Contextualizing the Effectiveness of a Collaborative Care Model for Primary Care Patients with Diabetes and Depression (Teamcare): A Qualitative Assessment Using RE-AIM

Can J Diabetes xxx (2015) 1e9

Contents lists available at ScienceDirect

Canadian Journal of Diabetes journal homepage: www.canadianjournalofdiabetes.com

Original Research

Contextualizing the Effectiveness of a Collaborative Care Model for Primary Care Patients with Diabetes and Depression (Teamcare): A Qualitative Assessment Using RE-AIM Lisa Wozniak MA a, Allison Soprovich MPH a, Sandra Rees BScPharm a, Fatima Al Sayah PhD a, Sumit R. Majumdar MD a, b, Jeffrey A. Johnson PhD a, c, * a

Alliance for Canadian Health Outcomes Research in Diabetes (ACHORD), University of Alberta, Edmonton, Alberta, Canada Department of Medicine, University of Alberta, Edmonton, Alberta, Canada c Department of Public Health Sciences, University of Alberta, Edmonton, Alberta, Canada b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 October 2014 Received in revised form 20 April 2015 Accepted 1 May 2015 Available online xxx

Objective: We evaluated the implementation of an efficacious collaborative care model for patients with diabetes and depression in a controlled trial in 4 community-based primary care networks (PCNs) in Alberta, Canada. Similar to previous randomized trials, the nurse care manager-led TeamCare intervention demonstrated statistically significant improvements in depressive symptoms compared with usual care. We contextualized TeamCare’s effectiveness by describing implementation fidelity at the organizational and patient levels. Methods: We used the Reach, Effectiveness, Adoption, Implementation and Maintenance (RE-AIM) framework to evaluate TeamCare. Qualitative methods used to collect data regarding the RE-AIM dimensions of Implementation and Effectiveness included interviews with PCN staff and specialists (n¼36), research team reflections (n¼4) and systematic documentation. We used content analysis, and Nvivo 10 for data management. Results: TeamCare was implemented as intended but with suboptimal fidelity. Deviations from the model included limited degrees of collaborative care practised within the PCNs, including varying physician participation, limited comfort in practising collaborative care and discontinuity of care managers. Despite suboptimal fidelity, respondents identified several implementation facilitators at the organizational level: training, ongoing implementation support, professional and personal qualities of the care manager and pre-existing relationships. Without knowledge of the effectiveness of the intervention in our controlled trial, respondents anticipated improved patient outcomes due to the main intervention components, including active patient follow up, specialist consultation and treat-to-target principles. Conclusions: Despite suboptimal implementation in Alberta’s primary care context, TeamCare resulted in improved outcomes similar to those demonstrated in previous randomized trials. A stronger culture of collaborative care would likely have yielded greater implementation fidelity and possibly better outcomes. Ó 2015 Canadian Diabetes Association

Keywords: collaborative care depression mixed methods qualitative research Reach, Effectiveness, Adoption, Implementation and Maintenance (RE-AIM) type 2 diabetes

r é s u m é Mots clés : soins en collaboration dépression méthodes mixtes recherche qualitative Reach, Effectiveness, Adoption, Implementation and Maintenance, soit

Objectif : Nous avons évalué la mise en œuvre d’un modèle efficace de soins en collaboration chez des patients souffrant de diabète et de dépression au cours d’un essai comparatif dans 4 réseaux de soins primaires (RSSP) communautaires de l’Alberta, au Canada. Comme dans les essais à répartition aléatoire précédents, l’intervention TeamCare menée par des gestionnaires en soins infirmiers a démontré des améliorations statistiquement significatives des symptômes de la dépression comparativement aux soins habituels. Nous avons contextualisé l’efficacité de la TeamCare en décrivant la fidélité de la mise en œuvre à l’échelon organisationnel et à l’échelon du patient.

* Address for correspondence: Jeffrey A. Johnson, PhD, 2040 Li KaShing Centre for Health Research Innovation, University of Alberta, Edmonton, Alberta T6G 2E1, Canada. E-mail address: [email protected] 1499-2671/$ e see front matter Ó 2015 Canadian Diabetes Association http://dx.doi.org/10.1016/j.jcjd.2015.05.004

2 portée, efficacité, adoption, mise en œuvre et maintien (RE-AIM) diabète de type 2

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

Méthodes : Nous avons utilisé le cadre Reach, Effectiveness, Adoption, Implementation and Maintenance, soit la portée, l’efficacité, l’adoption, la mise en œuvre et le maintien (RE-AIM) pour évaluer la TeamCare. Les méthodes qualitatives utilisées pour recueillir les données concernant les dimensions de la mise en œuvre (Implementation) et de l’efficacité (Effectiveness) du RE-AIM comprenaient les entrevues avec le personnel et les spécialistes (n ¼ 36) des RSSP, les réflexions de l’équipe de recherche (n ¼ 4) et la documentation systématique. Nous avons utilisé l’analyse de contenu et le NVivo 10 pour prendre en charge les données. Résultats : La TeamCare a été mise en œuvre comme prévu, mais a démontré une fidélité sous-optimale. Les déviations du modèle comprenaient le degré limité des soins en collaboration pratiqués dans les RSSP, dont la participation variable des médecins, le peu d’aisance dans la pratique des soins en collaboration et la discontinuité des gestionnaires de soins. En dépit de la fidélité sous-optimale, les répondants ont déterminé de nombreux facilitateurs de la mise en œuvre à l’échelon organisationnel: la formation, le soutien continu à la mise en œuvre, les qualités professionnelles et personnelles des gestionnaires de soins et les relations préexistantes. Sans connaître l’efficacité de l’intervention dans notre essai comparatif, les répondants ont anticipé l’amélioration des résultats cliniques des patients en raison des principales composantes de l’intervention, dont le suivi actif des patients, la consultation de spécialistes et les principes du treat-to-target (atteindre les objectifs du traitement). Conclusions : En dépit de la mise en œuvre sous-optimale dans le contexte des soins primaires en Alberta, la TeamCare a entraîné une amélioration des résultats cliniques qui est similaire à celle démontrée lors des essais à répartition aléatoire précédents. Une véritable culture des soins en collaboration aurait probablement généré une plus grande fidélité de la mise en œuvre et éventuellement de meilleurs résultats cliniques. Ó 2015 Canadian Diabetes Association

Introduction Comorbidity is a growing problem in Canada, with more than one-third of adults and one-half of seniors reporting 2 or more chronic conditions such as diabetes (1). Moreover, 75% of people with diabetes report having 2 or more additional chronic conditions, including mood disorders (1). People with type 2 diabetes have a 30% to 40% greater chance of experiencing depressive symptoms compared to the general population (2,3). Despite the reality of comorbidity, most chronic disease interventions focus on a single condition (4). Collaborative care, on the other hand, has been shown to improve depression outcomes in those with or without comorbidities (5e7). However, most of this evidence comes from the United States (8e10) and the United Kingdom (5,11,12). In addition, there is a lack of research concerning collaborative care conducted in primary care (13), including the innovative model of primary care networks (PCNs) in Alberta, Canada. Furthermore, few collaborative care interventions for diabetes care that have proven efficacious in randomized trials have been translated into practice (14). To address the need for comprehensive translational research on team-based models, we evaluated the implementation of a collaborative care model for depression and diabetes in 4 nonmetropolitan PCNs. TeamCare was a 12-month intervention led by a nurse care manager, who coordinated the care of patients with type 2 diabetes and depression or depressive symptoms. It was adapted from a model previously proven efficacious in randomized trials in the American-managed context called TEAMcare (8,9). In our nonrandomized pragmatic controlled trial (15), we recently demonstrated that the TeamCare intervention was clinically effective in improving depressive symptoms in people with type 2 diabetes in Alberta’s primary care context (16) and showed an effect size similar to that reported in other trials (5,6,8). Our goal was to present a more complete picture of the overall effectiveness of TeamCare. Therefore, we employed the Reach, Effectiveness, Adoption, Implementation and Maintenance (REAIM) framework (17e20) a priori and used a mixed-methods approach for a more comprehensive evaluation (21). To contextualize the evidence of the clinical effectiveness of TeamCare in Alberta’s primary care setting (16), we examined qualitative data collected during the intervention related to the RE-AIM dimensions of I (implementation) and E (perceived effectiveness). Given the

amount of data collected for our comprehensive evaluation (21), we report our findings related to the other RE-AIM dimensions elsewhere, in previous (16) and accompanying papers (22,23). Here, we describe the degree of implementation fidelity of the TeamCare intervention at the organizational level, including the delivery of intervention components, to determine whether its execution affected its effectiveness. Implementation fidelity refers to the extent to which an intervention is implemented as intended (24) or adheres to the original model (25). The degree to which an intervention is implemented with fidelity will affect its demonstrated outcomes or effectiveness. Also, we describe the perceived effectiveness of TeamCare in improving patients’ management of diabetes at the individual level from the perspectives of those responsible for its implementation because that could have implications for the future dissemination and sustainability of the intervention. For example, if the healthcare providers responsible for implementing an intervention do not think it will be effective, they may not adhere to the intervention model, thereby affecting its effectiveness. We might not recommend scaling-up an intervention in which the people delivering it believe it is ineffective, even if study outcomes show it is effective. Methods Setting: primary care networks Registered nurse care managers (CMs) implemented TeamCare in 4 nonmetropolitan PCNs in central Alberta. PCNs are similar to the medical home model in the United States (26). The 4 PCNs represented 140 family physicians serving 10 000 patients with type 2 diabetes. All 4 PCNs employed a centralized model with dedicated staff, office and clinic space to deliver PCN programs and services, including TeamCare. TeamCare intervention, training and qualifications We have described the study design and the TeamCare intervention elsewhere (15,16). Briefly, patients with depressive symptoms were identified through a mailed screening process using the Patient Health Questionnaire that was sent to those registered with the PCN as having type 2 diabetes. A CM worked with enrolled patients to develop individualized care plans over a

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

12-month period. The goals of TeamCare were to address depressive symptoms using a combination of medication and behavioural therapy, improve cardiometabolic (e.g. glycated hemoglobin [A1C] and blood pressure levels) outcomes and modify lifestyle behaviours (e.g. smoking cessation, weight loss). The CM coordinated communications among patients, their family physicians and consulting specialists (i.e. psychiatrists and endocrinologists or internists) associated with the study. The CM actively followed up with patients every 2 weeks in person or by telephone. Each CM consulted with specialists about their panel of patients weekly via telephone. The specialists suggested treatment options using evidence-based, locally endorsed algorithms, while family physicians made the final decisions regarding medications and other treatment choices. To promote fidelity across the sites, PCN staff and specialists responsible for implementing the TeamCare intervention received extensive training, including in-person workshops facilitated by the US developers of TEAMcare (8), monthly teleconferences with the US team, a project manual, one-on-one detailing and regular quality-assurance feedback, and ongoing informal implementation support through site visits, telephone and e-mail communications.

3

We conducted PCN site visits with the CMs and administrative personnel to review the TeamCare materials and resources, including the training manual adapted from TEAMcare (8,27), the web-based Care Manager Tracking System to manage patient information and track appointments, evidence-based and locally endorsed treatment algorithms, and available patient resources (28,29). We provide a summary of the training activities, materials and resources, and supports offered to ensure the consistent implementation of the TeamCare intervention across PCNs in Table 1. To evaluate the effectiveness of TeamCare, we compared the intervention in our study to a control group that did not receive the same level of intervention (16). Nonetheless, to identify eligible patients and meet ethical requirements, we initially screened patients into an active-control group, whereby notifications were sent to the family physicians of patients who screened positive for depressive symptoms. These active-control patients were also assessed for outcomes according to the study protocol (16). Despite this enhanced care of screening, notification and follow up, for the purpose of this article we refer to those patients as having usual care.

Table 1 Summary of training activities, materials and resources, and supports provided to members responsible for implementing the TeamCare intervention Training activities, materials/resources and supports

Members of the team

2-day training session on the intervention model facilitated by the developers of TEAMcare

-

Care managers Specialists Administrative personnel Family physician champions PCN administration

1-day booster session facilitated by the developers of TEAMcare

-

Care managers Specialists Administrative personnel Family physician champions PCN administration

Monthly teleconferences facilitated by the developers of TEAMcare

- Care managers - Specialists

Individual detailing sessions by research team to: Review TeamCare training manual Provide instruction and on-site training on CMTS Review evidence-based and locally endorsed treatment algorithms Review available patient resources Address PCN-specific issues (e.g. space, equipment)

- Care managers - Specialists (CMTS only) - Administrative personnel

Mock patient appointment scenarios offered by research nurse

- Care managers

Shadowing/observation of baseline patient appointment by research nurse at each PCN for quality assurance

- Care managers

Ongoing review of the CMTS by research team (nurse) for quality assurance

- Care managers - Specialists - Administrative personnel

Ongoing implementation support by research team via in-person PCN site visits, telephone, e-mail and quarterly bulletins

-

Reimbursement provided by research team to attend additional courses in diabetes or depression management (e.g. Diabetes Boot Camp, ASIST Suicide Training, Choices and Changes)

- Care managers

Clinic visits and PCN board presentations by research team or information letter sent from PCN administration describing TeamCare intervention and family physicians’ roles

- Family physicians

-

CMTS, Care Manager Tracking System; PCN, primary care network.

Care managers Specialists Administrative personnel Family physicians PCN administration

4

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

Data collection We collected qualitative data throughout the implementation of TeamCare, guided by the five RE-AIM dimensions (21). The primary sources of qualitative data included interviews, formal research teams’ reflections and systematic documentation (e.g. standardized checklist, field notes and meeting minutes). We used purposeful sampling and invited all key informants with expert knowledge of the implementation (I of RE-AIM) of TeamCare and its perceived effectiveness (E of RE-AIM) in Alberta’s primary care context to participate in the evaluation. For this qualitative assessment, the respondent groups included PCN management (i.e. Executive Directors [EDs], Chronic Disease Managers [CDMs] or their equivalents), CMs, participating specialists and our research team. We did not interview family physicians because they were not directly involved in implementing the intervention. In addition, each physician had only 1 to 2 patients participating in the study, some of whom may have been enrolled in the control group. Therefore, it would be unrealistic to ask physicians to comment on whether the intervention was implemented as intended or on perceived effectiveness. It is important to note that all respondents provided feedback on the implementation and/or perceived effectiveness of TeamCare prior to knowing the actual results of the local trial. PCN staff and specialists participated in interviews as many as 3 times (baseline, midpoint and postintervention) during the intervention, as appropriate. We conducted 36 in-person or telephone interviews, including 23 interviews with 14 PCN staff and 13 interviews with 7 participating specialists. Interviews were digitally recorded for subsequent analysis; in 1 case, detailed notes were taken for an individual who declined having the interview recorded. Interviews were transcribed verbatim by an independent transcriptionist and were verified for accuracy. In addition to in-person interviews, the EDs and CDMs in each PCN completed a standardized checklist at baseline (21). The purpose of the checklist was to document PCN characteristics, including perceived ability to implement TeamCare in the existing environments. We reviewed responses to the checklist during baseline interviews and asked the EDs and CDMs to expand upon their responses where necessary. As researchers, we documented our observations of and reflections on implementing TeamCare in each PCN, using a focusgroup format. We held 1 meeting for each participating PCN (n¼4). Team members (n¼5) received questions that allowed them to reflect upon the implementation of the intervention in each PCN, and the questions were discussed as a group. Formal group meetings lasted approximately 1 hour, and detailed notes were taken. Last, field notes and minutes from meetings were used to document the implementation of TeamCare in the PCN setting. The Health Ethics Research Board of the University of Alberta deemed this qualitative component of the larger TeamCare study exempt from review (21). Data analysis Content analysis (30,31) was used to analyze the data through an integrated approach. First, we used the RE-AIM framework as the initial coding structure. Data could be coded into multiple RE-AIM dimensions. Second, we used an inductive approach to identify emerging codes and concepts within each dimension. One researcher (LW) conducted the primary analysis. Emerging codes and concepts in each dimension were discussed as a team during formal meetings, and all discrepancies were resolved by consensus. Several verification strategies were used throughout the research process, from design to data collection and analysis, to ensure the reliability and validity of the results (32). These included ensuring methodologic coherence and appropriate sampling as

well as collecting and analyzing data concurrently (32). In addition, we participated in prolonged engagement (31,32) with the PCNs and member checking (33), and we used peer debriefing or review (30). We used study codes for highlighted quotes reflecting the respondent group, including PCN management (EDs and CDMs), CMs and specialists (psychiatrists and endocrinologists) so as to ensure participant and PCN confidentiality. Interview transcripts, checklists, notes and minutes were managed using Nvivo 10 (QSR International [Americas], Burlington, Massachusetts, USA) (34). Results Implementation Overall, the CMs delivered the TeamCare intervention components as intended across the 4 PCNs, as demonstrated through detailing and review of the Care Manager Tracking System by our research team: 1) the CMs actively managed patients; 2) the specialist consultations were held and 3) treat-to-target principles were used. However, TeamCare was not implemented with optimal fidelity, as there were deviations from the intended model related to the degree of collaboration practised (Table 2). Deviations included varied physician participation, limited comfort in practising collaborative care by team members, and discontinuity of CMs during the intervention, all of which were unintended or uncontrollable by the research team. Respondents (PCN management, CMs, specialists and researchers) reported varied physician participation in the TeamCare intervention (Table 2). For example, some physicians did not implement the specialists’ treatment recommendations or respond to the CMs’ communications. Respondents explained that the existing culture, including autonomy of physicians, physicianinitiated referral practices and traditional (i.e., hierarchical) physician-nurse relations, limited participation of some physicians in the intervention. Historically, family physicians have practised independently, and respondents reported that physicians were concerned about losing their autonomy in this collaborative care model: I think [the physicians] find the principles of the study [sound], it’s just that they feel they’re giving up a little bit of autonomy. (PCN management) The TeamCare model also circumvented the typical practice of family physicians’ initiating the referral of patients to a specialist of their choosing. In TeamCare, the CM provided family physicians with otherwise unsolicited treatment recommendations offered by participating study specialists. Respondents reported that some physicians were uncomfortable with this practice change: I know that some of the challenges, in terms of liaising with the specialistsdit’s outside the norm in terms of how physicians practice, to have a specialist provide input to a family physician, when they haven’t necessarily made a formal referral. I think some docs have had concerns about that and issues with it. (PCN management) As well, some respondents stated that traditional hierarchical physician-nurse relations impeded collaborative care in this setting: It’s always been an “us and them” thing between nurses and doctors for years and years. And most of my doctors are around my age, where when we first trained, we called them Sir and Doctor and, you know? And not making suggestions, just sort of telling them symptoms and helping them to make decisions. But

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

5

Table 2 Implementation of TeamCare: Summary of key findings TeamCare was implemented as intended but with suboptimal fidelity Components of the model delivered as intended U Active patient follow up U Specialist consultations held U Treat-to-target principles used

Deviations from delivery of model as intended related to degree of collaborative care ☒ Varying physician participation in model (e.g. not implementing specialist recommendations or responding to CM communications) - Existing culture in PCNs (i.e. autonomy, referral practice and traditional relationships) ☒ Varying comfort with collaborative care model by team members (e.g. CMs, physicians and specialists) ☒ Discontinuity of CM role (CM turnover in 2 PCNs; 2 CMs split role in 1 PCN)

Implementation facilitators and barriers, satisfaction with TeamCare intervention, and recommendations for improvement Implementation facilitators Implementation barriers U Training provided (e.g. Diabetes Boot Camp, initial training session with ☒ Finding appropriate CM for role was challenging (e.g. right mix of personal originators of TEAMcare, chronic disease management courses) and professional qualities; limited experience in either diabetes or mental U Mentorship/shadowing of CM health; communication skills) U Ongoing implementation support U Identified personal and professional qualities required of CMs: - Personal: Ability to learn quickly, effective communication skills, motivated, capable/confident, adaptable, well organized - Professional: Collaborative care background, diabetes and mental health experience, diabetes experience or mental health experience U Pre-existing relationship between participating specialists and physicians in 2 PCNs Satisfaction of team members Care managers: U Ability (i.e. time) to build relationships with patients U Enjoyed working for the PCN U Rewarding to see patients improve Specialists: U Enjoyed working on this project U Degree of satisfaction dependent on quality of relationship with CM

Dissatisfaction of team members Care managers: None reported

Specialists: ☒ Varying opinions by specialty around participating in model again: - Psychiatrists: yes - Endocrinologists: barriers include lack of time, lack of administrative support, medicolegal issues and compensation issues

Recommended improvements to TeamCare Colocation of CM and family physician similar to a decentralized PCN model may be better for implementing collaborative care models, like TeamCare (i.e. in-person communication between CM and physician) Employ changed management approach because collaborative care is not an established culture in Alberta CM, Care manager; PCN, primary care network.

they were always the boss; they were always one level above us. And a lot of my doctors are still having trouble transitioning to the team idea. (PCN CM)

training session led by the US team and the chronic disease management courses. PCN staff identified mentoring or shadowing another CM as being helpful in learning how to deliver TeamCare successfully:

Respondents reported varying ease among all team members (CMs, physicians and specialists) practising collaborative care, from very comfortable to not comfortable at all, and that it is not the standard of practice in Alberta:

[The CM] sat in on [patient] visits with [the other CMs] at first.huge mentoring role. I don’t think it’s something that you can just jump in and say I’ll do this. (PCN management)

There has to be some mentorship and some education, some ongoing facilitation. I mean, at least for an initial period of time, until [collaborative care] becomes a standard of care. (PCN management) Last, 2 nurses split the CM role in 1 PCN, as directed by the ED, and there was staff turnover of CMs in 2 PCNs. Respondents reported that the discontinuity of CMs made it problematic for the CMs to build trust and rapport with patients, specialists and physicians. Despite the suboptimal fidelity, respondents reported important facilitators to implementing TeamCare: training, ongoing implementation support, professional and personal qualities of the CMs, and pre-existing relationships between specialists and the PCN physicians. PCN staff identified didactic and experiential training as facilitators. The most valuable didactic training reported by PCN staff was the Diabetes Boot Camp course (35), the initial 2-day

Respondents cited ongoing support provided by the research team, including regular site visits and addressing implementation challenges, as essential to implementing TeamCare. They also identified personal and professional qualities required of CMs to facilitate its successful delivery. Respondents listed the following personal qualities as desirable: ability to learn quickly; effective communication skills; being motivated, capable or confident; being adaptable and well organized. Respondents also identified types of professional experience required of the CMs but varied in their opinions as to what experience was most important, ranging from a background in collaborative care to experience in both diabetes and mental health, to experience in diabetes or mental health. Last, respondents reported that pre-existing relationships between the participating specialists and PCN physicians in 2 of the 4 participating PCNs facilitated this collaborative care model. Although respondents identified several facilitators to implementing TeamCare, they reported only one barrier related to human resources: finding the appropriate CM with the right mix of

6

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

personal and professional qualities for the role. One respondent commented that the “superstars” of the nursing profession are needed to implement complex interventions like TeamCare successfully (PCN management). We assessed job satisfaction among the CMs and specialists tasked with implementing the intervention. This is an important aspect to measure because role dissatisfaction might result in high staff turnover, limiting its potential effectiveness and dissemination, regardless of demonstrated effectiveness. In general, the CMs reported a high degree of satisfaction with their role in implementing TeamCare. One CM stated, “It’s been really enjoyable. I have to say I think this is the best job I’ve had in my career” (PCN CM). Factors related to role satisfaction included time to build relationships with patients, working in the PCN environment and seeing patients improve. In contrast, the specialists reported variable satisfaction with their roles in TeamCare. Several specialists said that they enjoyed being part of the intervention. Some specialists commented that the degree of role satisfaction is dependent on the quality of relationship with the CM: “If you’re paired with a good care manager, it would be a very good experience. If you get a challenging care manager, it might not work well” (PCN endocrinologist). Last, the specialists differed by specialty in their opinions about whether they would participate in the TeamCare intervention again. The participating psychiatrists said they would participate again. However, the diabetes specialists identified barriers to continued participation, including

lack of time and administrative support as well as medicolegal and compensation issues: You would need salary support and some sort of administrative support to keep track medically and legally of all the recordsdyou really put yourself at risk by doing things at a distance. So I wouldn’t participate outside the confines of a study, I just wouldn’t do it. (PCN endocrinologist) Respondents recommended improvements to TeamCare that would increase the degree of collaboration practised. First, respondents advised having face-to-face collaboration between the CMs and family physicians (i.e. colocation) rather than the centralized model we instituted. Second, some PCN management recommended employing a change management approach to implementing TeamCare because the practice of collaborative care is not the established culture in Alberta’s primary care context. Perceived effectiveness of TeamCare Based on their experiences, and prior to knowing the results of our trial, most respondents (PCN staff, CMs and specialists) anticipated that the TeamCare intervention would result in improved patient outcomes compared to usual care (Table 3). Respondents perceived improvements in depressive symptoms and/or diabetes management among TeamCare patients. In addition, respondents

Table 3 Perceived effectiveness of TeamCare: Summary of key findings and supporting quotes Code

Subcodes

Supporting quotes

In general, PCN staff and specialists anticipated that TeamCare would result in improved patient outcomes at the midpoint and postintervention timelines Perceived effectiveness

U Perceived improvement in depressive symptoms

U Perceived improvement in quality of patient care - CM access to specialists - Efficient way for specialists to review panel of patients U Perceived improvement in diabetes management

Limited effectiveness

“We saw that there were clear results with the patients. We saw the reduction in the [PHQ-9] scores occurring every couple of weeks, which clearly indicated that what we were doing was making a difference.” (PCN psychiatrist) “One of the other successes was the consultation with the specialist. Because a lot of the times it seemed like the people that were depressed were not on a therapeutic dose of antidepressants, or they were really on the wrong thing.” (PCN CM) “I would say that based on my perceptions that it’s improved patient care in terms of.their overall diabetes care.” (PCN endocrinologist)

U Components of model contributing to improved outcomes: - Active patient follow up by a CM “I observed in a majority of cases that it was effective both for diabetes and depression, and I think it really had to do with the active faceto-face engagement with the [CM], somebody who was interested in caring for patients. So it was that active follow up and behavioural strategies around resistance, it really was the most effective.” (PCN psychiatrist) - Specialist consultation “And then the other part of it is the ease of access to the specialist. I think that was important in achieving those results.” (PCN CM) - Treat-to-target principles, especially for “[The specialist] offering suggestions to either titrate or change the depression management medication. He could just nail the right medication and if we got it right, it was like people’s mood lifted.. That was a significant change for a lot of people.” (PCN CM) ☒ Potential cointervention among usual-care patients “[The CM has] really struggled with trying to keep quiet about interventions [with usual-care patients].” (PCN management)

PCN staff and specialists reported varying opinions on increased collaborative care as a result of TeamCare at the midpoint and postintervention timelines No increase in collaborative care

Increase in collaborative care

☒

Because I/our team already practised collaborative care prior to TeamCare.

U Compared to typical working environment or previous role - CM access to specialist consultation was an example of increased collaborative care U TeamCare model promotes collaborative care

CM, Care manager; PCN, primary care network.

“I also access the other members of our team here at the primary care network. But we do that with all of our patients. So it wasn’t really different for TeamCare.” (PCN CM) “Well, certainly the [CM] collaboration with the specialist was different than what we have with usual care.” (PCN CM)

“Well, I think that it’s a great model in terms of that. And in terms of having a whole team like that.” (PCN CM)

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

identified improved quality of care because the CMs had access to specialists or because the model provided specialists with an efficient means of reviewing a panel of patients instead of consulting on a patient-by-patient basis. The factors respondents reported that contributed the most to improved patient outcomes were the 3 main components of the TeamCare model: active patient follow up, specialist consultation and following treat-to-target principles. In contrast, some respondents did not expect TeamCare to demonstrate effectiveness because they anticipated the difference in outcomes between the intervention and usual-care patients to be minimal as the result of cointervention. Examples of cointervention included usual-care patients’ being referred to existing PCN services or receiving components of the intervention from the CM during data-collection appointments. For example, one CM perceived that usual-care patients benefitted from their interactions during data collection appointments: I think the patients who were in the [usual-care] group benefitted from just being in the [usual-care] group and knowing that in 6 months, someone would be checking up on them. And all of themdand I can’t not answer questionsdtook advantage of the opportunity of having me one-on-one, even when I was doing the [data collection appointment] with them, to ask me questions about their disease and treatment and all of those things and their self-management of their diabetes. So I think they actually gained just from being in the [usual-care] group. (PCN CM) Respondents varied in their opinions as to whether or not collaborative care increased because of TeamCare. Some respondents perceived an increase in collaborative care compared to their usual work environments or previous roles. Respondents identified CM access to specialists as an example of improved collaborative care compared to usual care: “Well, certainly the collaboration with the specialist was different than what we have with usual care” (PCN CM). In addition, a few respondents stated that the TeamCare model promoted collaborative care. However, some PCN staff and specialists explained that collaborative care did not increase because they already practised this way, referring to their individual practices or the PCN setting: “The collaborative care was here [in the PCN] before” (PCN CM). Discussion We previously reported that TeamCare was clinically effective, showing statistically significant and clinically important improvements in depressive symptoms (16). Based on our comprehensive mixed-methods evaluation, we report that the TeamCare intervention components were implemented as intended across the PCNs but with suboptimal fidelity related primarily to the degree of collaborative care practised. As the Canadian equivalent of the patient-centred medical home, 1 of the major objectives of PCNs is to facilitate the use of multidisciplinary teams to provide primary care (36), so the TeamCare intervention appeared to be a good fit in Alberta’s context. Therefore, we assumed that PCNs represented high-functioning collaborative practices, with healthcare providers possessing the competencies to implement this team-based intervention. We also assumed that physicians belonging to PCNs would be supportive of, and interested in, participating in collaborative care models endorsed by their PCNs. Based on these assumptions, we did not train the healthcare providers responsible for implementing TeamCare in how to collaborate, beyond providing training in the model itself. However, bringing together health professionals from different disciplines does not guarantee that they will possess the knowledge, skills and attitudes necessary to work effectively as a team (37).

7

In our study, factors influencing the degree of collaboration practised included varied physician participation based on the existing culture (e.g. autonomy, hierarchical physician-nurse relations), limited comfort with practising collaborative care by team members, and discontinuity of CMs during the intervention. Similar barriers to collaborative care have been reported in other studies, including 2 conducted in Alberta; 1 examined nurses’ perspectives on factors influencing teamwork in the PCN environment (16) and the other explored core competencies of collaborative care (38). Barriers identified across studies, similar to our findings, included: degree of buy-in by staff (39) and physicians (16); fear of diluted professional identity (40) or threat to autonomy (41); disciplinary egos and unwillingness to accept feedback from team members in other disciplines (42); historical professional relationships (40,41); the socialization process of professional cultures (43); poor team communication (38,41,42,44), including hierarchical communication patterns (13), inadequate training or preparation among health professionals to practise collaboratively (38,40e42) and high staff turnover (42) or a low proportion of fulltime staff (44). Another factor influencing the degree of collaboration practised was appointing the right team members, with the right blend of personal and professional qualities, to implement TeamCare. The CM role was essential to achieving collaboration. Based on our findings, the CMs must have had experience in collaborative care and possess excellent communication skills, initiative and confidence. This combination of expertise requires hiring the “superstars” of the profession, making it difficult to find people with the necessary competencies. Recently, the Canadian Interprofessional Health Collaborative (CIHC) developed a national interprofessional competency framework (45). Their identification of core competencies is an important first step in consistent training of healthcare professionals in collaborative practice. Until then, the important role of mentorship by experienced nurses or other team members to achieve a high degree of collaborative care cannot be underestimated. In addition, system barriers to collaborative care, including varying models of compensation among healthcare providers and medicolegal issues, need to be recognized and addressed. Despite these challenges, respondents identified facilitators and provided recommendations for increasing the degree of collaborative care practised in TeamCare, including colocation of the CMs and family physicians and capitalizing on pre-existing clinical relationships between specialists and physicians when possible. In Alberta, the CMs delivered TeamCare via centralized PCN models, where they consulted with patients at the main PCN office instead of in physicians’ clinics. This differed from the original TEAMcare model, where the CM consulted with patients at their physicians’ clinics (8). Indeed, studies have shown that collaboration works best when team members are physically colocated (39,44) or have pre-existing clinical relationships (39). The successful delivery of collaborative care predicated on colocation and pre-existing relationships is likely to be mediated through mutual trust (42,46), which takes time to establish. The barriers and facilitators we observed appear to have influenced the degree of collaborative care practised within the TeamCare intervention in Alberta’s primary care context. However, the authors of a systematic review to identify best practices in collaborative care related to mental health in the primary care setting found that the degree of collaboration did not necessarily predict clinical outcomes (39). This may explain why most respondents in our study anticipated that TeamCare would result in improved patient outcomes or quality of care and the perception of increased collaborative care compared to usual care, despite the intervention’s suboptimal implementation. Indeed, respondents’ perceptions of the intervention’s effectiveness were concordant

8

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9

with the main results (16). The belief that TeamCare is effective and promotes or fits within collaborative practice by those implementing it bodes well for the successful dissemination of the intervention, provision-wide or in similar settings. Although the effectiveness of team-based care for depression has been proven in randomized trials (5e12), collaborative care remains an elusive concept outside of the trial setting for several reasons. It is a complex notion that is continuous and present to varying degrees (13). Often, the term collaboration is ambiguously defined (38) and is misused in research and practice settings (44,47). Researchers and practitioners use multiple terms, often interchangeably, such as collaborative care (5,8), patient care teams (48), interprofessional care (45), interdisciplinary care (44) and a variety of definitions to describe collaboration (5,39,48,49). In addition, the “active ingredients,” or mechanisms, of collaborative care that lead to positive results (e.g. clinical outcomes) remain unclear (38). Many practitioners agree that teamwork results in better care than that offered through single disciplines, but it is not always achieved easily in practice (44), and there appears to be a gap between rhetoric and reality when it comes to team-based care. The core strength of this study is contextualizing the proven effectiveness of TeamCare for patients with type 2 diabetes in primary care (16) by reporting implementation fidelity and perceived effectiveness. Regardless, our work has limitations. Although we sampled all individuals from the participating PCNs responsible for delivering TeamCare, our sample size was small. As with all qualitative research, there is limited transferability of the results beyond the present context (i.e. sample and setting). In conclusion, the collaborative care model implemented in Alberta’s primary care context resulted in reduced depressive symptoms in patients with type 2 diabetes, similar to those demonstrated in previous trials, despite suboptimal implementation. Collaborative care is a complex concept that cannot be reduced simply to 3 key intervention components in any given model. Some researchers have called for an end to researching collaborative care because its effects have been established (12). We agree with others (13), however, and recommend further research to establish the essential components and competencies of collaborative care, determine the right mix of partners, and identify the necessary conditions and supports for its success.

Acknowledgments JAJ is a Senior Scholar with Alberta Innovates-Health Solutions (AIHS) and a Centennial Professor at the University of Alberta. SRM is a Health Scholar funded by AIHS and holds the Endowed Chair in Patient Health Management funded by the Faculties of Medicine and Dentistry and Pharmacy and Pharmaceutical Sciences of the University of Alberta. This work was supported in part by grant from Alberta Health, and a CIHR Team Grant to the Alliance for Canadian Health Outcomes Research in Diabetes (reference # OTG88588), sponsored by the CIHR Institute of Nutrition, Metabolism and Diabetes (INMD). The authors thank Ana Mladenovic for reviewing and editing an earlier draft of this manuscript.

Author Contributions LW led the evaluation design, data analysis and drafted the manuscript. LW and AS developed the interview guide and conducted the interviews. AS, SR and SM actively contributed to data analysis and critically reviewed and revised the manuscript. FAS provided feedback on the manuscript. JAJ conceived of the study and was involved in data analysis and writing.

References 1. Broemeling AM, Watson DE, Prebtani F. Population patterns of chronic health conditions, co-morbidity and healthcare use in Canada: Implications for policy and practice. Healthc Q 2008;11:70e6. 2. Ali S, Stone MA, Peters JL, et al. The prevalence of co-morbid depression in adults with type 2 diabetes: A systematic review and meta-analysis. Diabet Med 2006;23:1165e73. 3. Goldney RD, Phillips PJ, Fisher LJ, Wilson DH. Diabetes, depression, and quality of life: A population study. Diabetes Care 2004;27:1066e70. 4. Wolff JL, Starfield B, Anderson G. Prevalence, expenditures, and complications of multiple chronic conditions in the elderly. Arch Intern Med 2002;162: 2269e76. 5. Gilbody S, Bower P, Fletcher J, et al. Collaborative care for depression: A cumulative meta-analysis and review of longer-term outcomes. Arch Intern Med 2006;166:2314e21. 6. van der Feltz-Cornelis CM, Nuyen J, Stoop C, et al. Effect of interventions for major depressive disorder and significant depressive symptoms in patients with diabetes mellitus: A systematic review and meta-analysis. Gen Hosp Psychiatr 2010;32:380e95. 7. Thota AB, Sipe TA, Byard GJ, et al. Collaborative care to improve the management of depressive disorders: A community guide systematic review and metaanalysis. Am J Prev Med 2012;42:525e38. 8. Katon WJ, Lin EH, Von Korff M, et al. Collaborative care for patients with depression and chronic illnesses. N Engl J Med 2010;363:2611e20. 9. Katon WJ, Von Korff M, Lin EHB, et al. The pathways study: A randomized trial of collaborative care in patients with diabetes and depression. Arch Gen Psychiatr 2004;61:1042e9. 10. Jacob V, Chattopadhyay SK, Sipe TA, et al. Economics of collaborative care for management of depressive disorders: A community guide systematic review. Am J Prev Med 2012;42:539e49. 11. Richards DA, Hill JJ, Gask L, et al. Clinical effectiveness of collaborative care for depression in UK primary care (CADET): Cluster randomised controlled trial. BMJ 2013;347:f4913. 12. Richards DA, Lovell K, Gilbody S, et al. Collaborative care for depression in UK primary care: A randomized controlled trial. Psychol Med 2008;38:279e87. 13. Schmitt MH. Collaboration improves the quality of care: Methodological challenges and evidence from US health care research. J Interprof Care 2001; 15:47e66. 14. Wagner EH, Glasgow RE, Davis C, et al. Quality improvement in chronic illness care: A collaborative approach. J Comm J Qual Improv 2001;27:63e80. 15. Johnson JA, Al Sayah F, Wozniak L, et al. Controlled trial of a collaborative primary care team model for patients with diabetes and depression: Rationale and design for a comprehensive evaluation. BMC Health Serv Res 2012; 12:258. 16. Johnson JA, Al Sayah F, Rees S, et al. Collaborative care vs. screening and follow-up for patients with depression and diabetes: Results of a primary-care based comparative effectiveness trial. Diabetes Care 2014;37:3220e6. 17. Glasgow RE, Vogt TM, Boles SM. Evaluating the public health impact of health promotion interventions: The RE-AIM framework. Am J Pub Health 1999;89: 1322e7. 18. Glasgow RE, McKay HG, Piette JD, Reynolds KD. The RE-AIM framework for evaluating interventions: What can it tell us about approaches to chronic illness management? Patient Educ Couns 2001;44:119e27. 19. Virginia Polytechnic Institute and State University. Reach effectiveness adoption implementation maintenance (RE-AIM). http://www.re-aim.org/. Accessed June 17, 2015. 20. Glasgow RE. Translating research to practice: Lessons learned, areas for improvement, and future directions. Diabetes Care 2003;26:2451e6. 21. Wozniak L, Rees S, Soprovich A, et al. Applying the RE-AIM framework to the Alberta’s Caring for Diabetes Project: A protocol for a comprehensive evaluation of primary care quality improvement interventions. BMJ Open 2012;2:5. 22. Wozniak L, Soprovich A, Rees S, et al. Challenges in identifying patients with type 2 diabetes for quality-improvement interventions in primary care settings and the importance of valid disease registries. Can J Diabetes [In press]. 23. Wozniak L, Soprovich A, Rees S, et al. Impact of organizational stability on adoption of quality-improvement interventions for diabetes in primary care settings. Can J Diabetes [In press]. 24. Breitenstein SM, Gross D, Garvey CA, et al. Implementation fidelity in community-based interventions. Res Nurs Health 2010;33:164e73. 25. Mowbray CT, Holter MC, Teague GB, Bybee D. Fidelity criteria: Development, measurement, and validation. Am J Eval 2003;24:315e40. 26. Rittenhouse DR, Shortell SM. The patient-centered medical home: Will it stand the test of health reform? JAMA 2009;301:2038e40. 27. University of Washington, Group Health Cooperative. TEAMcare: Managing depression, diabetes and coronary heart disease in primary care. http://www. teamcarehealth.org/Portals/0/Documents/TEAMcare%20Intervention%20Manual %202015.pdf. Accessed June 17, 2015. 28. Katon W, Ludman EJ, Simon G. The Depression Helpbook. 2nd edn. Boulder: Bull Publishing Group, 2008. 29. The Support Network. http://www.thesupportnetwork.com/. Accessed June 17, 2015. 30. Morse JM, Field PA. Qualitative research methods for health professionals. 2nd edn. Thousand Oaks: Sage, 1995. 31. Mayan MJ. Essentials of qualitative inquiry. Walnut Creek: Left Coast Press, 2009.

L. Wozniak et al. / Can J Diabetes xxx (2015) 1e9 32. Morse JM, Barret M, Mayan M, et al. Verification strategies for establishing reliability and validity in qualitative research. Int J Qual Meth 2002;1:2. 33. Guba EG, Lincoln YS. Fourth-generation evaluation. Newbury Park: Sage, 1989. 34. QSR International. Nvivo 10. http://www.qsrinternational.com/products_nvivo. aspx. Accessed June 17, 2015. 35. The National Diabetes Managment Strategy. Primary Care Diabetes Boot Camp. http://tndms.ca/bootcamp/diabetes_bootcamp.html. Accessed June 17, 2015. 36. Alberta Primary Care Networks. Alberta PCNs. https://www.pcnpmo.ca/ alberta-pcns/Pages/default.aspx. Accessed June 17, 2015. 37. Miller C, Ross N, Freeman M. Shared learning and clinical teamwork: New directions in education for multiprofessional practice. London: English National Board for Nursing, Midwifery and Health Visiting, 1999. 38. Suter E, Arndt J, Arthur N, et al. Role understanding and effective communication as core competencies for collaborative practice. J Interprof Care 2009;23:41e51. 39. Craven MA, Bland R. Better practices in collaborative mental health care: An analysis of the evidence base. Can J Psychiatr 2006;51(Suppl 1):S7e72. 40. Headrick LA, Wilcock PM, Batalden PB. Interprofessional working and continuing medical education. BMJ 1998;316:771e4. 41. Zwarenstein M, Reeves S. Working together but apart: Barriers and routes to nurse-physician collaboration. Jt Comm J Qual Improv 2002;28:242e7. 209.

9

42. Al Sayah F, Szafran O, Robertson S, et al. Nursing perspectives on factors influencing interdisciplinary teamwork in the Canadian primary care setting. J Clin Nurs 2014;23:2968e79. 43. Hall P. Interprofessional teamwork: professional cultures as barriers. J Interprof Care 2005;19(Suppl 1):S188e96. 44. Xyrichis A, Lowton K. What fosters or prevents interprofessional teamworking in primary and community care? A literature review. Int J Nurs Stud 2008;45: 140e53. 45. Canadian Interprofessional Health Collaborative. A national interprofessional competency framework. Vancouver: University of British Columbia, 2010. 46. Sommers LS, Marton KI, Barbaccia JC, Randolph J. Physician, nurse, and social worker collaboration in primary care for chronically ill seniors. Arch Intern Med 2000;160:1825e33. 47. Henneman EA, Lee JL, Cohen JI. Collaboration: A concept analysis. J Adv Nurs 1995;21:103e9. 48. Wagner EH. The role of patient care teams in chronic disease management. BMJ 2000;320:569e72. 49. Gunn J, Diggens J, Hegarty K, Blashki G. A systematic review of complex system interventions designed to increase recovery from depression in primary care. BMC Health Serv Res 2006;6:88.