|Home | About | Journals | Submit | Contact Us | Français|
Patients consider having a human connection with a physician to be an important aspect of end-of-life (EOL) care. We sought to develop and validate a measure of therapeutic alliance between advanced cancer patients and their physicians, and to evaluate the effects of therapeutic alliance on EOL experiences and care.
We developed The Human Connection (THC) scale to measure the extent to which patients felt a sense of mutual understanding, caring, and trust with their physicians. The scale was administered to 217 advanced cancer patients along with measures of attributes hypothesized to be related to therapeutic alliance, including emotional acceptance of terminal illness. EOL outcomes in 90 patients who died during the study were also examined.
The 16-item THC questionnaire was internally consistent (Cronbach’s α =.90) and valid, based on its expected positive association with emotional acceptance of the terminal illness (r=.31, P<.0001). THC scores were inversely related to symptom burden (r=−.19, P=.006), functional status (Karnofsky score, r=.22, P=.001), and mental illness (THC score 50.69 for patients with any DSM diagnosis versus 55.22 for those without, P=.03). THC scores were not significantly associated with EOL discussions (P=.68). Among patients who had died, EOL ICU care was inversely associated with therapeutic alliance (THC score 46.5 for those with ICU care versus 55.5 for those without, P=.002), such that patients with higher THC scores were less likely to spend time in the ICU during the last week of life.
The THC scale is a valid and reliable measure of therapeutic alliance between advanced cancer patients and their physicians. In addition, we found no evidence to suggest that EOL discussions harm patients’ therapeutic alliance. A strong therapeutic alliance is associated with emotional acceptance of a terminal illness and with decreased ICU care at the end of life among patients with advanced cancer.
Patient-physician communication is widely considered to be a core attribute of high quality care.1 In the end-of-life (EOL) setting in particular, substantial research and teaching efforts have focused on this specific aspect of the patient-physician relationship.2-11 But the patient-physician relationship encompasses more than communication alone. Patients who are nearing the EOL value being seen as a whole person by the physician,12,13 receiving care from one’s personal physician,13-15 having a physician with whom one can discuss one’s fears,13 having a relationship based on mutual respect, trust, and humility,14 and receiving care that is attentive to one’s emotions.13,15
Importantly, the patient-physician relationship has been shown to have positive effects on health outcomes for patients with chronic illnesses such as diabetes and HIV,16-19 and standards for relationship-building have been developed based on patient values.20-22 Similarly, the psychotherapy literature23,24 has identified the empathic bond between patient and therapist as the essential psychotherapeutic agent in the prediction of treatment response.25-31 We know little, however, about the human connection between advanced cancer patients and their physicians, and how this relationship affects the medical care patients receive at the EOL.
Physicians sometimes worry that frank discussions about prognosis can harm the physician-patient relationship.8,32-34 This concern may lead physicians to avoid conversations about prognosis and EOL care. Conversely, it also is possible that a strong patient-physician relationship can help patients and physicians to develop shared goals for the EOL period. We sought to take core components of the patient-physician relationship that have been validated in the psychotherapy literature and apply them to the oncologist and advanced cancer patient relationship. We developed The Human Connection (THC) scale to measure therapeutic alliance (TA), defined as the collaborative bond between the physician and patient. Based on theoretical frameworks for TA in the psychiatric literature,35 patient-reported priorities for EOL care from the medical literature,12-15 and the authors’ (RF, HGP) clinical and research experiences, the resulting scale was designed to evaluate five key aspects of the relationship: the extent to which the patient feels (1) that the oncologist listens to and understands the patient’s concerns about the illness;13,14 (2) that the relationship involves mutual caring and respect;12,13,15 (3) that the patient understands the information being shared by the oncologist;14 (4) that the patient trusts the oncologist;14 and (5) that the oncologist and patient work well together.13-15
We first evaluated the psychometric properties of the THC scale as a measure of TA, including relationships with socio-demographic characteristics, psycho-behavioral attributes, patients’ emotional acceptance36 of a terminal illness, and physical health among advanced cancer patients. Second, we examined relationships between TA and EOL care. Because of physician concerns that EOL discussions can negatively influence the patient-physician relationship, we also sought to examine TA among patients who had engaged in EOL discussions with their physicians.
Subjects were recruited as part of the Coping with Cancer Study, a multi-institutional investigation of advanced cancer patients and their primary (unpaid) caregivers. 36 The Coping with Cancer Study opened in August 2002, but The Human Connection (THC) scale was developed during the course of the study and therefore not administered to the entire cohort. Of 338 patients who participated in the Coping with Cancer Study, the THC scale was administered to all participants who enrolled between May 4, 2006, and July 20, 2007 (N=217). We present here results for patients who completed the THC scale.
Participating sites included Yale Cancer Center (New Haven, CT), Veterans’ Affairs Connecticut Healthcare System Comprehensive Cancer Clinics (West Haven, CT), Memorial Sloan-Kettering Cancer Center (New York, NY), Simmons Comprehensive Cancer Care Center (Dallas, TX), and Parkland Hospital Palliative Care Service (Dallas, TX).
Patients were eligible to participate if they had gastrointestinal or thoracic cancer with distant metastases and with failure of first- or second-line chemotherapy, if they were 20 years of age or older, if they were able to identify an unpaid caregiver, and if they had adequate stamina to complete the interview. Patient-caregiver dyads were excluded if either the patient or caregiver met criteria for dementia or delirium by neurobehavioral cognitive status examination,37 or if either the patient or caregiver was unable to speak English or Spanish. All study participants provided written informed consent.
Patients were asked to participate in a baseline interview, for which a $25 incentive was provided. Research staff from Yale University and Dana-Farber Cancer Institute trained the interviewers. Interviews were conducted in English or Spanish and took approximately 45 minutes to complete. Chart review was used at enrollment and after death to obtain additional clinical information, such as comorbidity at enrollment, location of death, and use of invasive measures at the end of life. In addition, a post-mortem questionnaire was administered within 2 weeks of death to a formal or informal caregiver (i.e., paid healthcare provider or family member, respectively) who had cared for the patient during his or her last week of life.
The institutional review boards of participating institutions approved study procedures.
The Human Connection scale (THC), administered in the patient’s baseline interview, was based on the authors’ clinical and research experience with terminally ill patients and relevant literature review. Core elements were developed from existing literature with extension to the care of advanced cancer patients.
The resulting THC scale is a 16-item questionnaire developed to assess therapeutic alliance between the patient and physician. Initially there were 30 potential items, three of which were open-ended and used primarily to validate core concepts. We eliminated selected questions, including items with the lowest degree of association with the scale as a whole, to focus on core concepts and limit redundancy. For example, one of the original questions asked patients how often the physician interrupted them. Responses to this question had little variability and poor correlation with the scale as a whole (item-total correlation of .07). In addition, other items addressed more directly the extent to which the patient felt that the physician listened to him or her. We therefore eliminated the item about interrupting and retained the other items about listening. All items had item-total correlations greater than 0.40. Item-total correlations were highest for questions about the extent to which the patient liked (r=.73), trusted (r=.77), and respected (r=.68) the physician, and the extent that the patient considered the physician to be thorough (r=.71), caring (r=.70), and open-minded (r=.68).
The 16 items are available in Table 1. Responses were measured on a 4-point Likert scale, with response categories relevant to question content. The 16-item THC scale had a high degree of internal consistency (Cronbach’s α =.90), and an eigenvalue of 7.37 when analyses specified a single factor. The scree plot further confirmed the presence of a single factor.
A THC score was created as a summary score of item responses such that a higher THC score indicated greater TA. Possible scores ranged from 16 to 64; in our sample, the mean value was 54.6 (SD 7.6, range 31-64).
Concurrent validity was assessed by examining hypothesized relationships between THC scores and disease characteristics, including symptom burden and functional status, and psycho-behavioral factors, such as mental health, spirituality and coping style. In parallel to the known role of TA in determining response to psychotherapy, we hypothesized that TA had the potential to affect the extent to which patients accepted their terminal illness on an emotional level, and that TA might thereby also be associated with choices for EOL care. Measures used in analyses are described below.
Subjects were asked to provide information on their gender, age, race/ethnicity, marital status, income, health insurance status, religion, and highest grade completed in school.
The Structured Clinical Interview for the Diagnostic and Statistical Manual (SCID) 38,39 was used to diagnose current major depressive disorder, generalized anxiety disorder, panic disorder, and post-traumatic stress disorder.
Spirituality was assessed by asking subjects “to what extent do you consider yourself a spiritual person?” (“very,” “moderately,” “slightly,” “not spiritual at all.”)
The 15-item Brief COPE40 was administered to define coping strategies, such as active, emotion-based, and avoidant coping. Relevant items were combined to create separate summative scores for use in analyses.
The 16-item McGill QOL scale measures quality of life in patients with advanced illness in four domains—physical symptoms, psychological symptoms, existential well-being, and support. Items were combined into summary scores for the relevant domains. Some individual items, such as reports of feeling depressed or feeling terrified about the future, were also used in analysis.
Symptom burden was measured using the physical symptom subscale of the McGill Quality of Life Questionnaire (MQOL).41,42 The Karnofsky 43 and Zubrod Scores,44 determined by the research interviewer at the time of enrollment, were used to describe the patients’ current state of health. The patient’s medical record, in consultation with the referring clinician, was used to determine the Charlson Index of Comorbidity score.45
Patients were asked to “describe your current health status,” with response options of “relatively healthy,” “seriously but not terminally ill,” and “seriously and terminally ill.” Patients who described themselves as “seriously and terminally ill” were considered to have cognitive acceptance of the terminal illness.
Patients were asked whether they and their physician had discussed any particular wishes about the care they would want to receive if they were dying.
The PEACE questionnaire36 was used to measure Peaceful Acceptance, which we also term emotional acceptance of a terminal illness.
Patients were asked if they had completed a do-not-resuscitate order, a living will, or a durable power of attorney for health care.
For patients who had died (N=90), a postmortem assessment, composed of chart review and caregiver interviews, was performed. The location of death (intensive care unit, hospital (non-intensive care), nursing home, inpatient hospice, or home) and any interventions used in the last week of life (care in the intensive care unit and use of a ventilator, a feeding tube, or chemotherapy) were recorded. Caregivers were asked to report, “In your opinion, how would you rate the overall quality of the patient’s death/last week of life?” with response options ranging from 0 (worst possible) to 10 (best possible.)
For binary independent variables (e.g., EOL discussion), relationships were assessed with non-paired t-tests comparing mean THC scores. When independent variables were continuous (such as age and functional status), relationships were assessed using linear regression or the Pearson correlation coefficient. When independent variables were ordinal (income and comorbidity), the Spearman correlation coefficient was used to describe relationships. We also tested for an interaction effect between EOL discussion and THC in the prediction of EOL outcomes.
Patient characteristics are available in Table 2. Deceased patients (N=90) who had post-mortem assessments available survived a median of 84 days (SD 94 days) from enrollment. Median follow-up on the remaining patients, who were either alive, or dead but without post-mortem assessments available, was 248 days from enrollment (SD 153 days). THC scores had no relationship with gender, race, marital status, income or education. Older patients tended to have lower THC scores (decrease in score by 0.23 per increasing year of age, P=.001) than younger patients.
THC scores were inversely associated with a SCID diagnosis of post-traumatic stress disorder (P=.0001, Table 3). Although THC scores were not significantly associated with a current diagnosis of major depression (P=.53) or generalized anxiety disorder (P=.12), scores were inversely associated with the presence of any current major psychiatric disorder (P=.03).
THC scores were also associated with some self-reported psychological states on the McGill QOL scale. For example, THC scores were inversely associated with feeling depressed (r=−.23, P=.0006) and terrified about the future (r=−.17, P=.01), and with the summary score for psychological symptoms (r=−.18, P=.007). In contrast, THC scores were positively associated with patients’ reports of support (r=.28, P<.0001) and existential well-being (r=.16, P=.02).
Spirituality was not significantly associated with THC score (P=.06). Some coping strategies were associated with THC scores, however. Patients who used emotion-based coping techniques tended to have higher THC scores, for example (r=.28, P<.0001), while avoidant coping was inversely associated with THC score (r=−.15, P=.02). THC scores were not associated with active coping techniques (r=.07, P=.28).
THC scores were not associated with comorbid conditions (r=−.09, P=.21) but were associated with functional status (Karnofsky score, r=.22, P=.001; Zubrod score, r=−.20, P=.003), with greater scores among those patients with better functional status. Similarly, patients with a higher burden of symptoms (McGill symptom burden, r=−.19, P=.006) tended to have lower THC scores. THC scores were not significantly related to proximity to death, however (r=.17, P=.10), among patients who died.
THC scores for patients who reported having discussions with their physicians about EOL care preferences (mean 54.97, SD 7.13) were no different from the scores of patients who did not report such discussions (mean 54.54, SD 7.66, P=.68, Table 4). Scores were also similar whether the patient did (55.11, SD 7.68) or did not (54.32, SD 7.19, P=.49) acknowledge having a terminal illness. Although THC was not associated with cognitive acceptance, it was associated with emotional acceptance of a terminal illness (PEACE score for Peaceful Acceptance, r=.31, P<.0001).
THC scores were not related to the presence of a do-not-resuscitate order at the time of questionnaire completion (P=.37), but patients with other forms of advance care planning such as living wills, health care proxies, or durable powers of attorney tended to have lower THC scores (53.97, SD 8.07) than patients who had not completed these types of advance care planning (57.05, SD 4.85, P=.002).
Among the 90 patients who had died, patients who received care in the intensive care unit in the last week of life had lower THC scores (mean 46.5, SD 8.26) than patients who did not receive ICU-based care in the last week (55.5, SD 6.33, P=.002, Table 5). Death in the hospital was also marginally, though not statistically significantly, associated with THC score (P=.06), with lower THC scores among patients who died in the hospital (50.36, SD 9.25) than among patients who died in other locations (55.68, SD 5.96). Caregiver-rated quality of patient death, however, was not associated with THC score (r=.17, P=.11).
A supportive and caring patient-physician relationship is a valued attribute of high quality EOL care;12-15 thus, therapeutic alliance, as measured by the THC scale, may be a profoundly important aspect of the patient’s experience of serious illness. The scale has demonstrated internal consistency and validity, including associations with attributes such as age and health state that are known to be associated with global perceptions of care.46 Construct validity was further affirmed by the scale’s associations with mental health, self-reported psychological state, and coping style; as expected, patients with psychological distress or certain mental health problems had lower TA. The THC scale’s core attributes revolve around mutual caring, respect, and trust.
We used our validated measure of TA to evaluate further the role of the patient-physician relationship in EOL care. Perhaps most importantly, we found no evidence that EOL discussions are associated with diminished therapeutic alliance. The coexistence of TA and EOL discussions also suggests that physicians who have close bonds with their patients are no less likely to engage in EOL conversations.
Notably, our data cannot exclude the possibility that physicians selectively engage in EOL discussions with patients who are most receptive to these conversations. In addition, although we sought specifically to determine whether EOL discussions harm TA, we should also consider the possibility that a successful TA should help patients to establish goals for care, including for EOL care. Our scale did not measure this phenomenon. However, the psychotherapeutic literature does suggest that goal-setting is an important component of TA, and this issue may apply directly to EOL discussions and planning. We plan to include attributes of goal-setting in future versions of the scale, which will allow us to re-evaluate associations with EOL discussions.
Although we found no associations between TA and EOL discussions, we did find that TA is associated with some EOL care outcomes, including greater existential well-being and emotional acceptance of a terminal illness, the latter of which is associated with decreased use of feeding tubes at the EOL.36 Like the THC scale, our measure of emotional acceptance is newly developed and validated.36 Further exploration of both issues is needed in future studies. However, this finding raises the question as to whether a human connection between patients and physicians can promote a sense of peaceful acceptance of one’s impending death.
In addition, our data suggest that patients with lower TA may be more likely to receive ICU-based care at the end of life. Because our sample was small, only a few patients received intensive, life-prolonging care, and other indicators of such care were not statistically associated with THC. This may speak to limited statistical power to detect such effects, and, again, may relate to the absence of goal-setting as a domain of the THC scale. Confirmation of these findings in future studies will be important, and relationships should be reconsidered with future versions of the scale.
In addition, we do not know whether relationships between TA and EOL care are causative or simply associated. We have speculated that the therapeutic relationship between patients and oncologists has the potential to shape the EOL care that patients receive. However, we do not know whether TA directly affects EOL decision-making, and establishing causation will be challenging even in carefully designed future studies. The results do allow us to recognize this as a possibility, while also recognizing TA as a care attribute that patients value.
Patients with poorer performance status, older age, or any major psychiatric disorder tended to have lower THC scores than patients without these attributes. This finding identifies several populations “at risk” of poorer TA. We might hypothesize that patients who are dealing with significant symptoms or with serious mental health issues may be less able to form TA with physicians because of communication barriers, diversion of energies to pressing medical issues, and psychiatric vulnerabilities that may impede the development of trust and connection. These special populations may require enhanced efforts by the oncologist to achieve connection, perhaps through co-management of the patient with other clinicians to address palliative and psychiatric issues and enhanced attention to treatment of psychiatric disorders.
However, it is also possible that physicians withdraw from their relationships with certain types of patients; according to previous literature, some patients feel abandoned by their physicians at the end of life.47,48 Some physicians, for a variety of reasons (eg, helplessness, guilt, frustration), may be reluctant to engage emotionally with patients who are in more distress. Importantly, we did not find an association between proximity to death and THC among patients who had died, and whether physician withdrawal in serious illness is indeed the cause of the lower THC scores among patients with more severe impairment cannot be adequately assessed by our study. Nonetheless, patients report systematically different therapeutic alliance in these situations, and this deserves attention and further study. Future research should also examine physicians’ experiences of THC as a possible mediator of patient TA responses and outcomes.
Our study had several limitations. For example, THC was assessed at one point in time, often several months before death. Little is known about how the patient-physician connection may change over time. In addition, although we have evaluated markers of life-prolonging care, some patients may wish to receive intensive measures at the EOL. In the best case, TA could allow patients and physicians to formulate and carry out shared goals for care. Future study should therefore consider attainment of shared goals at the EOL as an important potential outcome of therapeutic alliance.
We have developed and validated The Human Connection scale in advanced cancer patients. Our findings raise the possibility that the patient-physician relationship can positively affect the EOL experience for these patients, perhaps through its association with peaceful acceptance of a terminal illness. In addition, we found no evidence that EOL discussions impair TA. Understanding TA as a possible mediator of the EOL experience will require ongoing study. Nonetheless, the THC scale offers a way to quantify much of what is valuable about the patient-physician relationship, so that this phenomenon can be considered in future studies of care experiences among cancer patients.
This research was supported in part by the following grants to Dr. Prigerson: MH63892 from the National Institute of Mental Health and CA106370 from the National Cancer Institute; a Fetzer Religion at the End-of-Life Grant; the Center for Psycho-Oncology and Palliative Care Research, Dana-Farber Cancer Institute. Dr. Mack was supported by an American Cancer Society Mentored Research Scholar Grant.