Customize your JAMA Network experience by selecting one or more topics from the list below.
Dickinson GE, Tournier RE, Still BJ. Twenty Years Beyond Medical SchoolPhysicians' Attitudes Toward Death and Terminally Ill Patients. Arch Intern Med. 1999;159(15):1741–1744. doi:10.1001/archinte.159.15.1741
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
In response to consumer demands and recent changes in health care, the American Medical Association and the Association of American Medical Colleges have expressed concern about how physicians relate to patients, especially those who are seriously ill.
To determine the impact of 20 years of medical practice on the attitudes of physicians toward terminally ill patients and their families.
Data were gathered from questionnaires mailed in 1976 and again in 1996 to physicians who graduated from medical school between 1972 and 1975.
Responses were received from 71% and 63% of the 1664 and 1109 physicians surveyed in 1976 and 1996, respectively. Using a t test for paired variables, statistically significant differences were noted for physicians' responses to all of the 11 Likert-type attitudinal statements on death and terminally ill patients and their families. Physicians in 1996 were more willing to inform terminally ill patients of their prognosis and in general seemed more confident with dying patients than they were in 1976.
After 2 decades of practicing medicine, physicians' attitudes toward terminally ill patients seem to have changed; physicians appear to be more open to communicating with terminally ill patients and their families on issues concerning death and dying.
IN CONTEMPORARY teaching hospitals, most therapeutic relationships are fleeting (eg, emergency department visits, outpatient appointments, or short hospital stays).1 Students in medical school are taught that medicine is primarily about science and only secondarily about people.2 From time to time, practicing physicians in nearly every specialty must deal with patients who are dying or have a life-threatening disease.
Physicians are trained to investigate, diagnose, prolong life, and cure.3 When a patient has a terminal diagnosis, physicians often feel that they have little more to offer and equate the inability to cure the patient with failure. Currently, medical education trains physicians to treat disease aggressively until the patient improves or death is imminent. Physicians sometimes find themselves prolonging the process of dying without providing the patient with either an improved quality of life or time to prepare for death.4
Traditionally, medical schools have not allocated much time and effort to help future physicians relate to terminally ill patients and their families.5 Physicians receive limited training in communicating freely with patients, and nowhere are difficulties with communication more exposed than when physicians deal with patients who are very sick. A physician's body language and repression of emotional reactions can convey the message that concerns about dying and the inevitability of death are not within the scope of medical care.
As figures of authority, physicians have enormous power to help patients and their families. Physicians may feel uncomfortable with death and somewhat awkward responding to dying patients, but they have an obligation not to abandon a dying patient. If an isolated death is a pervasive fear, for example, the physician can reassure the patient that he or she will not die alone.6
Most studies about physicians' attitudes toward dying and death have been cross-sectional in design. Physicians have been studied while still students in medical school or after graduation, but rarely in a panel study. New physicians enter the medical profession with certain attitudes and feelings toward patients that will be shaped until they more or less comply with those of the medical profession as a whole.7 The student is molded into the medical profession by learning what is expected and then by doing it. Socialization continues throughout the life cycle; thus, attitudes and behavior change, more so in some individuals than in others. In light of the recently expressed concerns of the American Medical Association (AMA) and the Association of American Medical Colleges about how physicians relate to patients, especially those who are seriously ill, this study is an effort to determine whether the attitudes of physicians toward death and terminally ill patients and their families have changed significantly in the 2 decades since they completed medical school.
Graduates of the classes of 1972 through 1975 from 5 medical schools (University of Southern California, University of Colorado, Vanderbilt University, Pennsylvania State University, and Yale University) were selected to receive a questionnaire to determine their attitudes toward dying patients. These schools were selected from a 1975 survey of curricula on death education in the 113 medical schools in the United States.8 Of the 107 responding schools, these 5 programs were the only ones that had offered a thanatology course during the previous 4 years (only a minority of the students took the optional course). These 5 medical schools, while not randomly selected, provide a good representation of private and public schools and a broad geographical distribution.
Names of the graduates of these medical programs were obtained from the AMA in Chicago. The questionnaire was refined by a panel of 15 medical students and recent medical school graduates. It was then mailed to 1664 physicians in the fall of 1976; 3 follow-up mailings were sent to nonrespondents. In 1996, we used the AMA's Directory of Physicians in the United States, 35th edition, to locate current addresses of the participants in the 1976 survey and sent them the same questionnaire. Of the 1664 physicians in the original sample, we sent questionnaires to 1109. The questionnaires were accompanied by a cover letter and a self-addressed, stamped envelope.
Respondents were asked to indicate the degree to which they agreed or disagreed with each of 11 statements on a 5-point Likert scale ranging from 1 ("strongly agree") to 5 ("strongly disagree"); 3 was a neutral, mid-scale choice. The statements concerning attitudes toward death and reactions to dying patients and their families are listed with the corresponding number below:
When one of my patients dies, I always wonder if something could have been done to save him/her (wonder).
I feel as comfortable with a dying patient as I do with any other patient (comfortable).
I do not think about death very much (think).
Treating a dying patient is one of the most unpleasant aspects of my profession (unpleasant).
Whenever possible, I avoid a person who is dying from an irreversible condition (avoid).
I try to avoid telling a patient directly that he/she is dying (tell).
I find it more difficult to deal with the family of a dying patient than with families of my other patients (family).
A patient's death does not depress me when I know there was nothing I could do to save him/her (depress).
Telling a person he/she is going to die is difficult for me (difficult).
I believe physicians refer terminal patients to other physicians more than nonterminal patients in order to avoid having to deal with their dying (refer).
I think it is essential that a dying patient be told of his/her prognosis (patient told).
We calculated the mean and SD for each variable and ran a t test for each pair of variables to determine the statistical significance of changes in the Likert items over time. Additionally, we ran a t test for the independent variables of sex and medical specialty against each of the Likert items over time.
In 1976, 1664 questionnaires were mailed and 1093 usable questionnaires were returned (71% return rate) (124 questionnaires were undeliverable). In 1996, 1109 questionnaires were mailed and 696 usable questionnaires were returned (63% return rate) (28 returned questionnaires were unusable). In both surveys, 90% of the respondents were men and 10% were women, representative of the sex ratio of medical school graduates in the early 1970s.9 The mean age of the respondents to the 1996 survey was 48.57 years (median, 48 years; mode, 48 years; range, 44-70 years). In 1976, the mean age was 28.74 years (range, 24-45 years).
Statistically significant differences (P<.01) were noted for physician responses to all 11 attitudinal statements toward death and terminally ill patients and their families (Table 1). After 20 years of practice, physicians more strongly agreed with the statement, "When one of my patients dies, I always wonder if something could have been done to save him/her." On the other hand, those physicians were less likely to be depressed over a patient's death when they knew nothing could be done to save the patient.
Compared with 1976, physicians felt slightly less comfortable in 1996 with dying patients. Physicians in 1996 disagreed more strongly with the statement, "I do not think about death very much," than in 1976. Physicians in 1996 also reported stronger disagreement with the statement, "I believe physicians refer terminal patients to other physicians more often than nonterminal patients in order to avoid having to deal with their dying."
After 2 decades of practicing medicine, the physicians who participated in our survey did not consider treating a dying patient as unpleasant an aspect of the profession as they did soon after graduating from medical school. Likewise, physicians in 1996 reported that they were less likely to avoid a person dying from an irreversible condition than they were in 1976.
When a patient had been diagnosed as terminally ill, the physicians surveyed were less likely to "try to avoid telling" after 20 years of practice. Likewise, "telling a patient he/she is going to die" was less difficult for these physicians after 2 decades of practicing medicine. Overwhelmingly, these veteran physicians more strongly agreed that it is "essential that a dying patient be told of his/her prognosis" compared with 20 years earlier. Also, physicians in 1996 found it less "difficult to deal with the family of a dying patient than with families of my other patients." Overall, after 20 years of practicing medicine, the physicians surveyed communicated more openly with their terminally ill patients and their families and apparently did so with less stress.
When we examined physicians' attitudes toward terminally ill patients and their families over time by sex, the differences were greater in 1976 (Likert items 3, 4, 6, 8, and 10;P<.05) than in 1996 (Likert items 1 and 10;P<.05). Men changed their attitudes more over the 20-year period (all 11 Likert items;P<.01) than women (Likert items 1-3 and 11;P<.01). In 1996, more male than female physicians wondered if something could have been done to save the deceased patient. Also in 1996, more female than male physicians believed physicians refer terminally ill patients more often than nonterminally ill patients to avoid having to deal with their dying.
The medical specialties of the physicians who were surveyed were divided into 4 categories: (1) high probability of having terminally ill patients and direct patient contact, (2) high probability of having terminally ill patients but low probability of direct patient contact, (3) low probability of having terminally ill patients but high probability of direct patient contact, and (4) low probability of having terminally ill patients and direct patient contact. These categories were partially based on previous research in which a panel of physicians rated by medical specialty how frequently physicians deal with terminally ill patients.10
The high contact and the high probability/low contact specialties changed significantly in 8 of the 11 Likert items over time (P<.01), yet the low probability/low contact specialties changed in only 4 items (Likert items 1-3 and 11;P<.01). The 2 groups with high contact in 1996 were less likely to find that "telling the patient is difficult" and disagreed more strongly with the statement "physicians refer patients to other physicians to avoid having to deal with their dying." In the low contact group, responses to these 2 items (Likert items 6 and 10) were unchanged over time. In all 4 categories, the only item (Likert item 7) on which physicians' attitudes did not vary over time was, "I find it difficult to deal with the family of a dying patient."
In comparing responses to Likert items on physicians' attitudes toward death and terminally ill patients and their families over a 20-year interval, significant changes were observed. After 20 years of practice, the physicians we surveyed were less likely to avoid telling the dying patient of the prognosis, less likely to avoid a person who is dying from an irreversible condition, found that "telling a person he/she is going to die" was not as difficult, were more likely to feel that "it is essential that a dying patient be told of his/her prognosis," agreed more strongly that dealing with the family of a dying patient was less difficult, and more strongly felt that treating a dying patient was no longer one of the most unpleasant aspects of the medical profession. Perhaps over time death becomes more routine and familiar and therefore less unpleasant.
Early in their careers, physicians may have a stronger need to shield patients from issues concerning death and dying, whereas, with more experience, physicians may be less protective as they become aware of the resiliency of patients. The reactions of younger physicians to death and dying are expected; they are consistent with the reactions of the death-denying society into which physicians are initially socialized. After 20 years of practicing medicine, the physicians we surveyed seem to have become more accustomed to dealing with death and/or were forced to come to grips with their attitudes toward death. If this is the case, increased attention to death and dying education in medical schools could accelerate this transformation.
The changes in the attitudes of physicians toward terminally ill patients and their families are not surprising in light of the AMA's shift in policy in 1980 regarding physicians informing patients of their prognoses.11 This is the current position of the AMA:
the physician must properly inform the patient of the diagnosis and of the nature and purpose of the treatment undertaken or prescribed. The physician may not refuse to so inform the patient.11
the physician must properly inform the patient of the diagnosis and of the nature and purpose of the treatment undertaken or prescribed. The physician may not refuse to so inform the patient.11
Previously, the AMA had recommended that the decision to inform the patient be left to the discretion of the physician ("therapeutic privilege"). The AMA's new policy may simply reflect what our survey indicates was already taking place. In addition, our findings may reflect the general trend of the 1980s and 1990s—patients have increasingly demanded the right to know their prognoses and have demanded more humane medical care.
According to our survey, physicians in 1996 believed that physicians were less likely to refer dying patients to other physicians to avoid dealing with them, perhaps suggesting that physicians are now more at ease with terminally ill patients. Somewhat different from the other 10 Likert items, this statement requires that physicians report how they perceive their colleagues relating to terminally ill patients. We do not know whether fewer referrals are actually being made or whether physicians have bonded with one another after 20 years of practice to the point that they refuse to admit that a colleague cannot cope with a patient because a terminal prognosis implies physician failure. Whatever the case, the physicians we surveyed reported that in 1996 they more strongly disagreed with the statement, "Physicians refer terminal patients to other physicians more than nonterminal patients."
The physicians surveyed in 1996 "think about death" more than they did in 1976. This is in keeping with our finding that physicians experience increased discomfort with dying patients after 20 years of practice. Apparently, these physicians have become more anxious about death the longer they have been in practice. This increase in the frequency of thoughts about death may also be related to the mean age (48.57 years) of the 1996 respondents.12 This age falls within the US Census Bureau's definition of middle age (45-64 years). At this age, failing health, the death of one's parents, the loss of close friends, and changes in one's physical appearance caused by aging may contribute to a heightened awareness of death. As long as one's parents are living, a buffer is perceived to exist between the person and death (parents are supposed to die first). After one's parents die, however, the buffer is gone. Middle-aged people watch their peers die quick deaths from cardiovascular disease and prolonged deaths from cancer. Thus, thinking about death more at 49 years than at 29 years is normal, whether one is a physician or not.
The respondents to the 1996 survey more strongly disagreed with the statement, "When one of my patients dies, I always wonder if something could have been done to save him/her," yet they were less likely to become depressed over a patient's death when they knew nothing could be done to save the patient. The comfort level of these physicians in 1996 with dying patients, however, was still not the same as with other patients. This finding is consistent with the conclusions of a study of 99 oncologists that noted that the respondents with more years of experience as physicians and more years of experience caring for dying patients reported feeling less comfortable with dying patients than with other patients.13
Though the attitudinal differences between male and female physicians toward terminally ill patients and their families in 1976 were not significant, in 1996 the differences were even less significant, perhaps suggesting that occupational role socialization is more influential than sex role socialization for physicians. There appears to be a blending of attitudes over time in these male and female cohorts. This finding is supported by recent studies of both medical students and practicing physicians that have failed to find persistent sex differences in attitudes on various medical issues.14
In our examination of medical specialties, the attitudes of the low probability/low contact group changed very little over the 20-year period compared with the attitudes of the other 3 groups. In essence, our study had a quasi-experimental research design, with the low probability/low contact group representing a control group that was minimally exposed to terminally ill patients. As expected, without the stimuli of interaction with terminally ill patients, physicians' attitudes remained relatively stable. On the other hand, in the 3 groups with some probability of having contact with terminally ill patients, attitudes changed more over time. Across the board, responses from physicians in all 4 medical specialty categories were the same over time on dealing with the family of a dying patient (Likert item 7).
Our findings suggest that physicians' attitudes toward death and terminally ill patients and their families have changed over 2 decades. Physicians seemed to be more open to informing patients of their prognoses in 1996 than in 1976. In addition, these physicians seemed more secure with dying patients than they were at the beginning of their careers. Differences in attitudes toward terminally ill patients were apparently determined much less by sex in 1996 than they were in 1976. Physicians' attitudes toward terminally ill patients and their families changed over time in 3 of the medical specialty groups we examined, but not in the low probability and low contact group; physicians' attitudes in the latter group changed only slightly. Overall, the physicians we surveyed appear to be distancing themselves less from their dying patients and their families than in 1976.
Corresponding author: George E. Dickinson, PhD, Department of Sociology and Anthropology, College of Charleston, 66 George St, Charleston, SC 29424 (e-mail: email@example.com).
Accepted for publication December 21, 1998.
This study was supported by a grant from the Faculty Research and Development Board of the College of Charleston, Charleston, SC.