usual care patients reported antidepressant use or psychotherapy during the 12-month study period. Our findings suggest that despite this recent increase in antidepressant use, treatment of late-life depression in primary care remains challenging. At the 12-month follow-up, only 19% of usual care patients reported at least a 50% reduction in depressive symptoms from baseline and only 8% were completely free of depression symptoms.

Compared with these relatively modest effects of usual care treatment, intervention patients had significantly higher rates of depression treatment, greater satisfaction with depression care, and greater improvements in depression. Our findings are similar to earlier studies of collaborative care for mixed-aged adults with depression that integrated psychiatrists or psychologists into primary care settings and found greater improvements in depression in intervention than usual care patients.74,75 Our treatment effects in terms of number needed to treat to achieve a treatment response at 12 months (number needed to treat, 4; 95% confidence interval, 3-5) are similar to a number needed to treat of 4 reported in a Cochrane review of antidepressants compared with placebo or no treatment in medically ill adults.76 Subjects assigned to the IMPACT intervention also reported less health-related impairment in work, family, and social functioning and better quality of life than usual care patients, suggesting that the effects of this intervention on health extend beyond reducing depressive symptoms. We are particularly encouraged by the observation that differences between intervention and control patients in all health outcomes examined increased during the 12-month follow-up period. Longer-term follow-up will be needed to determine if these differences persist after discontinuation of the intervention resources at 12 months.

Our sample was recruited from 8 diverse health care organizations nationally, representing a wide variety of practices and patients.36 For example, the median household income of participants from the 8 organizations varied 5-fold ($8400 to $40 000 per year), and the proportion of patients with a high school education varied 3-fold (32% to 93%). We observed substantial intervention effects on depression at each of the 8 health care organizations, indicating that the IMPACT care model is feasible and effective in a range of primary care clinics that serve patients with widely diverse sociodemographic and clinical characteristics. We did not find significant interactions between intervention status and baseline depression diagnosis (major depression or dysthymic disorder) or between intervention status and recruitment method (screening or referral). We believe that our screening procedures identified a number of depressed older adults who might not have been recognized by their primary care practitioners and agree with the recent recommendation by the US Preventive Services Task Force that screening for depression in primary care is effective when coupled with systematic depression treatments such as those offered in our study.77

Despite substantial improvements in depression and quality of life, only approximately half of the intervention patients experienced a 50% reduction in depressive symptoms and only 25% to 30% became completely free of depressive symptoms. This may be due to greater medical comorbidity (a mean of 3.2 chronic medical illnesses and 65% with chronic pain), greater ambivalence about depression treatment among patients and practitioners, and lower treatment intensity in this effectiveness study conducted under naturalistic practice conditions compared with treatment efficacy studies with select samples in academic medical centers.36 It is also questionable whether complete freedom from symptoms, such as fatigue or lack of energy, is a realistic goal in older adults with multiple chronic medical illnesses. Further research is needed to examine the long-term outcomes of persistently depressed patients, to identify factors associated with treatment participation, adherence, and treatment resistance, and to develop effective interventions for this group. Possible strategies might include earlier and more aggressive use of in-person psychiatric consultation for nonresponders to antidepressants or psychotherapy in primary care or more aggressive use of additional treatments such as electroconvulsive therapy.

Most participants (51%) stated a preference for psychotherapy during the baseline interview before randomization, and 30% of intervention patients received a course of PST-PC in primary care. Given that only 8% of patients reported receiving counseling or psychotherapy in the 3 months before the baseline interview, it appears that the intervention program substantially increased the likelihood that patients received psychotherapy by offering this service in the primary care setting. However, when confronted with the need to travel to the clinic for PST-PC sessions, some patients may have opted to try antidepressants instead of PST-PC.

Our estimated 12-month intervention cost of $553 is consistent with cost estimates from an earlier study of collaborative care for depression using nurse care managers.78 It seems relatively modest given total annual Medicare spending of $5506 per enrollee in 199812,79 and the fact that health care costs for depressed, older adults are up to 50% higher than for older adults without depression.5,12 We plan to examine differences in total health care costs using administrative data from the participating organizations to compare the cost-effectiveness of the intervention to usual care.

Our study design may have biased our comparisons in favor of the usual care group.36 Referring practitioners were notified if a patient meeting study criteria was assigned to usual care, possibly resulting in treatment that would not have occurred in true usual care. Practitioners treated both intervention and usual care patients from 1999 to 2002; a spillover effect in which primary care practitioners applied improved skills learned from exposure to the intervention to the treatment of their usual care patients may have resulted. However, earlier studies23,80 have not found substantial effects of notification about depression status or physician training on usual care. Finally, we used a protocol to identify patients with thoughts of suicide during the follow-up interviews and referred them to appropriate clinical evaluation regardless of group assignment, possibly resulting in additional mental health care for the most depressed usual care participants. These biases may contribute to an underestimation of the effectiveness of the intervention compared with usual care outside a research setting. Additional limitations include our reliance on self-reports of chronic medical conditions and antidepressant and psychotherapy use. However, earlier research at 2 of our study sites61,62,81 found high rates of agreement between self-reported antidepressant use and prescription fill data from a pharmacy database.

The IMPACT model, a collaborative, stepped care management intervention for late-life depression, appears to be feasible and significantly more effective than usual care in a wide range of primary care practices.

Oxman TE, Barrett JE, Barrett J.  et al.  Symptomatology of late-life minor depression among primary-care patients.  Psychosomatics.1990;31:174-180.Google Scholar
Lyness JM, Caine ED, King DA.  et al.  Psychiatric disorders in older primary care patients.  J Gen Intern Med.1999;14:249-254.Google Scholar
Schulberg HC, Katon WJ, Simon GE.  et al.  Best clinical practice: guidelines for managing major depression in primary medical care.  J Clin Psychiatry.1999;60(suppl 7):19-26.Google Scholar
Alexopoulos GS, Chester JG. Outcomes of geriatric depression.  Clin Geriatr Med.1992;8:363-373.Google Scholar
Callahan CM, Hui SL, Nienaber NA.  et al.  Longitudinal study of depression and health services use among elderly primary care patients.  J Am Geriatr Soc.1994;42:833-838.Google Scholar
Cole MG, Bellavance F. The prognosis of depression in old age.  Am J Geriatr Psychiatry.1997;5:4-14.Google Scholar
Schulberg HC, Mulsant B, Schulz R.  et al.  Characteristics and course of major depression in older primary care patients.  Int J Psychiatry Med.1998;28:421-436.Google Scholar
Geerlings SW, Beekman AT, Deeg DJ.  et al.  Physical health and the onset and persistence of depression in older adults: an eight-wave prospective community-based study.  Psychol Med.2000;30:369-380.Google Scholar
Unützer J, Patrick DL, Diehr P.  et al.  Quality adjusted life years in older adults with depressive symptoms and chronic medical disorders.  Int Psychogeriatr.2000;12:15-33.Google Scholar
Luber MP, Hollenberg JP, Williams-Russo P.  et al.  Diagnosis, treatment, comorbidity, and resource utilization of depressed patients in a general medical practice.  Int J Psychiatry Med.2000;30:1-13.Google Scholar
Luber MP, Meyers BS, Williams-Russo PG.  et al.  Depression and service utilization in elderly primary care patients.  Am J Geriatr Psychiatry.2001;9:169-176.Google Scholar
Unützer J, Patrick DL, Simon G.  et al.  Depressive symptoms and the cost of health services in HMO patients aged 65 years and older.  JAMA.1997;277:1618-1623.Google Scholar
Ciechanowski PS, Katon WJ, Russo JE. Depression and diabetes: impact of depressive symptoms on adherence, function, and costs.  Arch Intern Med.2000;160:3278-3285.Google Scholar
Conwell Y, Duberstein PR, Cox C.  et al.  Relationships of age and axis I diagnoses of victims of completed suicide: a psychological autopsy study.  Am J Psychiatry.1996;153:1001-1008.Google Scholar
Bruce ML, Leaf P, Rozal G.  et al.  Psychiatry status and 9-year mortality data in the New Haven Epidemiologic Catchment Area Study.  Am J Psychiatry.1994;151:716-721.Google Scholar
Penninx BW, Geerlings SW, Deeg DJ.  et al.  Minor and major depression and the risk of death in older persons.  Arch Gen Psychiatry.1999;56:889-895.Google Scholar
Penninx BW, Beekman AT, Honig A.  et al.  Depression and cardiac mortality: results from a community-based longitudinal study.  Arch Gen Psychiatry.2001;58:221-227.Google Scholar
NIH Consensus Development Panel on Depression in Late Life.  Diagnosis and treatment of depression in late life.  JAMA.1992;268:1018-1024.Google Scholar
Lebowitz BD. Depression and treatment of depression in late life: an overview of the NIH consensus statement.  Am J Geriatr Psychiatry.1996;4(suppl 1):S3-S6.Google Scholar
Williams Jr JW, Barrett J, Oxman T.  et al.  Treatment of dysthymia and minor depression in primary care: a randomized controlled trial in older adults.  JAMA.2000;284:1519-1526.Google Scholar
Reynolds III CF, Frank E, Perel JM.  et al.  Nortriptyline and interpersonal psychotherapy as maintenance therapies for recurrent major depression: a randomized controlled trial in patients older than 59 years.  JAMA.1999;281:39-45.Google Scholar
Callahan CM, Hendrie HC, Dittus RS.  et al.  Improving treatment of late life depression in primary care: a randomized clinical trial.  J Am Geriatr Soc.1994;42:839-846.Google Scholar
Callahan CM. Quality improvement research on late life depression in primary care.  Med Care.2001;39:772-784.Google Scholar
Unützer J, Katon W, Russo J.  et al.  Patterns of care for depressed older adults in a large staff model HMO.  Am J Geriatr Psychiatry.1999;7:235-243.Google Scholar
Unützer J, Katon W, Sullivan M.  et al.  Treating depressed older adults in primary care: narrowing the gap between efficacy and effectiveness.  Milbank Q.1999;77:225-256.Google Scholar
Unützer J, Simon G, Belin TR.  et al.  Care for depression in HMO patients aged 65 and older.  J Am Geriatr Soc.2000;48:871-878.Google Scholar
Unützer J, Rubenstein L, Katon WJ.  et al.  Two-year effects of quality improvement for depression on medication management.  Arch Gen Psychiatry.2001;58:935-942.Google Scholar
Young AS, Klap R, Sherbourne CD.  et al.  The quality of care for depressive and anxiety disorders in the United States.  Arch Gen Psychiatry.2001;58:55-61.Google Scholar
Waxman HM, Carner EA. Physicians' recognition, diagnosis, and treatment of mental disorders in elderly medical patients.  Gerontologist.1984;24:593-597.Google Scholar
Shapiro S, Skinner EA, Kessler LG.  et al.  Utilization of health and mental health services: three ECA sites.  Arch Gen Psychiatry.1984;41:971-978.Google Scholar
Goldstrom ID, Burns BJ, Kessler LG.  et al.  Mental health services use by elderly adults in a primary care setting.  J Gerontol.1987;42:147-153.Google Scholar
Gurland B, Cross P, Katz S. Epidemiological perspectives on opportunities for treatment of depression.  Am J Geriatr Psychiatry.1996;4:7-14.Google Scholar
Ganguli M, Gilby J, Seaberg E.  et al.  Depressive symptoms and associated factors in a rural elderly population: the moVIES project.  Am J Geriatr Psychiatry.1995;3:144-160.Google Scholar
Cole MG, Yaffe MJ. Pathway to psychiatric care of the elderly with depression.  Int J Geriatr Psychiatry.1996;11:157-161.Google Scholar
Swartz MS, Wagner HR, Swanson JW.  et al.  Administrative update: utilization of services, I: comparing use of public and private mental health services: the enduring barriers of race and age.  Community Ment Health J.1998;34:133-144.Google Scholar
Unützer J, Katon W, Williams Jr JW.  et al.  Improving primary care for depression in late life: the design of a multicenter randomized trial.  Med Care.2001;39:785-799.Google Scholar
Wagner E. The role of patient care teams in chronic illness management.  BMJ.2000;320:569-572.Google Scholar
Von Korff M, Glascow R, Sharpe M. Organizing care of chronic illness.  BMJ.2002;325:92-94.Google Scholar
Hegel MT, Imming J, Cyr-Provost M.  et al.  Role of behavioral health professionals in a collaborative stepped care treatment model for depression in primary care: Project IMPACT.  Fam Syst Health.2002;20(1):265-277.Google Scholar
Saur CD, Harpole LH, Steffens DC.  et al.  Treating depression in primary care: an innovative role for mental health nurses.  J Am Psychiatr Nurses Assoc.2002;8:159-167.Google Scholar
Oishi SM, Shoai R, Katon W.  et al.  Impacting late life depression: integrating a depression intervention into primary care.  Psychiatr Q.In press.Google Scholar
Derogatis LR, Lipman RS, Covi L. SCL-90: an outpatient psychiatric rating scale--preliminary report.  Psychopharmcol Bull.1973;9:13-28.Google Scholar
Spitzer RL, Williams JB, Kroenke K.  et al.  Utility of a new procedure for diagnosing mental disorders in primary care: the PRIME-MD 1000 study.  JAMA.1994;272:1749-1756.Google Scholar
Williams JB, Gibbon M, First MB.  et al.  The Structured Clinical Interview for DSM-III-R (SCID): multisite test-retest reliability.  Arch Gen Psychiatry.1992;49:630-636.Google Scholar
First MB, Spitzer RL, Gibbon M, Williams JB. Structured Clinical Interview for DSM-IV Axis I Disorders (SCID)Washington, DC: American Psychiatric Press Inc; 1996.
Mayfield D, McLeod G, Hall P. The CAGE questionnaire: validation of a new alcoholism screening instrument.  Am J Psychiatry.1974;131:1121-1123.Google Scholar
Callahan CM, Unverzagt FW, Hui SL.  et al.  Six-item screener to identify cognitive impairment among potential subjects for clinical research.  Med Care.2002;40:771-781.Google Scholar
Unützer J.and the IMPACT Investigators.  IMPACT Intervention Manual. Los Angeles, Calif: Center for Health Services Research, UCLA Neuropsychiatric Institute; 1999.
Harpole L, Steffens D, Saur C.  et al.  Making an IMPACT: Improving Care for Late Life DepressionDurham, NC: Duke University Media Group; 1999.
Oishi S, Unützer J.and the IMPACT Study Investigators.  Making an IMPACT on Late Life Depression: Working With Your Health Care TeamLos Angeles, Calif: Center for Health Services Research, UCLA Neuropsychiatric Institute; 1999.
Hegel MT, Barrett JE, Oxman TE.  et al.  Problem-Solving Treatment for Primary Care (PST-PC): A Treatment Manual for Depression. Hanover, NH: Dartmouth University; 1999.
Arean PA, Hegel MT, Unützer J. Problem-Solving Therapy for Older Primary Care Patients: Maintenance Group Manual for Project IMPACT. Los Angeles: University of California, Los Angeles; 1999.
Arean PA, Hegel MT, Unützer J. Problem-Solving Treatment in Primary Care: Addendum to PST-PC Treatment Manual for Project IMPACT. Los Angeles: University of California, Los Angeles; 1999.
Mynors-Wallis LM, Gath DH, Day A.  et al.  Randomized controlled trial of problem solving treatment, antidepressant medication, and combined treatment for major depression in primary care.  BMJ.2000;320:26-30.Google Scholar
Hegel M, Barrett J, Oxman T. Training therapists in problem-solving treatment of depressive disorders in primary care: lessons learned from the treatment effectiveness project.  Fam Syst Health.2000;18:423-435.Google Scholar
Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure.  J Gen Intern Med.2001;16:606-613.Google Scholar
Unützer J, Choi Y, Cook I.  et al.  A web-based data management system to improve care for depression in a multicenter clinical trial.  Psychiatr Serv.2002;53:671-673, 678.Google Scholar
Sheehan DV, Harnett-Sheehan K, Raj BA. The measurement of disability.  Int Clin Psychopharmacol.1996;11(suppl 3):89-95.Google Scholar
Lin EH, VonKorff M, Russo J.  et al.  Can depression treatment in primary care reduce disability? a stepped care approach.  Arch Fam Med.2000;9:1052-1058.Google Scholar
Meyers BS, Sirey J, Bruce ML. Who goes where? clinical and sociodemographic predictors of service site in mental health outpatients. Poster presented at: National Institute of Mental Health Conference on Improving the Condition of People With Mental Illness; September 5, 1997; Washington, DC.
Unützer J, Katon W, Callahan CM.  et al.  Depression treatment in a sample of 1,801 depressed older adults in primary care.  J Am Geriatr Soc.In press.Google Scholar
Katon W, Robinson P, Von Korff M.  et al.  A multifaceted intervention to improve treatment of depression in primary care.  Arch Gen Psychiatry.1996;53:913-919.Google Scholar
Saunders K, Simon G, Bush T.  et al.  Assessing the feasibility of using computerized pharmacy refill data to monitor antidepressant treatment on a population basis: a comparison of automated and self report data.  J Clin Epidemiol.1998;51:883-890.Google Scholar
Littell RC, Milliken GA, Stroup WW.  et al.  SAS System for Mixed Models. Cary, NC: SAS Publications; 1996:chap 3.
Little RJ. Missing data adjustments in large surveys.  J Business Econ Stat.1988;6:287-301.Google Scholar
Bell R. Presentation at Depression PORT Methods Workshop (I). Santa Monica, Calif: RAND; 1999.
Rubin DB. Multiple Imputation for Non-response in SurveysNew York, NY: John Wiley & Sons; 1987.
Lavori P, Dawson R, Shera D. A multiple imputation strategy for clinical trials with truncation of patient data.  Stat Med.1995;14:1913-1925.Google Scholar
Coyne J, Katz IR. Improving the primary care treatment of late life depression: progress and opportunities.  Med Care.2001;39:756-759.Google Scholar
Simon GE, Katon WJ, VonKroff M.  et al.  Cost-effectiveness of a collaborative care program for primary care patients with persistent depression.  Am J Psychiatry.2001;158:1638-1644.Google Scholar
Olfson M, Marcus SC, Druss B.  et al.  National trends in the outpatient treatment of depression.  JAMA.2002;287:203-209.Google Scholar
Mamdani MM, Pariikh SV, Austin PC.  et al.  Use of antidepressants among elderly patients: trends and contributing factors.  Am J Psychiatry.2000;157:360-367.Google Scholar
Blazer DG, Hybels CF, Simonsick EM.  et al.  Marked differences in antidepressant use by race in an elderly community sample: 1986-1996.  Am J Psychiatry.2000;157:1089-1094.Google Scholar
Katon W, Von Korff M, Lin E.  et al.  Collaborative management to achieve treatment guidelines: impact on depression in primary care.  JAMA.1995;273:1026-1031.Google Scholar
Katon W, Von Korff M, Lin E.  et al.  Stepped collaborative care for primary care patients with persistent symptoms of depression: a randomized trial.  Arch Gen Psychiatry.1999;56:1109-1115.Google Scholar
Gill D, Hatcher S. A systematic review of treatment of depression with antidepressant drugs in patients who also have a physical illness.  J Psychosom Res.1999;47:131-143.Google Scholar
Pignone MP, Gaynes BN, Rushton JL.  et al.  Screening for depression in adults: a summary of the evidence for the US Preventive Services Task Force.  Ann Intern Med.2002;136:765-776.Google Scholar
Schoenbaum M, Unützer J, Sherbourne C.  et al.  Cost-effectiveness of practice-initiated quality improvement for depression: results of a randomized controlled trial.  JAMA.2001;286:1325-1330.Google Scholar
Center for Medicare and Medicaid Services.  Medicare and Medicaid statistical supplement 2000.  Health Care Financing Rev.June 2001:104.Google Scholar
Lin EH, Katon WJ, Simon GE.  et al.  Achieving guidelines for the treatment of depression in primary care: is physician education enough?  Med Care.1997;35:831-842.Google Scholar
Hunkeler EM, Meresman JF, Hargreaves WA.  et al.  Efficacy of nurse telehealth care and peer support in augmenting treatment of depression in primary care.  Arch Fam Med.2000;9:700-708.Google Scholar