Introduction

Bad news is any information that drastically changes a person’s view of the future in a negative way [4]. Examples include news that an illness or loss of function will be chronic or permanent [41]; that the necessary treatment will be burdensome, painful, or costly; or that the prognosis for continued life is shorter than expected [35]. In the case of an incurable illness such as cancer, physicians frequently must break bad news to patients and their loved ones [1], yet medical education typically offers little formal preparation for this daunting task [15, 25].

Traditionally, the field of medicine was guided by a paternalistic patient care model. Hippocrates, for example, recommended that physicians be wary of giving bad news because the patient may “take a turn for the worse” [18]. Percival gave a similar warning in Medical Ethics [33], as did the American Medical Association in its first code of medical ethics [41]. In more recent years, however, paternalism has been replaced by shared decision making, a doctor–patient relationship model that emphasizes patient autonomy and full disclosure [28, 30, 41].

In most Western countries today, only under rare circumstances is nondisclosure of bad news considered to be ethically permissible [6]. Most cancer patients expect to be fully informed of their diagnosis and involved in decisions about their medical care [8, 40], and physicians who fail to disclose information may lose the trust of these patients [37]. Indeed, a review of studies on patient preferences found that from 50 to 90% of terminally ill patients desired full disclosure [22]. Research indicates that patients have higher levels of satisfaction and lower levels of anxiety and distress when physicians deliver information honestly and effectively [2, 38]. It also indicates that honest disclosure allows patients to make health care decisions that are better informed and are consistent with their own goals and values [6].

Yet the recent shift toward shared decision making is still at a somewhat inchoate stage. Not only do many physicians have difficulty talking with their patients about a poor prognosis [9, 27], but because a sizable minority of patients still do not want full disclosure, each physician faces the difficulties of ascertaining how an individual patient would like to have bad news handled [31]. As a result, patients are sometimes frustrated about not obtaining the information they require, and doctors are often frustrated about patients not voicing their concerns and preferences about receiving information [21].

To effectively address the concerns of patients and doctors, it is important to gain a better understanding of the modes of discourse that are used in doctor–patient communication and that have a direct impact on how each person perceives and understands the other. Although there have been some investigations of patient preferences about obtaining information, few studies have shed light on how physicians actually talk with cancer patients and their family members. Our study was designed to employ qualitative methods to examine the language that oncologists, incurable cancer patients, and patients’ kin use when they talk about death and about treatment-related and disease-related prognosis.

Patients and methods

Study population and setting

Our study was part of a larger project that investigated whether giving cancer patients a prompt sheet would encourage them to ask questions during their initial visit with an oncologist [3, 17]. Both the larger project (which included 319 cancer patients and their 9 oncologists) and the current study (which included 29 of the 118 incurable cancer patients and their 6 oncologists) were approved by the Ethics Committees of the Central Sydney Area Health Service, the Western Sydney Area Health Service, and the University of Sydney. All participants provided written informed consent.

Patients were eligible to participate in the larger project if they had received a diagnosis of cancer and were scheduled for an initial visit with an oncologist who worked in one of the two outpatient clinics of a university teaching hospital in Sydney. Patients were excluded from participation if they were under 18 years old, did not speak English, had advanced incapacity, had a life-threatening illness other than cancer, or were not available for the follow-up duration.

With permission from participating patients, consultations were audiotaped to allow for analysis of the information. The researchers indicated that each patient would be given a copy of the audiotape within 1 week of the consultation. Only three patients refused audiotaping primarily because they were upset and felt it would be intrusive.

For the current study, we randomly selected transcripts from the total pool of 118 incurable cancer patients and ceased when we reached theoretical saturation (i.e., no new themes emerged during analysis). The 29 selected transcripts included verbatim statements from 29 incurable cancer patients, the 6 oncologists who consulted with them, and the patients’ kin (family members) who were present during the consultations. The transcripts were imported into Ethnograph (Qualis Research, Denver, CO) before the analysis began.

Development of coding criteria

Two study investigators created preliminary criteria for coding the discussions in the transcripts. To test the criteria, they independently applied them to the coding of numerous randomly chosen transcripts from incurable cancer patients. In cases in which there was disagreement between coders regarding coding of transcribed segments, a physician was consulted, and then the criteria were refined. This method was continued until the two coders reached 100% agreement, and a final manual of criteria was prepared and applied to all 29 transcripts.

Data coding and analysis

The standard unit of analysis was a conversational turn, defined as a unit of text in which one person spoke for any period of time, regardless of the length of time. We analyzed each turn for the presence or absence of “prognostic talk,” defined as any talk concerning outcomes related to the disease or its treatment. We coded the conversational turns in terms of various features, including who talked about the prognosis (patient, oncologist, or patient’s kin), what the subject was (treatment-related or disease-related outcome), when the prognosis was discussed (before or after the patient’s examination), how it was discussed (in explicit or implicit terms), and what the focus of discussion was (estimated time frame, anticipated life span, or projected survival). We defined explicit talk as utterances that included terms such as die, dying, dead, death, terminal, and kill. We defined implicit talk as utterances that included either euphemistic language about death (e.g., “pushing up daisies”) or indirect language about prognosis (e.g., “limited time frame,” “life expectancy,” or “long-term survival”).

In tabulating the number of transcripts and conversational turns that contained prognostic talk, we included talk about the cancer-related prognosis of the patient under consultation (e.g., “I don’t think you’re going to be dead in 6 months”) and talk about the cancer-related prognosis of patients under similar circumstances (e.g., “Without treatment, half of a group of people like you would die within 6 to 9 months”). We did not include prognostic talk if it was unrelated to cancer (e.g., “Your blood pressure will allow you to go on forever”) or if it was unrelated to the patient under consultation (e.g., “My mother died of lung cancer at 60”). If a conversational turn included both explicit and implicit talk about death, we applied a code for each type of talk. If a conversational turn focused on some combination of time frame, life span, and survival, we applied a code for each focus.

Results

Demographic data

Tables 1 and 2 show the characteristics of the 29 participating patients and their 6 oncologists. All of the patients were seeing the oncologist for the first time, and 23 of them were accompanied by kin.

Table 1 Sociodemographic and clinical characteristics of the 29 patientsa
Full size table
Table 2 Sociodemographic and occupational characteristics of the 6 oncologists
Full size table

Of the 29 patients, about half (51.7%) were men; most (62.0%) were 60 years or older (range, 38–83 years; mean, 61.3 years); over half (51.6%) had at least a high school diploma; most (65.5%) were married; and most (72.3%) were employed outside the home. Although the patients had various types of cancer, in all cases, the cancer was characterized as terminal and metastatic. In about a third (31.0%) of the cases, the disease was recurrent. While 17 (58.6%) patients were estimated to live a matter of months, 6 (20.7%) patients were estimated to live a matter of years, and the prognosis was unknown in the remainder. In most cases (65.5%), the goal was aggressive palliative treatment.

Of the six oncologists, five (83.3%) were men. The oncologists ranged in age from 37 to 58 years, with a mean of 45.7 years. One specialized in radiation oncology, and the rest specialized in medical oncology. In the current study, the number of patients seen by the oncologists ranged from three to nine, with a mean of 4.8 patients.

Prognostic talk

As shown in Table 3, prognostic talk occurred in 23 of the 29 visits.

Table 3 Frequencies and examples of categories of language used by patients, oncologists, and patients’ kin during the 29 visits
Full size table

In the six visits that excluded talk of prognosis, none of the patients had recurrent cancer. In five of the six visits, kin were present. The most common recommendation was for tests (e.g., x-rays or blood tests) to monitor the extent of cancer progression or spread. The aim of treatment recommendations was usually at palliative goals such as management of pain, fatigue, and weight loss. A “wait and see” approach to options such as chemotherapy or radiation therapy was often employed, particularly if there were concerns about severe health-compromising side effects. In one case, for example, the oncologist stated: “The side effects of chemotherapy are less to do with sickness and hair loss but more to do with diarrhea and mouth ulcers and skin rashes.... You’d be much better off employed keeping yourself well.” As is evident in this statement, the visits that excluded talk of prognosis focused on addressing and preserving the quality of life (QOL) as opposed to delaying the onset of death.

In the 23 visits that included talk of prognosis, 9 of the patients had recurrent cancer. In 18 of the 23 visits, kin were present. Explicit talk of death occurred in nearly half (52.2%) of the 23 visits, whereas implicit talk occurred in all (100%) of the 23 visits.

Explicit talk of death

Of the 23 visits that included prognostic talk, explicit talk of death was used in 12, and kin were present in 10 of the 12 visits. For the 12 visits, there were 36 coded examples of explicit talk of death: 27 attributable to the oncologist, 5 to the patient, and 4 to the kin.

In all ten visits during which the oncologist used explicit talk of death, we found that this talk occurred after completion of the physical examination. Explicit talk was more likely to occur in cases in which the cancer was both metastatic and recurrent. The explicit talk of death was often in response to a patient’s explicit comment or question. In one exchange, for example, the patient asked the oncologist, “Are you telling me that I’m going to die?” The oncologist responded, “Yes, this disease will ultimately kill you.” In this instance, the oncologist answered the patient’s question in terms that were blunt and direct and clearly conveyed that, although the timing of death is uncertain, the patient will die from the disease. In cases in which the oncologist was not responding to a patient’s comment or question, we usually did not find this pithy type of discourse.

In most of the visits during which patients or kin used explicit talk of death, they usually did so early in the clinic visit, even before the patient was examined. This talk tended to involve emotionally charged questioning about whether the patient was going to die from the disease. In the two examples given below, after the question of death was raised, the oncologist did not probe for further discussion of emotions, nor did the oncologist make emotionally supportive follow-up comments. Instead, the conversation was redirected toward history taking.

In the first case, describing the response to learning some test results from another doctor, the patient told the oncologist: “I’m just, just climbing the walls, and I thought, shit, I’m going to die tomorrow, or am I going to die next week? ... I suppose it’s like being hit with a brick.” The oncologist’s response to this patient was, “OK, at the moment have you had any trouble with weight loss, with appetite problems, diarrhea, constipation?” In the second case, a family member stated, “I don’t like to hear in 6 months he’ll be dead.” The oncologist’s response, directed toward the patient, was, “OK, when did you have the operation?” In both cases, when the oncologist completed the examination and then brought up the subject of death, no reference was made to the patient’s or kin’s earlier talk of death.

Although oncologists used explicit talk more frequently than did patients and kin, the oncologists tended to downplay death as an outcome by coupling their explicit talk with euphemistic or indirect talk of death. For example, one family member asked, “So you’re saying, [clears throat] excuse me, it’s definitely terminal disease?” The oncologist responded: “I prefer not to use the word terminal. It sounds ... more emotional than it has to be. But it is going to shorten a life expectancy of someone at the age of 44.” The oncologist did not express emotional support of the family’s or patient’s concerns about the future and instead used the phrase “shorten a life expectancy” perhaps as a distancing mechanism.

Implicit talk of death

Implicit talk of death occurred in all 23 of the visits that included any prognostic talk, and kin were present in 18 of the 23 visits. Oncologists, patients, and kin all used implicit talk generally after the physical examination occurred.

In all, we coded 149 examples. Of these, 74 (49.7%) concerned estimated time frame, 40 (26.8%) concerned anticipated life span, and 35 (23.5%) concerned projected survival. Most implicit talk focused on only one of these descriptors, although time frame was combined with projected survival 12 times and with life span once.

Descriptor 1: estimated time frame

Implicit talk of estimated time frame occurred during 16 visits and involved discussion regarding the remaining time or period of time an individual was expected to have in relation to the disease or cancer treatment. The following are several examples of talk about time frame: “How long will I have?... A few months?” “You’ll have a year.” “Expectation of years is realistic.”

Most commonly, an estimated time frame was discussed in terms of the treatment adding time and the disease decreasing time for the patient. Typically, talk of disease-related outcomes occurred in conjunction with talk of treatment-related outcomes, as in this exchange:

Oncologist: Stomach cancer is relatively sensitive to chemotherapy.... If it shrinks, then there’s a reasonable chance you’ll be around the place in 12 months’ time, maybe longer than that. If, on the other hand, nothing is done or the treatment doesn’t work,... the chances of you being around Christmas time are, you know, iffy.

Kin: [So] if the treatment doesn’t work, you might have 2 months, and if the treatment does work—

Oncologist: [If it] does work you’ll have a year...

Less frequently, discussions concerning an estimated time frame included only talk about treatment-related outcomes. Instances included specific talk of how long a patient can look forward to with treatment, how long the treatment would keep the disease at bay, or how long each phase of treatment would last. Talk about extending the time frame was evident in the following conversation:

Oncologist: Chemotherapy might be tried. I think we’d have to say that it is unlikely to be a curative procedure.

Kin: Just to gain time?

Oncologist: Yep.

In relatively few cases, discussions about time frame focused on the disease-related outcome alone, without mention of treatment. In these discussions, time was usually estimated in terms of a number of months or years. For example, a patient asked, “How long will I have?... A few months?” The oncologist responded, “If it is the lung cancer, maybe 12 months or so.”

Descriptor 2: anticipated life span

Implicit talk of the anticipated life span (also sometimes referred to as the average life span) occurred during 20 visits and involved discussion regarding the remaining life an individual was expected to live in relation to the disease or treatment of the disease. Talk of life span included the use of phrases such as “length of your life,” “life-threatening disease,” “being alive,” “prolong life,” and “life expectancy.”

Usually, talk of the anticipated life span considered disease-related outcomes to be contingent on treatment-related outcomes. For example, referring to a large group of cancer patients, an oncologist told the patient:

You may live much longer than the average person with advanced colon cancer, just because the biology of the disease is favorable.... Chemotherapy can improve the chances of living longer, and [the evidence is] from this large group of people with multiple different types of disease.

Less frequently, talk of anticipated life span focused only on treatment-related outcomes. For example, regarding chemotherapy, a patient asked, “If I go on the treatment, I might live a bit longer, a bit longer, will I?” Rarely was the possibility that treatment would shorten the life span discussed. However, one oncologist stated, “Chemotherapy is a polite name for poisoning you within an inch of your life.”

Relatively few comments concerning life span focused only on disease-related outcomes. In most cases, these comments were made by patients or kin. For example, one patient stated: “I’ve never laid awake at night worrying too much about it [cancer]. I’ve pretty much been prepared to hope for the best and accept each day as a bonus and go for your life.” In response to an oncologist telling a patient, “Your biology of your disease may not be as bad as others,” the patient responded, “I like life, and I wouldn’t mind a bit more of it.”

In eight visits, anticipated life span was discussed in reference to QOL. In one case, for example, the oncologist stated: “We can rarely cure this type of problem. What we can do, though, is shrink the mass. It will sometimes go away completely, and we can increase the life span and the quality of life, but the tumor will usually come back.”

Descriptor 3: projected survival

Implicit talk of projected survival occurred during 11 visits and involved discussion regarding the act or process of surviving in relation to the disease or cancer treatment. Talk of projected survival included the use of phrases such as “chance of surviving,” “impact on survival,” “survival improving,” and “long-term survival.”

Usually, talk of survival centered on talk of disease-related outcomes in conjunction with talk of treatment-related outcomes. For example, an oncologist stated: “You’ve had the operation, you’ve had the chemotherapy, and your survival may have been prolonged because of your chemotherapy. Unfortunately, the disease came back.... So I think there is evidence that we can improve your survival with chemotherapy.”

Less frequently, talk of survival focused only on treatment-related outcomes. For example, another oncologist stated, “If you’ve got one area that could just be lopped out surgically, then that’s something worth pursuing because it might give you a chance of long-term survival.”

Least often, talk of survival focused only on disease-related outcomes. For example, yet another oncologist stated, “The survival, the long-term survival isn’t as good ... because you can develop the disease elsewhere, in particular the lungs.”

Discussion

In our study of the language used in 29 first-time encounters of oncologists and patients with previously diagnosed, incurable cancer, we found that the majority of visits (23, or 79.3%) included some prognostic talk.

Explicit talk, defined as talk that used the word “terminal” or variations of the word “death,” occurred in only about half (52.2%) of the 23 visits. Explicit talk of death was used mostly by oncologists in cases in which a patient had metastatic and recurrent cancer. When patients or their kin used explicit talk of death, they usually did so in emotional terms early in the clinic visit, even before the patient was examined, suggesting that they were eager to know about the patient’s prognosis. Even then, physicians responded by redirecting the discussion to history taking.

Implicit talk occurred in all (100%) of the 23 visits. Such talk by the oncologist, patient, or patient’s kin involved euphemistic or indirect death references, usually to an estimated time frame (49.7%) but sometimes to an anticipated life span (26.8%) or projected survival (23.5%). Most participants used alternatives to the word “death” (a term suggesting an abrupt break with life). For example, they referred to the prospect of dying as “not going to live,” or they referred to the disease as “life-threatening” or one that “shortens life,” suggesting that they recognized death as a possible outcome but wished to focus the discussion on remaining life. Similarly, participants preferred to talk of “gaining time” or chances of “survival,” terms that emphasize a continuation of existence and create a sense of distance from death.

In implicit talk, study participants were more likely to discuss treatment-related prognoses than disease-related prognoses. Patients tended to view treatment as a means by which time can be extended, and oncologists tended to offer statistical averages such as 5-year survival rates for patients with the same form and stage of cancer. Although some patients discussed remaining time in terms of how it would be used to reach a personal goal (e.g., spending one last holiday with kin), it was more common for oncologists and patients to discuss remaining time in terms of how it would be used to undergo testing or treatment (e.g., chemotherapy). The outcomes discussed were usually biomedical (e.g., tumor shrinkage or symptom management) rather than psychosocial (e.g., QOL).

Our findings are consistent with results of earlier studies. Studies have shown that doctors break bad news to cancer patients in a predictable and routine way. Regardless of a patient’s individual information needs or desires, doctors discuss the prognosis after talking about diagnosis, relevant evidence, need for further investigations, and treatments being considered [25]. Other studies have shown that doctors focus on treatment-related prognosis more than disease-related prognosis [16], possibly because this information has more “action-relevance” [29] or because treatments often need to be undertaken rapidly to be effective. Further, others have shown that doctors often focus more on relieving patients’ bodily pain than on relieving their emotional distress [36].

For physicians, the historical emphasis on the biomedical model in medical training places more value on technical proficiency than communication skills. Therefore, many physicians have little or no formal training about how to disclose prognostic information, and may feel unprepared for the onus of doing so [7, 25, 36]. It is understandable that a doctor would delay discussing the prognosis until after an examination because it makes sense to gather as much evidence as possible before addressing such a ponderous issue. It is not clear, however, why doctors in our study often ignored a patient’s or kin’s prognostic questions and did not provide assurance that their questions were important and would be addressed after the examination.

Barriers to effective communication of prognosis include physicians’ fears of being blamed by the patient [26], of not knowing all of the answers sought by the patient, or of inflicting pain on the patient [26] and their fears of having failed the patient [9, 35]. Compounding the issue is the fact that patients often have multiple physicians, making it unclear who should disclose prognostic information [4, 11, 26].

For cancer patients, one factor contributing to communication problems is that they vary considerably in what prognostic information they would like to receive and how they prefer it to be presented [23, 28]. Qualitative studies about information needs have identified several consistent patient concerns such as conflicts between wanting information and fearing bad news [20]. Yet which concern is most important to any given patient is highly variable, and few patient characteristics accurately predict which will be most important [2, 20].

As Franks observes, presenting bad news “is not an isolated skill but a particular form of communication” [14]. One reason for using indirect or euphemistic language is the desire to put bad news in a positive light or at least a neutral light [17]. By using implicit language and coupling this language with recommendations for treatment [29], physicians can soften the impact of emotionally traumatic information and help instill hope. Implicit language, however, is equivocal language and may confuse patients and prevent them from taking appropriate actions with respect to weighing the risks and benefits of treatment and to putting their affairs in order [19, 43]. Studies show cancer patients have problems understanding prognostic information in the form in which it is usually presented [16, 23]. This may be due in part to the nature of the information and in part to the manner in which physicians deliver bad news [1, 39]. Implicit language may further contribute to patient anxiety and depression, both of which are more likely to occur in those who have unresolved concerns and perceive that they have been given inadequate information [12, 32].

Although our study involved more physicians than did previous qualitative studies concerning doctor–patient communication, it included only a small sample of oncologists and patients, and the oncologists practiced in two large centers in which the majority of patients were individuals living in nonrural areas. These factors limit the generalizability of our results concerning communication about death. However, we believe that replication in other settings with a larger and more diverse sample of patients and physicians is warranted because insights into how physicians address the problems of discussing prognostic information can help break down the barriers to disclosing bad news to patients.

Learning general communication skills can enable health care providers to break bad news in a manner that is more comfortable for them and more satisfying for patients and their families [11]. But unless the providers are given systematic training, the breaking of bad news and discussions of cancer prognosis are likely to fall short of patients’ needs and expectations [5, 17] and may even have negative psychological consequences for patients [13]. Numerous studies have demonstrated that focused educational interventions improve the ability of medical students and residents to deliver bad news [10, 15]. Multidisciplinary workshops have also been reported to be successful in helping doctors and nurses acquire key communication skills [24].

However, most of the guidelines and recommendations that have been presented for breaking bad news are not evidence-based. In fact, fewer than 25% of the publications on breaking bad news are based on studies reporting original data [37, 42], and the clinical efficacy of many standard recommendations has not been empirically demonstrated [34, 42]. We believe that our qualitative analysis of patient–physician communication provides valuable insights that can be used as a basis for further research to identify how health care providers actually deliver bad news, to develop better ways to deliver it, and to evaluate what works best under particular circumstances. The ultimate goal is to deliver prognostic information in a manner that is honest and compassionate but is also understandable and useful to patients and their kin.