Physician and Nursing Perspectives on Patient Encounters in End-of-Life Care

There is a strong impetus to better understand the contemporary experience of dying and to improve end-of-life care in the United States.¹ Effective communication between patients and their care providers has been promoted as a key element within this experience and as a critical factor for maintaining quality of life during end-of-life care.² Multiple factors can impede or facilitate communication in this arena: attitudes toward death and dying,³ provider and patient anxiety over diagnostic and prognostic discussions,⁴ and cultural, socioeconomic, and educational influences.⁵ These factors highlight the nature and context of the provider-patient relationship, which plays a central role in end-of-life medical care for primary care physicians.⁶ Although much emphasis has been given to improving providers’communication skills⁷ and to facilitating decision making,^8-10 there is a paucity of research on the actual dynamics of the provider-patient encounter in end-of-life care.

One theoretical model of physician attitudes and roles in their encounters with dying patients incorporated 3 dimensions: direct involvement with the patient, the physician’s own needs and development, and cooperation with other caregivers.¹¹ Subsequent qualitative research with family physicians supported these dimensions and emphasized the primacy of the patient relationship and the area of physician personal domain.^12,13 However, the applicability of these findings and the model they provide to other specialties and disciplines that are involved in end-of-life care remains unclear. Studies of critical care nurses also identified patient concerns and their own personal concerns as key areas and raised the issue of frustration in their limited role in the management of patients at the end of life.¹⁴ In a study of palliative care workers, most perceived themselves as open and sensitive to their patients, although many felt poorly supported by other staff members.¹⁵

A greater and more complete understanding of the provider’s perspective of the provider-patient encounters in end-of-life care could enhance communication and decision making. This area of inquiry is especially salient in light of the recent Commonwealth-Cummings project on the quality of care at the end of life, which has identified medical provider interventions as one of 4 critical components in a multifaceted framework for a good death.¹ In addition, facilitating the communication and decision-making processes that accompany care at the end of life has been suggested as a key role for family physicians and other generalist providers.¹⁶ We conducted a study using qualitative research methods to explore and describe the nature of provider-patient encounters at the end of life. Our specific aim with this study was to describe factors that both medical and nursing providers identify as influential in their positive and negative encounters with dying patients.

Methods

Design

Because of the exploratory nature of our study, we selected a qualitative research method (semi structured interviews) to gain a richer and more complete description of the multifaceted phenomena that occur between medical and nursing providers and their dying patients. Semi structured interviews are dialogues that are guided and open ended enough to produce transcripts of the discussion as primary data.¹⁷ Editing analysis, in which meaningful units or segments of text are identified into categories or codes that can be used to construct and interpret common themes and patterns, was chosen as a way to generate new understanding in the area of provider perceptions in end-of-life care encounters.¹⁷

Sampling

We used maximum variation sampling to obtain the broadest range of provider perspectives on end-of-life encounters.¹⁸ This approach allowed us to obtain a wide range of information and perspectives on provider perceptions of their encounters with dying patients. We therefore selected a diverse sample of physicians and nurses based on the following criteria:(1) providers in specialties that had the highest probability of encountering dying patients in their practice and (2) providers who had experience in multiple clinical settings, such as academic health centers, private practice, and community settings.

Participants

The Human Subjects Committee of the University of Kansas Medical Center approved our study. We interviewed 6 physicians and 6 nurses from a large Midwestern metropolitan area. The physicians represented the following specialties: geriatric medicine, cardiology, pulmonary medicine/critical care, internal medicine/hospice, oncology, and family practice. The nurse participants were selected from the following specialties: neurology, nephrology/dialysis unit, oncology, hospice, inpatient palliative care, and cardiology/critical care.

The average age of the participants in our sample was 44 years (range=35-54 years);2 were men, and 10 were women. The physician participants had been in practice an average of 10 years (range=5-18 years),while nurse participants had an average of 13 years’(range=5-20 years) experience.

Data Collection

We identified 12 potential participants who met the study criteria and were either known to us professionally or were suggested to us by other colleagues. These participants were contacted and invited to participate. There were no refusals. Semi structured interviews were conducted between September 1998 and April 1999 in either the participant ’s office or the investigator’s office and lasted from 1 hour to 1.5 hours. We conducted individual interviews with participants from our respective disciplines (S. A.F, nurses; T. P.D.,physicians).

Participants were asked to reflect on 2 contrasting end-of-life patient encounters that they had personally experienced in their practices. The encounters were to involve multiple contacts with a single patient who had since died. The first reflection focused on encounters that resulted in a very positive and rewarding outcome from the participant’s perspective. The second reflection involved encounters that were viewed by the participant as personally difficult and troubling. The interview began with the positive reflection, and the participants were allowed to speak freely with few interruptions. An interview guide (Appendix) provided the template for follow-up questions oriented around treatment decisions, the communication process, and participant attitudes, values, and beliefs regarding end-of-life care. All interview sessions were audiotaped and professionally transcribed.

Data Analysis

The interview transcripts were checked for accuracy and verified against the original audiotapes by the investigator who conducted the interview, and the transcripts were then formatted and entered into a qualitative software package (QSR NU*DIST).¹⁹ All data (text, codes, categories, and notes) were entered, retrieved, and analyzed using the computerized software. Independent meaningful units or segments of text that were relevant to the study purpose were identified, labeled, and organized simultaneously by the investigator/analyst in a process called coding.²⁰ After the initial coding was completed, we reviewed the data to evaluate emerging patterns and themes. This iterative process was used to search for systematic relationships and for contrasts and irregularities among the codes and categories, and it continued until consensus or agreement by the analysts was reached. Credibility was maintained by peer debriefing, a search for rival explanations in the data, and by member checking.^18,21 The interdisciplinary nature of the investigators and the iterative coding process enhanced peer debriefing and rival explanation searching, which occurred repeatedly. Member checking was conducted with 4 participants who agreed with the study findings.

Results

Participant interviews were grouped into 6 general categories:(1) provider characteristics,(2) patient characteristics,(3) interpersonal relationships,(4) prognostic certainty/clarity,(5) decision making, and (6) effective communication. Each of these categories is described below with verbatim quotes from the participants.

Provider Characteristics

Participants identified specific provider attitudes, knowledge, and skills as important components of positive encounters. Provider attitudes such as openness to discussions about death and dying, a willingness to be vulnerable, and death acceptance were associated with positive encounters. A heuristic perspective—one that incorporated a receptiveness to learning from patients, past experiences, and even painful mistakes—was inherent in these encounters.

A cardiologist shared his willingness to learn from an elderly patient who challenged his traditional treatment paradigm by refusing a recommendation for bypass surgery:

Maybe my bias was to do what I had been trained to do, which was to send her for bypass, put her through some misery, and maybe not have changed the outcome that much, you know maybe added a year or so, put her through quite a lot of trauma. But she removed my personal bias.

Vulnerability was typified by one participant who related:

This one was hard for me personally…. It was painful for me, because I really liked him, but rewarding on the other hand because I was able to really support him in his autonomous decision to do what he wanted to do.

Death attitudes, whether avoidant or accepting, were also key elements in this category. A nurse participant suggested that death avoidant attitudes were major factors in negative encounters:

But this particular physician can sidestep some of these tough conversations. And that doesn’t make him a bad physician, that’s just his style. He doesn’t do well with those conversations. He’s actually quite a good oncologist, just bad with end-of-life conversations.

A sufficient knowledge base and skill level in caring for dying patients also were important characteristics. Most participants stated their lack of formal education in palliative care as an impediment to providing good care in these settings. As one physician candidly shared, “We learn end-of-life care by the seat of our pants, by the mistakes that we make.”

Patient Characteristics

Patient characteristics such as attitudes, values, and knowledge were important components of participants’perceptions of positive and negative encounters. Within the context of positive encounters, patients were described as proactive, information-seeking, educated, having a clear vision, and focused on reality. Patients in these settings were perceived as facing reality from the onset of diagnosis and being capable of finding something positive in the face of death.

He [the patient] realized that not everything in life is fixable. He said, “This is no tragedy [death]; this is what happens in life.”But as treatments didn’t do any good anymore, he made a decision. He said, “I’ve got to get out and enjoy life.”

In contrast, patients in the negative encounters were perceived as fearful, demanding, untrusting, continually searching for ways to fight the disease, unwilling to give up, and living in chaotic family systems. One patient who had battled cancer for 6 years and spent the last weeks of his life in the intensive care unit was described by a nurse participant:

He [patient] did not want to give up, wanted to be positive, did not want to hear a negative thought, a negative piece of information, and had a physician who kind of went along with that …[which] created a big problem because here you have a man who’s dying basically, and they [family] don’t want to hear it. So that’s probably the problem in a nutshell.

Interpersonal Relationships

Interpersonal relationships were networks of family members, friends, and care providers that either facilitated or impeded the encounters. Participants identified multiple elements in positive encounters: the provider’s sense of connection and a congruent belief system with the patient, patients with supportive relationships with family members and friends, and patients with an existential or spiritual support system. The provider’s sense of connection was captured by an oncologist who related: “We connected on a personality level. I was a shepherd of his, so to speak…personally. I connected with him probably more than I do with most people.”

These positive relationships were also strengthened by common belief systems that were inclusive of patients’ existential or spiritual beliefs. One participant stated, “We could talk about spiritual issues, and maybe that was the thing that bonded us.”Participants also identified coherent and consistent patient belief systems in their positive encounters, “He talked openly about dying and that he wasn’t afraid. He felt that spiritually he was ready.” Also, in their depictions of positive encounters participants cited families and friends who provided emotional support yet allowed for the patient’s wishes to be honored.

Negative encounters were highlighted by strained uncomfortable provider-patient relationships characterized by a lack of trust. Participants identified patients and family members in conflict, hindering supportive relationships. These strained relationships were marked by a lack of acceptance of diagnostic and prognostic information by both patients and families. This lack of acceptance would be manifested in anger and hostility toward providers as one participant related:

He would burst into tears and say, ‘you can’t just let her die. You have to do everything.’ So there was never a no code on this woman even though she was clearly end stage.

Participants cited the lack of an existential orientation or spiritual reference in many patients with whom they had a negative encounter. An oncology nurse typified this perspective:

Some people are clearly afraid to die. I have seen the people who don’t seem to have much of a faith base, not religiously speaking but spiritually. I mean, if their heart and soul and mind aren’t all connected, and they don’t have a sense of purpose, of a beginning and an end, those people are very frightened to die. And some have a religious base, and some don’t. But it’s that sense of the beginning and the end. We all are born, and we’re all going to die. And people who don’t have that sense don’t do well when it comes time to die. Emotionally it’s very difficult for them.

Prognostic Certainty/Clarity

Participants identified the precision of the patient’s diagnosis and prognostic accuracy, as well as the concomitant therapeutic plan, as a facilitating factor in positive encounters. A nursing participant who works with amyotrophic lateral sclerosis (ALS) patients was representative of this orientation:

I guess we start out with the diagnosis basically [following a lengthy diagnostic work-up] and talk to them about what the diagnosis is, that it is a terminal disease.

Another participant shared that prognostic criteria provided clarity and assisted in anticipatory care planning. A pulmonologist/critical care physician shared:

[We] watch the pulmonary function decline, and they turn dyspneic enough—it’s clear, you know—we’ve got great criteria for respiratory failure, and those criteria are very black and white.

In positive encounters, both patients and physicians were able to discern with greater clarity the timing in transitioning care from cure to comfort. For some this was based on the clarity of prognosis, while others used prognosis and an analysis of the burdens and benefits of continued treatment within their conceptualization of quality of life. A family physician related the following:

I was on call, and there was a lady who was from out of town visiting a family, who had a pretty massive stroke…the family thought long and hard about “do we put a feeding tube in her”and the swallowing study showed she could swallow. But she wasn’t able to tell us what she wanted. And she wasn’t making a lot of progress…. They talked with her pastor, who had been her pastor for approximately 40 years…. They decided not to feed her or give her any intravenous fluids. They thought she had a good life, she’s had a lot of faith, she’s told them she’s ready to go and that type of stuff.

Prognostic clarity greatly enhanced positive encounters but was difficult to achieve even for experienced physicians, one of whom stated that, “you can’t outline, you know, the myriad different things that can happen.”However, participants perceived that for family members, clarity of ten was difficult to achieve, especially when interacting with multiple physicians. Patients in intensive care unit settings with multiple specialists and revolving call coverage made prognostic clarity difficult to achieve for some family members. Family members were depicted as misinterpreting different versions of the same story or being provided conflicting information:

He [the patient] had multiple physicians who were trading off and on day to day so there wasn’t the continuity…and so the family heard many different versions of what his prognosis was.

Decision Making

Participants described the degree of unanimity or being on the same page with patients and family members with regard to outcome expectations and treatment decisions as an essential element, one that differentiated positive and negative encounters. Being on the same page was facilitated by shared provider/patient characteristics, such as death attitudes and enriched interpersonal relationships, and by prognostic clarity.

Negative encounters were marked by conflict due to a discongruous assessment or care plan among providers or between family members and providers. One nursing participant described conflict with a physician son who limited the amount of pain medication administered to his dying mother. This unilateral decision to reduce pain medication left the participant feeling conflicted regarding the quality of the patient’s dying:

She had the abdominal surgery and her belly full of cancer, so she had lots of reason for pain, although his [son’s] perception was she had no reason for pain…. The whole thing just, just didn’t feel good to me.

Decision making was an ongoing process that included seeking, sharing, hearing, and processing information within one’s value system. Mismatched expectations for treatment outcomes were inherent in negative encounters and were promoted by a patient or family member’s inability to assimilate and process information accurately. One physician participant shared his perception of unrealistic expectations of family members due to technology:

Perhaps that [high-tech interventions in the intensive care unit] gave them a false sense of security that appropriate therapy had been given, and therefore there was a chance of survival. I did give them statistics saying that, you know, the mortality for this condition is 80%.I think patients whenever they hear that always think they’re the 20% that’s going to survive. They grasp onto the 20% rather than hone in on the adverse.

Effective Communication

Effective communication described the sharing of information in a direct and honest manner by an interdisciplinary team, as well as the exploration of goals, values, and feelings at multiple points in time. It was the primary category cited by all participants. Effective communication can be manifested as an evolutionary process of discussing treatment options at multiple points in time and sometimes over years. Time was a major component of effective communication, time to have complex discussions as well as selecting the optimal time to approach patients and families with difficult issues. One geriatrician noted the link between communication and time: “Communication is difficult and hard, especially if it is a sudden bolt out of the blue. There isn’t time for information to percolate. Families need time.”

Participants related that timing issues were influenced by provider and patient characteristics and by disease prognosis and clarity. of ten discussions were triggered by acute medical events or by ineffective therapeutic interventions. In positive encounters, frequent discussions occurred within the context of an established provider-patient relationship and without a sense of urgency. Communication in negative encounters was characterized by an atmosphere of fear and anger, a lack of trust, and a crisis orientation:

I just think there was no time to get used to the idea that we have a malignancy that is not going to be cured, and if you had the opportunity to build a professional relationship that can be a therapeutic relationship, you can work through some of those issues. But in her case there was absolutely no time to do it. She was extremely frightened at the beginning, and so there was no time to build; there was no trust.

The timing to approach end-of-life care discussions was difficult to judge, and participants had conflicting opinions about whether discussions should occur when patients were still healthy or when there was evidence of a decline. The variability and uncertainty of chronic disease trajectories despite functional status confounded the timing of discussions. An internist who serves as a hospice medical director reflected on a discussion she recently had with a patient:

The good news is you’ve been pretty functional, you feel pretty good. The bad news is you could die any-time. And you know, who is the decision maker ifyou can’t speak for yourself, what are your goals, and how aggressive should we be.

In most positive encounters the participants described spending time with patients and families, a task that was difficult and influenced by the realities of clinical practice in settings that were “overbooked in every slot,”as one participant explained.

A cohesive interdisciplinary team enhanced effective communication. Participants viewed social workers and chaplains as invaluable in assisting patients and families to assimilate information and to facilitate an acceptance of death. A common value system and knowledge base regarding treatment plans among care team members also marked positive encounters:

And some people get it right away and are very realistic. But for most people it takes several conversations over time, and it takes the whole team being on the same page basically. We all have to be saying the same thing.

In contrast, participants identified problems that occurred when interdisciplinary members functioned in independent roles, lacked a team concept and a shared understanding of treatment goals, and did not communicate among themselves.

Discussion

The purpose of our study was to explore the complex phenomena that occur between nursing and physician providers and their dying patients and to describe those factors that providers identify as important in those encounters. The results of our study have both theoretical and practical applications for providers of end-of-life care. Although our original aim was to identify factors, our results suggested that interrelationships existed among the categories. A descriptive conceptual framework was therefore built to graphically depict these relationships (Figure). A conceptual framework explains key categories either graphically or narratively and the presumed relationships among them.²² We constructed an initial framework and several iterations until consensus was reached.

The framework outlines a dynamic reorientation process of both patient and provider norms, values, and behaviors from a curative biomedical perspective to a palliative course that is centered around assisting patients in achieving a quality, comfortable death. This process has been characterized as socialization to dying.²³ In medical settings socialization is inclusive of the content and characteristics of learning social and cultural cues and adaptations, as well as the manner and process by which ideas and ideals are communicated and reinforced.²⁴

Our framework suggests that 4 content domains (provider and patient characteristics, interpersonal relationships, prognostic certainty, and clarity) and 2 process domains (effective communication, and decision making and planning) are key components for understanding the socialization process that providers and patients undergo in end-of-life care. Attitudinal variables such as death acceptance and openness to discussions about death and dying were the major common elements of both provider and patient characteristics in our study. Recent attitudinal research on end-of-life care has focused on the volatile issue of physician-assisted suicide and euthanasia.^25,26 The balance of these studies suggest that demographic and social variables such as sex, education level, and religiosity are tied to both physician and patient attitudes toward death and dying. Religiosity and sex—in addition to mental health status and general health status—are key variables in understanding death attitudes in elderly populations.^27,28

Structuration theory (the construction of meaning through social interaction) provides a useful way to view these findings and the larger current ambivalence and confusion in the United States of how to best understand and situate death.²⁹ This orientation maintains that social determinants such as education, religion, and culture are major elements that facilitate the interpretation and understanding of death and dying.³⁰ If death attitudes serve as proxies for an understanding of death and dying, our results and our framework are congruent with this perspective and have practical implications as well. Socially constructed death attitudes greatly contribute and may be predictive of the end-of-life care experience for both patients and providers. For example, patients or providers with death attitudes characterized by a fear of death or death anxiety may rarely consider—much less begin—the difficult transition from serious illness to dying. As a result, the quality-of-life for these dying patients would be greatly diminished and minimally affected by interventions that are not cognizant of this process. By providing care that is longitudinal, comprehensive, and patient-centered,³¹ family physicians and other primary care physicians are in a unique and advantageous position to assess the impact of social and cultural influences in their dying patients and to incorporate these determinants in their care plans.

Participants in our study cited the quality and character of interpersonal relationships in demarcating positive and negative encounters. Previous qualitative work with patients has also validated the importance of strengthening relationships with loved ones as a domain of quality end-of-life care.³² Our findings suggest that these relationships may be inclusive of provider-patient relationships, as well as an identified spiritual component. According to our study participants, the prognostic clarity of the disease facilitated positive encounters. There is widespread interest in developing and refining prognostic criteria in diseases involving chronic organ disease, although the clinical prediction of nonmalignant disease remains largely ineffective.³³ From a practical viewpoint, our study suggests that primary care providers may consider redirecting their attention away from exacting a disease prognosis toward identifying and enhancing supportive relationships for the patient and developing treatment plans based on personal values and quality of life.

Effective communication is the keystone of the framework and is intertwined with the additional process domain of decision making and planning. Participants amplified several tasks and characteristics of effective communication as central to guiding the patient through the critical transition of curative orientation—in the face of a life-limiting illness—to a dying one. The central place of effective communication in our framework highlights an additional role for family physicians as they care for patients at the end of life. Although disseminating information and empowering patients and family members have been promoted as key functions for providers,³⁴ participants in our study suggest that this process is more inclusive than these tasks. The continuous assessment and identification of patient goals, values, and feelings at multiple time points are functions that are of ten relegated to nonmedical or non-nursing (ie, social work, pastoral care) providers, yet participants in our study cited these responsibilities as vital.

Limitations

There are several limitations to our study. As an exploratory study, the sample size is small and the conceptual framework generally should be considered preliminary and open to modification. Qualitative studies are not intended to be statistically representative of any population but to provide an in-depth examination of complex phenomena. The frequency and validation of participant experiences were not determined. However, the strength of this investigation lies in the in-depth examination and the emerging conceptual framework. Although our categories may be self-evident as factors potentially influencing the outcome of provider-patient encounters in end-of-life care (eg, patient and provider death attitudes),our framework depicts them as a dynamic, complex interaction of modifiable and potentially nonmodifiable factors. Recent initiatives on provider training to improve the quality of care at the end of life (eg, the American Medical Association’s Education for Physicians on End-of-life Care project) are extremely important yet may simplify this complex interaction and the difficulty (or in some cases the impossibility) of assisting a patient’s socialization to dying. Finally, our theoretical framework is based on the provider’s perspective and does not address the patient, family, or caregiver view.

Conclusions

Our conceptual framework, characterized by a socialization process from a curative orientation to a dying orientation that is centered around the process of effective communication, is more inclusive and dynamic than previously described. Future research should test and refine the applicability of this framework and the interventions that facilitate socialization to dying.

Acknowledgments

Our work was supported by the Robert Wood Johnson Foundation Generalist Faculty Scholars Program (T. P.D),the John A. Hartford Foundation (S. A.H.,T. P.D.),and the School of Nursing office of Grants and Research and the Center on Aging at the University of Kansas Medical Center (S. A.H.,T. P.D.).We thank Ann Kuckelman Cobb, RN, PhD, and Anne Walling, MD, for their review of our manuscript.

There is a strong impetus to better understand the contemporary experience of dying and to improve end-of-life care in the United States.¹ Effective communication between patients and their care providers has been promoted as a key element within this experience and as a critical factor for maintaining quality of life during end-of-life care.² Multiple factors can impede or facilitate communication in this arena: attitudes toward death and dying,³ provider and patient anxiety over diagnostic and prognostic discussions,⁴ and cultural, socioeconomic, and educational influences.⁵ These factors highlight the nature and context of the provider-patient relationship, which plays a central role in end-of-life medical care for primary care physicians.⁶ Although much emphasis has been given to improving providers’communication skills⁷ and to facilitating decision making,^8-10 there is a paucity of research on the actual dynamics of the provider-patient encounter in end-of-life care.

One theoretical model of physician attitudes and roles in their encounters with dying patients incorporated 3 dimensions: direct involvement with the patient, the physician’s own needs and development, and cooperation with other caregivers.¹¹ Subsequent qualitative research with family physicians supported these dimensions and emphasized the primacy of the patient relationship and the area of physician personal domain.^12,13 However, the applicability of these findings and the model they provide to other specialties and disciplines that are involved in end-of-life care remains unclear. Studies of critical care nurses also identified patient concerns and their own personal concerns as key areas and raised the issue of frustration in their limited role in the management of patients at the end of life.¹⁴ In a study of palliative care workers, most perceived themselves as open and sensitive to their patients, although many felt poorly supported by other staff members.¹⁵

A greater and more complete understanding of the provider’s perspective of the provider-patient encounters in end-of-life care could enhance communication and decision making. This area of inquiry is especially salient in light of the recent Commonwealth-Cummings project on the quality of care at the end of life, which has identified medical provider interventions as one of 4 critical components in a multifaceted framework for a good death.¹ In addition, facilitating the communication and decision-making processes that accompany care at the end of life has been suggested as a key role for family physicians and other generalist providers.¹⁶ We conducted a study using qualitative research methods to explore and describe the nature of provider-patient encounters at the end of life. Our specific aim with this study was to describe factors that both medical and nursing providers identify as influential in their positive and negative encounters with dying patients.

Methods

Design

Because of the exploratory nature of our study, we selected a qualitative research method (semi structured interviews) to gain a richer and more complete description of the multifaceted phenomena that occur between medical and nursing providers and their dying patients. Semi structured interviews are dialogues that are guided and open ended enough to produce transcripts of the discussion as primary data.¹⁷ Editing analysis, in which meaningful units or segments of text are identified into categories or codes that can be used to construct and interpret common themes and patterns, was chosen as a way to generate new understanding in the area of provider perceptions in end-of-life care encounters.¹⁷

Sampling

We used maximum variation sampling to obtain the broadest range of provider perspectives on end-of-life encounters.¹⁸ This approach allowed us to obtain a wide range of information and perspectives on provider perceptions of their encounters with dying patients. We therefore selected a diverse sample of physicians and nurses based on the following criteria:(1) providers in specialties that had the highest probability of encountering dying patients in their practice and (2) providers who had experience in multiple clinical settings, such as academic health centers, private practice, and community settings.

Participants

The Human Subjects Committee of the University of Kansas Medical Center approved our study. We interviewed 6 physicians and 6 nurses from a large Midwestern metropolitan area. The physicians represented the following specialties: geriatric medicine, cardiology, pulmonary medicine/critical care, internal medicine/hospice, oncology, and family practice. The nurse participants were selected from the following specialties: neurology, nephrology/dialysis unit, oncology, hospice, inpatient palliative care, and cardiology/critical care.

The average age of the participants in our sample was 44 years (range=35-54 years);2 were men, and 10 were women. The physician participants had been in practice an average of 10 years (range=5-18 years),while nurse participants had an average of 13 years’(range=5-20 years) experience.

Data Collection

We identified 12 potential participants who met the study criteria and were either known to us professionally or were suggested to us by other colleagues. These participants were contacted and invited to participate. There were no refusals. Semi structured interviews were conducted between September 1998 and April 1999 in either the participant ’s office or the investigator’s office and lasted from 1 hour to 1.5 hours. We conducted individual interviews with participants from our respective disciplines (S. A.F, nurses; T. P.D.,physicians).

Participants were asked to reflect on 2 contrasting end-of-life patient encounters that they had personally experienced in their practices. The encounters were to involve multiple contacts with a single patient who had since died. The first reflection focused on encounters that resulted in a very positive and rewarding outcome from the participant’s perspective. The second reflection involved encounters that were viewed by the participant as personally difficult and troubling. The interview began with the positive reflection, and the participants were allowed to speak freely with few interruptions. An interview guide (Appendix) provided the template for follow-up questions oriented around treatment decisions, the communication process, and participant attitudes, values, and beliefs regarding end-of-life care. All interview sessions were audiotaped and professionally transcribed.

Data Analysis

The interview transcripts were checked for accuracy and verified against the original audiotapes by the investigator who conducted the interview, and the transcripts were then formatted and entered into a qualitative software package (QSR NU*DIST).¹⁹ All data (text, codes, categories, and notes) were entered, retrieved, and analyzed using the computerized software. Independent meaningful units or segments of text that were relevant to the study purpose were identified, labeled, and organized simultaneously by the investigator/analyst in a process called coding.²⁰ After the initial coding was completed, we reviewed the data to evaluate emerging patterns and themes. This iterative process was used to search for systematic relationships and for contrasts and irregularities among the codes and categories, and it continued until consensus or agreement by the analysts was reached. Credibility was maintained by peer debriefing, a search for rival explanations in the data, and by member checking.^18,21 The interdisciplinary nature of the investigators and the iterative coding process enhanced peer debriefing and rival explanation searching, which occurred repeatedly. Member checking was conducted with 4 participants who agreed with the study findings.

Results

Participant interviews were grouped into 6 general categories:(1) provider characteristics,(2) patient characteristics,(3) interpersonal relationships,(4) prognostic certainty/clarity,(5) decision making, and (6) effective communication. Each of these categories is described below with verbatim quotes from the participants.

Provider Characteristics

Participants identified specific provider attitudes, knowledge, and skills as important components of positive encounters. Provider attitudes such as openness to discussions about death and dying, a willingness to be vulnerable, and death acceptance were associated with positive encounters. A heuristic perspective—one that incorporated a receptiveness to learning from patients, past experiences, and even painful mistakes—was inherent in these encounters.

A cardiologist shared his willingness to learn from an elderly patient who challenged his traditional treatment paradigm by refusing a recommendation for bypass surgery:

Maybe my bias was to do what I had been trained to do, which was to send her for bypass, put her through some misery, and maybe not have changed the outcome that much, you know maybe added a year or so, put her through quite a lot of trauma. But she removed my personal bias.

Vulnerability was typified by one participant who related:

This one was hard for me personally…. It was painful for me, because I really liked him, but rewarding on the other hand because I was able to really support him in his autonomous decision to do what he wanted to do.

Death attitudes, whether avoidant or accepting, were also key elements in this category. A nurse participant suggested that death avoidant attitudes were major factors in negative encounters:

But this particular physician can sidestep some of these tough conversations. And that doesn’t make him a bad physician, that’s just his style. He doesn’t do well with those conversations. He’s actually quite a good oncologist, just bad with end-of-life conversations.

A sufficient knowledge base and skill level in caring for dying patients also were important characteristics. Most participants stated their lack of formal education in palliative care as an impediment to providing good care in these settings. As one physician candidly shared, “We learn end-of-life care by the seat of our pants, by the mistakes that we make.”

Patient Characteristics

Patient characteristics such as attitudes, values, and knowledge were important components of participants’perceptions of positive and negative encounters. Within the context of positive encounters, patients were described as proactive, information-seeking, educated, having a clear vision, and focused on reality. Patients in these settings were perceived as facing reality from the onset of diagnosis and being capable of finding something positive in the face of death.

He [the patient] realized that not everything in life is fixable. He said, “This is no tragedy [death]; this is what happens in life.”But as treatments didn’t do any good anymore, he made a decision. He said, “I’ve got to get out and enjoy life.”

In contrast, patients in the negative encounters were perceived as fearful, demanding, untrusting, continually searching for ways to fight the disease, unwilling to give up, and living in chaotic family systems. One patient who had battled cancer for 6 years and spent the last weeks of his life in the intensive care unit was described by a nurse participant:

He [patient] did not want to give up, wanted to be positive, did not want to hear a negative thought, a negative piece of information, and had a physician who kind of went along with that …[which] created a big problem because here you have a man who’s dying basically, and they [family] don’t want to hear it. So that’s probably the problem in a nutshell.

Interpersonal Relationships

Interpersonal relationships were networks of family members, friends, and care providers that either facilitated or impeded the encounters. Participants identified multiple elements in positive encounters: the provider’s sense of connection and a congruent belief system with the patient, patients with supportive relationships with family members and friends, and patients with an existential or spiritual support system. The provider’s sense of connection was captured by an oncologist who related: “We connected on a personality level. I was a shepherd of his, so to speak…personally. I connected with him probably more than I do with most people.”

These positive relationships were also strengthened by common belief systems that were inclusive of patients’ existential or spiritual beliefs. One participant stated, “We could talk about spiritual issues, and maybe that was the thing that bonded us.”Participants also identified coherent and consistent patient belief systems in their positive encounters, “He talked openly about dying and that he wasn’t afraid. He felt that spiritually he was ready.” Also, in their depictions of positive encounters participants cited families and friends who provided emotional support yet allowed for the patient’s wishes to be honored.

Negative encounters were highlighted by strained uncomfortable provider-patient relationships characterized by a lack of trust. Participants identified patients and family members in conflict, hindering supportive relationships. These strained relationships were marked by a lack of acceptance of diagnostic and prognostic information by both patients and families. This lack of acceptance would be manifested in anger and hostility toward providers as one participant related:

He would burst into tears and say, ‘you can’t just let her die. You have to do everything.’ So there was never a no code on this woman even though she was clearly end stage.

Participants cited the lack of an existential orientation or spiritual reference in many patients with whom they had a negative encounter. An oncology nurse typified this perspective:

Some people are clearly afraid to die. I have seen the people who don’t seem to have much of a faith base, not religiously speaking but spiritually. I mean, if their heart and soul and mind aren’t all connected, and they don’t have a sense of purpose, of a beginning and an end, those people are very frightened to die. And some have a religious base, and some don’t. But it’s that sense of the beginning and the end. We all are born, and we’re all going to die. And people who don’t have that sense don’t do well when it comes time to die. Emotionally it’s very difficult for them.

Prognostic Certainty/Clarity

Participants identified the precision of the patient’s diagnosis and prognostic accuracy, as well as the concomitant therapeutic plan, as a facilitating factor in positive encounters. A nursing participant who works with amyotrophic lateral sclerosis (ALS) patients was representative of this orientation:

I guess we start out with the diagnosis basically [following a lengthy diagnostic work-up] and talk to them about what the diagnosis is, that it is a terminal disease.

Another participant shared that prognostic criteria provided clarity and assisted in anticipatory care planning. A pulmonologist/critical care physician shared:

[We] watch the pulmonary function decline, and they turn dyspneic enough—it’s clear, you know—we’ve got great criteria for respiratory failure, and those criteria are very black and white.

In positive encounters, both patients and physicians were able to discern with greater clarity the timing in transitioning care from cure to comfort. For some this was based on the clarity of prognosis, while others used prognosis and an analysis of the burdens and benefits of continued treatment within their conceptualization of quality of life. A family physician related the following:

I was on call, and there was a lady who was from out of town visiting a family, who had a pretty massive stroke…the family thought long and hard about “do we put a feeding tube in her”and the swallowing study showed she could swallow. But she wasn’t able to tell us what she wanted. And she wasn’t making a lot of progress…. They talked with her pastor, who had been her pastor for approximately 40 years…. They decided not to feed her or give her any intravenous fluids. They thought she had a good life, she’s had a lot of faith, she’s told them she’s ready to go and that type of stuff.

Prognostic clarity greatly enhanced positive encounters but was difficult to achieve even for experienced physicians, one of whom stated that, “you can’t outline, you know, the myriad different things that can happen.”However, participants perceived that for family members, clarity of ten was difficult to achieve, especially when interacting with multiple physicians. Patients in intensive care unit settings with multiple specialists and revolving call coverage made prognostic clarity difficult to achieve for some family members. Family members were depicted as misinterpreting different versions of the same story or being provided conflicting information:

He [the patient] had multiple physicians who were trading off and on day to day so there wasn’t the continuity…and so the family heard many different versions of what his prognosis was.

Decision Making

Participants described the degree of unanimity or being on the same page with patients and family members with regard to outcome expectations and treatment decisions as an essential element, one that differentiated positive and negative encounters. Being on the same page was facilitated by shared provider/patient characteristics, such as death attitudes and enriched interpersonal relationships, and by prognostic clarity.

Negative encounters were marked by conflict due to a discongruous assessment or care plan among providers or between family members and providers. One nursing participant described conflict with a physician son who limited the amount of pain medication administered to his dying mother. This unilateral decision to reduce pain medication left the participant feeling conflicted regarding the quality of the patient’s dying:

She had the abdominal surgery and her belly full of cancer, so she had lots of reason for pain, although his [son’s] perception was she had no reason for pain…. The whole thing just, just didn’t feel good to me.

Decision making was an ongoing process that included seeking, sharing, hearing, and processing information within one’s value system. Mismatched expectations for treatment outcomes were inherent in negative encounters and were promoted by a patient or family member’s inability to assimilate and process information accurately. One physician participant shared his perception of unrealistic expectations of family members due to technology:

Perhaps that [high-tech interventions in the intensive care unit] gave them a false sense of security that appropriate therapy had been given, and therefore there was a chance of survival. I did give them statistics saying that, you know, the mortality for this condition is 80%.I think patients whenever they hear that always think they’re the 20% that’s going to survive. They grasp onto the 20% rather than hone in on the adverse.

Effective Communication

Effective communication described the sharing of information in a direct and honest manner by an interdisciplinary team, as well as the exploration of goals, values, and feelings at multiple points in time. It was the primary category cited by all participants. Effective communication can be manifested as an evolutionary process of discussing treatment options at multiple points in time and sometimes over years. Time was a major component of effective communication, time to have complex discussions as well as selecting the optimal time to approach patients and families with difficult issues. One geriatrician noted the link between communication and time: “Communication is difficult and hard, especially if it is a sudden bolt out of the blue. There isn’t time for information to percolate. Families need time.”

Participants related that timing issues were influenced by provider and patient characteristics and by disease prognosis and clarity. of ten discussions were triggered by acute medical events or by ineffective therapeutic interventions. In positive encounters, frequent discussions occurred within the context of an established provider-patient relationship and without a sense of urgency. Communication in negative encounters was characterized by an atmosphere of fear and anger, a lack of trust, and a crisis orientation:

I just think there was no time to get used to the idea that we have a malignancy that is not going to be cured, and if you had the opportunity to build a professional relationship that can be a therapeutic relationship, you can work through some of those issues. But in her case there was absolutely no time to do it. She was extremely frightened at the beginning, and so there was no time to build; there was no trust.

The timing to approach end-of-life care discussions was difficult to judge, and participants had conflicting opinions about whether discussions should occur when patients were still healthy or when there was evidence of a decline. The variability and uncertainty of chronic disease trajectories despite functional status confounded the timing of discussions. An internist who serves as a hospice medical director reflected on a discussion she recently had with a patient:

The good news is you’ve been pretty functional, you feel pretty good. The bad news is you could die any-time. And you know, who is the decision maker ifyou can’t speak for yourself, what are your goals, and how aggressive should we be.

In most positive encounters the participants described spending time with patients and families, a task that was difficult and influenced by the realities of clinical practice in settings that were “overbooked in every slot,”as one participant explained.

A cohesive interdisciplinary team enhanced effective communication. Participants viewed social workers and chaplains as invaluable in assisting patients and families to assimilate information and to facilitate an acceptance of death. A common value system and knowledge base regarding treatment plans among care team members also marked positive encounters:

And some people get it right away and are very realistic. But for most people it takes several conversations over time, and it takes the whole team being on the same page basically. We all have to be saying the same thing.

In contrast, participants identified problems that occurred when interdisciplinary members functioned in independent roles, lacked a team concept and a shared understanding of treatment goals, and did not communicate among themselves.

Discussion

The purpose of our study was to explore the complex phenomena that occur between nursing and physician providers and their dying patients and to describe those factors that providers identify as important in those encounters. The results of our study have both theoretical and practical applications for providers of end-of-life care. Although our original aim was to identify factors, our results suggested that interrelationships existed among the categories. A descriptive conceptual framework was therefore built to graphically depict these relationships (Figure). A conceptual framework explains key categories either graphically or narratively and the presumed relationships among them.²² We constructed an initial framework and several iterations until consensus was reached.

The framework outlines a dynamic reorientation process of both patient and provider norms, values, and behaviors from a curative biomedical perspective to a palliative course that is centered around assisting patients in achieving a quality, comfortable death. This process has been characterized as socialization to dying.²³ In medical settings socialization is inclusive of the content and characteristics of learning social and cultural cues and adaptations, as well as the manner and process by which ideas and ideals are communicated and reinforced.²⁴

Our framework suggests that 4 content domains (provider and patient characteristics, interpersonal relationships, prognostic certainty, and clarity) and 2 process domains (effective communication, and decision making and planning) are key components for understanding the socialization process that providers and patients undergo in end-of-life care. Attitudinal variables such as death acceptance and openness to discussions about death and dying were the major common elements of both provider and patient characteristics in our study. Recent attitudinal research on end-of-life care has focused on the volatile issue of physician-assisted suicide and euthanasia.^25,26 The balance of these studies suggest that demographic and social variables such as sex, education level, and religiosity are tied to both physician and patient attitudes toward death and dying. Religiosity and sex—in addition to mental health status and general health status—are key variables in understanding death attitudes in elderly populations.^27,28

Structuration theory (the construction of meaning through social interaction) provides a useful way to view these findings and the larger current ambivalence and confusion in the United States of how to best understand and situate death.²⁹ This orientation maintains that social determinants such as education, religion, and culture are major elements that facilitate the interpretation and understanding of death and dying.³⁰ If death attitudes serve as proxies for an understanding of death and dying, our results and our framework are congruent with this perspective and have practical implications as well. Socially constructed death attitudes greatly contribute and may be predictive of the end-of-life care experience for both patients and providers. For example, patients or providers with death attitudes characterized by a fear of death or death anxiety may rarely consider—much less begin—the difficult transition from serious illness to dying. As a result, the quality-of-life for these dying patients would be greatly diminished and minimally affected by interventions that are not cognizant of this process. By providing care that is longitudinal, comprehensive, and patient-centered,³¹ family physicians and other primary care physicians are in a unique and advantageous position to assess the impact of social and cultural influences in their dying patients and to incorporate these determinants in their care plans.

Participants in our study cited the quality and character of interpersonal relationships in demarcating positive and negative encounters. Previous qualitative work with patients has also validated the importance of strengthening relationships with loved ones as a domain of quality end-of-life care.³² Our findings suggest that these relationships may be inclusive of provider-patient relationships, as well as an identified spiritual component. According to our study participants, the prognostic clarity of the disease facilitated positive encounters. There is widespread interest in developing and refining prognostic criteria in diseases involving chronic organ disease, although the clinical prediction of nonmalignant disease remains largely ineffective.³³ From a practical viewpoint, our study suggests that primary care providers may consider redirecting their attention away from exacting a disease prognosis toward identifying and enhancing supportive relationships for the patient and developing treatment plans based on personal values and quality of life.

Effective communication is the keystone of the framework and is intertwined with the additional process domain of decision making and planning. Participants amplified several tasks and characteristics of effective communication as central to guiding the patient through the critical transition of curative orientation—in the face of a life-limiting illness—to a dying one. The central place of effective communication in our framework highlights an additional role for family physicians as they care for patients at the end of life. Although disseminating information and empowering patients and family members have been promoted as key functions for providers,³⁴ participants in our study suggest that this process is more inclusive than these tasks. The continuous assessment and identification of patient goals, values, and feelings at multiple time points are functions that are of ten relegated to nonmedical or non-nursing (ie, social work, pastoral care) providers, yet participants in our study cited these responsibilities as vital.

Limitations

There are several limitations to our study. As an exploratory study, the sample size is small and the conceptual framework generally should be considered preliminary and open to modification. Qualitative studies are not intended to be statistically representative of any population but to provide an in-depth examination of complex phenomena. The frequency and validation of participant experiences were not determined. However, the strength of this investigation lies in the in-depth examination and the emerging conceptual framework. Although our categories may be self-evident as factors potentially influencing the outcome of provider-patient encounters in end-of-life care (eg, patient and provider death attitudes),our framework depicts them as a dynamic, complex interaction of modifiable and potentially nonmodifiable factors. Recent initiatives on provider training to improve the quality of care at the end of life (eg, the American Medical Association’s Education for Physicians on End-of-life Care project) are extremely important yet may simplify this complex interaction and the difficulty (or in some cases the impossibility) of assisting a patient’s socialization to dying. Finally, our theoretical framework is based on the provider’s perspective and does not address the patient, family, or caregiver view.

Conclusions

Our conceptual framework, characterized by a socialization process from a curative orientation to a dying orientation that is centered around the process of effective communication, is more inclusive and dynamic than previously described. Future research should test and refine the applicability of this framework and the interventions that facilitate socialization to dying.

Acknowledgments

Our work was supported by the Robert Wood Johnson Foundation Generalist Faculty Scholars Program (T. P.D),the John A. Hartford Foundation (S. A.H.,T. P.D.),and the School of Nursing office of Grants and Research and the Center on Aging at the University of Kansas Medical Center (S. A.H.,T. P.D.).We thank Ann Kuckelman Cobb, RN, PhD, and Anne Walling, MD, for their review of our manuscript.

There is a strong impetus to better understand the contemporary experience of dying and to improve end-of-life care in the United States.¹ Effective communication between patients and their care providers has been promoted as a key element within this experience and as a critical factor for maintaining quality of life during end-of-life care.² Multiple factors can impede or facilitate communication in this arena: attitudes toward death and dying,³ provider and patient anxiety over diagnostic and prognostic discussions,⁴ and cultural, socioeconomic, and educational influences.⁵ These factors highlight the nature and context of the provider-patient relationship, which plays a central role in end-of-life medical care for primary care physicians.⁶ Although much emphasis has been given to improving providers’communication skills⁷ and to facilitating decision making,^8-10 there is a paucity of research on the actual dynamics of the provider-patient encounter in end-of-life care.

One theoretical model of physician attitudes and roles in their encounters with dying patients incorporated 3 dimensions: direct involvement with the patient, the physician’s own needs and development, and cooperation with other caregivers.¹¹ Subsequent qualitative research with family physicians supported these dimensions and emphasized the primacy of the patient relationship and the area of physician personal domain.^12,13 However, the applicability of these findings and the model they provide to other specialties and disciplines that are involved in end-of-life care remains unclear. Studies of critical care nurses also identified patient concerns and their own personal concerns as key areas and raised the issue of frustration in their limited role in the management of patients at the end of life.¹⁴ In a study of palliative care workers, most perceived themselves as open and sensitive to their patients, although many felt poorly supported by other staff members.¹⁵

A greater and more complete understanding of the provider’s perspective of the provider-patient encounters in end-of-life care could enhance communication and decision making. This area of inquiry is especially salient in light of the recent Commonwealth-Cummings project on the quality of care at the end of life, which has identified medical provider interventions as one of 4 critical components in a multifaceted framework for a good death.¹ In addition, facilitating the communication and decision-making processes that accompany care at the end of life has been suggested as a key role for family physicians and other generalist providers.¹⁶ We conducted a study using qualitative research methods to explore and describe the nature of provider-patient encounters at the end of life. Our specific aim with this study was to describe factors that both medical and nursing providers identify as influential in their positive and negative encounters with dying patients.

Methods

Design

Because of the exploratory nature of our study, we selected a qualitative research method (semi structured interviews) to gain a richer and more complete description of the multifaceted phenomena that occur between medical and nursing providers and their dying patients. Semi structured interviews are dialogues that are guided and open ended enough to produce transcripts of the discussion as primary data.¹⁷ Editing analysis, in which meaningful units or segments of text are identified into categories or codes that can be used to construct and interpret common themes and patterns, was chosen as a way to generate new understanding in the area of provider perceptions in end-of-life care encounters.¹⁷

Sampling

We used maximum variation sampling to obtain the broadest range of provider perspectives on end-of-life encounters.¹⁸ This approach allowed us to obtain a wide range of information and perspectives on provider perceptions of their encounters with dying patients. We therefore selected a diverse sample of physicians and nurses based on the following criteria:(1) providers in specialties that had the highest probability of encountering dying patients in their practice and (2) providers who had experience in multiple clinical settings, such as academic health centers, private practice, and community settings.

Participants

The Human Subjects Committee of the University of Kansas Medical Center approved our study. We interviewed 6 physicians and 6 nurses from a large Midwestern metropolitan area. The physicians represented the following specialties: geriatric medicine, cardiology, pulmonary medicine/critical care, internal medicine/hospice, oncology, and family practice. The nurse participants were selected from the following specialties: neurology, nephrology/dialysis unit, oncology, hospice, inpatient palliative care, and cardiology/critical care.

The average age of the participants in our sample was 44 years (range=35-54 years);2 were men, and 10 were women. The physician participants had been in practice an average of 10 years (range=5-18 years),while nurse participants had an average of 13 years’(range=5-20 years) experience.

Data Collection

We identified 12 potential participants who met the study criteria and were either known to us professionally or were suggested to us by other colleagues. These participants were contacted and invited to participate. There were no refusals. Semi structured interviews were conducted between September 1998 and April 1999 in either the participant ’s office or the investigator’s office and lasted from 1 hour to 1.5 hours. We conducted individual interviews with participants from our respective disciplines (S. A.F, nurses; T. P.D.,physicians).

Participants were asked to reflect on 2 contrasting end-of-life patient encounters that they had personally experienced in their practices. The encounters were to involve multiple contacts with a single patient who had since died. The first reflection focused on encounters that resulted in a very positive and rewarding outcome from the participant’s perspective. The second reflection involved encounters that were viewed by the participant as personally difficult and troubling. The interview began with the positive reflection, and the participants were allowed to speak freely with few interruptions. An interview guide (Appendix) provided the template for follow-up questions oriented around treatment decisions, the communication process, and participant attitudes, values, and beliefs regarding end-of-life care. All interview sessions were audiotaped and professionally transcribed.

Data Analysis

The interview transcripts were checked for accuracy and verified against the original audiotapes by the investigator who conducted the interview, and the transcripts were then formatted and entered into a qualitative software package (QSR NU*DIST).¹⁹ All data (text, codes, categories, and notes) were entered, retrieved, and analyzed using the computerized software. Independent meaningful units or segments of text that were relevant to the study purpose were identified, labeled, and organized simultaneously by the investigator/analyst in a process called coding.²⁰ After the initial coding was completed, we reviewed the data to evaluate emerging patterns and themes. This iterative process was used to search for systematic relationships and for contrasts and irregularities among the codes and categories, and it continued until consensus or agreement by the analysts was reached. Credibility was maintained by peer debriefing, a search for rival explanations in the data, and by member checking.^18,21 The interdisciplinary nature of the investigators and the iterative coding process enhanced peer debriefing and rival explanation searching, which occurred repeatedly. Member checking was conducted with 4 participants who agreed with the study findings.

Results

Participant interviews were grouped into 6 general categories:(1) provider characteristics,(2) patient characteristics,(3) interpersonal relationships,(4) prognostic certainty/clarity,(5) decision making, and (6) effective communication. Each of these categories is described below with verbatim quotes from the participants.

Provider Characteristics

Participants identified specific provider attitudes, knowledge, and skills as important components of positive encounters. Provider attitudes such as openness to discussions about death and dying, a willingness to be vulnerable, and death acceptance were associated with positive encounters. A heuristic perspective—one that incorporated a receptiveness to learning from patients, past experiences, and even painful mistakes—was inherent in these encounters.

A cardiologist shared his willingness to learn from an elderly patient who challenged his traditional treatment paradigm by refusing a recommendation for bypass surgery:

Maybe my bias was to do what I had been trained to do, which was to send her for bypass, put her through some misery, and maybe not have changed the outcome that much, you know maybe added a year or so, put her through quite a lot of trauma. But she removed my personal bias.

Vulnerability was typified by one participant who related:

This one was hard for me personally…. It was painful for me, because I really liked him, but rewarding on the other hand because I was able to really support him in his autonomous decision to do what he wanted to do.

Death attitudes, whether avoidant or accepting, were also key elements in this category. A nurse participant suggested that death avoidant attitudes were major factors in negative encounters:

But this particular physician can sidestep some of these tough conversations. And that doesn’t make him a bad physician, that’s just his style. He doesn’t do well with those conversations. He’s actually quite a good oncologist, just bad with end-of-life conversations.

A sufficient knowledge base and skill level in caring for dying patients also were important characteristics. Most participants stated their lack of formal education in palliative care as an impediment to providing good care in these settings. As one physician candidly shared, “We learn end-of-life care by the seat of our pants, by the mistakes that we make.”

Patient Characteristics

Patient characteristics such as attitudes, values, and knowledge were important components of participants’perceptions of positive and negative encounters. Within the context of positive encounters, patients were described as proactive, information-seeking, educated, having a clear vision, and focused on reality. Patients in these settings were perceived as facing reality from the onset of diagnosis and being capable of finding something positive in the face of death.

He [the patient] realized that not everything in life is fixable. He said, “This is no tragedy [death]; this is what happens in life.”But as treatments didn’t do any good anymore, he made a decision. He said, “I’ve got to get out and enjoy life.”

In contrast, patients in the negative encounters were perceived as fearful, demanding, untrusting, continually searching for ways to fight the disease, unwilling to give up, and living in chaotic family systems. One patient who had battled cancer for 6 years and spent the last weeks of his life in the intensive care unit was described by a nurse participant:

He [patient] did not want to give up, wanted to be positive, did not want to hear a negative thought, a negative piece of information, and had a physician who kind of went along with that …[which] created a big problem because here you have a man who’s dying basically, and they [family] don’t want to hear it. So that’s probably the problem in a nutshell.

Interpersonal Relationships

Interpersonal relationships were networks of family members, friends, and care providers that either facilitated or impeded the encounters. Participants identified multiple elements in positive encounters: the provider’s sense of connection and a congruent belief system with the patient, patients with supportive relationships with family members and friends, and patients with an existential or spiritual support system. The provider’s sense of connection was captured by an oncologist who related: “We connected on a personality level. I was a shepherd of his, so to speak…personally. I connected with him probably more than I do with most people.”

These positive relationships were also strengthened by common belief systems that were inclusive of patients’ existential or spiritual beliefs. One participant stated, “We could talk about spiritual issues, and maybe that was the thing that bonded us.”Participants also identified coherent and consistent patient belief systems in their positive encounters, “He talked openly about dying and that he wasn’t afraid. He felt that spiritually he was ready.” Also, in their depictions of positive encounters participants cited families and friends who provided emotional support yet allowed for the patient’s wishes to be honored.

Negative encounters were highlighted by strained uncomfortable provider-patient relationships characterized by a lack of trust. Participants identified patients and family members in conflict, hindering supportive relationships. These strained relationships were marked by a lack of acceptance of diagnostic and prognostic information by both patients and families. This lack of acceptance would be manifested in anger and hostility toward providers as one participant related:

He would burst into tears and say, ‘you can’t just let her die. You have to do everything.’ So there was never a no code on this woman even though she was clearly end stage.

Participants cited the lack of an existential orientation or spiritual reference in many patients with whom they had a negative encounter. An oncology nurse typified this perspective:

Some people are clearly afraid to die. I have seen the people who don’t seem to have much of a faith base, not religiously speaking but spiritually. I mean, if their heart and soul and mind aren’t all connected, and they don’t have a sense of purpose, of a beginning and an end, those people are very frightened to die. And some have a religious base, and some don’t. But it’s that sense of the beginning and the end. We all are born, and we’re all going to die. And people who don’t have that sense don’t do well when it comes time to die. Emotionally it’s very difficult for them.

Prognostic Certainty/Clarity

Participants identified the precision of the patient’s diagnosis and prognostic accuracy, as well as the concomitant therapeutic plan, as a facilitating factor in positive encounters. A nursing participant who works with amyotrophic lateral sclerosis (ALS) patients was representative of this orientation:

I guess we start out with the diagnosis basically [following a lengthy diagnostic work-up] and talk to them about what the diagnosis is, that it is a terminal disease.

Another participant shared that prognostic criteria provided clarity and assisted in anticipatory care planning. A pulmonologist/critical care physician shared:

[We] watch the pulmonary function decline, and they turn dyspneic enough—it’s clear, you know—we’ve got great criteria for respiratory failure, and those criteria are very black and white.

In positive encounters, both patients and physicians were able to discern with greater clarity the timing in transitioning care from cure to comfort. For some this was based on the clarity of prognosis, while others used prognosis and an analysis of the burdens and benefits of continued treatment within their conceptualization of quality of life. A family physician related the following:

I was on call, and there was a lady who was from out of town visiting a family, who had a pretty massive stroke…the family thought long and hard about “do we put a feeding tube in her”and the swallowing study showed she could swallow. But she wasn’t able to tell us what she wanted. And she wasn’t making a lot of progress…. They talked with her pastor, who had been her pastor for approximately 40 years…. They decided not to feed her or give her any intravenous fluids. They thought she had a good life, she’s had a lot of faith, she’s told them she’s ready to go and that type of stuff.

Prognostic clarity greatly enhanced positive encounters but was difficult to achieve even for experienced physicians, one of whom stated that, “you can’t outline, you know, the myriad different things that can happen.”However, participants perceived that for family members, clarity of ten was difficult to achieve, especially when interacting with multiple physicians. Patients in intensive care unit settings with multiple specialists and revolving call coverage made prognostic clarity difficult to achieve for some family members. Family members were depicted as misinterpreting different versions of the same story or being provided conflicting information:

He [the patient] had multiple physicians who were trading off and on day to day so there wasn’t the continuity…and so the family heard many different versions of what his prognosis was.

Decision Making

Participants described the degree of unanimity or being on the same page with patients and family members with regard to outcome expectations and treatment decisions as an essential element, one that differentiated positive and negative encounters. Being on the same page was facilitated by shared provider/patient characteristics, such as death attitudes and enriched interpersonal relationships, and by prognostic clarity.

Negative encounters were marked by conflict due to a discongruous assessment or care plan among providers or between family members and providers. One nursing participant described conflict with a physician son who limited the amount of pain medication administered to his dying mother. This unilateral decision to reduce pain medication left the participant feeling conflicted regarding the quality of the patient’s dying:

She had the abdominal surgery and her belly full of cancer, so she had lots of reason for pain, although his [son’s] perception was she had no reason for pain…. The whole thing just, just didn’t feel good to me.

Decision making was an ongoing process that included seeking, sharing, hearing, and processing information within one’s value system. Mismatched expectations for treatment outcomes were inherent in negative encounters and were promoted by a patient or family member’s inability to assimilate and process information accurately. One physician participant shared his perception of unrealistic expectations of family members due to technology:

Perhaps that [high-tech interventions in the intensive care unit] gave them a false sense of security that appropriate therapy had been given, and therefore there was a chance of survival. I did give them statistics saying that, you know, the mortality for this condition is 80%.I think patients whenever they hear that always think they’re the 20% that’s going to survive. They grasp onto the 20% rather than hone in on the adverse.

Effective Communication

Effective communication described the sharing of information in a direct and honest manner by an interdisciplinary team, as well as the exploration of goals, values, and feelings at multiple points in time. It was the primary category cited by all participants. Effective communication can be manifested as an evolutionary process of discussing treatment options at multiple points in time and sometimes over years. Time was a major component of effective communication, time to have complex discussions as well as selecting the optimal time to approach patients and families with difficult issues. One geriatrician noted the link between communication and time: “Communication is difficult and hard, especially if it is a sudden bolt out of the blue. There isn’t time for information to percolate. Families need time.”

Participants related that timing issues were influenced by provider and patient characteristics and by disease prognosis and clarity. of ten discussions were triggered by acute medical events or by ineffective therapeutic interventions. In positive encounters, frequent discussions occurred within the context of an established provider-patient relationship and without a sense of urgency. Communication in negative encounters was characterized by an atmosphere of fear and anger, a lack of trust, and a crisis orientation:

I just think there was no time to get used to the idea that we have a malignancy that is not going to be cured, and if you had the opportunity to build a professional relationship that can be a therapeutic relationship, you can work through some of those issues. But in her case there was absolutely no time to do it. She was extremely frightened at the beginning, and so there was no time to build; there was no trust.

The timing to approach end-of-life care discussions was difficult to judge, and participants had conflicting opinions about whether discussions should occur when patients were still healthy or when there was evidence of a decline. The variability and uncertainty of chronic disease trajectories despite functional status confounded the timing of discussions. An internist who serves as a hospice medical director reflected on a discussion she recently had with a patient:

The good news is you’ve been pretty functional, you feel pretty good. The bad news is you could die any-time. And you know, who is the decision maker ifyou can’t speak for yourself, what are your goals, and how aggressive should we be.

In most positive encounters the participants described spending time with patients and families, a task that was difficult and influenced by the realities of clinical practice in settings that were “overbooked in every slot,”as one participant explained.

A cohesive interdisciplinary team enhanced effective communication. Participants viewed social workers and chaplains as invaluable in assisting patients and families to assimilate information and to facilitate an acceptance of death. A common value system and knowledge base regarding treatment plans among care team members also marked positive encounters:

And some people get it right away and are very realistic. But for most people it takes several conversations over time, and it takes the whole team being on the same page basically. We all have to be saying the same thing.

In contrast, participants identified problems that occurred when interdisciplinary members functioned in independent roles, lacked a team concept and a shared understanding of treatment goals, and did not communicate among themselves.

Discussion

The purpose of our study was to explore the complex phenomena that occur between nursing and physician providers and their dying patients and to describe those factors that providers identify as important in those encounters. The results of our study have both theoretical and practical applications for providers of end-of-life care. Although our original aim was to identify factors, our results suggested that interrelationships existed among the categories. A descriptive conceptual framework was therefore built to graphically depict these relationships (Figure). A conceptual framework explains key categories either graphically or narratively and the presumed relationships among them.²² We constructed an initial framework and several iterations until consensus was reached.

The framework outlines a dynamic reorientation process of both patient and provider norms, values, and behaviors from a curative biomedical perspective to a palliative course that is centered around assisting patients in achieving a quality, comfortable death. This process has been characterized as socialization to dying.²³ In medical settings socialization is inclusive of the content and characteristics of learning social and cultural cues and adaptations, as well as the manner and process by which ideas and ideals are communicated and reinforced.²⁴

Our framework suggests that 4 content domains (provider and patient characteristics, interpersonal relationships, prognostic certainty, and clarity) and 2 process domains (effective communication, and decision making and planning) are key components for understanding the socialization process that providers and patients undergo in end-of-life care. Attitudinal variables such as death acceptance and openness to discussions about death and dying were the major common elements of both provider and patient characteristics in our study. Recent attitudinal research on end-of-life care has focused on the volatile issue of physician-assisted suicide and euthanasia.^25,26 The balance of these studies suggest that demographic and social variables such as sex, education level, and religiosity are tied to both physician and patient attitudes toward death and dying. Religiosity and sex—in addition to mental health status and general health status—are key variables in understanding death attitudes in elderly populations.^27,28

Structuration theory (the construction of meaning through social interaction) provides a useful way to view these findings and the larger current ambivalence and confusion in the United States of how to best understand and situate death.²⁹ This orientation maintains that social determinants such as education, religion, and culture are major elements that facilitate the interpretation and understanding of death and dying.³⁰ If death attitudes serve as proxies for an understanding of death and dying, our results and our framework are congruent with this perspective and have practical implications as well. Socially constructed death attitudes greatly contribute and may be predictive of the end-of-life care experience for both patients and providers. For example, patients or providers with death attitudes characterized by a fear of death or death anxiety may rarely consider—much less begin—the difficult transition from serious illness to dying. As a result, the quality-of-life for these dying patients would be greatly diminished and minimally affected by interventions that are not cognizant of this process. By providing care that is longitudinal, comprehensive, and patient-centered,³¹ family physicians and other primary care physicians are in a unique and advantageous position to assess the impact of social and cultural influences in their dying patients and to incorporate these determinants in their care plans.

Participants in our study cited the quality and character of interpersonal relationships in demarcating positive and negative encounters. Previous qualitative work with patients has also validated the importance of strengthening relationships with loved ones as a domain of quality end-of-life care.³² Our findings suggest that these relationships may be inclusive of provider-patient relationships, as well as an identified spiritual component. According to our study participants, the prognostic clarity of the disease facilitated positive encounters. There is widespread interest in developing and refining prognostic criteria in diseases involving chronic organ disease, although the clinical prediction of nonmalignant disease remains largely ineffective.³³ From a practical viewpoint, our study suggests that primary care providers may consider redirecting their attention away from exacting a disease prognosis toward identifying and enhancing supportive relationships for the patient and developing treatment plans based on personal values and quality of life.

Effective communication is the keystone of the framework and is intertwined with the additional process domain of decision making and planning. Participants amplified several tasks and characteristics of effective communication as central to guiding the patient through the critical transition of curative orientation—in the face of a life-limiting illness—to a dying one. The central place of effective communication in our framework highlights an additional role for family physicians as they care for patients at the end of life. Although disseminating information and empowering patients and family members have been promoted as key functions for providers,³⁴ participants in our study suggest that this process is more inclusive than these tasks. The continuous assessment and identification of patient goals, values, and feelings at multiple time points are functions that are of ten relegated to nonmedical or non-nursing (ie, social work, pastoral care) providers, yet participants in our study cited these responsibilities as vital.

Limitations

There are several limitations to our study. As an exploratory study, the sample size is small and the conceptual framework generally should be considered preliminary and open to modification. Qualitative studies are not intended to be statistically representative of any population but to provide an in-depth examination of complex phenomena. The frequency and validation of participant experiences were not determined. However, the strength of this investigation lies in the in-depth examination and the emerging conceptual framework. Although our categories may be self-evident as factors potentially influencing the outcome of provider-patient encounters in end-of-life care (eg, patient and provider death attitudes),our framework depicts them as a dynamic, complex interaction of modifiable and potentially nonmodifiable factors. Recent initiatives on provider training to improve the quality of care at the end of life (eg, the American Medical Association’s Education for Physicians on End-of-life Care project) are extremely important yet may simplify this complex interaction and the difficulty (or in some cases the impossibility) of assisting a patient’s socialization to dying. Finally, our theoretical framework is based on the provider’s perspective and does not address the patient, family, or caregiver view.

Conclusions

Our conceptual framework, characterized by a socialization process from a curative orientation to a dying orientation that is centered around the process of effective communication, is more inclusive and dynamic than previously described. Future research should test and refine the applicability of this framework and the interventions that facilitate socialization to dying.

Acknowledgments

Our work was supported by the Robert Wood Johnson Foundation Generalist Faculty Scholars Program (T. P.D),the John A. Hartford Foundation (S. A.H.,T. P.D.),and the School of Nursing office of Grants and Research and the Center on Aging at the University of Kansas Medical Center (S. A.H.,T. P.D.).We thank Ann Kuckelman Cobb, RN, PhD, and Anne Walling, MD, for their review of our manuscript.