|Home | About | Journals | Submit | Contact Us | Français|
Recommendations for improving care include increased patient-clinician collaboration, patient empowerment, and greater relational continuity of care. All rely upon good clinician-patient relationships, yet little is known about how relational continuity and clinician-patient relationships interact, or their effects on recovery from mental illness.
Individuals (92 women, 85 men) with schizophrenia, schizoaffective disorder, affective psychosis, or bipolar disorder participated in this observational study. Participants completed in-depth interviews detailing personal and mental health histories. Questionnaires included quality of life and recovery assessments and were linked to records of services used. Qualitative analyses yielded a hypothesized model of the effects of relational continuity and clinician-patient relationships on recovery and quality of life, tested using covariance structure modeling.
Qualitative data showed that positive, trusting relationships with clinicians, developed over time, aid recovery. When “fit” with clinicians was good, long-term relational continuity of care allowed development of close, collaborative relationships, fostered good illness and medication management, and supported patient-directed decisions. Most valued were competent, caring, trustworthy, and trusting clinicians who treated clinical encounters “like friendships,” increasing willingness to seek help and continue care when treatments were not effective and supporting “normal” rather than “mentally ill” identities. Statistical models showed positive relationships between recovery-oriented patient-driven care and satisfaction with clinicians, medication satisfaction, and recovery. Relational continuity indirectly affected quality of life via satisfaction with clinicians; medication satisfaction was associated with fewer symptoms; fewer symptoms were associated with recovery and better quality of life.
Strong clinician-patient relationships, relational continuity, and a caring, collaborative approach facilitate recovery from mental illness and improved quality of life.
Recent recommendations for improving the delivery of health care include suggestions for substantive collaboration between patients and clinicians, increased patient empowerment and activation, and greater continuity of care, including seeing the same clinician over time (relational continuity) (Hibbard, Stockard, Mahoney, & Tusler, 2004; Institute of Medicine et al., 2006; Institute of Medicine & Committee on Quality Health Care in America, 2001; Reid, Haggerty, & McKendry, 2002; US Department of Health and Human Services, 1999; World Health Organization, 2006). Although these factors are generally addressed as independent indicators of care quality, that conceptualization may not accurately reflect real-world care delivery, for these practices and behaviors occur within the context of clinician-patient relationships, and good relationships are a likely prerequisite for producing the kind of collaboration and activated patients envisioned in these recommendations. Thus, these quality-of-care indicators may be more appropriately conceptualized as interrelated indicators of the quality of clinician-patient relationships. Berwick makes this point eloquently, arguing that “health care tends to regard human interactions more as a toll or price than as a goal or product,” and proceeds as though interactions between clinicians and patients are the “burden it must bear so that it can deliver the care,” (pp. 50–51) when, in reality, interactions themselves are the care, and “To perfect care, we must perfect interactions” (p. 52) (Berwick, 2002).
For these reasons, improved “continuity of care,” in and of itself, may not directly improve the results of mental health or other care. For example, some continuity of care indicators measure whether or not an individual has had a visit with any clinician over a particular period of time, while others measure whether outpatient care was received after a hospitalization, or whether an individual’s care was received within the same clinic over a specific time period, irrespective of the number of clinicians that person visited within the clinic (Adair et al., 2003; Freeman & Sheppert, 2000; Johnson, Prosser, Bindman, & Szmukler, 1997; Reid et al., 2002). While it is clear that these forms of continuity have an important place in understanding how care is delivered, none addresses the human relationships and interactions inherent in good care, or the knowledge that is, or is not, brought to bear when relational continuity is present or absent. For example, when coordination between clinicians is lacking or minimal, “continuous” care may be provided, but with little information passed from one clinician to the next. In such situations, patients and family members must tell their stories over and over, and little “real” continuity exists (see Wyden’s [Wyden, 1998] powerful account of how he often provided the only continuity in his son’s long-term treatment for schizophrenia).
Given the large differences in current continuity measures, and the complexities of the interactions and processes they are attempting to assess, it is not surprising that research evaluating the effects of “continuity of care” has produced inconsistent results (Adair et al., 2003; Adair et al., 2005; Freeman & Sheppert, 2000). In addition, most studies of care continuity have followed people for relatively short periods (12 months or less) (Adair et al., 2005), yet the kinds of clinician-patient relationships necessary for improving care may take significant time and experience to develop. Thus, what we have come to call “continuity of care” could more realistically be viewed as a group of somewhat-related measures that serve as proxies for a more complex set of interrelationships that include: (a) coordination of care between clinicians working simultaneously with one patient, (b) smooth transitions in care from one clinician or service setting to another, c) clinician-patient interaction patterns, d) learning, education, and information exchange on the part of clinicians and patients, and, perhaps most importantly, e) the construction of a shared history of illness and recovery and the development of important interpersonal ties (Chien, Steinwachs, Lehman, Fahey, & Skodol-Wilson, 2000; Donaldson, 2001; Freeman & Sheppert, 2000; Johnson et al., 1997).
For these reasons, to truly understand the effects of different forms of continuity of care, we must focus on the processes by which particular types of continuity affect specific outcomes. To date, few studies have addressed such process-related questions among individuals with serious mental health problems (Johnson et al., 1997), even though such individuals are likely to have greater need for continuity of care because of symptoms, memory problems, and cognitive deficits that may interfere with their ability to provide information about their treatment history. Moreover, despite recent recognition of the importance of including patient perspectives in research (Corrigan & Ralph, 2005), studies of continuity of care have rarely included patient or consumer experiences, or assessed their preferences (Crawford, de Jonge, Freeman, & Weaver, 2004; Joyce et al., 2004).
An exception to these criticisms is work by Ware et al. (Ware, Tugenberg, Dickey, & McHorney, 1999) who interviewed and observed a small number of clinicians (n = 16) and patients (n = 16), carefully detailing processes by which clinicians and other staff worked to improve various types of continuity. They found that clinicians a) stepped outside of prescribed roles to meet patient needs, b) anticipated potential trouble spots and took action to prevent problems, c) worked to smooth transitions by creating service or provider overlap or by delaying some transitions to reduce the impact of others, d) tried to be flexible about access to care, and e) prodded the system to improve response times. When clinicians had shared clinical histories with clients, they were able to situate current circumstances or behavior in relation to prior circumstances or behavior, often helping to redefine current struggles as improvements over prior difficulties. Yet, though these processes occurred in the context of relationships between clinicians and clients, Ware et al.’s descriptions focused primarily on clinician activities and processes, and did not directly address clinician-patient relationships. More recently, Ware and her team examined the perspectives of low-income consumers on their relationships with clinicians (Ware, Tugenberg, & Dickey, 2004), reporting that consumers appreciated receiving “extra things”–things normally outside the constrained clinician-client relationship, that consumers wanted to feel that they had some form of common ground with their clinicians, such as race or religion, and valued feeling known by their clinician. The opportunity to talk in a context of acceptance and warmth, and without fear, was also important, as was being treated as an individual and having the opportunity to provide input into treatment. Finally, consumers identified practitioner availability and flexibility for scheduling appointments as important aspects of these relationships. These later analyses, however, did not examine the connection between relational continuity of care, as explored in the former work, and these clinician-patient relationships.
Our work addresses this gap in the literature by extending Ware et al.’s work. We use triangulated qualitative and quantitative data to examine clinician-patient relationships in the context of relational continuity. We also examine the associations between relational continuity, clinician-patient relationships, quality of life, and recovery. We report results from the STARS Study (Study of Transitions and Recovery Strategies), an exploratory, longitudinal, mixed-methods study of recovery from serious mental illness.
The setting for this study was Kaiser Permanente Northwest (KPNW), a nonprofit, prepaid, group model, integrated health plan serving about 480,000 members in Oregon and Washington State. KPNW provides outpatient and inpatient medical, mental health, and addiction treatment, and maintains an integrated electronic medical record that contains comprehensive administrative and treatment data on all its members. Clinicians are salaried employees of either the health plan or the Permanente Medical Group.
STARS is a mixed-methods, exploratory, longitudinal study of recovery among individuals with serious mental health problems (schizophrenia, schizoaffective disorder, bipolar disorder, or affective psychosis). To participate, individuals agreed to complete two in-depth interviews at baseline, a baseline paper-and-pencil questionnaire, another interview and questionnaire 12 months following study enrollment, and a final interview and questionnaire 24 months post-enrollment. In-depth interviews at baseline were designed to explore mental health history, personal experiences affecting mental health and recovery as well as experiences with mental illness, symptoms, and mental health care. We also sought information about relationships with family and friends, current life circumstances, and about experiences, individuals, and role models that had been influential in participants’ lives or recovery process. Baseline questionnaires assessed quality of life, happiness, psychiatric symptoms, recovery, stigmatizing experiences, alcohol and drug use, regular activities, living situation, and socio-demographic characteristics. Questionnaire data were linked to health plan records of diagnoses, service use and clinician information. Results presented here were derived from information collected at baseline, so reflect historical information as well as baseline status and service use during the 12 months prior to study enrollment. The study was approved and monitored by KPNW’s Institutional Review Board and Research Subjects Protection Office. All study participants provided informed consent prior to participation.
Study participants were identified using health plan membership and diagnostic records. Inclusion criteria were: having a diagnosis (for a minimum of 12 months) of schizophrenia, schizoaffective disorder, bipolar disorder, or affective psychosis; having at least 12 months of health plan membership prior to study enrollment; being age 16 years or older; and planning to stay in the area for at least 12 months. We excluded those with diagnoses of dementia, mental retardation, or organic brain syndrome because such conditions interfere with individuals’ ability to provide informed consent, participate in interviews, and complete paper-and-pencil questionnaires. We also excluded those whose mental health clinician felt they were unable to participate.
After extracting a pool of potential participants (n = 1827) we prepared recruitment letters to be signed by the principal investigator and the member’s mental health clinician (or primary care provider when we found no specialty mental health visits). We asked 213 providers to sign letters for individuals they believed able to participate. Of letters reviewed, 15.8% were screened out and 15 clinicians did not return letters at all (representing 17 patients). Potential participants who met inclusion criteria were stratified based on diagnosis (mood vs. schizophrenia spectrum) and gender, and were selected randomly within these groups to achieve roughly equal representations of men and women, and of individuals with mood (bipolar disorder, affective psychosis) or schizophrenia spectrum (schizophrenia, schizoaffective disorder) disorders.
Recruitment letters were mailed in small batches over a period of 10 months and followed by a telephone call if we received no response to the letter. We surpassed our recruitment goal of 170 after attempting to recruit 418 people. Of these 418 attempts, we successfully contacted 350 individuals, received 127 refusals, and found 22 individuals ineligible for the study. Overall, we enrolled 184 of the 418 persons we attempted to recruit, 46% of those who were eligible. Of these, 3 did not complete both baseline interviews and 4 were excluded because study staff determined that medical record diagnoses had been in error. Data from these latter 4 individuals were not included in analyses.
Study participants were 177 KPNW members, 93 women (52.2%) and 85 men (47.8%) with serious mental illness. The average age of participants at baseline was 48.8 (SD = 14.8) years, ranging from 16 to 84 years. The enrolled sample distributions for age and sex, within diagnosis, did not differ from the study-eligible population of health plan members. Additional demographic and descriptive information may be obtained from the senior author.
All participants completed two baseline in-depth interviews, averaging four hours in total. Interviews were semi-structured and included questions addressing mental health history and diagnoses, significant life events, current symptoms and symptom management, experiences with mental health care, relationships with clinicians, changes wished for in mental health care, and other questions. All interviews were audio taped and transcribed, verbatim. Study investigators and interviewers reviewed transcripts weekly throughout the data collection period to ensure transcript accuracy and appropriate interviewing techniques, and to develop a general descriptive coding scheme. Interviews were then coded using the software program Atlas.ti (Muhr, 2004).
Following completion of preliminary coding, three of the authors (CG, S), DC) developed a secondary, detailed coding scheme, specific to relationships with clinicians, by reviewing all material assigned to codes for relationships with mental health care providers and continuity of care, and identifying themes and ideas present in the text. This secondary coding scheme contained 23 sub codes; relevant interview text was coded a second time using the secondary coding scheme. We then extracted common themes from codes with more complex pattern content (e.g., mutuality, comfort with provider), and identified examples illustrative of codes with relatively unidimensional content (e.g., availability/access to care as needed). In the sections that follow, we describe the most commonly discussed and important themes derived from these interviews. We completed check coding throughout this process to ensure coder consistency; inconsistencies were discussed and resolved by the team, and code definitions were revised when needed.
Based on our qualitative analyses, we generated an inductively derived theoretical model of how relationships with clinicians and clinician approach affect quality of life and recovery. We tested this model using questionnaire, health plan, and interview-based data in a covariance structure model using AMOS Version 6 (Arbuckle, 2005). Missing data from the questionnaire were few so were estimated using the full information maximum likelihood procedure available in AMOS. Results of both qualitative and mixed-methods analyses are presented in the sections that follow.
173 of the 177 participants completed baseline paper-and-pencil questionnaires. Using KPNW’s automated databases, we linked questionnaire and interview data to records of services used and to the clinicians who provided each service. All measures were included in a covariance structure model testing the processes hypothesized following the qualitative analyses. Definitions of quantitative measures follow:
Based on Steinwachs (1979), we calculated UPC for the 12 months prior to the date each participant was enrolled in the study. We calculated three UPC measures for these 12 months, each measured as the number of visits to the most-seen provider in three provider categories–mental health prescribers (psychiatrists and nurse practitioners), counselors, and primary care providers–divided by the total number of visits to all providers seen within each category. UPC is calculated only when an individual has more than one visit in a provider category. Because participants received mental health care from different types of providers, not all had a UPC score for each provider category. To manage this problem, we created one combined UPC variable. This single variable was populated hierarchically in the following order: Mental health prescribers (70.3%), mental health counselors (5.7%), and primary care providers (17.7%). Thus, if participants did not have a mental health prescriber, but did have a counselor, this variable codes continuity with their counselor. Similarly, when participants had no mental health prescriber or counselor, this variable codes continuity with their primary care provider. After reviewing the distribution for the combined measure, we receded UPC to a binary indicator of 100% continuity of care (coded 1) vs. less than 100% continuity of care (coded 0). We made this decision because 66.5% had seen their mental health care provider 100% of the time in the prior year, which would have violated assumptions of normality for statistical analyses.
We included three self-reported quality-of-life indicators in our analyses: The first was a single item from the Wisconsin Quality of Life Index rating overall quality of life during the prior four weeks, scaled from 1 to 10. The second was from the National Opinion Research Center’s (NORC) U.S. General Social Survey 1972–1996 rating general happiness from “not at all happy” to “very happy.” The third measure was the Wisconsin Quality of Life Index General Satisfaction Scale (Becker, Diamond, & Sainfort, 1993). This scale is importance-weighted and measures overall life satisfaction with neighborhood, housing, food, clothing, transportation, sex life, personal safety, and with how time is spent during the day. We removed one item typically included in the scale because it measures satisfaction with mental health services and we were examining, in part, relationships between satisfaction with mental health services and quality of life and recovery outcomes.
Our primary measure of mental health symptoms was the Colorado Symptoms Inventory, measuring anxiety/depression, psychoticism, and dangerousness to self/others (Shern et al., 1994). We also used the SF-ia Mental Health and Role-Emotion sub-scales, assessing general mental health and emotional functioning (Ware, Kosinski, & Keller, 1995; Ware, Jr., Kosinski, Turner-Bowker, & Gandek, 2002).
Satisfaction with clinicians was measured using a 5-point scale from “very satisfied” to “very dissatisfied,” asked separately for medication prescribers, counselors, and primary care providers. The three questions assessed a) “the personal interest and attention your [provider type] gives you?” b) “your [provider type’s] competence, skill, and ability?” and c) “the amount of information and explanation your [provider type] gives you?” As with the continuity of care measure, we used a hierarchical approach in populating these variables. We began with satisfaction with prescribers (the most commonly seen practitioner for mental health care). If participants did not have a prescriber, then the variable was populated with information about satisfaction with counselors, followed by primary care providers.
We collected information about medication satisfaction among those taking medications (n = 158) with three questions: a) “Do you have side effects (unwanted effects) from them?” (None, Slight, Mild, Moderate, or Severe) b) “Do you feel the medication helps control your symptoms?” (Not at all, Some, A fair amount, Quite a bit, Eliminates all symptoms) c) “How do you feel about taking your psychiatric medications?” (Very Dissatisfied to Very Satisfied on a 7-point scale).
We had three indicators of recovery-oriented, patient-directed care: a) “My mental health care provider(s) believes I can recover,” b) “When I ask for help I get the kind of help I want,” and c) “How often have you discussed your recovery with your current mental health care provider(s)?” The former two questions are rated on a 5-point scale from “strongly agree” to “strongly disagree.” Responses to the latter question were “never,” “sometimes,” and “frequently.”
We used two measures of recovery in analyses, both subscales of the Recovery Assessment Scale (RAS) (Corrigan, Giffort, Rashid, Leary, & Okeke, 1999). They included Personal Confidence & Hope (based on 9 items) and Goal & Success Orientation (based on 5 items).
We derived a quantitative measure of participants’ perceived fit and comfort with their clinician (at baseline) using descriptions of their relationships present in our qualitative data. Coders rated relationships as having either “poor,” “moderate,” or “good” fit and comfort. Unqualified positive or negative assessments of current clinician-patient relationship were coded “good” or “poor,” respectively. Relationships described as having both positive and negative aspects, or that were neutral, were coded as having “moderate” fit and comfort. This resulted in a three-point scale ranging from poor to good. Check coding of 24 transcripts indicated coder agreement for “fit” at 87.5%.
Average lifetime health plan membership was 16.2 years (sd = 10.3 years) and ranged from 1.8 to 44.8 years. Participants had been seeing their current clinician for an average of 5.2 years (sd = 4.2 years) with a range from o to 18 years. Slightly over 46% of the sample had been visiting their current clinician for 5 years or more. Despite these longstanding relationships, the majority of study participants also had experiences with other clinicians that they could bring to bear when discussing their care experiences.
Our discovery of these long-term relationships, and their importance to participants, stimulated the work reported in this paper. As part of our analyses, we extracted eight primary themes from the interview data. These are described below, as are important sub-themes. Themes were generally interwoven, however, with individual themes rarely appearing alone.
The need for a good “fit” and to feel comfortable with clinicians was a common theme in participants’ descriptions of what made for good relationships. Most often, participants made relatively quick assessments upon meeting a clinician about whether or not the “fit” was good and most opted for change when it seemed poor. For example:
I tried to go to this mental health counselor… and my God,…it was like…once I had eye contact with her, if I tried to glance away or look down, anything, I’d look up and …I was impaled on her eyes. I think I went to her twice, and I could not go back to her again.
Participants described multiple factors as contributing to the fit and comfort they felt with clinicians. Many mentioned some shared experience or characteristic, including shared values, experiences, gender, age, or religious or world views. Also of great importance were the following: the clinician attending carefully to the patient and accepting the patient unconditionally; having the right balance of closeness or distance in the relationship; the clinician not threatening the patient (as with involuntary hospitalization); the clinician being able to have fun, a sense of humor, or a positive attitude; the clinician providing honest feedback and pushing the patient to take on more responsibility when it was appropriate; the clinician’s ability to stay calm during difficult times; and the clinician being willing to not just “do it by the book.” Often, the best “fit” included a sense of comfort across several of these dimensions.
Interviewer: Could you describe things you’d look for in a clinician that would be a good match for you?
Participant: Somebody who would look at you and listen to you, and would work with me on medication, and that I would feel comfortable with, comfortable meaning at ease, and not nervous, and somebody that would remember what I had told them about my family, or about my life and wouldn’t have to ask me every time I came in, somebody I felt was personally interested in me, somebody that would return my calls within a day, and, or let me know some way why they didn’t return my call. I like [PSYCHIATRIST] having [NURSE] as his nurse because I can go through her, and right away…she’s available.
The following passage illustrates another aspect of comfort in clinical relationships–interpersonal distance. It suggests that the clinician was able to provide care with the interpersonal distance this participant preferred. While other participants wanted closer relationships with their clinicians, this participant was more comfortable with greater distance.
Now Doctor [NAME] is kind of a hands-off type of a guy. You walk in and he says, “Hi, how are you doing? I know you’ve got the procedure down well.” [I say] Three months ago I had to take that extra half a pill, but everything has been all right. Then the rest of the conversation is, “Do you have any holiday plans?” Of course, he’s looking at me and he’s evaluating me through these conversations, but nobody is trying to get inside my head to figure out if I have had a neurosis all of my life. It’s just very comfortable. It’s just the way that I would want it to be.
When patients were not comfortable with their clinicians, some participants mentioned that it could lead to poor treatment adherence. For example, one participant said “I had never encountered anyone that was like her [clinician]. She was really odd…lt really discouraged me from going to see the doctor and taking my medicine. I didn’t want to take the things that she thought I should take. It was really a hindrance to my recovery….” Conversely, comfortable relationships encouraged communication about decisions and concerns that could affect mental health outcomes in important ways:
I am thinking of trying some Dexatrim, or something like that because I take [medications] which increase my appetite…! want them [clinicians] all to know about it because my behavior can change with the addition of an appetite suppressant, or whatever. I’m not going to do it without a doctor’s supervision. I’m going to consider it. I’m going to try it, but I feel comfortable enough to ask them about it instead of going behind their backs. That’s an important thing to me.
A caring, compassionate approach on the part of the clinician was essential in facilitating the development of the long-term partnerships we encountered. This was the most commonly discussed and prominent theme in participants’ descriptions of good and bad relationships with mental health care providers. Patients described many ways clinicians could show they cared, while conversely, the absence of these practices was often seen as indicating that a clinician did not care. Key indicators that a clinician was caring and compassionate included the clinician listening carefully, providing hope and reassurance, returning calls during off hours, spending extra time when it was needed, and checking in to make sure all was well during difficult periods. Overall, individuals had a strong sense of whether or not a clinician cared about them and wanted to help. The following passages illustrate caring practices and the strength of feelings that resulted when such care was received:
Providers who were perceived as not caring were disliked, avoided, and left behind, sometimes leading to gaps in care or less frequent visits than were needed. Some participants described clinicians who were harsh or judgmental while other clinicians were seen as impersonal and businesslike. Such clinicians were generally perceived as uncaring. The following passage shows how what might be deemed appropriate care from a medical standpoint can be perceived as uncaring, lead to reduced consistency in visits, and poor adherence to prescribed treatment.
Interviewer: Tell me why your psychiatrist is not helpful.
Participant: She’s very meticulous, she’s very businesslike, she’s very short with you. All she wants to hear about is how the medication is working and do we need to change it, that’s it. She has no compassion. Not that I go in there with a problem that I’d ever discuss with her, but it would be nice to feel comfortable enough to say this isn’t working right, or that isn’t working right, because she’s just so quick, let’s try this, let’s try that. We have tried so many different things.
Interviewer: Is there a reason why you’re sticking with the same psychiatrist?
Participant: No, I’ve been debating, because I’m finding myself putting off going to see her, putting it off and putting it off.
A common subtheme within the desire for a caring and compassionate approach was the importance of clinicians listening, understanding, believing, and truly knowing their patients. One participant said “He [psychiatrist] mostly controlled my medication. He didn’t really know me as a person.” Relationships with clinicians were undermined when they did not listen well, believe, or seem to understand their patients, or, perhaps worst, even remember their patients or what they had said during previous consults:
Unexpectedly, a large number of participants described relationships with their clinicians more as friendships than as clinical relationships. Moreover, these friendships were described as mutual–with clinicians sharing and contributing to the relationship in ways that are often considered outside the purview of clinical encounters, such as by sharing some personal or family information (e.g., births of children or grandchildren). Finally, more “friendly” relationships helped people feel “normal,” rather than “mentally ill.” For example:
Participants also talked often about the importance of being able to trust their clinician and, equally as important, to have their clinician trust them. Such trust was developed in the context of long-term relationships and could provide important boosts to self-esteem and feelings of self-worth. The following quotations illustrate discussions about the importance of trust-that of the patient for the clinician and of the clinician for the patient.
Another theme stressed by many participants was the importance of having a collaborative relationship with their clinician and, as part of that relationship, having some control over treatment decisions.
Participant: Actually, I just met her recently in the last month, and so far out of all of the doctors that I have seen here she seems to be the best. She is the most informed, the most willing to talk and work with me and what I want. Of course, this is what she wants me to do. We work together. I think she’s really good so far.
Interviewer: So…working together to figure out what’s going to be best for you has been helpful?
Participant: Uh huh. Then I get to keep some of that control.
Many of these discussions addressed how important it was for participants to have knowledge about medications, support from clinicians in controlling and making day-to-day medication adjustments, and clinicians responsive to requests for medication changes when symptoms worsened or side-effects were problematic.
Interviewer: How has your mental health care…affected your progress toward making your life what you want it to be?
Participant: It helped me feel good about myself, especially with Doctor [NAME] because he talked to me as an equal. He didn’t talk down to me. He didn’t talk to me as if I were some kind of animal. He discussed the medications with me. He trusted me. He trusted me to medicate myself or to make a choice. He listened to me. Any questions that I had, he answered me openly and truthfully. He didn’t beat around the bush about anything. Anything that I would ask point blank, he would answer me. I’m really impressed with him as a doctor. I like him.
Participants reported that seeing the same clinician over time provided the historical background and experience necessary to get good care and to form a good working partnership.
I think the main reason I’ve stuck with him [for 25 years] is because I tend to be someone that if I find something that works…then I’ll stick with it, and there’s a lot to be said for having a provider who knows your history, knows where you’ve been. …He definitely always lets me share my ideas, and his approach is we work things through together and he listens to me, so…that’s the reason I stuck with him.
When individuals had good and longstanding relationships with their clinicians, losing access to their clinician could be devastating.
Interviewer: And what was that like just not being able to see him [valued clinician]…?
Participant: I just felt like a plant ripped up by its roots.
Almost as common as discussions of the importance of a caring approach were discussions about the clinician’s competence, flexibility, and ability to solve problems creatively. These characteristics were uniformly appreciated and valued:
When such competencies were absent, or clinicians appeared to be too passive, patients lost confidence and were unhappy about their relationship with their clinician.
All he would do was sit there and look like he was partially hypnotized himself, and go, uh huh, and that was about it. He rarely had any solutions, or anything, for my problems.
Throughout these accounts and stories was an overarching theme present in the good clinical relationships we found: A strong sense of hope arising from clinicians’ positive attitudes and confidence about recovery and about treatment and available alternatives. This included powerful messages indicating clinicians were confident that a treatment would be found that would work, even when patients were having difficulties with current treatments.
He actually makes me feel like part of the decision, as far as what treatment to proceed with. If I feel that I want to try something else to improve conditions…he may give me some suggestions and he has always said, “Do you want to try to do this?” He says, don’t worry. If it doesn’t work we can do this and this and the other, but do you want to try it first this way?
Qualitative analyses suggested that the factors most important to producing a recovery-promoting relationship between clinicians and their patients are the following: A caring, collaborative, and competent clinician, who listens well, believes and understands what the client says, is reassuring about future recovery and effects of treatment, and is willing to engage in a relationship that has at least some of the qualities of a friendship, including mutual trust.
If fit and comfort with a clinician are acceptable to a patient, and continuity of care is possible, these relationship factors appear to work together to facilitate recovery.
Based on results of our qualitative analyses, we hypothesized a model explaining how continuity of care might interact with relationships with mental health care clinicians to affect recovery and quality of life. Questionnaire-derived measures were available for all hypothesized concepts except fit with clinician; to obtain an indicator of fit we returned to our qualitative data and coded participants’ relationships with current clinicians as having good, moderate, or poor fit (based on participants’ perspectives).
Figure 1 includes results of a covariance structure model testing our hypothesized model among participants who were taking medications at the time they completed the questionnaire (n = 158). Those not taking medications (n = 19) were not included in these analyses because medication satisfaction was hypothesized as central to the model for the majority of participants.
Table 1 describes measures of latent variables and associated factor loadings, reflected by standardized regression weights. Recovery-oriented, patient-directed care, satisfaction with clinician, satisfaction with medication, and quality of life were each measured by three items; recovery was measured by two items. Fit with clinician and relational continuity of care were measured by single items (latent variables with multiple measures are indicated by ovals in Figure 1; measured variables are indicated by rectangles). The weights of the items and their corresponding latent constructs were all strong and in the expected directions. Dotted lines in Figure 1 indicate paths that were hypothesized but not statistically significant. These paths were removed; coefficients presented are standardized and statistically significant at p ≤ .05.
Table 2 shows model fit indices–all indicating that the model has good fit. The Normed Fit Index (NFI) and Comparative Fit Index (CFI) provide percent improvement measures in comparison to the worst possible model fit; both indicate that our model is very close to the best-possible fitting model. Parsimony-adjusted CFI adjusts for degrees of freedom and shows that good model fit was not achieved by sacrificing parsimony. The Root Mean Square Error of Approximation (RMSEA) also adjusts for degrees of freedom; values less than .08 are considered acceptable fit, values under .05 indicate good fit (McDonald & Ho, 2002). The chi-square/degrees of freedom ratio should be less than three, thus also indicating good model fit. Finally, the percent of variance explained by the model is high, with 66% of the variance explained in the quality of life factor and 56% of the variance explained in the recovery factor.
The findings of our statistical model are consistent with our qualitative findings–that recovery-oriented, patient-directed mental health care, in the context of a good relationship with a clinician, plays an important role in recovery and quality of life among individuals with serious mental illness. Greater relational continuity of care was associated with greater satisfaction with clinicians, and indirectly through satisfaction, with better quality of life (total effect of continuity of care on quality of life was 0.03). A recovery-oriented and patient-directed approach to care was associated with greater satisfaction with mental health medications, leading to fewer mental health symptoms and to better quality of life and recovery. Recovery-oriented care was also directly associated with improved recovery outcomes, and indirectly through clinician satisfaction, with quality of life. The total effect of patient-directed recovery-oriented approach to care on recovery was 0.57, and on quality of life 0.37.
To our knowledge, this is the first study reporting consumers’ views and experiences with long-term relational continuity of care in the context of mental health services, or of the effects of that type of care continuity on quality of life and recovery from serious mental illness. Our triangulated qualitative and quantitative findings indicate that individuals with serious mental illness who receive recovery-oriented, patient-directed care in the context of long-term, close, collaborative relationships with their clinicians are more satisfied with their care and have fewer psychiatric symptoms, better recovery outcomes, and enhanced quality of life. Clinicians who adopt rehabilitation principles and a healing model (Anthony, 2004; Cruz & Pincus, 2002)–who are kind, caring, understanding, supportive, and interested in their patients as “persons”– facilitate development of strong and trusting relationships that improve care, provide the context for relational continuity over time, and increase consultation and requests for help during periods of increased symptoms.
In addition to these benefits, our work illuminates some of the processes by which the collaborative approaches found in other settings (Bauer et al., 2006; Dietrich et al., 2004; Druss Rohrbaugh, Levinson, & Rosenheck, 2001; Hedrick et al., 2003; Katon et al., 1999; Saultz & Albedaiwi, 2004; Saultz & Lochner, 2005; Unutzer et al., 2002; Wagner et al., 2001) affect outcomes. For example, mental health clinicians who collaborated with their patients to develop and evaluate plans of care that were acceptable to both clinician and patient, particularly regarding medications, were seen as more helpful and more competent, while empowering their patients to participate in and take responsibility for their own care.
Relationships that were deemed most helpful and collaborative took significant time to establish, however, with mutual experience and trust developing over a number of years. Such findings suggest that high clinician turnover in the current publicly funded mental health system, and clinician training practices that are based on rotating residencies, may negatively affect patients.
Our results also address consumers’ preferences regarding “boundaries” in clinician-patient relationships. Despite professionals’ concerns about maintaining appropriate limits in clinical interactions (Nadelson & Notman, 2002; Smith & Fitzpatrick, 1995), clinician-patient relationships that provided at least some of the support and mutuality of friendships were more highly valued by patients, experienced as more helpful, and seen as supportive of better mental health. These consumer preferences are consistent with recent recognition that appropriate boundaries may differ as a result of patients’ personal and cultural characteristics, past experiences, and specific illness or treatment effects (Nadelson & Notman, 2002). Our findings are also consistent with contemporary acknowledgments that provider boundaries may need to be adjusted to accommodate different needs and situations (Smith & Fitzpatrick, 1995), that some boundary “crossings” can be helpful to patients (e.g., touching a patient’s hand when s/he is upset about a loss) (Nadelson & Notman, 2002), and that self-disclosure can play an important role in therapeutic relationships when it is managed appropriately and is intentional (Psychopathology Committee of the Group for the Advancement of Psychiatry, 2001). Our findings also suggest that clinicians’ decisions about boundaries may be informed in important ways by the trust and shared understanding developed over time in a close clinician-patient relationship. That is, a shared history can help a clinician determine the best approach in a particular situation, and also help the patient understand what is meant (or not meant) when a traditional boundary is “crossed.” Moreover, our results indicate that crossing some boundaries may provide needed interactions and interpersonal feedback that can affirm self-worth among individuals whose mental illness has affected their interpersonal functioning.
Additionally, our work suggests that psychiatry’s movement toward a purely disease management, rather than a rehabilitation or healing model (Anthony, 2004; Cruz & Pincus, 2002), may be counterproductive. Most of the effective clinician-patient relationships we found could best be characterized as coming from a healing model, with education and disease management information delivered within the context of the healing relationship. Moreover, we found that relationships that were purely focused on disease management, even when the clinician was perceived as being very competent, were not as valued by consumers or perceived as helpful as those characterized by a more caring approach. Interestingly, our work was carried out in a managed care setting (a context which many might consider less conductive to producing relational continuity or supporting a healing model), suggesting that managed care per se does not preclude healing approaches. In fact, the setting studied here likely facilitated such long-term relationships because clinicians are salaried and staff turnover is low.
Our covariance structure model was tested in the sample from which the theoretical model was derived. Such triangulation lends strong support for our qualitative findings, but this model should be validated in other samples in the future. In addition, in order to include satisfaction with medications in the statistical model, we excluded 15 individuals who were not taking medications. Although these participants were included in the qualitative analyses, we do not know if the covariance structure model findings accurately represent those individuals. In addition, because of limited variability in our measures of fit with clinician and relational continuity of care, it is likely that our estimates of the strength/importance of these relationships are overly conservative.
The generalizability of our work may be limited because our sample differs from those in most public sector settings–study participants were more likely to be married, have higher education and income levels, and be more likely to be employed. One explanation for this difference is that individuals with such characteristics are more likely to have access to a private health plan. Yet, it is also possible that the good clinical relationships and long-term continuity of care we found have helped improve such outcomes. The fact that many participants had received care in the public system prior to obtaining care in Kaiser Permanente suggests that this may be true for as least some individuals. Another limitation to generalizability results from the low staff turnover in the health plan. We believe that this difference represents a benefit of this study site, however, as it provides important information about how care could be provided in publicly funded settings if staff turnover could be reduced.
Other limitations include our decision to have mental health clinicians screen participants for ability to participate, because it is possible that clinicians screened out individuals with whom they had poor relationships. Although possible, clinicians were instructed to screen out those patients they felt were unable to participate (but not to screen them out for other reasons), and the information clinicians received indicated, accurately, that the study was addressing recovery from serious mental illness, not that it was a study of clinician-patient relationships (this latter work emerged from our qualitative analyses).
We found that focusing on clinician-patient relationships and a caring, collaborative, healing-based approach can facilitate recovery from mental illness. Moreover, our findings show that such approaches are viable in environments where cost containment and a focus on disease management remain necessary. Much of the discussion since the President’s New Freedom Commission (President’s New Freedom Commission on Mental Health, 2003) has focused on major systems transformations. Our results suggest that more modest improvements at the micro level, improving relational continuity of care and increasing the number of clinicians who adopt collaborative care approaches, could enhance recovery efforts. The challenge then becomes to increase continuity and collaboration in systems plagued by high staff turnover. Worth noting is that the Chief of Mental Health in the system studied here has targeted such turnover problems by requiring a 3-year commitment from clinicians prior to their hire (Lauretta Young, MD, personal e-mail communication, September 22, 2006).
This research was supported by a grant from the national institute of mental health (recoveries from severe mental illness, Ro1 Mho6232l). the authors would like to thank Michael Leo for analytic support, Elizabeth Shuster and Jeff Showell for help with data extraction, and interviewers Sue Leung, Alison Firemark and Micah Yarborough for their excellent interviews, their help in developing the qualitative coding scheme, and for coding interviews.
Carla A. Green, Center for Health Research, Kaiser Permanente Northwest, Portland, OR; Oregon Health & Science University.
Michael R. Polen, Center for Health Research, Kaiser Permanente Northwest, Portland, OR.
Shannon L. Janoff, Center for Health Research, Kaiser Permanente Northwest, Portland, OR.
David K. Castleton, Center for Health Research, Kaiser Permanente Northwest, Portland, OR.
Jennifer P. Wisdom, New York State Psychiatric Institute.
Nancy Vuckovic, Center for Health Research, Kaiser Permanente Northwest, Portland, OR.
Nancy A. Perrin, Oregon Health & Science University; Center for Health Research, Kaiser Permanente Northwest, Portland, OR.
Robert I. Paulson, Florida Mental Health Institute, University of South Florida.
Stuart L. Oken, Kaiser Permanente Northwest, Portland, OR.