Design and implementation of the Telemedicine-Enhanced Antidepressant Management Study

General Hospital Psychiatry 28 (2006) 18 – 26

Design and implementation of the Telemedicine-Enhanced Antidepressant Management Study John C. Fortney, Ph.D.a,b,c,T, Jeff M. Pyne, M.D.a,b,c, Mark J. Edlund, M.D., Ph.D.a,b,c, Dean E. Robinson, M.D.c,d,e, Dinesh Mittal, M.D.a,b,c, Kathy L. Henderson, M.D.a,b,c,f a

VA Health Services Research and Development (HSR&D), Center for Mental Health and Outcomes Research, Central Arkansas Veterans Healthcare System, North Little Rock, AR 72114, USA b Division of Health Services Research, Department of Psychiatry, College of Medicine, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA c VA South Central Mental Illness Research Education and Clinical Center, North Little Rock, AR 72114, USA d Mental Health Service, Overton Brooks VA Medical Center, Shreveport, LA 71101, USA e Department of Psychiatry, School of Medicine, Louisiana State University Health Sciences Center, Shreveport, LA 71130, USA f Mental Health Product Line, South Central Veterans Health Care Network, North Little Rock, AR 72114, USA Received 25 March 2005; accepted 7 July 2005

Abstract Objective: Evidence-based practices designed for large urban clinics are not necessarily transportable into small rural practices. Implementing collaborative care for depression in small rural primary care clinics presents unique challenges because it is typically not feasible to employ on-site mental health specialists. The purpose of the Telemedicine-Enhanced Antidepressant Management (TEAM) study was to evaluate a collaborative care model adapted for small rural clinics using telemedicine technologies. The purpose of this paper is to describe the TEAM study design. Method: The TEAM study was conducted in small rural Veterans Administration community-based outpatient clinics with interactive video equipment available for mental health, but no on-site psychiatrists/psychologists. The study attempted to enroll all patients whose depression could be appropriately treated in primary care. Results: The clinical characteristics of the 395 study participants differed significantly from most previous trials of collaborative care. At baseline, 41% were already receiving primary care depression treatment. Study participants averaged 5.5 chronic physical health illnesses and 56.5% had a comorbid anxiety disorder. Over half (57.2%) reported that pain impaired their functioning extremely or quite a bit. Conclusions: Despite small patient populations in rural clinics, enough patients with depression can be successfully enrolled to evaluate telemedicine-based collaborative care. D 2006 Elsevier Inc. All rights reserved. Keywords: Depression; Rural; Telepsychiatry

1. Introduction Intervention studies have demonstrated the effectiveness of collaborative care models designed to improve depression outcomes in primary care settings [1– 9]. Collaborative care involves primary care providers (PCPs) working in conjunction with a depression care team comprising non-

T Corresponding author. VA Health Services Research and Development, Center for Mental Health and Outcomes Research (152/NLR), North Little Rock, AR 72114, USA. Tel.: +1 501 257 1726. E-mail address: [email protected] (J.C. Fortney). 0163-8343/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.genhosppsych.2005.07.001

physicians (e.g., nurses, pharmacists) and mental health specialists (e.g., psychologists, psychiatrists). The demonstrated cost-effectiveness of collaborative care has led to implementation efforts to promote adoption in routine practice [10 –12]. Interventions designed and tested in large urban clinics are not necessarily applicable in small rural practices [13]. Implementing collaborative care in small rural primary care practices presents unique challenges because it is typically not feasible to employ mental health specialists on site. In fact, only 25% of primary care practices nationwide have on-site mental health specialists [14]. Although 21% of the

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

U.S. population lives in rural areas according to the 2000 Census, only one previous study of collaborative care (QuEST) recruited a substantial number of rural primary care practices as study sites [4]. In the QuEST study, on-site primary care nurses were trained to provide depression care management, and psychiatrists were available for telephone consults [4]. The QuEST collaborative care model design has the advantage that nurse care managers have established therapeutic relationships with their patients, have access to on-site paper medical records and have open communication channels with PCPs. However, the QuEST design has the potential disadvantage that nurse care managers cannot specialize in depression treatment, lack meaningful access to mental health supervision/consultation and must incorporate care management activities into busy routines with competing demands. The purpose of the Telemedicine-Enhanced Antidepressant Management (TEAM) study was to adapt the collaborative care model for small rural primary care practices using telemedicine technologies without altering the nature/content of the collaborative care model itself. Telemedicine (e.g., telephone, interactive video, electronic medical records, and internet) facilitates communication between a centrally located depression care team and PCPs practicing in geographically diverse clinic locations. We chose to conduct this first telemedicine-based collaborative care trial in rural areas served by the Veterans Administration (VA) healthcare system. The VA is a particularly suitable setting for telemedicine-based interventions because of the widespread standardized use of interactive video technology and electronic medical records [e.g., Computerized Patient Record System (CPRS)]. The objectives of the TEAM study are to compare processes and outcomes among patients with depression treated at intervention and matched control sites, and to determine whether the intervention was cost-effective in routine practice. The purpose of this paper is to describe the design of the TEAM study, including (1) methods used to enroll study participants, (2) usual depression care in the VA, (3) the TEAM intervention and (4) methods for evaluating the effectiveness/cost-effectiveness of the intervention. In addition, we describe the socioeconomic and clinical characteristics of the study participants and discuss our rational for the study design and the resulting strengths and weaknesses.

2. Methods 2.1. Study sites Veterans Administration is organized into 21 Veterans Integrated Service Networks (VISNs). TEAM was conducted in VISN 16, one of the largest and most rural of the networks. The study was conducted in community-based outpatient clinics (CBOCs), of which there are 674 currently in operation across the nation and 34 in VISN 16.

19

Community-based outpatient clinics are satellite facilities, usually located a long distance from their bparentQ VA Medical Centers (VAMC) that maintain administrative responsibility. To be eligible for this study, CBOCs must have (1) treated 1000–5000 patients in Fiscal Year 2000, (2) no on-site psychiatrists/psychologists and (3) interactive video equipment available for mental health. To be eligible, parent VAMCs must have had at least two CBOCs meeting these inclusion criteria. Seven CBOCs and three parent VAMCs in VISN 16 met these criteria, and all were recruited as study sites. 2.2. Patient enrollment 2.2.1. Sampling frame We sought to enroll all CBOC patients with depression who could be appropriately treated by PCPs (i.e., no significant psychiatric comorbidity). In addition, patients already receiving VA specialty mental health treatment were excluded because they were not expected to benefit from collaborative care. We screened patients from 12 to 18 months at each clinic to generate a large enough sample and to include infrequent users of services. Screening for short periods yields samples that overrepresent frequent users [15,16] who are more likely to report depression [18]. The VA’s patient database (VISTA) was used to identify all scheduled primary care appointments with a bdueQ annual depression screen. Appointments made, rescheduled and canceled were downloaded from VISTA every night during the enrollment period (N = 57,838). Walk-in patients (4.4%) and patients who scheduled appointments less than 3 days in advance (2.2%) were excluded. 2.2.2. Screening Fig. 1 presents the flow of potentially eligible patients from the scheduling of their primary care appointment to enrollment. Prior to their appointment, patients were mailed postcards stating that because their clinic was participating in a study, their annual preventive health questionnaire (smoking, alcohol use, and depression) would be administered by telephone and that, depending on their answers, they could be asked to participate in a study. The postcard provided a toll-free number for patients who did not want to be contacted. Research assistants attempted to screen patients up to 3 weeks prior to their appointment. Of the 24,882 patients with a due depression screen, 73.6% (n = 18,306) were successfully screened prior to their appointment. Reasons for unsuccessful screens included unable to contact (9.7%), refusal (5.8%), impaired/intoxicated (5.5%) and phone disconnected or wrong number (5.3%). Screening was conducted using Computer-Assisted Telephone Interview (CATI) software. The PHQ2 was administered, followed by the remaining PHQ9 items for patients positive on the PHQ2 (score z 3) [17]. Results of the depression screen were entered into CPRS the day before the appointment for patients in both the intervention and usual care groups. Notes were entered using standard

20

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

those screened, 6.9% had PHQ9 scores z 12. Patients not appropriately treated by a PCP were excluded in a hierarchical manner. First, VISTA data were used to exclude patients who had a specialty mental health encounter in the previous 6 months or a future specialty mental health appointment (23.8%), or a schizophrenia diagnosis (0.7%). Patients previously or currently treated by a PCP for depression were not excluded. Second, patients endorsing suicidal ideation on the PHQ9 and who subsequently endorsed current suicide ideation were excluded if the study psychiatrist contacting them did not believe that the patient could be appropriately treated in primary care (2.2%). Otherwise, potentially eligible veterans were informed about the study and asked to complete a brief eligibility telephone assessment. Ten percent refused and 0.8% could not be recontacted or were too impaired/intoxicated. To minimize response burden, exclusion criteria were applied hierarchically during the assessment such that once an exclusion criterion was met, no further eligibility questions were asked. Forty percent were ineligible for the following reasons: life event preventing participating (0.3%), bereavement [18,19] (4.5%), pregnant/breastfeeding (0%), court-appointed guardian (1.7%), alcohol dependence [20] (10.1%), drug dependence [20] (1.9%), bipolar disorder [20] (14.0%) and cognitive impairment [21] (7.6%). 2.2.4. Enrollment Of those eligible, 91.3% agreed to participate and were administered the baseline phone interview, and then mailed a letter about the study. On-site clinic personnel trained by the PI (JCF) conducted the informed consent process. Of those verbally agreeing to participate in the study, 91.9% attended their appointment and were consented. Only 1.4% refused to consent. Three hundred ninety-five patients were enrolled. 2.2.5. Randomization The randomization unit was the CBOC. Communitybased outpatient clinics were matched by parent VAMC, with one CBOC within each pair randomized to the intervention. Preintervention patterns of care (e.g., frequency of primary care visits, percent diagnosed with depression and percent referred) were similar across the intervention and matched control sites. 2.3. Usual care Fig. 1. Enrollment flowchart.

templates (e.g., the two-stage clinical reminder described below), and the patient’s PCP was designated as an additional signer. 2.2.3. Eligibility Patients with PHQ9 scores z 12 were eligible for the study. This cutoff score has 98% sensitivity and 96% specificity for DSM-IV major depressive disorder [17]. Of

Veterans Administration treats over a half a million veterans for depression each year and has implemented several quality improvement initiatives targeting depression. For example, the VA has widely disseminated its depression treatment guidelines. In addition, when the study began in 2003, there were national performance measures requiring annual screens for depression (instituted in 1998) and follow-up assessments of all positive screens (instituted in 2002). In VISN 16, a two-stage clinical reminder was integrated into CPRS to facilitate depression screening and assessment. This reminder was designed with input from the

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

TEAM investigators and was integrated into the subject enrollment procedures described above. The clinical reminder included the PHQ2 as the first stage screen and the PHQ9 as the second stage assessment. The clinical reminder also provided information and decision aids to PCPs and automatically generated a progress note with screening/assessment results and a checklist-generated treatment plan. Although the national screening and assessment performance measures were replaced by a HEDIS-based depression performance measure (3 followup visits in 3 months) midway through the enrollment period, VISN 16 continued to recommend annual depression screens and follow-up assessments. The VA has a pharmacy formulary that includes all common antidepressants, although some (e.g., venlafaxine) are restricted to psychiatrists. Patients prescribed antidepressants can either pick up their medications (typically a 1-month supply) at the clinic or have them mailed to their residence. Patients can call the VA pharmacy to order refills for up to 12 months after their initial prescription and request to have them mailed to their home. Outpatient visits are free for most veterans. Medications are free for highpriority veterans, while others are required to make a $7 copayment for each medication. At large VAMCs, access to specialty mental health care is excellent and specialty referrals are common. In VISN 16, all of the larger CBOCs lacking on-site psychiatrists had interactive video equipment installed in 2002–2003 to facilitate telepsychiatry referrals to the mental health clinics of the parent VAMC. The interactive video equipment was identical at all CBOCs and VAMCs, and comprised the MHPL Polycom videoconferencing unit, a PC-based Polycom iPower 9800 with a single 32U XGA monitor, and Power Cam plus NTSC camera with auto voice tracking. The units retailed for $26,176, including turnkey installation, and 1-year maintenance. This system used the H.323 protocol and 384 kb of bandwidth. These PC-based units were chosen because they support software applications (e.g., PowerPoint presentations) and interfaced with medical devices. At the time, the less expensive applicationbased systems did not have these capabilities. In addition, the following low-cost intervention components were provided to both intervention and usual care sites: (1) 1 h CME presentation about managing depression in primary care via interactive video; (2) PCPs were informed about the TEAM website (www.va.gov/TEAM), which contained a link to the MacArthur Foundation Depression Tool Kit, and (3) patients were mailed the depression educational pamphlet designed for the Partners in Care Study and were informed about the TEAM website that contained links to educational materials from the National Institute of Mental Health. Recent data indicate that quality of usual depression care in the VA is higher than in private practice [22]. However, there are still major barriers to providing evidence-based treatments in the VA. Only 9% of VA patients with

21

diagnosed major depressive disorder have 3 follow-up visits in the 3 months following a new treatment episode, and only 34% receive z 90 days of supply of antidepressants [23]. Treatment barriers are even greater in rural CBOCs that do not have on-site psychiatrists or on-site mental health specialists providing evidence-based psychotherapy. Moreover, because of long travel distances to parent VAMCs, referrals for face-to-face specialty mental health care are often not feasible. 2.4. TEAM intervention The structure and components of the TEAM intervention were designed to be similar to other collaborative care interventions. What makes the TEAM intervention unique is the mode of delivery (i.e., telemedicine technologies are used to facilitate the collaboration between the on-site PCPs and the off-site depression care team). The structure of the TEAM intervention follows the multifaceted, multitargeted, stepped-care collaborative model of depression treatment (Table 1). The intervention is multifaceted because it uses complementary strategies and multitargeted because it targets both providers and patients. The stepped-care intervention begins with a standard level of care (steps 1a or 1b in Table 1) for all patients and increases treatment intensity for patients failing to respond to lower levels of care. Each additional level of stepped-care involved a greater number of intervention personnel with increasing mental health expertise. Patients received the intervention for 12 months of acute phase treatment or until a 6-month continuation phase was complete (without a relapse), whichever came first. The intervention involved five types of providers: (1) PCPs located at CBOCs, (2) consult telepsychiatrists located at parent VAMCs, (3) an off-site depression nurse care manager (RN), (4) an off-site clinical pharmacist (PharmD) and (5) an off-site supervisor psychiatrist. Telemedicine technologies facilitated communication among patients, PCPs, telepsychiatrists and the depression care team. The consult telepsychiatrist accepted consultations or referrals from PCPs. The supervising psychiatrist (JMP or MJE) provided clinical supervision to the care manager and clinical pharmacist via weekly face-to-face meetings. If implemented in routine care, supervision of the depression care team could be the responsibility of the consult telepsychiatrist. Patients and providers could choose either watchful waiting (Step 1a) or treatment with an antidepressant medication (Step 1b), although providers were encouraged to initiate guideline-concordant antidepressant treatment in the positive depression screen CPRS progress note. By phone, the care manager conducted patient education, activation and barrier assessment/resolution activities, as well as the symptom and medication monitoring components of the intervention. To ensure fidelity and to facilitate portability, encounters were scripted and used standardized instruments. Scripts and instruments were administered

22

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

Table 1 Structure of TEAM intervention Step

Intervention component

Provider type

Mode of communication

Step 1a

Watchful waitinga Barrier assessment and resolution Patient education and activation Symptom monitoring Antidepressant therapy Barrier assessment and resolution Patient education and activation Symptom and medication monitoring Medication history Recommended antidepressant therapy Patient education and activation Symptom and medication monitoring Medication management Psychiatric consultation Recommended antidepressant therapy Patient education and activation Symptom and medication monitoring Medication management Referral to mental health at parent VAMC

PCP Nurse Nurse Nurse PCP Nurse Nurse Nurse Pharmacist PCP Nurse Nurse Pharmacist Telepsychiatrist PCP Nurse Nurse Pharmacist PCP

Face-to-face Telephone Telephone Telephone/CPRS Face-to-face Telephone Telephone Telephone/CPRS Telephone/CPRS Face-to-face Telephone Telephone/CPRS Telephone/CPRS Interactive video/CPRS Face-to-face Telephone Telephone/CPRS Telephone/CPRS Face-to-face

Step 1b

Step 2

Step 3

Step 4 a

Although watchful waiting was not recommended by TEAM personnel, it could have been chosen by the patient and/or provider.

using CATI software, which facilitated skip patterns, automated scoring and branching scripts. During the initial encounter, patients were (1) administered the PHQ9 symptom monitoring tool, (2) educated and activated using a semistructured script [4] and (3) assessed for treatment barriers, with semistructured follow-up scripts for endorsed barriers [4]. Symptom and medication monitoring were conducted every 2 weeks during the acute treatment phase and every 4 weeks during watchful waiting or continuation treatment phases. Monitoring followed a structured script [3] to assess response, current medication use and severity of antidepressant side effects. Nonadherent patients or those experiencing severe side effects were administered a semistructured script [3] by the care manager to address common problems. All responses were provided to the PCP using CPRS progress notes. Progress notes containing information about failed trials requested an electronic signature from the PCP to alert them to the progress note. Intervention staff could track whether progress notes had been signed. An antidepressant trial was considered to have failed in the acute phase of treatment if the patient (1) was nonadherent to the medication, (2) experienced severe side effects during two consecutive medication monitoring encounters, (3) experienced z 5 points of increase in their PHQ9 score, sustained for two monitoring encounters, or (4) did not respond after 8 weeks of antidepressant therapy (50% decrease in baseline severity). Criteria 1–3 were included in our definition of a failed trial in order to encourage PCPs to revise treatment plans in a timely manner when their patients experienced problems. If the patient did not respond to the initial antidepressant, the clinical pharmacist conducted detailed medication histories and provided pharmacotherapy recommendations to PCPs (step 2). The pharmacist also provided nonscripted medication management to patients. Pharmacist encounters were

conducted by phone, and treatment recommendations were provided to the PCP using CPRS progress notes requesting an electronic signature. We chose to have a pharmacist instead of a psychiatrist provide medication management and make treatment recommendations in step 2 in order to contain costs and because pharmacists have been shown to be effective at improving depression outcomes [6,7]. If the patient did not respond to two antidepressants trials, the depression care team recommended a telepsychiatry consultation (step 3). 2.5. Intervention fidelity Fidelity was tracked by the care manager using CATI software. The following activities were tracked: (1) timeliness of an initial care manager encounter, (2) frequency and timeliness of care manager follow-up encounters, (3) frequency of PCPs signing CPRS progress notes when requested and (4) whether care was stepped up after each failed trial. 2.6. Data collection and analysis 2.6.1. Measures The primary outcomes are (1) whether or not the patient was prescribed an antidepressant, (2) medication adherence, measured by patient self-report and pharmacy records, (3) treatment response and remission, (4) health-related quality of life and (5) satisfaction with care. Depression severity was measured using a different instrument (Hopkins Symptom Checklist SCL-20) than the symptom monitoring instrument (PHQ9) in order to avoid the measurement bias, which would result if intervention patients altered their responses over time due to feedback from the nurse care manager following changes in their depression severity. Response is measured dichotomously as a 50% improvement in depression severity and remission is

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

23

Table 2 Constructs and instruments Construct

Instrument

Baseline

Follow-up

Demographics Physical comorbidity Depression history Major depressive disorder Depression severity Dysthymia Minor depressive disorder Panic disorder Generalized anxiety disorder At-risk drinking Posttraumatic stress disorder Depression beliefs Treatment preferences Social support Health-related quality of life Health-related quality of life Medication adherence Service utilization Satisfaction with care

Depression Outcomes Module [18,19] Depression Outcomes Module [18,19] Depression Outcomes Module [18,19] Mini International Neuropsychiatric Interview [24,25] SCL-20 [26,27] Mini International Neuropsychiatric Interview [24,25] Mini International Neuropsychiatric Interview [24,25] Mini International Neuropsychiatric Interview [24,25] Mini International Neuropsychiatric Interview [24,25] Mini International Neuropsychiatric Interview [24,25] Mini International Neuropsychiatric Interview [24,25] TEAM Depression Beliefs Inventory Quality Improvement for Depression [4,5] Duke Social Support and Stress Scale [28,29] Quality of Well Being Scale [30 –33] SF-12V [34,35] TEAM Medication Adherence Assessment Quality Improvement for Depression [4,5] Experience of Care and Health Outcomes Survey [36]

Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No No

No No No Yes Yes Yes Yes No Yes Yes No Yes No Yes Yes Yes Yes Yes Yes

measured dichotomously as a decrease in severity below 0.5. The following case-mix factors were measured: age, gender, race, marital status, education, income, employment, chronic health, health-related quality of life, prior depression treatment, family history of depression, age of onset, number of depression episodes, psychiatric comorbidity, social support and perceived need for, effectiveness of and barriers to treatment. Computer-Assisted Telephone Interview software was administered at baseline (prior to the appointment), and at 6 and 12 months. Study participants were reimbursed $40 for each interview. Table 2 provides the instruments used to measure each construct. Veterans Administration service utilization, prescriptions and costs will be captured from administrative databases. 2.6.2. Sample/attrition weights Sampling and attrition weights were calculated to adjust for the potential bias associated with nonparticipation and/or loss to follow-up. Using data about all potential study participants from VISTA (e.g., age, gender, race, marital status and percent service connected), we specified four logistic regressions predicting: (1) being successfully screened, (2) completing the eligibility assessment, (3) completing the baseline assessment and (4) attending appointment and being consented. For each study participant, the predicted probability associated with each of these events was calculated, the four probabilities were multiplied together, the product was inverted to generate the sample weight, and standardized to have a mean of one. Attrition weights were calculated by specifying a logistic regression predicting completion of each follow-up interview using case-mix data collected at baseline, and inverting the predicted probability. 2.6.3. Effectiveness analysis Patients will be the unit of the intent-to-treat analysis, and hypotheses will be tested using multilevel models. Due to

the large number of available case-mix adjusters, multivariate analysis will include only those case-mix factors found to significantly predict the dependent variable at the P V.2 level. The cost-effectiveness analysis will be conducted from the perspectives of the VA and society, and will examine both depression specific and generic outcomes (i.e., quality adjusted life years). The power analysis is based on an a significance level of .05 and a final sample size of 340 subjects with complete data (assuming 85% completed the 12-month follow-up). Assuming zero intraclass correlation among subjects within sites, there will be 97% power to detect a 20% difference (e.g., 50% vs. 30%) in the proportion responding to treatment. If the clinic-level intraclass correlation is .02, statistical power will decrease to 78%. The study is underpowered to detect interaction effects that might moderate the impact of the intervention.

3. Results The socioeconomic and clinical characteristics of the study participants are provided in Table 3. Eighty-two percent met criteria for major depressive disorder. Of particular note is the extremely high level of disease burden. Virtually, all (99.2%) of the study participants reported having at least one serious chronic health condition, and the average number was 5.5 [e.g., diabetes (32.9%), heart disease (32.2%), lung disease (20.3%), stroke (18.2%) and cancer (12.7%)]. SF12V physical health and mental health component scores were well below the general population, and veterans using VA primary care services [37]. Over half (57.2%) of study participants reported that pain impaired their functioning extremely or quite a bit on the SF12V. Behavioral health comorbidity was common with 56.5% having a current anxiety disorder and 13% were at-risk

24

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

Table 3 Baseline socioeconomic and clinical characteristics of study participants Variables

Means or percentage

P value

Overall (n = 395)

Intervention group (n = 177)

Usual care group (n = 218)

Socioeconomic Age Male Caucasian African American Native American Other race/ethnicity Annual household income b$20,000 Married High school graduate Employed Social support (0 –1)

59.2 91.7% 74.7% 18.2% 3.5% 3.6% 51.7% 62.3% 76.0% 21.9% 0.42

58.4 93.8% 76.3% 15.3% 4.0% 4.4% 44.4% 62.7% 74.6% 24.4% 0.42

59.8 89.9% 73.4% 20.6% 3.2% 2.8% 55.6% 61.9% 77.1% 20.0% 0.42

.90 .87 .57 .32 .79

Clinical PHQ9 (depression screen score) SCL-20 (depression severity score) PCS MCS Quality of well being score Chronic physical illnesses Family history of depression Age depression onset b 18 Prior depression episodes Prior depression treatment Current depression treatment Antidepressants acceptable Current major depressive disorder Current dysthymia Current panic disorder Current generalized anxiety disorder Current posttraumatic stress disorder Were at-risk drinkers

16.4 1.8 30.0 36.5 0.4 5.5 45.2% 17.2 3.7 65.7% 40.9% 79.4 82.0% 4.1% 9.6% 50.7% 23.8% 12.9%

16.3 1.9 30.4 36.1 0.4 5.3 46.6% 15.6 3.7 66.5% 35.2% 79.9% 83.1% 2.8% 9.6% 45.9% 24.9% 13.0%

16.4 1.8 29.7 36.9 0.4 5.7 44.2% 19.2 3.6 65.1% 45.4% 78.9% 81.2% 5.1% 9.6% 54.1% 22.9% 12.8%

.77 .76 .62 .56 .65 .16 .64 .34 .79 .78 .04 .28 .63 .27 .99 .69 .66 .96

.24 .17 .39

PCS, physical component score; MCS, mental health component score.

drinkers. Study participants had multiple prior depression episodes, two thirds had received prior depression treatment and nearly half (41%) were currently receiving depression treatment at baseline. Receiving treatment at baseline was the only case-mix factor to vary significantly across the intervention and usual care groups. The clinical characteristics of the study participants recruited from this publicly funded health system differ significantly from most previous trials of collaborative care and from most antidepressant clinical trials. 4. Conclusions The strengths and weaknesses of the TEAM study design were impacted by many decisions and tradeoffs. The first decision concerned the unit of randomization. We chose not to randomize at the patient level because of potential contamination among providers treating both intervention and usual care patients. We chose not to randomize at the provider level because of potential turnover and contamination across providers. By randomizing matched clinics, we avoid potential contamination problems. The disadvantage of clinic level randomization is

the potential variation in practice patterns not equalized through randomization. Although the matched CBOCs have similar observable preintervention practice patterns, there may be unobserved differences. The second decision concerned identifying a meaningful comparison group. We chose a comparison group that reflected usual care enhanced by low-intensity intervention components. We considered a comparison group representing a lower standard of care such as that provided in small rural CBOCs without interactive video. However, we believed that determining the most cost-effective way to treat depression in small rural primary clinics (i.e., telepsychiatry referral versus telemedicine-based collaborative care) was the more policy relevant question. Moreover, because low intensity intervention components were provided to both the intervention and comparison groups, results will indicate whether the high intensity intervention components focusing on medication management are needed to improve outcomes. The third major decision concerned inclusion and exclusion of various intervention components. Because the VA does not have the resources to offer evidence-based psychotherapy to large numbers of veterans, we excluded

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26

this potential intervention component. In fact, we deliberately instructed the care manager to not provide counseling or social support beyond that necessary to develop and maintain a therapeutic alliance. Thus, unlike other recent trials of collaborative care (e.g., Partners in Care, Project IMPACT and PROSPECT) [5,8,38], the TEAM intervention did not increase access to psychotherapy. In addition, we excluded patient assessment (e.g., personal history, stressors, comorbidity) and self-management (e.g., goal setting, exercise programs, etc.). Although the exclusion of these collaborative care components may have reduced intervention effectiveness, our results will help determine whether antidepressant management alone contributes significantly to improved outcomes. The fourth study design issue was not so much a decision as it was a consequence of conducting the study in small rural clinics, which made it necessary to recruit as many patients as possible in order to have sufficient statistical power. As a result, a major strength is that the rigorous attention to sampling and recruitment resulted in the vast majority of eligible patients from the clinics being enrolled in the study. Thus, the results should be highly generalizable to the target population. Finally, our decision to evaluate telemedicine-based collaborative care in the VA resulted in strengths and weaknesses. Although the VA is one of the largest managed care organizations in the United States, results may not generalize to other settings/populations. However, there are many advantages to conducting the initial evaluation of telemedicine-based collaborative care in the VA. First, the VA healthcare system has strong linkages between primary care and mental health, making collaboration organizationally feasible despite geographic barriers. Second, the VA employs standardized interactive video and electronic medical record technology, which facilitated the delivery of the intervention. Although many of these technologies are currently unique to the VA, they are likely to be adopted in other settings in the near future. In addition, VA leadership is committed to widespread implementation of collaborative care in response to the President’s New Freedom Commission on Mental Health. One of the goals of the VA’s mental health strategic plan entitled Achieving the Promise: Transforming Mental Health Care in VA is to develop a collaborative care dissemination package for national rollout. Thus, if found to be effective, the TEAM intervention has the potential to be widely implemented across the VA healthcare system. Acknowledgments This research was supported by VA IIR 00-078-3 grant to Dr. Fortney, VA NPI-01-006-1 grant to Dr. Pyne, the VA HSR&D Center for Mental Health and Outcomes Research and the VA South Central Mental Illness Research Education and Clinical Center. Drs. Pyne and Edlund were supported by VA HSR&D Research Career Awards.

25

Dr. Mittal was supported by VISN 16 South Central Network Research Career Development Grant Program. The authors would like to thank Dr. Kathryn Rost, Dr. Lisa Rubenstein, Dr. Greg Simon and Dr. Morris Weinberger for commenting on a previous draft of this manuscript.

References [1] 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(12):1109 – 15. [2] 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(10):924 – 32. [3] Simon GE, VonKorff M, Rutter C, Wagner E. Randomised trial of monitoring, feedback, and management of care by telephone to improve treatment of depression in primary care. BMJ 2000; 320(7234):550 – 4. [4] Rost K, Nutting P, Smith J, Werner J, Duan N. Improving depression outcomes in community primary care practice: a randomized trial of the QuEST intervention. Quality Enhancement by Strategic Teaming. J Gen Intern Med 2001;16(3):143 – 9. [5] Wells KB, Sherbourne C, Schoenbaum M, et al. Impact of disseminating quality improvement programs for depression in managed primary care: a randomized controlled trial. JAMA 2000; 283(2):212 – 20. [6] Finley PR, Rens HR, Pont JM, et al. Impact of collaborative care model upon depression in primary care: a randomized controlled trial. Pharmacotherapy 2003;23:1175 – 85. [7] Adler DA, Bungay KM, Wilson IB, et al. The impact of a pharmacist intervention on 6-month outcomes in depressed primary care patients. Gen Hosp Psychiatry 2004;26(3):199 – 209. [8] Unutzer J, Katon W, Callahan CM, et al. Collaborative care management of late-life depression in the primary care setting: a randomized controlled trial. JAMA 2002;288(22):2836 – 45. [9] Hedrick SC, Chaney EF, Felker B, et al. Effectiveness of collaborative care depression treatment in Veteran’s Affairs primary care. J Gen Intern Med 2003;18(1):9 – 16. [10] Wagner EH, Glasgow RE, Davis C, et al. Quality improvement in chronic illness care: a collaborative approach. Joint Comm J Qual Improv 2001;27(2):63 – 80. [11] VA Office of Research and Development — Health Services Research and Development Service. Improving treatment for depression in primary care. Available at http://www.hsrd.research.va.gov/ publications/internal/QUERIQuarterly_VO6 _ NO3.pdf. VA QUERI Quarterly: Newsletter of the Quality Enhancement Research Initiative, 2004;6(3):1,4. [12] Oxman TE, Dietrich AJ, Williams Jr JW, Kroenke K. A threecomponent model for reengineering systems for the treatment of depression in primary care. Psychosomatics 2002;43(6):441 – 50. [13] Rost K, Fortney J, Fischer E, Smith J. Use, quality and outcomes of care for mental health: the rural perspective. Med Care Res Rev 2002; 59(3):231 – 65. [14] Williams Jr JW, Rost K, Dietrich AJ, et al. Primary care physicians’ approach to depressive disorders. Effects of physician specialty and practice structure. Arch Fam Med 1999;8:58 – 67. [15] Lee M, Yano E, Wang M, Simon B. What patient population does visit-based sampling in primary care settings represent? Med Care 2002;40(9):761 – 70. [16] Shepard DS, Neutra R. A pitfall in sampling medical visits. Am J Public Health 1977;67(8):743 – 50. [17] Kroenke K, Spitzer RL. The PHQ-9: a new depression diagnostic and severity measure. Psychiatr Ann 2002;32(9):509 – 15. [18] Smith Jr G.R., Burnam A, Burns BJ, Cleary P, Rost KM. Depression Outcomes Module (DOM). In: First MB, editor. Handbook of

26

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

J.C. Fortney et al. / General Hospital Psychiatry 28 (2006) 18 – 26 psychiatric measures. Washington (DC)7 American Psychiatric Association; 2000. p. 213 – 5. Kramer TL, Smith GR, D’Arezzo KW, Card-Higginson P. Depression outcomes module. The guide to behavioral health outcomes management systems. Little Rock (AR): American Psychiatric Association; 2000. p. 71 – 83. Kessler RC, Andrews G, Mroczek D, Ustun B, Wittchen HU. The World Health Organization Composite International Diagnostic Interview Short Form (CIDI-SF). Int J Methods Psychiatr Res 2004;7:171 – 85. Katzman R, Brown T, Fuld P, et al. Validation of a short orientationmemory-concentration test of cognitive impairment. Am Jsychiatry 1983;140(6):734 – 9. Asch SM, McGlynn EA, Hogan MM, et al. Comparison of quality of care for patients in the Veterans Health Administration and patients in a national sample. Ann Intern Med 2004;141(12):938 – 45. VA treatment of depression: are patients receiving adequate followup? Oral Presentation, VA QUERI National Meeting: enhancing impact through collaboration and integration. Washington, DC; 2003. Lecrubier Y, Sheehan DV, Weiller E, et al. The Mini International Neuropsychiatric Interview (MINI). A short diagnostic structured interview: reliability and validity according to the CIDI. Eur Psychiatry 1997;12(5):224 – 31. Sheehan DV, Lecrubier Y, Sheehan KH, et al. The validity of the Mini International Neuropsychiatric Interview (MINI) according to the SCID-P and its reliability. Eur Psychiatry 1997;12(5):232 – 41. Derogatis LR, Lipman RS, Rickels K, Uhlenhuth EH, Covi L. The Hopkins symptom checklist (HSCL): a measure of primary symptom dimensions. Pharmacopsychiatry 1974;7:79 – 110. Derogatis LR, Lipman RS, Rickels K, Uhlenhuth EH, Covi L. The Hopkins symptom checklist (HSCL): a self-report symptom inventory. Behav Sci 1974;19:1 – 15. Parkerson Jr GR, Michener JL, Wu LR, et al. Associations among family support, family stress, and personal functional health status. J Clin Epidemiol 1989;42(3):217 – 29.

[29] Parkerson Jr GR, Broadhead WE, Tse CJ. Quality of life and functional health of primary care patients. J Clin Epidemiol 1992; 45(11):1303 – 13. [30] Kaplan RM, Anderson JP. The general health policy model: an integrated approach. In: Spiker B, editor. Quality of life and pharmacoeconomics in clinical trials. Philadelphia7 Lippincott-Raven Publishers; 1996. p. 309 – 21. [31] Kaplan RM, Bush JW. Health-related quality of life measurement for evaluation research and policy analysis. Health Psychol 1982; 1(1):61 – 80. [32] Pyne JM, Patterson TL, Kaplan RM, et al. Assessment of the quality of life of patients with major depression. Psychiatr Serv 1997; 48(2):224 – 30. [33] Pyne JM, Patterson TL, Kaplan RM, et al. Preliminary longitudinal assessment of quality of life in patients with major depression. Psychopharmacol Bull 1997;33(1):23 – 9. [34] Jones D, Kazis L, Lee A, et al. Health status assessments using the Veterans SF-12 and SF-36: methods for evaluating outcomes in the Veterans Health Administration. J Ambul Care Manage 2001; 24(3):68 – 86. [35] Kazis LE, Miller DR, Clark J, et al. Health-related quality of life in patients served by the Department of Veterans Affairs: results from the Veterans Health Study. Arch Intern Med 1998;158(6):626 – 32. [36] Beebe TJ, Harrison PA, McRae Jr JA, Asche SE. Evaluating behavioral health services in Minnesota’s Medicaid population using the Experience of Care and Health Outcomes (ECHO) Survey. J Health Care Poor Underserved 2003;14(4):608 – 21. [37] Kazis LE, Miller DR, Skinner KM, et al. Patient-reported measures of health: the Veterans Health Study. J Ambul Care Manage 2005; 27(1):70 – 83. [38] Alexopoulos GS, Katz IR, Bruce ML, et al. Remission in depressed geriatric primary care patients: a report from the PROSPECT Study. Am J Psychiatry 2005;162(4):718 – 24.