Hospice & Palliative Medicine

Modern medicine has dramatically improved the length and quality of life for countless persons but has also created problems in integrating these advances with end-of-life care. Although 70% of Americans want to die in their own home supported by family, 74% of Americans currently die in institutions.^1,2 Clearly, there is a disparity between patient preferences and the end-of-life services provided in America.

Most recommendations for the provision of quality end-of-life care are based on unresearched expert opinion that identifies problems and offers solutions.^3-5 Studies have explored communication strategies,⁶ satisfaction levels with care,⁷ perceptions of management issues,⁸ and attitudes toward both death and dying patients.⁹ Yet little research addresses how practicing community clinicians develop the appropriate relationships and integrate the requisite knowledge, skills, and attitudes to provide quality end-of-life care.

The relationship between clinician, patient, and family and the personal meaning of events for each participant greatly influence end-of-life care.¹⁰ Although most commentary on physician roles and relationships in end-of-life care also reflects unresearched expert opinion,^11-13 Steinmetz and Gabel^14,15 theorized a model that was later validated against the experience of practicing family physicians. The role of the clinician in the successful provision of quality care has not been systematically researched and remains a major challenge to improving end-of-life care. We describe an exploratory qualitative study to determine the perceptions of practicing clinicians regarding quality end-of-life care.

Methods

With approval from the Human Subjects Committee of the University of Washington School of Medicine, faculty were recruited from 9 community programs in the Affiliated Family Practice Residency Network of the Department of Family Medicine. Twenty semistructured long interviews were conducted from a convenience sample at 3 residency sites.¹⁶ An additional 8 semistructured interviews and 2 focus groups (10 participants) further explored the data. Thus, a total of 38 clinicians participated in the study. Demographic characteristics of the study cohort have been published previously.¹⁰

We conducted the initial 20 interviews using open-ended questions designed to uncover faculty perceptions of quality end-of-life care. The questions explored may be found in the Table. Nonspecific prompts such as “tell me more” were used to enrich data and avoid interviewer bias. The interview was rehearsed with a medical school faculty expert in ethnographic research, field tested, and adjusted before final use. The interviews were audio-recorded and transcribed into qualitative research software.¹⁷

Throughout the interview process, the investigators met weekly to compare findings, discuss emerging categories, and jointly code the transcripts for model development using the process of grounded theory described by Glaser and Strauss.¹⁸ Common themes, communication and educational issues, and roles and relationships were identified. Data were constantly compared and winnowed to facilitate manageability; data collection and analysis occurred simultaneously.^19,20 Common themes are published elsewhere.¹⁰ We report and discuss the analysis of the data pertaining to clinician roles and relationships.

We employed the following 3 strategies to assess the methodologic validity of the study and face validity of the themes and models emerging from the data analysis.

TABLE
INTERVIEW QUESTIONS

Methodologic validation

The themes, models, and 2 interview transcripts were reviewed by 2 expert consultants with experience in qualitative research on end-of-life care and chronic debilitating disease. They assessed the validity of the research process and model development.^21,22 These experts confirmed the methodologic approach, affirmed that the transcripts supported the coded model, and noted that the results were consistent with their own past research experience.

Face validity faculty interviews

Long interviews were conducted with a convenience sample of 8 faculty at 2 other residency sites. Researchers described the previous interview process and presented a written summary of the roles and relationship model coded from the data. Dialogue was encouraged with nonspecific prompts to enrich the data and avoid interviewer bias. Interviews were recorded, transcribed, and coded by the researchers to assess face validity and further develop the roles and relationship model using a grounded theory approach.

Face validity focus groups

Two focus groups were conducted with a total of 10 faculty at 2 additional sites. The entire model (themes, communication, roles, and relationships) was presented, and dialogue was encouraged. The researchers recorded, transcribed, and reviewed the focus groups to further assess face validity and refine the model.

Content validity

In a manner identical to that of the focus groups, the researchers presented the entire model at 2 of the 3 sites where faculty provided the initial 20 interviews. These sessions were audiotaped, transcribed, and reviewed by the researchers. The model presented was considered a valid, clinically plausible summary of the content of the initial interviews.

Results

Content analysis of the roles and relationships described by the study cohort revealed 3 distinct clinician roles in the provision of end-of-life care that we termed “consultant,” “collaborator,” and “guide.” All roles were viewed as important; none was more valuable than another; and the roles appeared to build on each other, often merging. Roles were implemented fluidly, with clinicians moving from one role to another as circumstances dictated. Although clinicians tended to describe successful cases, they freely discussed the challenges of end-of-life care, noting that less-than-ideal results were frequent. Thus, the models were described as “best practice” rather than routine care.

Consultant

The consultant provides expert medical information to the patient and family based on the biomedical model and the disease process. The power of the consultant role emanates from the clinician’s medical authority and special knowledge. The consultant presents information to help the patient and family understand the diagnosis, prognosis, and treatment. Once this knowledge is understood, the patient and family determine its meaning to them and decide on treatment. If the patient or family cannot decide or understand the medical implications of their particular situation, the consultant decides based on the medical facts.

A pediatrician illustrated the application of the consultant role in 2 situations involving newborns with severe heart anomalies. Here the physician describes the difficulty of using only medical facts to assist the family in deciding whether to discontinue the ventilator: “I remember saying: ‘The only thing that’s keeping the baby alive at this point is the ventilator. How do you feel about making the decision to turn that off?’ I remember it was a very cruel thing to have them make the decision.”

The physician continued: “Next time I said: ‘We have to sit down and talk about how your baby is doing because he died, and I’m going to turn off the machines now.’ I don’t think we should put them in the position of having to make this decision. We should make this decision and tell them what we’re going to do.” In this situation the clinician employed biomedical expertise to determine the appropriate medical care and outcome for the patient and family. In both cases the decision centered on continuing or discontinuing a medical intervention (respirator) and avoided discussion of the parents’ affective experience. Hallmarks of the consultant role include a biomedical focus, disease-centered decision-making processes, and the clinician’s assumption of authority based on biomedical expertise.

Collaborator

The collaborator exchanges information with the patient and family to promote a common understanding of the diagnosis and illness experience, working with them to choose a treatment path. The collaborator incorporates all components of the consultant role and additionally requires the clinician to understand the patient/family experience. The collaborator considers patient and family issues that need to be addressed to understand the medical facts, appreciates their past experiences with serious illness, determines what information would be difficult for them to accept but would benefit them if they were challenged to confront it, and recognizes that the patient and family hold ultimate responsibility for making treatment decisions.

The following quotes exemplify collaborative clinical care. In the first example, the physician steps outside his comfort zone to provide medical care based on the expressed wishes of a 30-year-old man dying of an advanced brain tumor: “The issue for me was letting go of control. He was going 4 hours away to a fishing cabin and going on a boat. I was really nervous…what if he all of a sudden crashed there? It took me a while to get used to the idea that he needs to be able to do what he enjoys doing, and everybody knows that there’s a risk.”

In a second example, the physician describes collaborating with the family of an 85-year-old woman suffering a massive stroke: “Having heard from the family that they understand that Grandma has had a big stroke and isn’t going to survive…then what wishes do Grandma and the family have? Have they ever discussed this sort of situation?”

In both these examples an understanding of the patient/family experience directs the clinician toward appropriate end-of-life care and is a major source of the collaborator’s power.

Guide

As a guide, the clinician actively and personally seeks solutions for the patient based not only on the medical facts (consultant) and the patient’s values and preferences (collaborator) but also on the guide’s greater understanding of the medical context. In essence, the guide not only knows why and where the patient and family prefer to go but also how to get there.

The following quote demonstrates a clinical application of the guide role. The physician first collaborates to understand the resistance of a terminally ill patient to hospice care:

“They had been very resistant to hospice for reasons that I wasn’t quite clear on. It wasn’t until I was in the home and listened to them talk that I realized they viewed hospice as ‘people are giving up on him.’ I think nobody really was talking to the patient about whether he was willing to give up or if he was ready to die.

“Once patient and family concerns were understood, the guide role was employed: I told the wife and family, ‘We’ve maxed out our medical therapies. There’s not much more we can do for him physically, but perhaps there’s something we can do for him spiritually and emotionally.’ When it was presented to them that way, they were much more understanding. We talked about hospice philosophy and looking at death as a part of life, saying: ‘That we’re not going to resuscitate does not mean do not treat.’ That’s where I came in more as the guide and said: ‘This is what I think is reasonable. What are your expectations? What do you want, Billy? And what do you want, as far as [his] wife and kids?”

The guide role requires that the clinician interpret the patient’s experience, integrate this interpretation with the clinician’s understanding of the clinical situation, and make a recommendation based on the guide’s personal and professional understanding of the situation. The power of the guide role emanates from the clinician’s understanding of how to use the medical system to see that the goals of patient and family are realized.

Discussion

The results of our study are consistent with reports over the past 3 decades by researchers, educators, and social critics who have explored how health professionals provide end-of-life care.^23-29 Despite their varying perspectives, all reflect a common theme: the need to provide care based on the unique illness experience and values of the patient and family. The majority of clinicians in our study spoke of the importance of their relationships with patients and families. More than 50% of all interview commentary addressed clinician issues of relationship and personal meaning when providing end-of-life care.

The descriptions of consultant,³⁰ collaborator,^31,32 and guide³³ confirm previous theoretical discussions regarding the nature of roles and relationships between patients and physicians. The guide is the most complex of the roles described by our study cohort and lends itself to ambivalence on the part of clinicians, in light of its potential to be misunderstood as paternalism. Yet, given the vulnerability and dependence of many patients who are terminally ill, the guide provides these patients with structure, safety, support, and care—based on the patients’ values and goals—reminiscent of the role of a nurturing parent.³⁴

The personal nature of the decisions the guide facilitates reflects the reality of medicine as a moral enterprise.^34,35 Having generally witnessed many more deaths than the patient, the guide has knowledge regarding the processes of dying and medical systems that is instrumental in assuring that the desires of the patient and family are realized. A knowledge of patient and family, an appreciation of the futility of the medical situation, and an insight into the process and systems of dying afford the clinician an opportunity to shape the death experience; as Nuland³⁶ described: “Each of us needs a guide who knows us as well as he knows the pathways by which we can approach death.”

Facility with these roles may be a function of personal talent, introspection, and experience. The physicians studied reported that competence with these roles grew over time through delivering endof-life care and learning from patients, families, and other caregivers. Less experienced clinicians tended to describe the consultant and collaborator roles only. Clinicians describing the guide role had been in practice at least a decade and thus were at least 17 years into their training and practice in medicine. Apparently, expertise in all 3 roles requires not only excellent technical diagnostic and treatment skills but also the complex integrated skills of relational knowledge and caring gained through experience.

Our data indicate that relationship, meaning, and roles are primary moderators of the organization of exemplary end-of-life care. Yet, it appears to take nearly 2 decades for physicians to gain the confidence in their knowledge, skills, and attitudes necessary to comfortably guide patients and families through the nuances of end-of-life care. How to successfully educate clinicians to use such a model earlier is beyond our scope but appears crucial to improving training efforts. Current attempts to teach end-of-life care do not develop and explore the complex integrative domains of relational knowledge^37-40 described in our study. Studies of educational interventions that stress the importance of the relational aspects of end-of-life care appear warranted.

To facilitate skill acquisition, the authors propose the steps diagrammed in the Figure to assist clinicians in providing quality end-of-life care. We believe the time spent determining patient-centered goals and the roles and relationships required by the clinician should at least equal the time spent determining which tests and treatments to provide. At the end of life, when tests and treatments result in fewer benefits and greater risks and burdens, determining patient and family needs and goals becomes increasingly important as management changes from cure to care.

FIGURE
FLOW DIAGRAM FOR CLINICIAN ROLES

Limitations

This study is limited by a study sample of primary care, family practice faculty who geographically represent all practice in the Pacific Northwest and are predominantly Euro-American in ethnicity. The majority are family physicians, with the remainder pediatricians, internists, and a small number of nurse practitioners and physician assistants. Although no significant differences in terms of roles, relationships, and personal meanings between professional groups were noted, whether these findings are transferable to other primary care clinicians in community practice is unknown. Whether non–primary care specialists and clinicians of differing ethnicity or geographic region would respond differently is unknown. Generalizing these findings to non–primary care clinicians and clinicians of color requires further research.

Conclusions

Given the enormity of social and cultural values that make death a taboo topic, it is unclear how any single reform can easily overcome the multiple barriers to improved end-of-life care. Whether it is possible to teach attitudes and values, such as empathy and self-reflection, is uncertain, though promising curricula and research exist.^41-42 If primary care physicians and other professionals improve their knowledge, skills, and attitudes in the delivery of such care, the ultimate effect on improving the experience of patients and families will require further study.

Acknowledgments

We wish to thank the Emily Davie and Joseph S. Kornfeld Foundation and The Nathan Cummings Foundation, New York, which provided generous grants that made our study and article possible. Additionally, we thank Phyllis Silverman, PhD, Harvard University, for her early counsel and advice; Tom Taylor, MD, PhD, University of Washington School of Medicine, for his thoughtful input into the study design, his sage guidance, and his review of the manuscript; Jeanne Quint Benoliel, DNS, Professor Emeritus, University of Washington School of Nursing, for her inspiration and review of the study design and results; and Greg Guldin, PhD, Pacific Lutheran University, for his thoughtful review of the manuscript.

Modern medicine has dramatically improved the length and quality of life for countless persons but has also created problems in integrating these advances with end-of-life care. Although 70% of Americans want to die in their own home supported by family, 74% of Americans currently die in institutions.^1,2 Clearly, there is a disparity between patient preferences and the end-of-life services provided in America.

Most recommendations for the provision of quality end-of-life care are based on unresearched expert opinion that identifies problems and offers solutions.^3-5 Studies have explored communication strategies,⁶ satisfaction levels with care,⁷ perceptions of management issues,⁸ and attitudes toward both death and dying patients.⁹ Yet little research addresses how practicing community clinicians develop the appropriate relationships and integrate the requisite knowledge, skills, and attitudes to provide quality end-of-life care.

The relationship between clinician, patient, and family and the personal meaning of events for each participant greatly influence end-of-life care.¹⁰ Although most commentary on physician roles and relationships in end-of-life care also reflects unresearched expert opinion,^11-13 Steinmetz and Gabel^14,15 theorized a model that was later validated against the experience of practicing family physicians. The role of the clinician in the successful provision of quality care has not been systematically researched and remains a major challenge to improving end-of-life care. We describe an exploratory qualitative study to determine the perceptions of practicing clinicians regarding quality end-of-life care.

Methods

With approval from the Human Subjects Committee of the University of Washington School of Medicine, faculty were recruited from 9 community programs in the Affiliated Family Practice Residency Network of the Department of Family Medicine. Twenty semistructured long interviews were conducted from a convenience sample at 3 residency sites.¹⁶ An additional 8 semistructured interviews and 2 focus groups (10 participants) further explored the data. Thus, a total of 38 clinicians participated in the study. Demographic characteristics of the study cohort have been published previously.¹⁰

We conducted the initial 20 interviews using open-ended questions designed to uncover faculty perceptions of quality end-of-life care. The questions explored may be found in the Table. Nonspecific prompts such as “tell me more” were used to enrich data and avoid interviewer bias. The interview was rehearsed with a medical school faculty expert in ethnographic research, field tested, and adjusted before final use. The interviews were audio-recorded and transcribed into qualitative research software.¹⁷

Throughout the interview process, the investigators met weekly to compare findings, discuss emerging categories, and jointly code the transcripts for model development using the process of grounded theory described by Glaser and Strauss.¹⁸ Common themes, communication and educational issues, and roles and relationships were identified. Data were constantly compared and winnowed to facilitate manageability; data collection and analysis occurred simultaneously.^19,20 Common themes are published elsewhere.¹⁰ We report and discuss the analysis of the data pertaining to clinician roles and relationships.

We employed the following 3 strategies to assess the methodologic validity of the study and face validity of the themes and models emerging from the data analysis.

TABLE
INTERVIEW QUESTIONS

Methodologic validation

The themes, models, and 2 interview transcripts were reviewed by 2 expert consultants with experience in qualitative research on end-of-life care and chronic debilitating disease. They assessed the validity of the research process and model development.^21,22 These experts confirmed the methodologic approach, affirmed that the transcripts supported the coded model, and noted that the results were consistent with their own past research experience.

Face validity faculty interviews

Long interviews were conducted with a convenience sample of 8 faculty at 2 other residency sites. Researchers described the previous interview process and presented a written summary of the roles and relationship model coded from the data. Dialogue was encouraged with nonspecific prompts to enrich the data and avoid interviewer bias. Interviews were recorded, transcribed, and coded by the researchers to assess face validity and further develop the roles and relationship model using a grounded theory approach.

Face validity focus groups

Two focus groups were conducted with a total of 10 faculty at 2 additional sites. The entire model (themes, communication, roles, and relationships) was presented, and dialogue was encouraged. The researchers recorded, transcribed, and reviewed the focus groups to further assess face validity and refine the model.

Content validity

In a manner identical to that of the focus groups, the researchers presented the entire model at 2 of the 3 sites where faculty provided the initial 20 interviews. These sessions were audiotaped, transcribed, and reviewed by the researchers. The model presented was considered a valid, clinically plausible summary of the content of the initial interviews.

Results

Content analysis of the roles and relationships described by the study cohort revealed 3 distinct clinician roles in the provision of end-of-life care that we termed “consultant,” “collaborator,” and “guide.” All roles were viewed as important; none was more valuable than another; and the roles appeared to build on each other, often merging. Roles were implemented fluidly, with clinicians moving from one role to another as circumstances dictated. Although clinicians tended to describe successful cases, they freely discussed the challenges of end-of-life care, noting that less-than-ideal results were frequent. Thus, the models were described as “best practice” rather than routine care.

Consultant

The consultant provides expert medical information to the patient and family based on the biomedical model and the disease process. The power of the consultant role emanates from the clinician’s medical authority and special knowledge. The consultant presents information to help the patient and family understand the diagnosis, prognosis, and treatment. Once this knowledge is understood, the patient and family determine its meaning to them and decide on treatment. If the patient or family cannot decide or understand the medical implications of their particular situation, the consultant decides based on the medical facts.

A pediatrician illustrated the application of the consultant role in 2 situations involving newborns with severe heart anomalies. Here the physician describes the difficulty of using only medical facts to assist the family in deciding whether to discontinue the ventilator: “I remember saying: ‘The only thing that’s keeping the baby alive at this point is the ventilator. How do you feel about making the decision to turn that off?’ I remember it was a very cruel thing to have them make the decision.”

The physician continued: “Next time I said: ‘We have to sit down and talk about how your baby is doing because he died, and I’m going to turn off the machines now.’ I don’t think we should put them in the position of having to make this decision. We should make this decision and tell them what we’re going to do.” In this situation the clinician employed biomedical expertise to determine the appropriate medical care and outcome for the patient and family. In both cases the decision centered on continuing or discontinuing a medical intervention (respirator) and avoided discussion of the parents’ affective experience. Hallmarks of the consultant role include a biomedical focus, disease-centered decision-making processes, and the clinician’s assumption of authority based on biomedical expertise.

Collaborator

The collaborator exchanges information with the patient and family to promote a common understanding of the diagnosis and illness experience, working with them to choose a treatment path. The collaborator incorporates all components of the consultant role and additionally requires the clinician to understand the patient/family experience. The collaborator considers patient and family issues that need to be addressed to understand the medical facts, appreciates their past experiences with serious illness, determines what information would be difficult for them to accept but would benefit them if they were challenged to confront it, and recognizes that the patient and family hold ultimate responsibility for making treatment decisions.

The following quotes exemplify collaborative clinical care. In the first example, the physician steps outside his comfort zone to provide medical care based on the expressed wishes of a 30-year-old man dying of an advanced brain tumor: “The issue for me was letting go of control. He was going 4 hours away to a fishing cabin and going on a boat. I was really nervous…what if he all of a sudden crashed there? It took me a while to get used to the idea that he needs to be able to do what he enjoys doing, and everybody knows that there’s a risk.”

In a second example, the physician describes collaborating with the family of an 85-year-old woman suffering a massive stroke: “Having heard from the family that they understand that Grandma has had a big stroke and isn’t going to survive…then what wishes do Grandma and the family have? Have they ever discussed this sort of situation?”

In both these examples an understanding of the patient/family experience directs the clinician toward appropriate end-of-life care and is a major source of the collaborator’s power.

Guide

As a guide, the clinician actively and personally seeks solutions for the patient based not only on the medical facts (consultant) and the patient’s values and preferences (collaborator) but also on the guide’s greater understanding of the medical context. In essence, the guide not only knows why and where the patient and family prefer to go but also how to get there.

The following quote demonstrates a clinical application of the guide role. The physician first collaborates to understand the resistance of a terminally ill patient to hospice care:

“They had been very resistant to hospice for reasons that I wasn’t quite clear on. It wasn’t until I was in the home and listened to them talk that I realized they viewed hospice as ‘people are giving up on him.’ I think nobody really was talking to the patient about whether he was willing to give up or if he was ready to die.

“Once patient and family concerns were understood, the guide role was employed: I told the wife and family, ‘We’ve maxed out our medical therapies. There’s not much more we can do for him physically, but perhaps there’s something we can do for him spiritually and emotionally.’ When it was presented to them that way, they were much more understanding. We talked about hospice philosophy and looking at death as a part of life, saying: ‘That we’re not going to resuscitate does not mean do not treat.’ That’s where I came in more as the guide and said: ‘This is what I think is reasonable. What are your expectations? What do you want, Billy? And what do you want, as far as [his] wife and kids?”

The guide role requires that the clinician interpret the patient’s experience, integrate this interpretation with the clinician’s understanding of the clinical situation, and make a recommendation based on the guide’s personal and professional understanding of the situation. The power of the guide role emanates from the clinician’s understanding of how to use the medical system to see that the goals of patient and family are realized.

Discussion

The results of our study are consistent with reports over the past 3 decades by researchers, educators, and social critics who have explored how health professionals provide end-of-life care.^23-29 Despite their varying perspectives, all reflect a common theme: the need to provide care based on the unique illness experience and values of the patient and family. The majority of clinicians in our study spoke of the importance of their relationships with patients and families. More than 50% of all interview commentary addressed clinician issues of relationship and personal meaning when providing end-of-life care.

The descriptions of consultant,³⁰ collaborator,^31,32 and guide³³ confirm previous theoretical discussions regarding the nature of roles and relationships between patients and physicians. The guide is the most complex of the roles described by our study cohort and lends itself to ambivalence on the part of clinicians, in light of its potential to be misunderstood as paternalism. Yet, given the vulnerability and dependence of many patients who are terminally ill, the guide provides these patients with structure, safety, support, and care—based on the patients’ values and goals—reminiscent of the role of a nurturing parent.³⁴

The personal nature of the decisions the guide facilitates reflects the reality of medicine as a moral enterprise.^34,35 Having generally witnessed many more deaths than the patient, the guide has knowledge regarding the processes of dying and medical systems that is instrumental in assuring that the desires of the patient and family are realized. A knowledge of patient and family, an appreciation of the futility of the medical situation, and an insight into the process and systems of dying afford the clinician an opportunity to shape the death experience; as Nuland³⁶ described: “Each of us needs a guide who knows us as well as he knows the pathways by which we can approach death.”

Facility with these roles may be a function of personal talent, introspection, and experience. The physicians studied reported that competence with these roles grew over time through delivering endof-life care and learning from patients, families, and other caregivers. Less experienced clinicians tended to describe the consultant and collaborator roles only. Clinicians describing the guide role had been in practice at least a decade and thus were at least 17 years into their training and practice in medicine. Apparently, expertise in all 3 roles requires not only excellent technical diagnostic and treatment skills but also the complex integrated skills of relational knowledge and caring gained through experience.

Our data indicate that relationship, meaning, and roles are primary moderators of the organization of exemplary end-of-life care. Yet, it appears to take nearly 2 decades for physicians to gain the confidence in their knowledge, skills, and attitudes necessary to comfortably guide patients and families through the nuances of end-of-life care. How to successfully educate clinicians to use such a model earlier is beyond our scope but appears crucial to improving training efforts. Current attempts to teach end-of-life care do not develop and explore the complex integrative domains of relational knowledge^37-40 described in our study. Studies of educational interventions that stress the importance of the relational aspects of end-of-life care appear warranted.

To facilitate skill acquisition, the authors propose the steps diagrammed in the Figure to assist clinicians in providing quality end-of-life care. We believe the time spent determining patient-centered goals and the roles and relationships required by the clinician should at least equal the time spent determining which tests and treatments to provide. At the end of life, when tests and treatments result in fewer benefits and greater risks and burdens, determining patient and family needs and goals becomes increasingly important as management changes from cure to care.

FIGURE
FLOW DIAGRAM FOR CLINICIAN ROLES

Limitations

This study is limited by a study sample of primary care, family practice faculty who geographically represent all practice in the Pacific Northwest and are predominantly Euro-American in ethnicity. The majority are family physicians, with the remainder pediatricians, internists, and a small number of nurse practitioners and physician assistants. Although no significant differences in terms of roles, relationships, and personal meanings between professional groups were noted, whether these findings are transferable to other primary care clinicians in community practice is unknown. Whether non–primary care specialists and clinicians of differing ethnicity or geographic region would respond differently is unknown. Generalizing these findings to non–primary care clinicians and clinicians of color requires further research.

Conclusions

Given the enormity of social and cultural values that make death a taboo topic, it is unclear how any single reform can easily overcome the multiple barriers to improved end-of-life care. Whether it is possible to teach attitudes and values, such as empathy and self-reflection, is uncertain, though promising curricula and research exist.^41-42 If primary care physicians and other professionals improve their knowledge, skills, and attitudes in the delivery of such care, the ultimate effect on improving the experience of patients and families will require further study.

Acknowledgments

We wish to thank the Emily Davie and Joseph S. Kornfeld Foundation and The Nathan Cummings Foundation, New York, which provided generous grants that made our study and article possible. Additionally, we thank Phyllis Silverman, PhD, Harvard University, for her early counsel and advice; Tom Taylor, MD, PhD, University of Washington School of Medicine, for his thoughtful input into the study design, his sage guidance, and his review of the manuscript; Jeanne Quint Benoliel, DNS, Professor Emeritus, University of Washington School of Nursing, for her inspiration and review of the study design and results; and Greg Guldin, PhD, Pacific Lutheran University, for his thoughtful review of the manuscript.

Modern medicine has dramatically improved the length and quality of life for countless persons but has also created problems in integrating these advances with end-of-life care. Although 70% of Americans want to die in their own home supported by family, 74% of Americans currently die in institutions.^1,2 Clearly, there is a disparity between patient preferences and the end-of-life services provided in America.

Most recommendations for the provision of quality end-of-life care are based on unresearched expert opinion that identifies problems and offers solutions.^3-5 Studies have explored communication strategies,⁶ satisfaction levels with care,⁷ perceptions of management issues,⁸ and attitudes toward both death and dying patients.⁹ Yet little research addresses how practicing community clinicians develop the appropriate relationships and integrate the requisite knowledge, skills, and attitudes to provide quality end-of-life care.

The relationship between clinician, patient, and family and the personal meaning of events for each participant greatly influence end-of-life care.¹⁰ Although most commentary on physician roles and relationships in end-of-life care also reflects unresearched expert opinion,^11-13 Steinmetz and Gabel^14,15 theorized a model that was later validated against the experience of practicing family physicians. The role of the clinician in the successful provision of quality care has not been systematically researched and remains a major challenge to improving end-of-life care. We describe an exploratory qualitative study to determine the perceptions of practicing clinicians regarding quality end-of-life care.

Methods

With approval from the Human Subjects Committee of the University of Washington School of Medicine, faculty were recruited from 9 community programs in the Affiliated Family Practice Residency Network of the Department of Family Medicine. Twenty semistructured long interviews were conducted from a convenience sample at 3 residency sites.¹⁶ An additional 8 semistructured interviews and 2 focus groups (10 participants) further explored the data. Thus, a total of 38 clinicians participated in the study. Demographic characteristics of the study cohort have been published previously.¹⁰

We conducted the initial 20 interviews using open-ended questions designed to uncover faculty perceptions of quality end-of-life care. The questions explored may be found in the Table. Nonspecific prompts such as “tell me more” were used to enrich data and avoid interviewer bias. The interview was rehearsed with a medical school faculty expert in ethnographic research, field tested, and adjusted before final use. The interviews were audio-recorded and transcribed into qualitative research software.¹⁷

Throughout the interview process, the investigators met weekly to compare findings, discuss emerging categories, and jointly code the transcripts for model development using the process of grounded theory described by Glaser and Strauss.¹⁸ Common themes, communication and educational issues, and roles and relationships were identified. Data were constantly compared and winnowed to facilitate manageability; data collection and analysis occurred simultaneously.^19,20 Common themes are published elsewhere.¹⁰ We report and discuss the analysis of the data pertaining to clinician roles and relationships.

We employed the following 3 strategies to assess the methodologic validity of the study and face validity of the themes and models emerging from the data analysis.

TABLE
INTERVIEW QUESTIONS

Methodologic validation

The themes, models, and 2 interview transcripts were reviewed by 2 expert consultants with experience in qualitative research on end-of-life care and chronic debilitating disease. They assessed the validity of the research process and model development.^21,22 These experts confirmed the methodologic approach, affirmed that the transcripts supported the coded model, and noted that the results were consistent with their own past research experience.

Face validity faculty interviews

Long interviews were conducted with a convenience sample of 8 faculty at 2 other residency sites. Researchers described the previous interview process and presented a written summary of the roles and relationship model coded from the data. Dialogue was encouraged with nonspecific prompts to enrich the data and avoid interviewer bias. Interviews were recorded, transcribed, and coded by the researchers to assess face validity and further develop the roles and relationship model using a grounded theory approach.

Face validity focus groups

Two focus groups were conducted with a total of 10 faculty at 2 additional sites. The entire model (themes, communication, roles, and relationships) was presented, and dialogue was encouraged. The researchers recorded, transcribed, and reviewed the focus groups to further assess face validity and refine the model.

Content validity

In a manner identical to that of the focus groups, the researchers presented the entire model at 2 of the 3 sites where faculty provided the initial 20 interviews. These sessions were audiotaped, transcribed, and reviewed by the researchers. The model presented was considered a valid, clinically plausible summary of the content of the initial interviews.

Results

Content analysis of the roles and relationships described by the study cohort revealed 3 distinct clinician roles in the provision of end-of-life care that we termed “consultant,” “collaborator,” and “guide.” All roles were viewed as important; none was more valuable than another; and the roles appeared to build on each other, often merging. Roles were implemented fluidly, with clinicians moving from one role to another as circumstances dictated. Although clinicians tended to describe successful cases, they freely discussed the challenges of end-of-life care, noting that less-than-ideal results were frequent. Thus, the models were described as “best practice” rather than routine care.

Consultant

The consultant provides expert medical information to the patient and family based on the biomedical model and the disease process. The power of the consultant role emanates from the clinician’s medical authority and special knowledge. The consultant presents information to help the patient and family understand the diagnosis, prognosis, and treatment. Once this knowledge is understood, the patient and family determine its meaning to them and decide on treatment. If the patient or family cannot decide or understand the medical implications of their particular situation, the consultant decides based on the medical facts.

A pediatrician illustrated the application of the consultant role in 2 situations involving newborns with severe heart anomalies. Here the physician describes the difficulty of using only medical facts to assist the family in deciding whether to discontinue the ventilator: “I remember saying: ‘The only thing that’s keeping the baby alive at this point is the ventilator. How do you feel about making the decision to turn that off?’ I remember it was a very cruel thing to have them make the decision.”

The physician continued: “Next time I said: ‘We have to sit down and talk about how your baby is doing because he died, and I’m going to turn off the machines now.’ I don’t think we should put them in the position of having to make this decision. We should make this decision and tell them what we’re going to do.” In this situation the clinician employed biomedical expertise to determine the appropriate medical care and outcome for the patient and family. In both cases the decision centered on continuing or discontinuing a medical intervention (respirator) and avoided discussion of the parents’ affective experience. Hallmarks of the consultant role include a biomedical focus, disease-centered decision-making processes, and the clinician’s assumption of authority based on biomedical expertise.

Collaborator

The collaborator exchanges information with the patient and family to promote a common understanding of the diagnosis and illness experience, working with them to choose a treatment path. The collaborator incorporates all components of the consultant role and additionally requires the clinician to understand the patient/family experience. The collaborator considers patient and family issues that need to be addressed to understand the medical facts, appreciates their past experiences with serious illness, determines what information would be difficult for them to accept but would benefit them if they were challenged to confront it, and recognizes that the patient and family hold ultimate responsibility for making treatment decisions.

The following quotes exemplify collaborative clinical care. In the first example, the physician steps outside his comfort zone to provide medical care based on the expressed wishes of a 30-year-old man dying of an advanced brain tumor: “The issue for me was letting go of control. He was going 4 hours away to a fishing cabin and going on a boat. I was really nervous…what if he all of a sudden crashed there? It took me a while to get used to the idea that he needs to be able to do what he enjoys doing, and everybody knows that there’s a risk.”

In a second example, the physician describes collaborating with the family of an 85-year-old woman suffering a massive stroke: “Having heard from the family that they understand that Grandma has had a big stroke and isn’t going to survive…then what wishes do Grandma and the family have? Have they ever discussed this sort of situation?”

In both these examples an understanding of the patient/family experience directs the clinician toward appropriate end-of-life care and is a major source of the collaborator’s power.

Guide

As a guide, the clinician actively and personally seeks solutions for the patient based not only on the medical facts (consultant) and the patient’s values and preferences (collaborator) but also on the guide’s greater understanding of the medical context. In essence, the guide not only knows why and where the patient and family prefer to go but also how to get there.

The following quote demonstrates a clinical application of the guide role. The physician first collaborates to understand the resistance of a terminally ill patient to hospice care:

“They had been very resistant to hospice for reasons that I wasn’t quite clear on. It wasn’t until I was in the home and listened to them talk that I realized they viewed hospice as ‘people are giving up on him.’ I think nobody really was talking to the patient about whether he was willing to give up or if he was ready to die.

“Once patient and family concerns were understood, the guide role was employed: I told the wife and family, ‘We’ve maxed out our medical therapies. There’s not much more we can do for him physically, but perhaps there’s something we can do for him spiritually and emotionally.’ When it was presented to them that way, they were much more understanding. We talked about hospice philosophy and looking at death as a part of life, saying: ‘That we’re not going to resuscitate does not mean do not treat.’ That’s where I came in more as the guide and said: ‘This is what I think is reasonable. What are your expectations? What do you want, Billy? And what do you want, as far as [his] wife and kids?”

The guide role requires that the clinician interpret the patient’s experience, integrate this interpretation with the clinician’s understanding of the clinical situation, and make a recommendation based on the guide’s personal and professional understanding of the situation. The power of the guide role emanates from the clinician’s understanding of how to use the medical system to see that the goals of patient and family are realized.

Discussion

The results of our study are consistent with reports over the past 3 decades by researchers, educators, and social critics who have explored how health professionals provide end-of-life care.^23-29 Despite their varying perspectives, all reflect a common theme: the need to provide care based on the unique illness experience and values of the patient and family. The majority of clinicians in our study spoke of the importance of their relationships with patients and families. More than 50% of all interview commentary addressed clinician issues of relationship and personal meaning when providing end-of-life care.

The descriptions of consultant,³⁰ collaborator,^31,32 and guide³³ confirm previous theoretical discussions regarding the nature of roles and relationships between patients and physicians. The guide is the most complex of the roles described by our study cohort and lends itself to ambivalence on the part of clinicians, in light of its potential to be misunderstood as paternalism. Yet, given the vulnerability and dependence of many patients who are terminally ill, the guide provides these patients with structure, safety, support, and care—based on the patients’ values and goals—reminiscent of the role of a nurturing parent.³⁴

The personal nature of the decisions the guide facilitates reflects the reality of medicine as a moral enterprise.^34,35 Having generally witnessed many more deaths than the patient, the guide has knowledge regarding the processes of dying and medical systems that is instrumental in assuring that the desires of the patient and family are realized. A knowledge of patient and family, an appreciation of the futility of the medical situation, and an insight into the process and systems of dying afford the clinician an opportunity to shape the death experience; as Nuland³⁶ described: “Each of us needs a guide who knows us as well as he knows the pathways by which we can approach death.”

Facility with these roles may be a function of personal talent, introspection, and experience. The physicians studied reported that competence with these roles grew over time through delivering endof-life care and learning from patients, families, and other caregivers. Less experienced clinicians tended to describe the consultant and collaborator roles only. Clinicians describing the guide role had been in practice at least a decade and thus were at least 17 years into their training and practice in medicine. Apparently, expertise in all 3 roles requires not only excellent technical diagnostic and treatment skills but also the complex integrated skills of relational knowledge and caring gained through experience.

Our data indicate that relationship, meaning, and roles are primary moderators of the organization of exemplary end-of-life care. Yet, it appears to take nearly 2 decades for physicians to gain the confidence in their knowledge, skills, and attitudes necessary to comfortably guide patients and families through the nuances of end-of-life care. How to successfully educate clinicians to use such a model earlier is beyond our scope but appears crucial to improving training efforts. Current attempts to teach end-of-life care do not develop and explore the complex integrative domains of relational knowledge^37-40 described in our study. Studies of educational interventions that stress the importance of the relational aspects of end-of-life care appear warranted.

To facilitate skill acquisition, the authors propose the steps diagrammed in the Figure to assist clinicians in providing quality end-of-life care. We believe the time spent determining patient-centered goals and the roles and relationships required by the clinician should at least equal the time spent determining which tests and treatments to provide. At the end of life, when tests and treatments result in fewer benefits and greater risks and burdens, determining patient and family needs and goals becomes increasingly important as management changes from cure to care.

FIGURE
FLOW DIAGRAM FOR CLINICIAN ROLES

Limitations

This study is limited by a study sample of primary care, family practice faculty who geographically represent all practice in the Pacific Northwest and are predominantly Euro-American in ethnicity. The majority are family physicians, with the remainder pediatricians, internists, and a small number of nurse practitioners and physician assistants. Although no significant differences in terms of roles, relationships, and personal meanings between professional groups were noted, whether these findings are transferable to other primary care clinicians in community practice is unknown. Whether non–primary care specialists and clinicians of differing ethnicity or geographic region would respond differently is unknown. Generalizing these findings to non–primary care clinicians and clinicians of color requires further research.

Conclusions

Given the enormity of social and cultural values that make death a taboo topic, it is unclear how any single reform can easily overcome the multiple barriers to improved end-of-life care. Whether it is possible to teach attitudes and values, such as empathy and self-reflection, is uncertain, though promising curricula and research exist.^41-42 If primary care physicians and other professionals improve their knowledge, skills, and attitudes in the delivery of such care, the ultimate effect on improving the experience of patients and families will require further study.

Acknowledgments

We wish to thank the Emily Davie and Joseph S. Kornfeld Foundation and The Nathan Cummings Foundation, New York, which provided generous grants that made our study and article possible. Additionally, we thank Phyllis Silverman, PhD, Harvard University, for her early counsel and advice; Tom Taylor, MD, PhD, University of Washington School of Medicine, for his thoughtful input into the study design, his sage guidance, and his review of the manuscript; Jeanne Quint Benoliel, DNS, Professor Emeritus, University of Washington School of Nursing, for her inspiration and review of the study design and results; and Greg Guldin, PhD, Pacific Lutheran University, for his thoughtful review of the manuscript.

I was confused by Dr Gross’s case discussion¹ in The Art of Medicine until I realized that his moral dilemma was not whether he should use warfarin to prevent strokes in an elderly patient with atrial fibrillation and gastrointestinal (GI) bleeding. His real dilemma was the conflict between what our medical school training tells us to do and what our hearts tell us to do. Fortunately, palliative medicine is now teaching us that we can follow our hearts and let go of these dilemmas. I believe the case presented is a perfect example.

In the case discussion a woman, Mrs Lopez, had multiple strokes. GI bleeding was a complication of warfarin therapy and a possible malignancy. Dr Gross’s heart told him not to be too aggressive in his testing and treatments. Mrs Lopez and her family agreed “not to embark on an elaborate search.” What was missed, perhaps, was the opportunity to more fully discuss her directives for care. What was also missed at the time was the identification that Mrs Lopez was terminally ill. She was terminal because her life was clearly coming to an end. The ability to cure and fix was no longer realistic. Some of her directives were clear: Dr Gross mentioned that Mrs Lopez accepted life’s uncertainties. Her philosophy seemed to be summed up in the phrase: What will be, will be. So a reasonable choice was palliative care, with an aggressive focus on keeping her comfortable and controlling distressing symptoms. GI bleeding is emotionally distressing and can cause weakness, fatigue, dyspnea, and chest pain, if coronary artery disease is present. The moral principle of “do no harm” would uphold the decision not to give her anything (such as warfarin) that increased the risk of GI bleeding. Aspirin might have been included in this decision, especially given aspirin’s limited possible benefit. Transfusions were appropriate treatment for the anemia resulting from the bleeding. Repeated strokes confirmed the terminal nature of her condition. She might die with the next stroke or she might live to have another one. As Mrs Lopez said, “Si dios quiere.”

The goals to reduce suffering, to control her symptoms, and to maximize her comfort could have made Dr Gross’s dilemma disappear. With these goals in mind he might have reacted to the next stroke in a different way. The goals and the ethical principles of beneficence and doing no harm would have supported his decision not to admit her to the intensive care unit again. The time for cure and fix was past. Central lines and nasogastric tubes were inappropriate for the patient, who was terminal and now hoped for a peaceful, comfortable, dignified death. Dr Gross did come to this realization. But he could have avoided the emotionally charged decision to discontinue Mrs Lopez’s treatments by not starting them.

Death need not be the enemy physicians hide from at all costs. Death is part of life. Physicians can help their patients and families a great deal by recognizing when the dying process is starting. The dilemma described by Dr Gross reflects physicians’ reluctance to let go of curing and fixing as the only goals of care. The principles and philosophy of palliative medicine can help physicians get out of apparent moral clinical dilemmas, especially those based on different treatments aimed at fixing what cannot be fixed. Palliative medicine goals remove the dilemma by upholding the physician’s role in looking at the whole patient, the whole picture, and recognizing the need for appropriate end-of-life care. Planning a good peaceful death may create the perfect outcome.

George F. Davis, MD
Albany, New York

REFERENCE

• Gross PR. A rock and a hard place. J Fam Pract 2000; 49:863-64.

The preceding letter was referred to Dr Gross who responded as follows:

I applaud Dr Davis for drawing attention to the value of palliative care. I’m afraid, however, that in his eagerness to prescribe palliative care to resolve the quandry I described, he missed the exact location of the “rock and hard place” I spoke of.

My dilemma was not so much a conflict between “what our medical school training tells us to do and what our hearts tell us to do”; rather, it was an uncertainty within my own heart and mind as to which course of action would lead to a better life for my patient.

In this situation—an elderly patient with atrial fibrillation and gastrointestinal bleeding—both heart and medical training speak with 1 voice: Do not anticoagulate. The dilemma is that this clear choice carries its own substantial risk—a stroke that can cripple if it does not kill. After she suffered 1 stroke, then 2, I worried that Mrs Lopez’s next event could lead not only to death but to months of hemiparesis before she perished—and wondered whether I was doing all I could to avert that disaster. Therefore, the thrust of my piece was that clinical uncertainty leads us to second-guess high-stakes decisions, and even wise choices can lead to unhappy outcomes.

I wholeheartedly agree with Dr Davis that palliative care should become our focus as patients near the end of their lives. Perhaps I could have been more exhaustive in my discussions with Mrs Lopez and her family. They did know that she had a terminal condition; we had discussed her wishes regarding resuscitation. Had I recognized at the moment of her last stroke that her situation would quickly deteriorate it would have been easy to spare her the nasogastric tube and central line. Given that she had just bounced back from a similar episode 2 weeks before, I found that call more difficult to make.

Our work as physicians offers us limitless opportunities to establish high standards for ourselves as we minister to our patients. Having set the bar, we may find that reaching it in every situation can be extraordinarily difficult. Just as we look for ways to comfort our patients, I hope that physicians can find ways to share, listen, and comfort one another when we try our best and still fall short.

Paul R. Gross, MD
St. Joseph’s Medical Center
Yonkers, New York

I was confused by Dr Gross’s case discussion¹ in The Art of Medicine until I realized that his moral dilemma was not whether he should use warfarin to prevent strokes in an elderly patient with atrial fibrillation and gastrointestinal (GI) bleeding. His real dilemma was the conflict between what our medical school training tells us to do and what our hearts tell us to do. Fortunately, palliative medicine is now teaching us that we can follow our hearts and let go of these dilemmas. I believe the case presented is a perfect example.

In the case discussion a woman, Mrs Lopez, had multiple strokes. GI bleeding was a complication of warfarin therapy and a possible malignancy. Dr Gross’s heart told him not to be too aggressive in his testing and treatments. Mrs Lopez and her family agreed “not to embark on an elaborate search.” What was missed, perhaps, was the opportunity to more fully discuss her directives for care. What was also missed at the time was the identification that Mrs Lopez was terminally ill. She was terminal because her life was clearly coming to an end. The ability to cure and fix was no longer realistic. Some of her directives were clear: Dr Gross mentioned that Mrs Lopez accepted life’s uncertainties. Her philosophy seemed to be summed up in the phrase: What will be, will be. So a reasonable choice was palliative care, with an aggressive focus on keeping her comfortable and controlling distressing symptoms. GI bleeding is emotionally distressing and can cause weakness, fatigue, dyspnea, and chest pain, if coronary artery disease is present. The moral principle of “do no harm” would uphold the decision not to give her anything (such as warfarin) that increased the risk of GI bleeding. Aspirin might have been included in this decision, especially given aspirin’s limited possible benefit. Transfusions were appropriate treatment for the anemia resulting from the bleeding. Repeated strokes confirmed the terminal nature of her condition. She might die with the next stroke or she might live to have another one. As Mrs Lopez said, “Si dios quiere.”

The goals to reduce suffering, to control her symptoms, and to maximize her comfort could have made Dr Gross’s dilemma disappear. With these goals in mind he might have reacted to the next stroke in a different way. The goals and the ethical principles of beneficence and doing no harm would have supported his decision not to admit her to the intensive care unit again. The time for cure and fix was past. Central lines and nasogastric tubes were inappropriate for the patient, who was terminal and now hoped for a peaceful, comfortable, dignified death. Dr Gross did come to this realization. But he could have avoided the emotionally charged decision to discontinue Mrs Lopez’s treatments by not starting them.

Death need not be the enemy physicians hide from at all costs. Death is part of life. Physicians can help their patients and families a great deal by recognizing when the dying process is starting. The dilemma described by Dr Gross reflects physicians’ reluctance to let go of curing and fixing as the only goals of care. The principles and philosophy of palliative medicine can help physicians get out of apparent moral clinical dilemmas, especially those based on different treatments aimed at fixing what cannot be fixed. Palliative medicine goals remove the dilemma by upholding the physician’s role in looking at the whole patient, the whole picture, and recognizing the need for appropriate end-of-life care. Planning a good peaceful death may create the perfect outcome.

George F. Davis, MD
Albany, New York

REFERENCE

• Gross PR. A rock and a hard place. J Fam Pract 2000; 49:863-64.

The preceding letter was referred to Dr Gross who responded as follows:

I applaud Dr Davis for drawing attention to the value of palliative care. I’m afraid, however, that in his eagerness to prescribe palliative care to resolve the quandry I described, he missed the exact location of the “rock and hard place” I spoke of.

My dilemma was not so much a conflict between “what our medical school training tells us to do and what our hearts tell us to do”; rather, it was an uncertainty within my own heart and mind as to which course of action would lead to a better life for my patient.

In this situation—an elderly patient with atrial fibrillation and gastrointestinal bleeding—both heart and medical training speak with 1 voice: Do not anticoagulate. The dilemma is that this clear choice carries its own substantial risk—a stroke that can cripple if it does not kill. After she suffered 1 stroke, then 2, I worried that Mrs Lopez’s next event could lead not only to death but to months of hemiparesis before she perished—and wondered whether I was doing all I could to avert that disaster. Therefore, the thrust of my piece was that clinical uncertainty leads us to second-guess high-stakes decisions, and even wise choices can lead to unhappy outcomes.

I wholeheartedly agree with Dr Davis that palliative care should become our focus as patients near the end of their lives. Perhaps I could have been more exhaustive in my discussions with Mrs Lopez and her family. They did know that she had a terminal condition; we had discussed her wishes regarding resuscitation. Had I recognized at the moment of her last stroke that her situation would quickly deteriorate it would have been easy to spare her the nasogastric tube and central line. Given that she had just bounced back from a similar episode 2 weeks before, I found that call more difficult to make.

Our work as physicians offers us limitless opportunities to establish high standards for ourselves as we minister to our patients. Having set the bar, we may find that reaching it in every situation can be extraordinarily difficult. Just as we look for ways to comfort our patients, I hope that physicians can find ways to share, listen, and comfort one another when we try our best and still fall short.

Paul R. Gross, MD
St. Joseph’s Medical Center
Yonkers, New York

I was confused by Dr Gross’s case discussion¹ in The Art of Medicine until I realized that his moral dilemma was not whether he should use warfarin to prevent strokes in an elderly patient with atrial fibrillation and gastrointestinal (GI) bleeding. His real dilemma was the conflict between what our medical school training tells us to do and what our hearts tell us to do. Fortunately, palliative medicine is now teaching us that we can follow our hearts and let go of these dilemmas. I believe the case presented is a perfect example.

In the case discussion a woman, Mrs Lopez, had multiple strokes. GI bleeding was a complication of warfarin therapy and a possible malignancy. Dr Gross’s heart told him not to be too aggressive in his testing and treatments. Mrs Lopez and her family agreed “not to embark on an elaborate search.” What was missed, perhaps, was the opportunity to more fully discuss her directives for care. What was also missed at the time was the identification that Mrs Lopez was terminally ill. She was terminal because her life was clearly coming to an end. The ability to cure and fix was no longer realistic. Some of her directives were clear: Dr Gross mentioned that Mrs Lopez accepted life’s uncertainties. Her philosophy seemed to be summed up in the phrase: What will be, will be. So a reasonable choice was palliative care, with an aggressive focus on keeping her comfortable and controlling distressing symptoms. GI bleeding is emotionally distressing and can cause weakness, fatigue, dyspnea, and chest pain, if coronary artery disease is present. The moral principle of “do no harm” would uphold the decision not to give her anything (such as warfarin) that increased the risk of GI bleeding. Aspirin might have been included in this decision, especially given aspirin’s limited possible benefit. Transfusions were appropriate treatment for the anemia resulting from the bleeding. Repeated strokes confirmed the terminal nature of her condition. She might die with the next stroke or she might live to have another one. As Mrs Lopez said, “Si dios quiere.”

The goals to reduce suffering, to control her symptoms, and to maximize her comfort could have made Dr Gross’s dilemma disappear. With these goals in mind he might have reacted to the next stroke in a different way. The goals and the ethical principles of beneficence and doing no harm would have supported his decision not to admit her to the intensive care unit again. The time for cure and fix was past. Central lines and nasogastric tubes were inappropriate for the patient, who was terminal and now hoped for a peaceful, comfortable, dignified death. Dr Gross did come to this realization. But he could have avoided the emotionally charged decision to discontinue Mrs Lopez’s treatments by not starting them.

Death need not be the enemy physicians hide from at all costs. Death is part of life. Physicians can help their patients and families a great deal by recognizing when the dying process is starting. The dilemma described by Dr Gross reflects physicians’ reluctance to let go of curing and fixing as the only goals of care. The principles and philosophy of palliative medicine can help physicians get out of apparent moral clinical dilemmas, especially those based on different treatments aimed at fixing what cannot be fixed. Palliative medicine goals remove the dilemma by upholding the physician’s role in looking at the whole patient, the whole picture, and recognizing the need for appropriate end-of-life care. Planning a good peaceful death may create the perfect outcome.

George F. Davis, MD
Albany, New York

REFERENCE

• Gross PR. A rock and a hard place. J Fam Pract 2000; 49:863-64.

The preceding letter was referred to Dr Gross who responded as follows:

I applaud Dr Davis for drawing attention to the value of palliative care. I’m afraid, however, that in his eagerness to prescribe palliative care to resolve the quandry I described, he missed the exact location of the “rock and hard place” I spoke of.

My dilemma was not so much a conflict between “what our medical school training tells us to do and what our hearts tell us to do”; rather, it was an uncertainty within my own heart and mind as to which course of action would lead to a better life for my patient.

In this situation—an elderly patient with atrial fibrillation and gastrointestinal bleeding—both heart and medical training speak with 1 voice: Do not anticoagulate. The dilemma is that this clear choice carries its own substantial risk—a stroke that can cripple if it does not kill. After she suffered 1 stroke, then 2, I worried that Mrs Lopez’s next event could lead not only to death but to months of hemiparesis before she perished—and wondered whether I was doing all I could to avert that disaster. Therefore, the thrust of my piece was that clinical uncertainty leads us to second-guess high-stakes decisions, and even wise choices can lead to unhappy outcomes.

I wholeheartedly agree with Dr Davis that palliative care should become our focus as patients near the end of their lives. Perhaps I could have been more exhaustive in my discussions with Mrs Lopez and her family. They did know that she had a terminal condition; we had discussed her wishes regarding resuscitation. Had I recognized at the moment of her last stroke that her situation would quickly deteriorate it would have been easy to spare her the nasogastric tube and central line. Given that she had just bounced back from a similar episode 2 weeks before, I found that call more difficult to make.

Our work as physicians offers us limitless opportunities to establish high standards for ourselves as we minister to our patients. Having set the bar, we may find that reaching it in every situation can be extraordinarily difficult. Just as we look for ways to comfort our patients, I hope that physicians can find ways to share, listen, and comfort one another when we try our best and still fall short.

Paul R. Gross, MD
St. Joseph’s Medical Center
Yonkers, New York

BACKGROUND: Acute mountain sickness results in symptoms such as headache, loss of appetite, nausea, fatigue, insomnia, and increased mortality in the first few days of exertion at high altitudes. Acetazolamide and dexamethasone have been recommended as treatments to prevent acute mountain sickness, although their efficacy and tolerability have not been well established.

DATA SOURCE: Thirty-three clinical trials comparing any pharmacologic treatment with placebo for the prevention of acute mountain sickness were identified through a comprehensive search of MEDLINE, EMBASE, the Cochrane Library, a high altitude bibliography Web site, and the bibliographies of retrieved reports and review articles. Of these trials, 18 reported complete prevention of acute mountain sickness as an outcome and as related to ascents to 4050 to 5890 meters (13300 to 19300 feet). These were included in a qualitative analysis of efficacy for any pharmacologic treatment (all 18 trials) and statistical meta-analyses for dexamethasone (8 trials representing 161 subjects) and acetazolamide (9 trials representing 295 subjects).

STUDY DESIGN AND VALIDITY: Only one author determined whether studies met inclusion criteria. All 3 authors read the included articles for validity, applying the 3-item 5-point Oxford score. The mean quality score was 3 (range=2-5). A total of 17 of the 33 trials used an adequate method of blinding. Validity was not a criteria for inclusion in the review. Following the qualitative analysis of efficacy, relative benefit (relative risk [RR] of prevention of mountain sickness) and numbers needed to treat (NNTs) were calculated for 2 interventions, dexamethasone and acetazolamide. Although the studies were rated for validity using previously tested criteria, no information on study limitations is given with which to judge the direction of effect. Using more than one author to determine eligibility would have minimized investigator bias. Although included studies varied in their definitions of acute mountain sickness, the results were homogenous. The investigators limited the outcome to complete prevention of symptoms. This conservative approach would underestimate the efficacy of the treatments making any significant benefit more likely to be real. Unfortunately, no information is provided on the subjects in the trials, including information such as age, sex, training, smoking status, and other risk factors for acute mountain sickness. This information would have helped generalize the study results.

OUTCOMES MEASURED: The primary outcome measured was complete prevention of acute mountain sickness according to the definition used in each of the original studies. Although reported for some of the studies, the reduction in symptom severity was not considered a criterion for efficacy in this study. Secondary outcomes included prevention of headache, nausea, insomnia, and dizziness, and the development of adverse drug reactions.

RESULTS: In 8 trials of dexamethasone (n=161) and 9 of acetazolamide (n=295), all with exposure above 4000 meters, single daily doses of dexamethasone 8 mg, 12 mg, and 16 mg or acetazolamide 750 mg were all more efficacious than placebo (NNT=3 for both). Smaller doses (dexamethasone 0.5 and 2 mg in one trial, acetazolamide 500 mg in 3 trials) did not prevent mountain sickness over placebo. Five trials of acetazolamide reported symptom end points including insomnia, headache, nausea, and dizziness; all results favored acetazolamide (NNT=3-6). Both dexamethasone and acetazolamide demonstrated a trend toward greater efficacy (lower NNT) with faster ascent rates. There was no relative benefit (RR for prophylaxis=1) for trials with ascent rates below approximately 500 meters per day; but there was a five- to six-fold relative benefit for prophylaxis at the highest ascent rates. Regarding adverse effects, subjects in 2 of 5 studies were more likely to experience depression on weaning of dexamethasone (overall RR=4.5). Acetazolamide was associated with polyuria (RR=4.2) and paraesthesia (RR=4.0).

BACKGROUND: Acute mountain sickness results in symptoms such as headache, loss of appetite, nausea, fatigue, insomnia, and increased mortality in the first few days of exertion at high altitudes. Acetazolamide and dexamethasone have been recommended as treatments to prevent acute mountain sickness, although their efficacy and tolerability have not been well established.

DATA SOURCE: Thirty-three clinical trials comparing any pharmacologic treatment with placebo for the prevention of acute mountain sickness were identified through a comprehensive search of MEDLINE, EMBASE, the Cochrane Library, a high altitude bibliography Web site, and the bibliographies of retrieved reports and review articles. Of these trials, 18 reported complete prevention of acute mountain sickness as an outcome and as related to ascents to 4050 to 5890 meters (13300 to 19300 feet). These were included in a qualitative analysis of efficacy for any pharmacologic treatment (all 18 trials) and statistical meta-analyses for dexamethasone (8 trials representing 161 subjects) and acetazolamide (9 trials representing 295 subjects).

STUDY DESIGN AND VALIDITY: Only one author determined whether studies met inclusion criteria. All 3 authors read the included articles for validity, applying the 3-item 5-point Oxford score. The mean quality score was 3 (range=2-5). A total of 17 of the 33 trials used an adequate method of blinding. Validity was not a criteria for inclusion in the review. Following the qualitative analysis of efficacy, relative benefit (relative risk [RR] of prevention of mountain sickness) and numbers needed to treat (NNTs) were calculated for 2 interventions, dexamethasone and acetazolamide. Although the studies were rated for validity using previously tested criteria, no information on study limitations is given with which to judge the direction of effect. Using more than one author to determine eligibility would have minimized investigator bias. Although included studies varied in their definitions of acute mountain sickness, the results were homogenous. The investigators limited the outcome to complete prevention of symptoms. This conservative approach would underestimate the efficacy of the treatments making any significant benefit more likely to be real. Unfortunately, no information is provided on the subjects in the trials, including information such as age, sex, training, smoking status, and other risk factors for acute mountain sickness. This information would have helped generalize the study results.

OUTCOMES MEASURED: The primary outcome measured was complete prevention of acute mountain sickness according to the definition used in each of the original studies. Although reported for some of the studies, the reduction in symptom severity was not considered a criterion for efficacy in this study. Secondary outcomes included prevention of headache, nausea, insomnia, and dizziness, and the development of adverse drug reactions.

RESULTS: In 8 trials of dexamethasone (n=161) and 9 of acetazolamide (n=295), all with exposure above 4000 meters, single daily doses of dexamethasone 8 mg, 12 mg, and 16 mg or acetazolamide 750 mg were all more efficacious than placebo (NNT=3 for both). Smaller doses (dexamethasone 0.5 and 2 mg in one trial, acetazolamide 500 mg in 3 trials) did not prevent mountain sickness over placebo. Five trials of acetazolamide reported symptom end points including insomnia, headache, nausea, and dizziness; all results favored acetazolamide (NNT=3-6). Both dexamethasone and acetazolamide demonstrated a trend toward greater efficacy (lower NNT) with faster ascent rates. There was no relative benefit (RR for prophylaxis=1) for trials with ascent rates below approximately 500 meters per day; but there was a five- to six-fold relative benefit for prophylaxis at the highest ascent rates. Regarding adverse effects, subjects in 2 of 5 studies were more likely to experience depression on weaning of dexamethasone (overall RR=4.5). Acetazolamide was associated with polyuria (RR=4.2) and paraesthesia (RR=4.0).

BACKGROUND: Acute mountain sickness results in symptoms such as headache, loss of appetite, nausea, fatigue, insomnia, and increased mortality in the first few days of exertion at high altitudes. Acetazolamide and dexamethasone have been recommended as treatments to prevent acute mountain sickness, although their efficacy and tolerability have not been well established.

DATA SOURCE: Thirty-three clinical trials comparing any pharmacologic treatment with placebo for the prevention of acute mountain sickness were identified through a comprehensive search of MEDLINE, EMBASE, the Cochrane Library, a high altitude bibliography Web site, and the bibliographies of retrieved reports and review articles. Of these trials, 18 reported complete prevention of acute mountain sickness as an outcome and as related to ascents to 4050 to 5890 meters (13300 to 19300 feet). These were included in a qualitative analysis of efficacy for any pharmacologic treatment (all 18 trials) and statistical meta-analyses for dexamethasone (8 trials representing 161 subjects) and acetazolamide (9 trials representing 295 subjects).

STUDY DESIGN AND VALIDITY: Only one author determined whether studies met inclusion criteria. All 3 authors read the included articles for validity, applying the 3-item 5-point Oxford score. The mean quality score was 3 (range=2-5). A total of 17 of the 33 trials used an adequate method of blinding. Validity was not a criteria for inclusion in the review. Following the qualitative analysis of efficacy, relative benefit (relative risk [RR] of prevention of mountain sickness) and numbers needed to treat (NNTs) were calculated for 2 interventions, dexamethasone and acetazolamide. Although the studies were rated for validity using previously tested criteria, no information on study limitations is given with which to judge the direction of effect. Using more than one author to determine eligibility would have minimized investigator bias. Although included studies varied in their definitions of acute mountain sickness, the results were homogenous. The investigators limited the outcome to complete prevention of symptoms. This conservative approach would underestimate the efficacy of the treatments making any significant benefit more likely to be real. Unfortunately, no information is provided on the subjects in the trials, including information such as age, sex, training, smoking status, and other risk factors for acute mountain sickness. This information would have helped generalize the study results.

OUTCOMES MEASURED: The primary outcome measured was complete prevention of acute mountain sickness according to the definition used in each of the original studies. Although reported for some of the studies, the reduction in symptom severity was not considered a criterion for efficacy in this study. Secondary outcomes included prevention of headache, nausea, insomnia, and dizziness, and the development of adverse drug reactions.

RESULTS: In 8 trials of dexamethasone (n=161) and 9 of acetazolamide (n=295), all with exposure above 4000 meters, single daily doses of dexamethasone 8 mg, 12 mg, and 16 mg or acetazolamide 750 mg were all more efficacious than placebo (NNT=3 for both). Smaller doses (dexamethasone 0.5 and 2 mg in one trial, acetazolamide 500 mg in 3 trials) did not prevent mountain sickness over placebo. Five trials of acetazolamide reported symptom end points including insomnia, headache, nausea, and dizziness; all results favored acetazolamide (NNT=3-6). Both dexamethasone and acetazolamide demonstrated a trend toward greater efficacy (lower NNT) with faster ascent rates. There was no relative benefit (RR for prophylaxis=1) for trials with ascent rates below approximately 500 meters per day; but there was a five- to six-fold relative benefit for prophylaxis at the highest ascent rates. Regarding adverse effects, subjects in 2 of 5 studies were more likely to experience depression on weaning of dexamethasone (overall RR=4.5). Acetazolamide was associated with polyuria (RR=4.2) and paraesthesia (RR=4.0).