Secondary Logo

Journal Logo

“It was haunting …”: Physicians’ Descriptions of Emotionally Powerful Patient Deaths

Jackson, Vicki A. MD, MPH; Sullivan, Amy M. EdD; Gadmer, Nina M. MHA; Seltzer, Deborah; Mitchell, Ann M. PhD, RN; Lakoma, Mathew D.; Arnold, Robert M. MD; Block, Susan D. MD

Featured Topic Research Report

Purpose To understand the emotional experiences of physicians who care for dying patients and to identify educational opportunities for improving patient care and physician well-being.

Method Between 1999–2001, physicians at two quaternary care medical centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania, participated in 90-minute, semistructured personal interviews on their most emotionally powerful patient death. Quantitative data was obtained through face-to-face surveys rated on ten-point scales that asked physicians about emotional characteristics of and emotional responses to the death. In the qualitative portion of the survey, physicians were asked to describe the details of the most emotionally powerful patient death, the types and sequence of their emotional reactions, their methods of coping, and subsequent changes in behavior.

Results Physicians had powerful experiences with death during all stages of their careers. Experiences with patient death generally fit into one of three types: “good,” “overtreated,” or “shocking/unexpected.” Housestaff often described coping in isolation with the disturbing emotions generated in the care of dying patients. Physicians learned how to care for and cope with dying patients from their experiences with patients whose deaths were most emotionally powerful and reported changes in their clinical behavior and career paths as a result.

Conclusions Physicians’ emotional reactions to patient death can affect patient care and the personal lives of physicians. Supervising physicians have an opportunity to improve both the care of dying patients and house-staff coping with these deaths by using the “teachable moments” that are present for trainees as they care for the dying.

Dr. Jackson is instructor of medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts.

Dr. Sullivan is instructor of psychiatry, The Dana-Farber Cancer Institute, and the Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts.

Dr. Block is chief, Division of Psychosocial Oncology and Palliative Care, Department of Medical Oncology; co-director, Center for Palliative Care; and associate professor of psychiatry (medicine), Harvard Medical School, Boston, Massachusetts.

Ms. Gadmer is a nurse practitioner student, Massachusetts General Hospital Institute of Health Professions, Boston, Massachusetts.

Ms. Seltzer is research administrator, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania.

Dr. Mitchell is assistant professor of nursing, University of Pittsburgh School of Nursing, Pittsburgh, Pennsylvania.

Mr. Lakoma is clinical research coordinator, Dana-Farber Cancer Institute, Boston Massachusetts.

Dr. Arnold is professor of medicine, University of Pittsburgh Medical Center, University of Pittsburgh, Pittsburgh, Pennsylvania.

Please see the end of this report for information about the authors.

Correspondence should be addressed to Dr. Jackson, Massachusetts General Hospital, 55 Fruit Street, Founders 600, Boston, MA 02114; e-mail: 〈〉.

Clinicians who care for dying patients experience significant levels of stress.1 Many factors contribute to this stress including inadequate training in both communication skills and medical management issues common to the care of dying patients.2,3 Furthermore, although caring for patients who die is an intrinsic part of being a physician, health care providers work in a medical culture where death is often viewed as a physician’s failure.4,5

Caring for dying patients forces physicians to confront their own mortality and personal experiences with death, dying, and life-threatening illness.4 Unlike psychiatrists and psychologists, internists and oncologists are not trained to recognize and discuss the powerful emotional reactions evoked by caring for patients. As a result, the distressing emotions experienced in the care of dying patients are often unrecognized and unexpressed.5–7

Recent studies have identified the negative consequences of coping with the stress of caring for the dying. Whippen and Canellos3 studied burnout in oncologists and found that 56% of respondents reported frustration and a sense of personal failure in their work. The authors conclude that coping with the challenges of providing end-of-life care was the single most important qualitative factor related to burnout. Recently, Shanafelt and colleagues8 found burnout was common among residents, and associated with suboptimal patient care attitudes, including having little emotional reaction to the death of a patient.

Little is known about physicians’ emotional reactions to the deaths of their patients. We chose to study physicians’ descriptions of their most emotionally powerful patient death because physicians remember particular patient deaths, both good and bad, often as a “critical incident” in their career.9 The study of physicians’ “critical incidents” is particularly useful in understanding important interactions with patients that can shape how physicians care for patients in the future. By examining the characteristics of the most emotionally powerful deaths, we sought to understand the emotional experiences of physicians who care for dying patients and to identify educational interventions to improve patient care and physician well-being.

Back to Top | Article Outline


We chose to use both quantitative and qualitative methods to provide a richer description of the physicians’ knowledge, beliefs, and experiences.

Back to Top | Article Outline


Our study of physicians’ descriptions of their most emotionally powerful patient death was part of a larger study of physicians’ emotional reactions to randomly selected patient deaths on the medical services at two quaternary care medical centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania.10 We reviewed charts weekly to obtain information on all inpatients who had died on the internal medicine services during the previous week. Participants were physicians who had cared for the dying patients at the time of death. Between June 1999 and September 2001, we scheduled all physicians who agreed to participate for a 90-minute, semistructured personal interview regarding their experiences caring for their most recent patient who died (randomly selected) and their most emotionally powerful patient death.

Both hospitals’ institutional review boards approved the study. To promote physicians’ willingness to honestly relate the story of their most emotionally powerful death, we obtained a certificate of confidentiality from the National Institutes of Health to protect respondents from any potential liability associated with their disclosures.

Back to Top | Article Outline

Instrument development

We developed our survey instrument, including the adjective checklist, after a review of the existing literature. Additionally, we conducted two focus groups of medical residents in which they described their experiences with patients’ deaths and we analyzed the transcripts. The instrument had closed-ended questions and open-ended questions with follow-up probes to assist physicians in relating the story of their most emotionally powerful patient death.

Back to Top | Article Outline


Quantitative interview.

We asked physicians to rate their experiences using a ten-point scale (1 = least, 10 = most) and to answer closed-ended questions on the physicians’ phase of professional development in which the death occurred, the length of time the physician provided care, and the time of the physician’s last personal contact with patient.

The interviews were conducted by one of us (NG) and by two other interviewers who are not authors of this article. Interviewers asked participants to choose from a checklist of adjectives one that best described their experience with their most recent and most emotionally powerful patient death (mundane, timely, spiritual, isolated, violent, natural, degrading, full of suffering, avoidable, dignified, peaceful, overtreated, gory, routine, a sense of closure, time-consuming, chaotic, accidental, unexpected) and how the patient death made them feel (sad, guilty, tired, depressed, stressed, disgusted, moved, relieved, competent, proud, shocked, uncertain, worried, inadequate, satisfied, frustrated, conflicted).

Back to Top | Article Outline

Qualitative interview.

We used semistructured, in-depth interviews to collect the qualitative data because this methodology is especially useful in understanding participants’ beliefs, knowledge, and experience.11–13 We explored aspects of each physician’s most emotionally powerful death including the cause of the patient’s death, the types and sequence of emotional reactions to the death, the coping response, subsequent changes in behavior, level of identification, nature and extent of communication about death and dying and any regrets with regard to the care of the patient. All interviews were audio taped and transcribed.

Back to Top | Article Outline



We completed univariate analyses using the Statistical Analysis System Version 8.0 (SAS Institute GmbH, Heidelberg, Germany) for all structured questions. We used general linear modeling for analysis of variance to test for differences in mean values of the ten-point scales across level of training and gender. Sample size precluded testing for significance among the subsample of physicians whose transcripts we qualitatively analyzed.

We used principal components analysis (PCA) to reduce the adjective checklists to key thematic clusters. PCA is a technique that is useful in gaining a better understanding of interrelationships among variables and to discover underlying structure by combining correlated variables into a single component.14 Adjectives with eigenvalues greater than 1 were included in the principal component. Eigenvalues represent the percentage variance that can be attributed to each variable. We removed those factors that accounted for the least amount of variance in our determination of the final principal components for each group. The process concluded when the factors retained accounted for as much total variance as possible and the eigenvectors revealed a positive close correlation.

Back to Top | Article Outline


Using quantitative data to inform a qualitative analysis is a form of triangulation. Triangulation is a method of safeguarding validity by using data derived from multiple sources.11–14 We used the thematic clusters identified from the PCA as the basis for initial coding of the qualitative portion of the interview. The themes derived from the PCA were then validated and expanded using a grounded theory methodological approach to the qualitative data. Grounded theory is a process of identifying analytical categories as they emerge from the data.11–14 Analysts included physicians, nurses, psychologists, and palliative care researchers. We analyzed the transcripts using open coding (the process of breaking down, comparing, conceptualizing, and categorizing data), axial coding (the process of reassembling data into groupings) and selective coding (the process of developing a core theme in the data and relating it to other themes).14 Through discussion at serial meetings, all readers contributed to consensus-coding taxonomies.

Initially, transcripts were randomly chosen from the 135 total transcripts. As the analysis progressed, certain types of patient deaths were oversampled to better understand the nuances of these stories. Final sample size for analysis of interviews (n = 51) was determined when theoretical saturation was reached. In qualitative research, this occurs when none of the analysts recognize new, unique content. The remaining 84 transcripts were coded into one of the three categories of deaths but not analyzed for further content.

Back to Top | Article Outline


Our study population was a subset of a larger data set in which all participants were asked to complete three parts of the study (one quantitative and two qualitative: most recent and most emotionally powerful patient deaths).10 In this larger study, 246 participants were eligible and 196 agreed to participate yielding a response rate of 80%. Of the 196 participants who agreed to participate in the larger study, a subset of 33 were asked only to complete the most recent death portion of the study while the remaining 163 were asked to complete the entire study. Of these 163 participants, 19 had missing quantitative data yielding a final quantitative analytic sample of 144 participants. Of these, nine had missing qualitative data yielding a final qualitative analytic sample of 135. We completed a full qualitative analysis on the transcripts of 51 of the 144 participants who had been asked to complete the most emotionally powerful death portion of the study and who had complete quantitative data.

Of the 144 physicians in the quantitative subset, 44 (31%) were attending physicians and 100 (69%) were housestaff. Of the 100 housestaff, 47 were residents and 53 were interns. Ninety-one (63%) of the sample were men and almost exclusively white. The mean age was 29 years for housestaff and 42 years for attending physicians. The demographics of participants and nonparticipants were similar.

Of the 51 participants whose interview transcripts we analyzed, 32 (63%) were housestaff and 19 (37%) were attending physicians. Thirty-five (68%) were men. Other than having more male physicians in our qualitative subsample, we found no differences between the quantitative group (n = 144) and the qualitative subsample (n = 51).

Thirteen (68%) of the attending physicians in the qualitative subsample reported that their most emotionally powerful death occurred during their years in practice. Of the 19 (60%) housestaff who reported it occurred during their internship, seven (23%) said the death occurred before or during medical school, and six (17%) during residency. We found no significant differences between physicians based on gender or level of training for emotional characteristics of the patient death (e.g., degree of conflict).

Attending physicians cared for the patient for the longest time; nine (46%) reported caring for the patient for more than six months. Thirty (93%) housestaff reported caring for the patient for less than one month. We found no significant difference in the time of the last contact with the patient according to level of training with 25 (50%) physicians reporting the last contact to be within hours of the patient’s death.

Back to Top | Article Outline

Patient cases

The diagnoses of the 135 patients whose deaths were most emotionally powerful were 74 (55%) cancer and 61 (45%) noncancer. Sixty-two (46%) were women, 73 (54%) were men. Fifteen of the patients had undergone bone marrow transplantation.

Back to Top | Article Outline

The emotionally powerful deaths

It was haunting … it continues to be haunting.

—Male resident

The entire sample of 144 physicians rated these memorable patient deaths to be emotionally powerful. In response to the question “How emotionally powerful was this death?” respondents gave a mean rating (SD) of 8.5 (1.2) on a ten-point scale (1 = least emotionally powerful death, 10 = most emotionally powerful death). The principal components analysis of the adjective checklists produced three typologies all with eigenvalues greater than 2: the “shocking or unexpected death,” the “good death,” and the “overtreated death.” The responses to questions about the emotional characteristics of the most emotionally powerful patient death for the entire sample (n = 144) as well as for the physicians whose narratives were analyzed (n = 51) are shown in Table 1.

Table 1

Table 1

Back to Top | Article Outline

Common themes

Several themes were consistently present throughout all the cases (see Table 2).

Table 2

Table 2

Back to Top | Article Outline

Personal experiences with loss.

Physicians’ personal experiences influenced their responses to and perceptions of patients’ deaths. Physicians recalled seriously ill or dying family members of their own while caring for dying patients. One resident whose father had recently died related his most emotionally powerful patient death:

… since I was his intern someone said why don’t you go ahead and pronounce him, and I sort of walked over to do that and I just burst into tears and had to leave … which I think was related more to being upset about my father.

Back to Top | Article Outline

Identification with patients and families.

Physicians who identified with the patient or a family member of the patient reported a closer relationship and more intense emotional reactions. Frequently this identification made the work both more rewarding and more difficult. One male intern described a 30-year-old patient with amyotrophic lateral sclerosis who was intubated for progression of his disease:

… these people could have been your neighbors … it scares you and makes you think of yourself. That scared me a couple of times. Wow, I can’t talk to this patient right now because I am associating myself with him a little closely.

Certain types of patients elicited more intense emotional reactions. Young patients, those undergoing bone marrow transplantation, and those who died from a traumatic event evoked descriptions of more powerful emotional responses from the physicians that made coping especially challenging. A male resident described his emotional reaction to the death of a young father who died of varicella pneumonia:

This was the hardest death I ever had to watch…. I identified with him because I have a one-year-old and he was almost my same age. I was thinking what if I were in his shoes and it was my daughter standing at my bed.

Back to Top | Article Outline

Closeness of the patient–physician relationship.

Physicians who felt a sense of connection with the patient or a family member of the patient reported amplified positive and negative emotional responses. The valence of the emotion varied with the type of death the physician described. For example, “good” patient deaths were associated with primarily positive feelings such as satisfaction at a job well done, while “shocking/unexpected” patient deaths were associated with negative feelings of distress as described by the following male intern:

It was horrible because … I was able to understand who he was as a person and I actually found myself … speaking with him—not necessarily about his illness but just about anything he wanted to. … [I]t is one thing to have someone come to you who is intubated and basically brain dead, but it is another thing to have a person who is alive … a good person and [you] have nothing to give him.

Back to Top | Article Outline

Sense of responsibility for the death.

Participants who felt a greater degree of responsibility for the patient death experienced amplified emotional responses. Unlike senior clinicians with more experience, trainees were more likely to attribute the cause of death to their lack of competence rather than the progression of the disease. A female intern described a patient with end-stage liver disease and severe pulmonary hypertension who died suddenly:

I felt totally responsible for his care…. I ran the code and everything went well. People were very quick with their interventions and then he died and I lost it. I just started crying uncontrollably…. I think you always question yourself —is there something I could have done differently. I sort of wanted him to have an autopsy so I would know the answer but on the other hand I didn’t want him to because I didn’t want to know the answer.

In contrast, a senior attending physician who, despite the patient dying during an elective procedure that he was performing, described little emotional reaction to the death of a patient:

… it wasn’t really the kind of thing that I needed emotional support for. It was just an unfortunate circumstance.

Back to Top | Article Outline

Intellectual and emotional closure.

Physicians reported greater distress from deaths of patients for whom the cause of death could not be determined. “Intellectual closure” occurred when the cause of death was well understood and was associated with the physician feeling competent in caring for the patient. Without this closure, physicians often reported guilt about some aspect of their patient care. One resident physician described caring for a patient who died suddenly:

I was shocked…. I really believed that we had, at least on an intellectual level, approached everything very, very elegantly … I thought we had addressed every issue…. We did everything right and he still died … I felt that I had let him down.

Physicians struggled with their own emotions as well as the emotions of their patients. “Emotional closure” occurred when the physician coped effectively with the strong personal emotions that arose in caring for the patient and achieved a sense of acceptance about the patient’s death. Routes to acceptance included discussion with colleagues, family, and friends; appreciation expressed by the patient’s family members that mitigated the sense of failure; and statements from patients “absolving” the physician of responsibility for the patient’s decline. A male attending physician described the death of a patient from leukemia:

I definitely cried … I still remember it … I went to the wake. I always felt that I wanted to do more, to cure him. I knew I was providing emotional support but I didn’t know how important that was until I went to the wake and heard it from everyone there. I felt really good about that.

Participants who did not feel a sense of “emotional closure” described continued powerful emotional responses despite the passage of time. A female attending described the death of a long-term patient 15 years earlier who died after a disturbing respiratory arrest in an outpatient area:

[Y]ou know this pit-in-your-stomach feeling and not sleeping and feeling miserable and horrible for weeks and weeks, I’m sure. I still cringe when I think about it.

Back to Top | Article Outline

Coping and lessons learned.

The lessons learned during the care of patients whose deaths were emotionally powerful were often critical in shaping doctors’ future care of patients and personal coping styles. One female intern described how an unexpected death changed her care of patients:

I think that I am more inclined to touch people physically, ‘cause I really hated that feeling of just watching. I think had I at least been able to touch his shoulder or hold his hand, or something, that I would have felt like a more human connection … [now] I touch people a lot.

One male intern described how he has changed his behavior as the result of his first patient death due to a complication from a thoracentesis that he performed:

I try not to form as much of an emotional bond to patients. Just because at that point, if something happens it hurts more and is harder to deal with. I think I … just don’t get close anymore.

Back to Top | Article Outline

Isolation and lack of attending physician guidance.

Trainees often described managing the emotions generated by the most emotionally powerful patient death alone or by speaking with friends or family outside of medicine.

One resident described coping with the first patient death during his internship; when asked if he spoke with anyone after the death he stated:

No, I sort of figured it comes with the territory of being a doctor. You are going to have to learn to deal with it one way or another.

Occasionally, trainees sought guidance from a colleague or a more senior resident; however, we found only two descriptions of attending physicians’ guiding trainees in navigating the powerful emotional reactions to the death of the patient.

Back to Top | Article Outline

Typologies of memorable deaths

Physicians often experience complex emotional reactions after the death of a patient. Our typologies provide one useful framework in understanding these complicated experiences. Most of the patient deaths fit one of the three categories; “shocking or unexpected,” “good,” or “overtreated.” However, 15 of 135 physicians (11%) in the qualitative analytic sample described patient deaths that were best described by more than one category. For example, some narratives began with a conflicted story typical of the “overtreated” patient descriptions but were followed by resolution and a peaceful “good” patient death.

Back to Top | Article Outline

“Shocking or unexpected” deaths

[I]t was only community-acquired pneumonia. She should have gotten better … then she went into afib…. [A]s a team we really struggled with the decision to anticoagulate her … it was a gray zone …. [S]he was clearly improving from the pneumonia standpoint and we were getting ready to send her home.… I came in the next day and she was unresponsive! … it was a big stroke … it was very tough because [of] that feeling of guilt.

—Male resident

The “Shocking or unexpected” death component was associated with the following adjectives: avoidable, uncertain, unexpected, shocking, inadequate, and guilty. This component had an eigenvalue of 2.4 that explained .40 of the variance with eigenvectors that were positively correlated (0.34–0.45). Fifty-nine of the 135 interviews (44%) were coded as “shocking or unexpected”; of those, 14 (24%) were interviews of attending physicians and 45 (76%) were of physicians in training. In comparison with the other typologies, physicians whose stories were categorized as “shocking or unexpected” on qualitative analysis reported that while these deaths were not characterized by conflict, they were more emotionally powerful and disturbing. Respondents reported less satisfaction from patient care and a more distant relationship with the patient (see Table 1).

Interns and residents were more likely to rank deaths as shocking or unexpected than were attending physicians. Moreover, many of these stories were of deaths from common causes, such as pneumonia or stroke. For physicians-in-training, the type of death, the intensity of emotion or the degree of suffering were new, disturbing, and often overinterpreted to be the result of a personal failure.

Only six senior clinicians characterized their most emotionally powerful patient death as shocking or disturbing. Half of these deaths occurred during the physician’s internship, while the remaining three deaths occurred during the attending physicians’ years in practice. The three deaths that occurred during the attending physicians’ years in practice were unusually powerful and included the death of a pregnant woman and the fetus, the death of a 19-year-old from Guillan-Barre, and the sudden death of a primary care patient in the physician’s waiting room.

“Shocking or unexpected” deaths were associated with physicians’ questioning their personal competence. They expressed disillusionment with the capabilities of medicine. A female attending described a beloved patient whom she had known for 15 years who died after a disturbing respiratory arrest in an outpatient area:

I felt very inadequate. I felt very sad, but I think my primary feeling was, at that point, that I’d somehow failed her family and her. That I had treated her for so long and yet at that final moment I couldn’t do what needed to be done. And that also the system had failed, that the hospital had failed her.

Participants frequently experienced a lack of emotional and intellectual closure, feeling responsible for some aspect of the patient’s suffering or the ultimate unforeseen death of the patient. Housestaff described intense emotional responses associated with these deaths that were unlikely to be validated or fully understood by others, including more senior clinicians, who cared for the patient. One resident described discussing intense feelings of guilt with her colleagues after the death of a patient who was thought to have died of a pulmonary embolism in the setting of subtherapeutic anticoagulation but whose postmortem examination revealed no obvious cause of death:

Well their automatic response is going to be, “Of course this is not your fault, you know, things happen, this is a cancer service.” But regardless, it wasn’t going to help me, because in my mind it was my fault.

As physicians gained experience, the tendency to overinterpret data or to be shocked by common medical scenarios lessened. One female intern stated:

I realized that … sometimes things just go wrong … death is almost a natural part of being a doctor on the wards. And you just can’t control it … it takes a couple of deaths before you realize that.

Back to Top | Article Outline

“Good” deaths

…[H]e spent much of the day sleeping because of his pain medications … he awakened and thanked me. So that was the most emotionally powerful time of it, when he was leaving. … All the family members thanked me and the daughter gave me a hug. … I knew I was never going to see him and his family again and it was just very sad.

—Male resident

The “good” death component was associated with the following adjectives: natural, spiritual, peaceful, sense of closure, moving, and satisfied. This component had an eigenvalue of 3.2 that explained .46 of the variance with eigenvectors that were positively correlated (0.31–0.43). Thirty-eight of the 135 physician (28%) interviews were coded as “good”; 20 (53%) of those interviews were of attending physicians and 18 (47%) were of physicians-in-training. In comparison with the other typologies and the overall sample, physicians whose stories were categorized as “good” deaths on qualitative analysis reported these deaths to be less emotionally powerful, less disturbing, and less conflict-laden. Physicians reported less patient suffering, more satisfaction with patient care, and less of a need for support (see Table 1).

The qualitative analysis supported these findings. “Good” patient deaths were perceived to have involved little suffering and were often described as peaceful and spiritual. Physicians who cared for these patients and their families readily developed close relationships associated with powerful positive emotional responses. Physicians were unlikely to question the care of the patient or to describe conflict, which supported successful emotional and intellectual closure. The negative emotional responses present focused on issues of loss and grief.

The patients whose deaths were described as “good” were typically “good patients.” “Good patients” and their families communicated openly with physicians about death. They understood the physicians’ descriptions of the medical facts, were involved in decision making and accepted recommendations. The patients and their families were grateful and often expressed this to the physicians caring for them.

Participants generally felt competent in their ability to provide good symptom relief and psychosocial support. In short, “good” deaths made the physician feel good. A female attending physician described a patient dying of lung disease:

…[W]e kept him comfortable. … We tried to create an environment in the room for his death. I was struck by the patient’s wishes being met, and I was struck by the sense that he was dying and people didn’t abandon him. … [W]hen you went into that room there was a sense of joy. … I felt good that the family had had me there. I felt really good about that.

Back to Top | Article Outline

“Overtreated” deaths

The problem was that she had a young husband and parents who were pretty much in complete denial. We were trying to be aggressive up to the end. To the point that we actually hung a new form of chemotherapy about four hours before she actually died, even though everybody knew except her immediate family that she was going to die within the next four to eight hours.

—Male resident

The “overtreated” death component was associated with the following adjectives: full of suffering, isolated, overtreated, frustrating, and degrading. This component had an eigenvalue of 2.0 that explained .41 of the variance with eigenvectors that were positively correlated (0.41–0.49). Twenty-three of the 135 interviews (17%) were coded as “overtreated”; of those, six (26%) were interviews of attending physicians and 17 (74%) were of physicians in training. In comparison with the other typologies and the overall sample, physicians whose stories were categorized as “overtreated death” on qualitative analysis reported these deaths to be more emotionally powerful, more disturbing, more conflict-laden, and filled with greater patient suffering (see Table 1).

Care of overtreated patients was frequently described in terms of conflict that interfered with emotional closure and resulted in long-lasting anger and frustration. Conflict could be either within the medical team or between the medical team and the family. One male resident described the cardiopulmonary resuscitation of a 35-year-old woman who had received a bone marrow transplant. The team felt the patient to be gravely ill and believed aggressive efforts would be futile, but the attending physician would not allow patients on the bone marrow transplant service to be made “do-not-resuscitate”:

… [W]e had done everything we could five times over and it was over. … And I left the room and just couldn’t make sense of the whole situation. And started crying in a relatively inconsolable fashion … I felt that there was nothing that anyone could say to me or that I could say to anyone else that was going to change how upsetting this experience was.

Participants described “overtreated” patients undergoing aggressive care with a limited understanding of their disease, its prognosis, and the implications of treatment, often enduring great physical and emotional suffering. Participants frequently stated this was due to patient, family, or physician denial about the gravity of the situation. Patients and families were often described as demanding and troublesome. One male resident described a patient who was dying after complications from a bone marrow transplant:

I think her parents actually were a barrier to her having the most peaceful death…. I think a lot of the stuff she allowed us to do to her in the last three days of her life were specifically for her parents. Some of which may have caused her more pain.

Housestaff described advocating for patients, but being unable to change the course of the care while they were required to carry out orders with which they did not agree. One male intern described caring for a 37-year-old patient with amyotrophic lateral sclerosis who was coerced into being intubated against his previously stated wishes:

I felt like a liar and a cheat…. As a student there are many people above me, but if you know [the patient] and you get along with them, you have a responsibility. That was a little troubling for me. It kind of ate away at me.

Back to Top | Article Outline


In our study, physicians described powerful experiences with death during all stages of their careers, but the deaths that occurred early in training were particularly vivid and intense. Participants reported experiencing a wide array of emotions in the context of three types of deaths: the “unexpected or shocking” death, the “good” death, and the “overtreated” death. Throughout the narratives, physicians’ emotional responses were influenced by their personal experiences with loss, the degree to which they identified with or felt close to the patient, and the degree to which they felt responsible for the death.

Our data indicate that physicians-in-training experience deaths differently than do more senior clinicians. Physicians-in-training in our study were more likely to describe a death as “shocking or unexpected.” Physicians-in-training described common patient deaths caused by pneumonia or stroke as disturbing whereas experienced clinicians did not relate these types of deaths as upsetting. “Unexpected or shocking” deaths were associated with persistent disturbing emotions that lasted for days to weeks and, at times, even years after the death. These deaths contained elements of trauma, helplessness, and guilt and generated questions of personal competence. As a result of these experiences, some trainees reported changes in the clinical care of their patients as well as their chosen career path. Physicians experienced similar emotions in the “overtreated” deaths; however, this type of death was the least commonly reported (see List 1).

List 1 Lessons for Medical Educators Supervising Physicians-in-Training Who Care for Dying Patients, Formulated during a Study Conducted at Two Medical Centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania, 1999–2001

List 1 Lessons for Medical Educators Supervising Physicians-in-Training Who Care for Dying Patients, Formulated during a Study Conducted at Two Medical Centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania, 1999–2001

We were struck by the degree to which physicians-in-training suffered their most emotionally powerful death in isolation from more senior physicians. Although these emotionally powerful patient deaths could have been opportunities for learning from more senior clinicians about the principles of pain and symptom management, ways to communicate with patients and their families about terminal illness, and self-care and personal coping strategies when caring for dying patients, we only found two descriptions of trainees discussing the death with an attending physician. There are a number of possible reasons why attending physicians do not use their trainees’ emotional reactions to patient deaths as “teachable moments.” First, attending physicians may be unaware that a particular death was emotionally powerful. Clinical experience appears to attenuate the intensity of certain physician emotional responses to patient deaths,10 and thus the attending physician may not understand how powerful the experience is for physicians-in-training. Second, attending physicians may not spend enough time with housestaff or students to know about disturbing interactions or personal identification with the patient or family that may contribute to powerful emotional responses to a death. Third, housestaff may feel uncomfortable initiating a conversation about such powerful emotions with attending physicians. Finally, attending physicians may be aware of, but unprepared to deal with, these powerful emotions and therefore ignore them. Attending physicians who wish to take advantage of these teaching opportunities should systematically inquire about the impact of a particular death for a trainee (see List 2).

List 2 Examples of Questions That Attending Physicians Can Use to Facilitate Discussion with Trainees After a Patient Death, Formulated during a Study Conducted at Two Medical Centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania, 1999–2001

List 2 Examples of Questions That Attending Physicians Can Use to Facilitate Discussion with Trainees After a Patient Death, Formulated during a Study Conducted at Two Medical Centers in Boston, Massachusetts, and Pittsburgh, Pennsylvania, 1999–2001

By capitalizing on these “teachable moments,” attending physicians allow housestaff to benefit from their experience in the care of the dying. Attending physicians have the opportunity to model excellent medical and psychosocial care. Equally important, these experiences can provide an opportunity for housestaff and attending physicians to reflect on the often- intimate relationship between physicians and the dying, and the power of the human spirit as they bear witness to the suffering and death of another. Some medical educators suggest a routine structured mechanism for reflection on powerful feelings elicited by patients as a way of lessening the traumatic effects of these experiences and promoting humanistic qualities in trainees.15–27

The three types of patient deaths described in our study give a framework for understanding physicians’ experiences and can serve as a guide in the discussion after a powerful patient death. Trainees who perceive a death as “shocking or unexpected” are likely to feel inappropriately responsible for a bad outcome and to have lingering questions about their competence or the appropriateness of their emotional responses to the death. Exploration of feelings of guilt or regret provides an opportunity for a senior clinician to elicit and clarify concerns about personal fallibility or the limits of medicine. Teaching in an “overtreated” death might focus on dealing with team conflicts, the sense of moral responsibility felt when acting against one’s own conscience, or when the family is seen as making a “wrong” decision. Finally, a discussion after a “good” death could focus on the normality of the feelings of grief and sadness that are evoked or the usefulness of writing a condolence letter as a method of bringing closure.

To our knowledge, this is the first study of physicians’ emotional reactions to their patients’ deaths using a random sample of patients to identify participants. One previous study by Saunderson and Ridsdale28 of general practitioners’ beliefs and attitudes about patient death also found frequent expressions of guilt and little influence of medical training on the management of dying patients. Another study of physicians’ day-to-day emotional experiences on inpatient medical wards by Kasman and colleagues25 found trainees to experience a variety of emotions including consistent descriptions of anxiety and guilt associated with perceived responsibility for untoward outcomes, as did our study. Our findings extend the findings of others in areas of burnout humanism, the importance of critical incidents, medical errors, and posttraumatic stress disorder among trainees.5–8,16,19,29

Our study had several limitations. First, it is possible that the physicians had recall bias; however, the reality of the clinical scenario appears to be less important than the perceived or actual effect of the emotional experience on the physician which was well-elucidated from the in-depth interview. Second, this study was conducted with primarily white physicians at two quaternary institutions, which may limit generalizability to other populations of physicians. Lastly, physicians who experience the death of a patient are likely to feel many emotions that will defy categorization into one of the three types of deaths we have described. The typologies are useful heuristic constructs, and should serve as guidelines for exploration and further discovery.

Supervising physicians have an opportunity to improve both the care of dying patients and housestaff coping with these deaths by using the “teachable moments” that are present for trainees as they care for the dying. Faculty development courses and readings are available to help interested faculty obtain these skills.30,31 Further research is needed on the effectiveness of educational interventions targeted at house officers’ distress after a patient dies and supporting their well-being, as well as the effects of such an intervention on patient care.

This work was supported by a grant from the Nathan Cummings Foundation. During the preparation of this manuscript, Dr. Jackson was supported by the Department of Veteran’s Affairs and the National Cancer Institute. Dr. Arnold is a Faculty Scholar of the Open Society Institute’s Project on Death in America and is supported by the LAS Trust.

Back to Top | Article Outline


1Vachon M. The Stress of Professional Caregivers. Oxford Textbook of Palliative Care. 2nd ed. New York: Oxford University Press, 1998:919–29.
2Kvale J, Berg L, Groff J, Lange G. Factors associated with Residents’ attitudes toward dying patients. Fam Med. 1999;31:691–96.
3Whippen D, Canellos G. Burnout syndrome in the practice of oncology: results of a random survey of 1,000 oncologists. J Clin Oncol. 1991;9:1916–20.
4Lowry F. Does doctor’s own fear of dying hinder palliative care? CMAJ. 1997;157:301–2.
5Meier D, Morrison S, Back A. The inner life of physicians and care of the seriously ill. JAMA. 2001;286:3007–14.
6Zinn W. Doctors have feelings too. JAMA. 1988;259:3296–98.
7Quill T. Caring for Patients at the End-of-Life: Facing an Uncertain Future Together. New York: Oxford University Press, 2001.
8Shanafelt T, Bradley K, Wipf J, Back A. Burnout and self-reported patient care in an internal medicine residency program. Ann Intern Med. 2002;136:358–67.
9Branch W, Pels R, Lawrence R, Arky R. Becoming a doctor—critical-incident reports from third year medical students. N Engl J Med. 1993;329:1130–32.
10Redinbaugh EM, Sullivan AM, et al. Doctors’ emotional reactions to recent death of a patient: cross sectional study of hospital doctors. BMJ. 2003;327:185–91.
11Pope C, Ziebland S. Mays N. Qualitative research in health care. Analysing qualitative data. BMJ. 2000;320:114–16.
12Mays N, Pope C. Qualitative research in health care. Assessing quality in qualitative research. BMJ. 2000;320:50–52.
13Strauss A, Corbin J. Basics of Qualitative Research: Grounded Theory Procedures and Techniques. Newbury Park: Sage Publications, 1990.
14Afifi A, Clark V. Computer-Aided Multivariate Analysis.New York: Van Nostrand Reinhold Company, 1984.
15Coulehan J, Williams P. Vanquishing virtue: the impact of medical education. Acad Med. 2001;76: 598–605.
16Klamen D, Grossman L, Kopacz D. Posttraumatic stress disorder symptoms in resident physicians related to their internship. Acad Psychiatry. 1995;19:142–49.
17Novack D, Suchman A, Clark W, Epstein R, Najberg E, Kaplan C. Calibrating the physician: personal awareness and effective patient care. JAMA. 1997;278:502–9.
18Quill TE, Williamson PR. Healthy approaches to physician stress. Arch Intern Med. 1990;150:1857–61.
19Novack DH, Epstein RM, Paulsen EH. Toward creating physician healers: fostering medical students’ self-awareness, personal growth, and well-being. Acad Med. 1999;74:516–20.
20Stange K, Miller W, Crabtree B, O’Connor P, Zyzanski S. Multi-method research: approaches for integrating qualitative and quantitative measures. J Gen Intern Med. 1994;9:278–81.
21Wu AW, Folkman S, McPhee SJ, Lo B. Do house officers learn from their mistakes? JAMA. 1991;265:2089–94.
22Brady D, Corbie-Smith G, Branch W. “What’s important to you?”: the use of narratives to promote self-reflection and to understand the experiences of medical residents. Ann Intern Med. 2002;137:220–23.
23Branch T, Kern D, Haidet P, Weissman P, Gracey C, Mitchell G, Inui T. Teaching the human dimensions of care in clinical settings. JAMA. 2001;286:1067–74.
24Epstein R. Mindful practice. JAMA. 1999;282:833–39.
25Kasman DL, Fryer-Edwards K. Braddock CH 3rd. Educating for professionalism: trainees’ emotional experiences on IM and pediatrics inpatient wards. Acad Med. 2003;78:730–41.
26Block S, Billings JA. Nurturing humanism through teaching palliative care. Acad Med. 1998;73:763–65.
27Billings JA, Block S. Palliative care in undergraduate medical education. Status report and future directions. JAMA. 1997;278:733–38.
28Saunderson E, Ridsdale L. General practitioners’ beliefs and attitudes about how to respond to death and bereavement: qualitative study. BMJ. 1999;319:293–96.
29Peters A. Death and medicine: a personal account. Am J Med. 1990;89:81–82.
30Fallowfield L, Jenkins V, Farewell V, Saul J, Duffy A, Eves R. Efficacy of a cancer research UK communication skills training model for oncologists: a randomised controlled trial. Lancet. 2002;359:650–56.
31Program in Palliative Care Education and Practice. The Harvard Medical School Center for Palliative Care 〈〉. Accessed on 20 May 2005.
© 2005 Association of American Medical Colleges