Care Management for Depression in Primary Care Practice: Findings From the RESPECT-Depression Trial

Abstract

PURPOSE This qualitative study examined the barriers to adopting depression care management among 42 primary care clinicians in 30 practices.

METHODS The RESPECT-Depression trial worked collaboratively with 5 large health care organizations (and 60 primary care practices) to implement and disseminate an evidence-based intervention. This study used semistructured interviews with 42 primary care clinicians from 30 practice sites, 18 care managers, and 7 mental health professionals to explore experience and perceptions with depression care management for patients. Subject selection in 4 waves of interviews was driven by themes emerging from ongoing data analysis.

RESULTS Primary care clinicians reported broad appreciation of the benefits of depression care management for their patients. Lack of reimbursement and the competing demands of primary care were often cited as barriers. These clinicians at many levels of initial enthusiasm for care management increased their enthusiasm after experiencing care management through the project. Psychiatric oversight of the care manager with suggestions for the clinicians was widely seen as important and appropriate by clinicians, care managers, and psychiatrists. Clinicians and care managers emphasized the importance of establishing effective communication among themselves, as well as maintaining a consistent and continuous relationship with the patients. The clinicians were selective in which patients they referred for care management, and there was wide variation in opinion about which patients were optimal candidates. Care managers were able to operate both from within a practice and more centrally when specific attention was given to negotiating communication strategies with a clinician.

CONCLUSIONS Care management for depression is an attractive option for most primary care clinicians. Lack of reimbursement remains the single greatest obstacle to more widespread adoption.

INTRODUCTION

Depression is a common condition in primary care practice,^1,2 with a prevalence of 5% to 9%.¹ Despite a solid evidence base for enhanced primary care of depression,^3–11 substantial deficits in the adequacy of care remain. A principal component of evidence-based approaches to improving depression outcomes in primary care includes care management.¹² Although there is considerable evidence for its effectiveness in improving depression outcomes in a variety of primary care settings,^13–16 primary care clinicians have been slow to adopt care management.^17–19

Care management for depression consists of a combination of assessing and monitoring a patient’s depression status; determining preferences, barriers, and progress; providing patient education and development of treatment and self-management plans; coordinating care with primary care and specialty clinicians; encouraging treatment adherence; and maintaining close communication with primary care and specialty clinicians.^18,20 There is some understanding of organizational-level barriers to care management,^17–19 but much less is known about clinician-level barriers, with the existing literature hinting at lack of time,^17,19 lack of experience with longitudinal care,¹⁸ lack of critical mass and flexibility to assign care management roles,¹⁹ and physician belief that depression is difficult to treat and of lower priority than medical problems.²¹

A great deal is still not understood about why primary care clinicians hesitate to adopt a care management model. This study examines the experience of primary care clinicians and care managers in primary care practices affiliated with 5 health care organizations (HCO-A, HCO-B, HCO-C, HCO-D, and HCO-E) who participated in the RESPECT-Depression Project.^22,23 A principal component of the intervention involved care management provided to patients according to a protocol, although a dissemination phase after the formal trial encouraged HCOs and practices to adapt the protocol to fit their specific setting. This report summarizes the experience with integrating care management into the enhanced care of depressed patients across a range of settings, based on structured interviews with clinicians and care managers.

METHODS

Care Management Context

The RESPECT-Depression project enrolled 224 patients who were known by their primary care clinicians to have depression and who agreed to participate in a care management program. They received a telephone call from a practice- or centrally-based care manager 1, 4, and 8 weeks after the initial visit, and every 4 weeks thereafter until depression remission. Telephone calls averaged 10 minutes and served to identify and address barriers to treatment adherence and to measure response to treatment using the PHQ-9.^24–27 The clinicians received reports about patients’ progress, including the PHQ-9 scores and care managers’ action taken.

Care managers regularly discussed their patient contacts with a mental health professional from their HCO. Based on the information discussed, the mental health professionals provided feedback (which was passed on to the clinicians) if they had suggestions about management. Most feedback was relatively simple and was incorporated into the care manager’s next report to the clinician. For more complex issues the mental health professional placed a telephone call to the primary care clinician. Clinicians received brief training on guideline-concordant depression care, but they followed their own clinical judgment in care for the patients enrolled in the project.

Participants and Settings

Clinicians were selected from the practices participating in the RESPECT trial and practices to whom the intervention was introduced during the 6-month dissemination phase after the trial. Clinicians were initially selected from those who actively participated in the trial and from practices that were relatively small, had local autonomy for making care changes, and had a mixed payer base. Subsequently clinicians were selected based on an iterative analysis process (described below) that varied the issues addressed and clinician characteristics according to the need to explore emerging themes. Level of clinician participation in care management was determined from the care manager referral logs, and clinician enthusiasm for care management was determined by interviews with care managers. All care managers and mental health professionals providing care manager oversight were interviewed.

Most of the participating clinicians were family physicians, with only a few general internists, nurse-practitioners, and physician’s assistants. The sample sizes and structure of the interviews did not permit meaningful comparisons among the clinician categories.

Data Collection and Analysis

Telephone interviews were conducted by one of the authors (P.A.N., K.G., or K.R.), audio-recorded, and transcribed. Transcripts were entered into FolioViews, version 4.2 (Open Market, Inc, Salt Lake City, Utah) and ATLAS.ti, version 5.2 (ATLAS.ti GmbH, Berlin, Germany) for reading, coding, and analysis. Data were collected and analyzed in 4 waves, with purposeful sampling of clinicians and care managers and modification of interview guides to explore emerging themes. Three interim analyses informed 4 distinct waves of data collection, and the iterative sampling strategy is summarized in Table 1 ▶.

Table 1.

Sampling Strategy in 4 Waves of Interviews

Wave	Sampling Strategy	Themes Explored
Wave 1	12 Clinicians who were active users of care manager	Perceived value of care management Patterns of communication with care manager
	8 Care managers	Perceived value of care management Patterns of communication with clinicians Perception of value of mental health supervision
	6 Psychiatrists and 1 clinical psychologist	Perceived value of care management Perception of their role in supervising care managers
Wave 2	8 Clinicians who had used care manager but referred only a few patients 4 Clinicians reported by care managers to be less enthusiastic	Perceived value of care management Reservations about care management
	4 Clinicians who were active users of care management	How to integrate care management into overall pattern of care Value of suggestions received from mental health consultants
Wave 3	4 Clinicians who were active users of care management 5 Care managers and one repeat interview	Perceived barriers to care management Which patients benefited from care management
	6 Clinicians who worked with an in-practice care manager 5 Care managers who were collocated in 1 or more specific practice sites (HCO-B and HCO-D)	Pros/cons of in-practice vs central location of care manager Which patients benefited from care management Value of suggestions received from mental health consultants
Wave 4	4 Repeat interviews with care managers 4 Clinicians who worked with care managers in the dissemination phase of the project 6 Second interviews with clinicians identified by care managers as enthusiastic after working with care manager	Effect of care manager exposure on current enthusiasm

HCO = health care organization.

After approximately one-third of the interviews had been accomplished, it was clear that we were hearing consistently enthusiastic reports from the clinicians. In wave 2 we specifically sought clinicians who had referred fewer patients and those identified by the care managers as less enthusiastic. We also continued to interview enthusiastic clinicians and pursued in more detail how they integrated care management into their overall scheme of care, what kinds of suggestions they received from psychiatrist, and what care management activities were most helpful.

In the second interim analysis, barriers to integrating care management began to emerge from less enthusiastic clinicians, along with the emerging theme that both clinicians and care managers believed some patients benefited more than others. We pursued this issue in wave 3 interviews with additional clinicians and care managers. We also began to observe varying clinician perceptions of care managers located in the practices or working from a central location. In wave 3 we focused clinician and care manager interviews in HCO-B and HCO-D, where both on-site and central care managers were being tried during the dissemination phase of the project.

In the third interim analysis, we learned that care managers believed many clinicians were more supportive of care management after they had worked together during the RESPECT trial. As a result, one author (P.A.N.) conducted additional interviews (including a few repeat interviews of clinicians) to describe clinicians’ initial and current enthusiasm for care management and what features of their experience led to a change.

The initial analyses focused separately on data from the clinicians, care managers, and mental health consultants. Themes emerged separately from the 3 groups, and our analysis sought to identify themes that were consistent across groups or that differed among groups in important ways. Throughout the iterative data collection and analysis, the team specifically sought disconfirming evidence both in the existing data and in subsequent interviews. In general, the mental health consultant interviews were all conducted in wave 1 and provided consistent information about the consultants’ role with care management. Early analysis of clinician data was conducted using all interview data, and subsequent analyses focused on the clinician interviews conducted according to the sampling strategy summarized in Table 1 ▶. Care manager interviews were conducted and analyzed in a similar manner. Finally, data were analyzed using both clinician and care manager data as the themes were fleshed out and disconfirming evidence was sought. Because the data supporting many themes were drawn from a subset of the interviews (as a result of the iterative pattern of analysis and data collection), we have indicated the approximate number of participants contributing data to each theme in the Results section.

A coding scheme was developed for interviews with clinicians and care managers, and 1 team member (K.R.) coded data using ATLAS.ti. A reliability check was performed in which another team member (P.A.N.) independently coded 5 clinician and 5 care manager interview transcripts, using the “add codes” merging feature of ATLAS.ti. Agreement among the duplicate coding was above 95% for both groups.

The 3 primary data analysts represent varied backgrounds—a primary care physician (P.A.N.), a sociologist (K.G.), and a public health analyst (K.R.). Only the physician had conducted previous research on depression care management and believed strongly in its value before the data analysis.

RESULTS

We interviewed 42 clinicians in 30 practices, and among the 5 HCOs we interviewed 18 care managers, and 7 mental health specialists. Table 2 ▶ compares the characteristics of the 5 HCOs, and Table 3 ▶ shows the distribution of care managers, clinicians, and mental health specialists interviewed within each HCO. Most practices were relatively small (2 to 4 clinicians), and most had wide latitude in organizing their practices and patterns of care.

Table 2.

Characteristics of 5 Participating Health Care Organizations (HCO)

Characteristic	HCO-A	HCO-B	HCO-C	HCO-D	HCO-E
Organizational type	Medical group	Medical group	Medical Group	Behavioral health network	Insurer
Insurance products	Fee for service	None	None	Capitated	Capitated, fee for service
Affiliated practices, No.	139	21	87	65	296
Practices owned, %	30	100	100	0	0
Total clinicians, No.	400	160	186	921	>700
Approximate patient populations, No.	>500,000 served	>100,000 served	320,000 served	>100,000 enrolled	2.3 million enrolled
Mental health carve-out, %	25	60	70	100	0
Practices from HCO participating in the study, No.	19	10	14	8	9
Among participating practices
Clinicians per practice, No. (range)	3.0 (2–6)	3.6 (2–7)	2.9 (1–6)	8.8 (2–11)	2.6 (1–4)
Clinicians participating, %	77	95	69	43	83
Practices with on-site mental health services, No. (%)	4 (21)	4 (40)	0 (0)	8 (100)	0 (0)

Table 3.

Number of Interviews Conducted by Health Care Organization (HCO) and Role in Project

Role	HCO-A	HCO-B	HCO-C	HCO-D	HCO-E	Total
Care managers	1	9	1	5	2	18
Mental health specialists	1	1	1	2	2	7
Clinicians	8	11	10	7	6	42
Total	10	21	12	14	10	67

The care managers in this project had a wide range of training and experience. Many of the care managers were registered nurses (1 had a background in psychiatric nursing) or social workers. A medical assistant was able to perform successfully as a care manager in an HCO-B practice.

Additional supporting material can be found in the Supplemental Appendix online-only at http://www.annfammed.org/cgi/content/full/6/1/30/DC1 or by correspondence with the first author.

High Clinician Acceptance of Care Management

Among the clinicians there was widespread endorsement of the value of care management for patients with depression. Aside from projected financial difficulty in paying for a care manager and the perceived challenges of making changes in practices, clinicians nearly universally believed that the care manager improved their care of depression. Most clinicians mentioned the value of regular feedback on the patient’s adherence to treatment and general progress between offices visits. Most also appreciated the care manager’s support of the patient challenges in obtaining care, including helping the patient sort through the complexities of arranging the logistics of care across several medical, mental health, and social problems. Clinicians generally believed that working with the care manager promoted a team-approach to depression care and expanded both the information available to the clinician and the scope of care the patients received. When asked, most clinicians responded that the care manager enhanced, rather than intruded, on the clinician-patient relationship.

What I think was helpful was her contact with the patient, giving me feedback on how patients are doing, suggesting when patients needed to contact me or come back sooner than I had initially recommended. I think those updates were very helpful to me (HCO-D, practice 24, clinician 30).

So there is somebody on staff who’s job it is to check in and see how it’s going and troubleshoot. It only makes for improvement in the way that depression is treated (HCO-B, practice 11, clinician 15).

Variation in Perception of Patient Benefit

After the formal RESPECT trial was completed, clinicians generally became more selective in which patients they referred for care management. During the wave 3 interviews we focused on asking an additional 10 clinicians and 10 care managers what characteristics of patients they thought would benefit most from care management. Many clinicians believed that patients with severe depression and with depression occurring within the context of other medical problems or complex social challenges would be good candidates and benefit from care management. A substantial number of clinicians also suggested that patients who are at risk of loss to follow-up, including those who had not fully accepted their diagnosis of depression, would be good candidates. Similarly, clinicians believed that patients with a change in treatment plan (including a change in medication dosage) would benefit from care manager follow-up. Many clinicians suggested that patients who seemed not be getting better were both candidates for care manager follow-up as well as mental health referral with the care manager coordinating the care. Several clinicians and care managers cited examples of the critical role the care manager played in coordinating care and communicating between primary and mental health specialty care.

Oh, yes, [care manager name] has been very helpful in both arranging referral to the counselors, making sure they actually get there, and keeping me in the loop. I’m actually getting letters back from them [mental health] now on most of my referrals (HCO-A, practice 6, clinician 8).

Certainly, if someone had a change in their management, a change in medication or a new referral, that’d be a patient I’d want help from the care manager (HCO-B, practice 8, clinician 10).

Clinician Reservations With Care Management

Although nearly all clinicians endorsed the value of the care management for their patients, 12 cited lack of reimbursement, and 1 noted that for this study the care manager and all training of practice staff had been provided by the project.

We’re lucky that the project is paying for [care manager name], and we don’t have to think about the cost. If we were going to do this on our own, I’m not sure my whole group would think it was worth the cost (HCO-C, practice 22, clinician 28).

Only 3 clinicians voiced skepticism that care management would improve the care and outcomes of their patients, and that the effect was worth the effort. Two other clinicians pointed out that making single-disease improvements for every condition of interest is not tenable in primary care. Two additional clinicians also noted how hard it is to effect change in their practice or to add to their current workload.

I simply can’t take on more. I get home about 7:00 as it is. Something else will have to give, and I don’t know what I’m doing that’s less important than this (HCO-C, practice 20, clinician 26).

After the formal trial was over, clinicians in one HCO could offer the care management services only to patients affiliated with the participating health plan. For several clinicians this limitation was a major ethical problem.

I refuse to only refer patients of [health plan name] for something that represents better care. We try very hard to meet the patient’s needs regardless of their coverage or ability to pay (HCO-E, practice 28, clinician 40).

Even those clinicians who were enthusiastic about incorporating the care manager into their practice style commented on the additional time required to add another member of the practice team, specifically to refer and communicate with the care manager.

Increased Clinician Enthusiasm After Trying Care Management

Midway through wave 3 we began to detect an emerging theme suggesting that clinicians with initial reservations about care management were more enthusiastic after working with a care manager. To pursue this finding, we interviewed 4 care managers a second time, and they identified 6 clinicians for a follow-up interview, as well as 4 new clinicians not previously interviewed.

I honestly didn’t think that the care manager would add much to my practice. I had agreed to participate in the project and this was part of it … so I went along. But after seeing some of my patients get better and like the care they got, I took it more seriously. I’d say I’m a big fan now (HCO-C, practice 23, clinician 29).

Psychiatric Oversight of Care Management as Major Support for Clinician and Care Manager

Oversight of the care manager and review of specific patients by the mental health specialist was a very popular component of the intervention. All care managers and 22 of 26 clinicians in waves 2 and 3 generally reported that the assurance and support from the mental health specialist was a major boost to their confidence in dealing with a range of patients. Medication management and detection of other psychiatric comorbidities were ranked high among the topics for which advice was given. No clinician expressed any resentment at receiving suggestions through the care manager (as opposed to directly from the specialist).

To actually have the recommendation about, “You might want to try this and this.” It’s been really useful (HCO-B, practice 11, clinician 13).

I found that there were a few (patients) where the psychiatrist consulting through the case manager made very good recommendations. And, those are certainly recommendations I would not have had otherwise (HCO-C, practice 15, clinician 20).

The mental health consultants (6 psychiatrists and 1 clinical psychologist) were unanimously enthusiastic about providing oversight for the care managers, and believed oversight was an efficient way for them to support depression management in primary care.

Location of Care Manager

For the RESPECT trial all the care management was provided in each HCO by 1 or 2 individuals who worked from a central location with multiple practices. In the dissemination phase after the trial, however, HCO-A, HCO-B, and HCO-D began to experiment with care managers located within the practice and other hybrid arrangements. For example, in HCO-B care managers located within the practice served patients from that practice as well as patients from 1 or 2 other practices. There were also care managers who split their time among several practices, making a point to be on the site of each practice during part of every week. HCO-A began to combine care management across more than 1 chronic condition and located in-house care managers in some of the larger practices. HCO-D continued to use predominantly central care managers, but it developed experience with a care manager located in a practice who also served patients in other locations. During wave 3 we interviewed 5 additional care managers and 6 clinicians who had experience with both an on-site and a centrally located care manager. In general, when the financial implications for the practice of supporting a care manager were set aside, clinicians appeared to prefer working with a care manager who was part of the practice and largely located within the practice. These clinicians believed that an in-house arrangement facilitated communication with the patient and with the clinicians.

… it’s just so much easier. She can stop me here immediately when she has a question, and we just hand the charts back and forth. We don’t have to have separate forms, … plus, we’ve found the patients very, very accepting of it when I see them and I prescribe a drug and I say, “[care manager name] is going to call you and see how you’re doing.” They know who it is and there doesn’t have to be a lot of explanation or permission or anything (HCO-B, practice 7, clinician 9).

Fundamental Importance of Establishing Clinician–Care Manager Relationship

Information on communication and developing the clinician–care manager relationship was discussed during all the interviews. Many clinicians and all care managers emphasized the importance of fostering a strong relationship between the care manager and the clinician for the care management process to work. The clinicians believed it was important to know the care manager well and to be confident that the care manager’s interaction with the patient would be consistent with the norm for their practice.

I need to feel comfortable … confident that she will interact with my patients as though she was a member of our practice (HCO-C, practice 21, clinician 27).

The care managers emphasized the importance of establishing an early face-to-face relationship with the clinicians. Several reported asking each of their clinicians how they wanted to communicate, providing a range of options (eg, fax, telephone, e-mail), including communicating through the clinician’s nurse.

It was very important to put in the work early on to understand the way each doctor wants to communicate with me (HCO-B, care manager 10).

DISCUSSION

The RESPECT-Depression Project provided the context and infrastructure for a qualitative study of implementing care management in small, autonomous, mixed-payer primary care practices. The project permitted the clinicians to experience care management and attempt to integrate it into their routine care without a great deal of up-front cost and effort. The great majority of clinicians saw value in care management and believed it enhanced the quality of their depression care. Most clinician concerns about care management related to reimbursement and competing demands on their time. Only 3 clinicians expressed skepticism that care management would improve care for their patients. It is encouraging that a number of initially skeptical clinicians became enthusiastic about care management after having the opportunity to work with a care manager, a finding consistent with Kilbourne et al’s observation that lack of exposure to longitudinal care is an important barrier to adoption of care management.¹⁸ We conclude from our analysis that the major barriers to more widespread use of care management in depression are largely economic and related less to attitudes and preferences of primary care clinicians. This finding further underscores the urgency of more enlightened reimbursement policy.

Nonetheless, our analysis points to some important enhancements and design options that need to be explored further. First, oversight of the care manager activities by a mental health consultant is a feature of some^23,28–30 but not all^3,7,10,31,32 forms of depression care management. There was a strong consensus from those interviewed that the role of the mental health specialist, acting through supervision of the care manager, (1) was an effective and efficient role for the mental health specialist, (2) provided the care manager with confidence to proceed with difficult patients, and (3) resulted in clinical suggestions that were appreciated by the clinicians. In all the clinician interviews, there was no evidence of resentment or lack of credibility of clinical suggestions passed to the clinician from the mental health specialist via the care manager. Several of the HCOs continued to support the mental health consultant in this role after RESPECT project resources ended. Routine use of mental health supervision of the care manager will have implications on both the cost and effectiveness of depression care management and deserves further research.

Many of our clinician interviews reflected the extreme time and economic pressure so common in small primary care practices.¹⁷ Competing demands for clinician time has been described in the literature^33–35 and was an often-cited barrier to use of the care manager. Clinicians appeared to be sensitive to any task, however small, that added to their workload. Even the seemingly modest commitment of time required to add the care manager to the team and to communicate with her or him presented a disincentive for some clinicians, even though many of the same clinicians cited important benefits of care management as well. Some of these clinicians even became fans and major users of care management after RESPECT offered exposure to it.

The physical location and affiliation of the care manager were perceived as important to many clinicians. Although many of the clinicians would have preferred to have a care manager based in the practice, interacting directly with the immediate care team, it is clear that economies of scale can be gained with centrally-based care managers who invest the time and effort to develop a strong relationship with the primary care clinicians. Oxman et al¹⁹ also note that on-site care managers are more vulnerable to the financial barrier and more likely to be discontinued after specific project funding ends.

We were surprised by the variation in clinician perception of which patients would benefit from care management. Individual clinician beliefs would certainly influence the number and types of patients that would receive care management. Whereas previous research suggests that care management may be particularly effective in rural and underserved populations,^36–38 there is little information to help clinicians prioritize individual depressed patients from their practice for care management, nor is it possible to determine the specific components of care management services that lead to improved outcomes in these patients. Further effectiveness research is needed in this area.

Methodologic limitations of the study include the possibility of bias, both in the selection and completion of interviews and in the inherent biases of the research team. We were able to interview only those clinicians who agreed to participate in the trial, then agreed to participate in an interview, and finally made themselves available for an interview. We were able to interview all the care managers and mental health specialists, but were only able to interview 42 of the 70 (60%) clinicians we identified as candidates. Although the research team attempted to separate their own biases about depression care from the objectivity of the interview data and often sought disconfirming evidence of major findings, this source of interpretation bias remains a possibility.

The study also has considerable strengths, including diversity in organizational and practice setting and the ongoing analysis with 4 separate waves of subject selection. Although the HCOs all volunteered for the study, not all practices, and certainly not all clinicians shared the initial enthusiasm, and we were able to talk with clinicians with a representative range of views about care management. We also had the advantage of getting perceptions of clinicians who might not otherwise have been exposed to care management. Interviews with clinicians and care managers who were introduced to the project only during the dissemination phase provided perspectives representing those given more latitude in adapting care management to their circumstances. The geographic, size, and variation in ownership among the practices enhanced the generalizability of the findings.

Our study adds to a growing body of evidence that care management for depression is effective, feasible, and acceptable to patients and clinicians. The major obstacles that remain are directly attributable to inadequate reimbursement for primary care management of depression. Adding a care manager to the primary care team has additional costs associated with the improved outcomes it can achieve, but the social and health care costs of remaining stuck in the status quo is unacceptable. There are no longer any excuses for retaining a harmful mental health reimbursement policy that restricts effective services.