SEARCH

SEARCH BY CITATION

Keywords:

  • cancer;
  • oncologists;
  • communication;
  • truth telling;
  • professional norms;
  • culture

Abstract

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

Objective: To evaluate the attitudes of the European Oncologists to information disclosure to patients with advanced cancer, their self-reported behaviors, and the factors that influence both attitudes and behaviors.

Methods: ESMO members were invited to complete an online questionnaire to evaluate both attitudes and clinical behaviors relating to the disclosure of information to patients with advanced cancer. Data were analyzed to evaluate demographic, educational and social factors influencing attitudes and behaviors.

Results: Two hundred and ninety-eight completed surveys were returned. The survey demonstrated strong internal consistency construct validity. The responses indicate that individual clinicians generally display a range of behaviors including non-disclosive as well as disclosive behaviors depending on the dynamics of individual interactions between oncologist and specific patient. Although regional cultural norms influence oncologists' attitudes toward disclosure and, indirectly, their self-reported behaviors, the impact is influenced by other factors: in particular, perceived institutional professional norms, the degree of training in breaking bad news and the frequency of exposure to requests by family members to withhold information from the patient.

Conclusions: Positive attitudes regarding disclosure of information to patients and disclosive behaviors can be encouraged, even in non-Western countries, by the development of strong professional norms and education in breaking bad news and coping with the emotional responses of patients. Copyright © 2010 John Wiley & Sons, Ltd.

Communication is central to the task of caring for patients with advanced and incurable cancer. It is among the most challenging aspects of the oncologist's professional role 1, 2 and a source of substantial professional stress 3–6. Patients with advanced and incurable cancer are dealing with the emotional impact of a life-threatening illness, complex treatment decisions with often limited likelihood of major benefit, and the challenges of balancing the maintenance of hope with realistic goals 7–9.

The ethical axiom of respect for persons is generally taken to imply that physicians ought to provide patients with as much information as they desire to facilitate participation in decision-making to the degree desired by the patient 10. Disclosure, in this context, refers to the imparting of information necessary to make informed decisions about ongoing care. The key elements of information necessary for patients to make informed decisions include: the diagnosis; the extent of disease; the range of therapeutic options available; the likelihood of benefit from each of the treatment options; the anticipated scope of benefit; and the likelihood of adverse effects or harm from the treatment and the potential severity of such adverse effects. Disclosure of information, therefore, is critical to the oncologist–cancer patient relationship.

Although patients vary in the degree to which they want to be medically informed and to thereby participate in the decision-making process, surveys of patient preferences indicate that the overwhelming majority of patients in Western countries 11–19, and substantial proportion of patients in non-Western countries 13, 20–38, want disclosure of information to participate in the decision-making process. Universally, recent survey data of patient preferences demonstrate substantial heterogeneity of preferences that are not individually predictable by geography, culture, age, race, sex or educational level 11–15, 19–37, 39. Furthermore, studies that have evaluated the concordance between patient preferences on one hand and family member perception of patient preferences on the other hand have demonstrated that family members commonly underestimate the amount of information that patients desire about their condition and the degree to which they want to be involved in decision-making 32, 37, 40–44.

Based on this data, there is a very wide consensus of ethical 45–47, medical 15, 23, 48–51, psychological 52, legal 53, and anthropological 54 opinion that physicians should ask patients about their individual preferences regarding disclosure of information and then act in accordance with the patient's opinion unless there are compelling contraindications.

In situations where patients do request information, physician non-disclosure or less than full-disclosure may be motivated by a number of reasons: concern for harm as a result of excessive information or full exposure to untoward information; concern that disclosure may be culturally inappropriate; requests by family members to limit the scope of disclosure; and professional role expectations 55–58. Other confounding factors may include lack of time to devote to these complex dialogs 59 and personal difficulty in dealing with ‘bad news’ dialogs either because they don't know how or because they may have emotional difficulty with such conversations 60, 61.

Previous surveys of oncologists' attitudes and practices regarding disclosure have indicated that oncologists vary substantially in the disclosure practices 55, 61–67; further, and the degree of disclosure with any one patient is highly influenced by individual factors 65–68. Overall, survey data indicate that oncologists from Western countries are more disclosive than those practicing in non-Western countries 61, and that other factors such as sex 61, age 61, training in the communication of bad news 4, 69–72, and frequent requests from family members for non-disclosure 40, 55, 73 may affect attitudes and actual clinical behaviors.

Even in Western countries, where most oncologists indicate a high willingness to disclose, patient-derived data indicate that such disclosure is often less than complete 74–78 and less than patients want 75, 76. As adverse consequences of poor communication in this clinical setting are substantial—including psychological distress to the patient and their family, costs to the health care system derived from unnecessary treatment or overly aggressive treatment, and indirect system distress, including burnout, stress and conflicts within the health care team 79—it is paramount that we further refine this element of care.

To better understand the barriers to disclosure of information by oncologists and factors that may improve disclosure practices, the Palliative Care Working Group and the executive of ESMO undertook to study European Oncologists' attitudes toward information disclosure to patients with advanced cancer, their self-reported behaviors in this clinical setting, and the factors that influence both attitudes and behaviors.

Method

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

A survey tool to evaluate both attitudes and clinical behaviors was developed based on a focus group of oncologists participating in the Palliative Care Working Group. The tool was reviewed by oncologists within the working group, acknowledged experts in communication and psycho-oncology, and the executive of ESMO for content validity.

The final version of the survey tool (Appendix 1) addressed demographic factors (questions 1–7); requests for non-disclosure of diagnostic and prognostic information either by the patient or by the patient's family members (question 9); four representative clinical scenarios (questions 8, 10, 11, and 12); single questions relating to information aids, enquiries abut emotional issues, second opinions and divergent opinions (questions 13, 14, 15 and 16); and 31 attitude items (question 17).

The four clinical scenarios were created to evaluate clinical behaviors that would apply during typical oncologist–patient interactions: when disclosing the diagnosis of an incurable cancer (question 8); when family members request non-disclosure of a diagnosis of advanced cancer to a competent patient (question 10); when chemotherapy is not working and the likelihood of benefit of further therapies is small (question 11); and when discussing treatment options with a patient suffering from advanced incurable cancer that is relatively resistant to chemotherapy (question 12).

Among the 31 items in question 17 were four embedded questions that did not relate to individual physician attitudes: two relating to oncologist education in breaking the bad news (17.18 and 17.30); one relating to local professional norms of practice regarding information transfer (17.9); and one relating to perceived patient satisfaction with communication (17.24).

In order to best represent the variability of practice by individual oncologists, multiple options were presented for each of the clinical scenarios; respondents were asked to indicate the frequency with which they would use each of the options (in items 8, 10 and 12), or the likelihood that they would use each response (item 11) using a 5-point verbal rating scale. Attitude questions and embedded questions were also structured using a 5-point verbal rating scale to identify the strength of agreement or disagreement.

For the case scenarios, clinical behavior items and attitudes, responses endorsing disclosure and shared decision-making were scored positively, while options endorsing non-disclosure and/or paternalism were scored negatively on a 5-point scale of −2, −1, 0, +1 and +2.

All members of ESMO (n=3792), which includes oncologists from all parts of the world but predominantly from Europe, were invited by the President of ESMO (by email) to participate in the survey with a link leading them to the online survey form. The survey was presented in English, which is the official language of the organization. The invitation was followed by reminder letters every 2 weeks over a 2-month period in 2006. Data were coded and entered into an Excel © 2003 file in preparation for statistical analysis.

Frequency distributions were calculated for demographic parameters, responses to clinical behavior questions, attitudes, requests for non-disclosure of diagnostic and prognostic information either by the patient or by the family members, and education. The internal construct convergence validity of the questions was evaluated by calculating Cronbach's alpha for the clinical behavior questions (items 8, 10, 11, and 12); for the 27 attitude items; for local family norms of non-disclosure (items 9.3 and 9.4); and for education items (17.18 and 17.30). Individual items with suboptimal internal convergence validity were identified post hoc and withdrawn from subsequent analyses and from the development of pooled summary scores.

Predicated on the demonstrated internal consistency construct validity, pooled summary scores for ‘attitudes,’ ‘clinical behavior,’ ‘training,’ and ‘family norms’ were calculated for each respondent by averaging the scores for each sub-item.

A post-hoc analysis of the poolability of responses from oncologists of different geographic regions was undertaken, based on the analysis of the respondents' summary scores for ‘attitudes’ and ‘clinical behavior.’ Finally, factors contributing to attitudes and clinical behaviors were evaluated in a stepwise regression analysis.

In this report we present the results regarding the internal consistency construct validity of the survey tool and the findings and analysis of the factors that influenced the attitudes and behaviors of the responding oncologists regarding the truthful disclosure of information to patients with advanced and incurable cancer.

Results

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

Demographics

Two hundred and ninety-eight completed surveys were returned (response rate: 7.9%). The distribution by region was: Western Europe 112 responses (37.6%), Southern (Mediterranean) Europe 52 responses (17.4%), Eastern (Former Soviet Union and satellite nations) European 45 responses (15.1%), South America 39 responses (13.1%), Middle East 22 responses (7.4%), United States 5 responses (1.7%), Australasia 5 responses (1.7%), Other 16 responses (6%). Most of the responding oncologists were male (n=217 respondents; 72.8%). The median respondent age was 42 years, and the median experience in oncology was 10–14 years. The largest proportion of respondents worked in teaching hospitals (n=114 respondents; 38.3%). Other respondents worked in comprehensive cancer centres (n=79 respondents; 25.5%), community hospitals (n=56 respondents; 19.7%), or private oncology practice (n=42 respondents; 14.1%).

Overall, most respondents were involved in the care of patients with advanced cancer: two hundred and seven respondents (69.7%) reported that the management of patients with advanced and incurable cancer represented a substantial proportion of their practice, and another 71 respondents (22.0%) reported that such patients constituted most of their practice. Few respondents reported that patients with advanced and incurable cancer represented a small proportion of their practice (6.4%), and very few oncologists had no involvement in the management of patients with advanced cancer (0.3%).

Response frequencies

Attitudes

Frequency distributions of the 27 attitude items are presented in Table 1. Based on the frequency distribution of responses, attitude items are presented in three groups: 13 items in which there was substantial affirmative consensus (>60% agree or agree strongly), six items in which there was substantial negative consensus (>60% disagree or disagree strongly), and eight items in which there was little consensus and substantial diversity of opinion.

Table 1. Frequency distribution for the 27 attitude items incorporated in question 17
 Agree strongly (%)Agree (%)Don't know (%)Disagree (%)Disagree strongly (%)
Attitudes items with substantial affirmative consensus (>60 agree or agree strongly)
Competent patients should decide who receives information about their diagnosis60.435.02.51.40.7
Most of my patients expect me to tell them what is the most appropriate treatment38.955.42.92.90
How I talk to my patients is just as important as the information that needs to be conveyed43.249.65.00.71.4
Patients values and life philosophy should be taken in to consideration in considering treatment planning41.850.76.11.10.4
When the outlook is poor, family members should be told.35.756.44.62.90.4
I try to guide patients to the treatment that I feel is best for them33.957.95.42.90
I encourage my patients to be actively involved in their treatment decisions28.660.36.83.90.4
Supportive touch helps me to convey a message of care even at the most difficult of times28.256.411.43.20.7
When the outlook is poor, this should be explained to the patient.22.961.18.66.41.1
I control my irritation and anger when a patient is angry with me19.364.311.13.91.4
It is best to give information to patients stepwise rather than ‘all at once’35.747.18.97.50.7
When discussing prognosis, I try to be as direct, clear and specific as possible20.755.412.19.62.1
Most of my patients want to be actively involved in the decision making about their treatment15.450.410.722.51.1
Most family concerns about disclosure of information to the patient are exaggerated15.450.020.011.82.8
Attitude items with substantial negative consensus (>60 disagree or disagree strongly)
Its not my job to have to talk about palliative care2.13.22.530.061.1
I feel irritated by patients who come with lists of prepared questions1.16.87.157.127.9
I hesitate to refer my patients to a mental health professional since it may offend them3.611.111.449.324.6
Family members, who know the patient best, should decide how much information the patient should receive.3.916.48.647.523.6
The more patients know about a bad prognosis, the worse they cope5.416.117.944.518.2
What patients don't know, does not hurt them2.513.921.138.623.9
Attitude items without overall consensus
When discussing prognosis, I try to be as general as possible7.548.913.227.52.9
It is important to convey hope for recovery, irrespective of the situation18.237.912.526.84.6
Good quality information derived from the internet or other sources helps patients make good decisions about their care11.843.624.317.13.2
The treating physician should decide how much information the patient should receive10.443.210.728.97.9
Patients who know too much about their disease are difficult to work with.5.426.814.343.99.6
When the outlook is poor, family members should be asked if the patient should be told7.123.912.139.317.5
I don't have enough time to deal with my patients emotional issues5.025.79.338.922.1
Clinical scenarios

Scenario 1 (item 8) asked about the frequency of use of three strategies in disclosing a new diagnosis of advanced or incurable cancer. The most commonly endorsed strategy was to tell the patient and family together (63%often/always, 14% rarely/never). Less frequently respondents endorsed telling the diagnosis initially to the patient alone, and then to the family, if the patient permits (41% often/always, 26% occasionally, 33% rarely/never), or telling the diagnosis initially to the family and then to the patient, only if the family permits (26% often/always, 15% occasionally, 59% rarely/never).

Scenario 2 (item 10) asked about the frequency of use of five strategies when family members request non-disclosure of a diagnosis of advanced cancer to a competent patient. Most respondents endorsed the response ‘I try to explain that it is the patient's best interest to be given any information he/she requests and that concealment usually harms more than it helps’ (80% often /always) and refuted the options ‘I try to respect their wishes and tell the patient they don't have cancer’ (they have an infection or some other chronic condition) (85% rarely/never) or ‘I try to respect their wishes and avoid discussing the issue with the patient’ (73% rarely/never). There was substantial heterogeneity regarding the frequency with which respondents would ask the patient what he/she would like to know (53% often/always, 20% occasionally, 27% rarely/never) or give the patient partial information using obfuscating language (27% often/always, 25% occasionally, 47% rarely/never)

Scenario 3 (item 11) asked about the frequency of use of seven communication strategies when discussing ongoing care when chemotherapy is not working and that the likelihood of benefit of further therapies is small. The four most commonly endorsed strategies were to: ‘Explain the situation and offer a range of options including palliative care, alternative chemotherapy, and experimental therapies (if available)’ (73% often/always, 18% occasionally), ‘Explain the situation and ask how the patient and his family are coping and what their major concerns are’ (64% often/always, 20% occasionally), ‘Tell the patient that the chemotherapy is not working and suggest an approach of palliative care only since the likelihood of harm is greater than the likelihood of benefit from further chemotherapy’ (60% often/always, 27% occasionally) and ‘Explain that conventional treatments are no longer working and suggest consideration of an experimental treatment (if one is available)’ (40% often/always, 46% occasionally). A smaller proportion of respondents commonly endorsed strategies to: ‘Explain that you are sorry that here is nothing more that you can do and suggest a hospice or palliative care referral’ (17% often/always, 29% occasionally) or ‘Avoid a confrontation that may undermine the patient's hope and offer and suggest another course of chemotherapy’ (7% often/always, 20% occasionally).

Scenario 4 (item 12) asked about the likelihood of use of seven strategies in a discussion with a 45-year-old man with metastatic cancer of the pancreas with multiple liver metastases. Most respondents reported that they would ‘Ensure that the patient and his wife understand the nature and extent of the disease’ (88% often/always), ‘Explain that chemotherapy may make him feel better’ (85% often/always), ‘To explain that metastatic pancreas cancer cannot be cured’ (79% often/always) and not ‘To prescribe chemotherapy without an explanation’ (94% rarely/never). These was substantial diversity of practice among respondents as to whether they would ‘explain that the major aim of the treatment is palliation and that this can be achieved with or without chemotherapy’ (62% often/always, 18% occasionally, 20% rarely/never), ‘ensure that the patient not give up and tell him that you will help him get well’ (48% often/always, 14% occasionally, 38% rarely/never), ‘explain that this condition is best treated with a trial of chemotherapy’ (51% often/always, 32% occasionally and 16% rarely/never) or to ‘explain that his life expectancy is unlikely to be different with or without chemotherapy’ (39% often/always, 28% occasionally and 33% rarely/never).

Requests for collusion in non-disclosure: family norms (Question 9)

Overall, respondents reported that the requests by family members to withhold information about prognosis (55% often/always) or diagnosis from the patient (49% often/always) were much more common than requests by the patient to withhold similar information from family members (<5% often/always)

Training

Oncologists' training regarding issues of disclosure to patients with advanced and incurable cancer were addressed in two items. The item ‘in my oncology training I received it good training in breaking bad news’ (17.18) addressed cognitive elements of training. The item ‘I don't feel trained to deal with my patients' emotional problems’ (17.30) addressed the affective elements of communication education. Response frequencies for these items are presented in Table 2.

Table 2. Frequency distribution of the embedded items in question 17 regarding education in communication of bad news
  Agree strongly (%)Agree (%)Don't know (%)Disagree (%)Disagree strongly (%)
17.18In my oncology training, I received good training in breaking bad news6.833.911.830.417.1
17.30I don't feel trained to deal with my patients emotional problems1.418.218.647.913.9
Professional norms

Perceived local professional norms were evaluated in a single item (17.9), embedded within the bank of attitude questions. The item stated ‘In my work setting I am expected to discuss matters first with the family and only later with the patient’ and asked respondents how strongly they agree or disagree with that assertion on a 5-point verbal rating scale (Table 3).

Table 3. Frequency distribution of the embedded item in question 17 regarding perceived local professional norms regarding disclosure of information to the patient
  Agree strongly (%)Agree (%)Don't know (%)Disagree (%)Disagree strongly (%)
17.9In my work setting, I am expected to discus matters first with the family and only later with the patient6.119.35.035.434.3

Internal consistency construct validity and summary score calculation for attitudes, clinical behavior, training and family norms

Attitudes

Cronbach's alpha for the 27 attitude items (excluding the embedded questions) was 0.79.

Clinical behaviors

For the four clinic clinical scenario questions (8, 10, 11, and 12), items with suboptimal internal correlation were identified and excluded from the analysis (8.3, 11.4, and 12.8). After exclusion of those items, the Cronbach alphas for the four individual clinical scenarios were 0.62. 0.79, 0.53 and 0.69, respectively. The responses for all of the non-excluded items from the four clinical scenarios were highly correlated, with a Cronbach alpha of 0.76.

Training

The two items that addressed the adequacy of training (17.18 and 17.30) were modestly correlated (Cronbach alpha 0.5, Spearman rho =0.3).

Family norms to request withholding of information: Items 9.3 and 9.4 addressed how frequently respondents experienced requests from family members to withhold information about the diagnosis or prognosis from the patient. These items had a Cronbach alpha of 0.9.

Poolability of regions into distinct clusters

The correlation of the responses from respondents form different regions to ‘clinical response’ and ‘attitude’ summary scores was evaluated.

By univariate analysis of variance (ANOVA), there was no significant difference in the summary scores for ‘clinical response’ and ‘attitude’ summary scores between Western Europe, USA and Australasia. These were pooled as a ‘Western’ cluster (Table 4). Similarly, there was no significant difference in the summary scores ‘clinical response’ and ‘attitude’ between respondents from Eastern Europe, Southern Europe, South America and Middle East, and these were pooled as a ‘non-Western’ cluster (Table 4). Univariate ANOVA comparing the Western and non-Western clusters for ‘clinical response’ and ‘attitude’ demonstrated highly significant differences between the two regional clusters (p<0.0001; Figure 1)

thumbnail image

Figure 1. Histograms for summary scores of attitudes and clinical behaviors of respondents from Western and non-Western region clusters

Download figure to PowerPoint

Table 4. ANOVA clinical behaviors and attitudes summary scores to evaluate the poolability of responses from regions into WEST and non-WEST clusters
 Clinical behaviors indexAttitudes index
RegionNMeanMedianSDMeanMedianSD
West
W Europe1121.0281.0880.3520.6780.7410.322
US51.2321.3330.2990.6150.6300.150
Australasia51.1551.1330.1180.8740.8880.093
   ANOVA P=0.33  ANOVA P=0.35 
Non-West
Eastern Europe520.4960.4130.4200.3900.4070.259
Mediterranean Europe450.5340.6250.6770.3170.4070.390
S America390.3960.4330.4250.2880.2960.272
Mid East220.2310.3170.4510.2330.2030.336
   ANOVA P=0.11  ANOVA P=0.31 

Responses from the 16 respondents identified as ‘Other’ region were heterogeneous (showing a bimodal distribution of responses) and could not be pooled into either region.

Multivariate analyses

Stepwise regression analyses of the factors contributing to ‘attitude’ and ‘clinical behavior’ summary scores were performed.

Attitudes

The model included demographic variables of age, sex, years of experience, work setting, proportion of work dealing with advanced cancer, region (West vs. Non-West), perceived local professional norms (17.9), and indices for family norms of non-disclosure and education in breaking bad news. Overall, the model accounted for 44% of the observed variance. Significant contributing factors to having positive attitudes toward disclosure included (in descending order), local professional norms, region WEST, training index, frequency of exposure to patients with advanced cancer, younger age, and less frequent exposure to family members requesting non-disclosure. Results are presented in Table 5.

Table 5. Multivariate stepwise ANOVA of factors influencing to attitudes index
 Coef.Std. Errort95 Conf intervalp
  1. R2=0.4412.

Local professional norms paternalism−0.1280.0167.95−0.096−0.159<0.0000
Region WEST0.0770.0441.73−0.0110.1640.0002
Training Index0.0640.0183.580.0290.0990.0042
Frequent exposure to pts with Adv Cancer0.0910.032.990.0310.1510.0078
Age−0.0040.002−2.26−0.00700.0276
Family requests not to disclose−0.0440.022−2.03−0.087−0.0010.0402
Clinical behaviors

The model included demographic variables of age, sex, years of experience, work setting, proportion of work dealing with advanced cancer, region (West vs. Non-West), attitudes summary score, perceived local professional norms (17.9), and indices for family norms of non-disclosure and education in breaking bad news. Overall, the model accounted for 63% of the observed variance. Significant contributing factors to the clinical behavior index for disclosiveness included (in descending order), local professional norms, attitudes, less frequent exposure to family members requesting non-disclosure and the frequency of exposure to patients with advanced cancer. Results are presented in Table 6.

Table 6. Multivariate stepwise ANOVA of factors influencing to clinical behaviors index
 Coef.Std. Errort95 Conf intervalp
  1. R2=0.6324.

Local professional norms paternalism−0.1640.0217.91−0.123−0.204<0.0000
Attitudes0.5830.0738.020.4400.726<0.0000
Family requests not to disclose−0.0820.025−3.31−0.132−0.0340.0008
High exposure to patients with advanced cancer0.0790.0372.150.0070.1520.0306

Discussion

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

The low overall response rate, and the possibility of selection bias, makes it impossible to derive definitive data regarding practices and attitudes in different regions of Europe. Nonetheless, the data collected from the near 300 individual respondents was sufficiently substantial to enable an analysis of the factors that influence the individual attitudes of oncologists relating to disclosure practices and their self-reported patterns of behavior.

The research tool developed for this study underwent extensive pre-utilization peer review for face validity. Post-hoc evaluation indicated strong internal consistency construct validity for each of the major subsections, in particular the 27 attitude items and the four clinical scenarios that were the basis for the clinical behavior summary scores. Compared with previously published survey tools 40, 61–64, 80 this tool provided substantially greater detail regarding attitudes and self-reported clinical behaviors.

The major findings of this study are that individual clinicians generally display a range of behaviors including non-disclosive and disclosive behaviors depending on the dynamics of individual interactions, and those non-disclosive behaviors are widely prevalent. These data also demonstrate that clinician behaviors are strongly influenced by a range of attitudes or values including perceived institutional professional norms, the norms of the country in which they work, the extent of training in breaking bad news, the frequency of exposure to requests by family members to withhold information from the patient, the oncologist's experience in working with patients with advanced cancer and age. The strength of these influences is indicated in the high proportion of variance (R2) that is accounted for in the multivariate models for attitudes and clinical behaviors of 0.44 and 0.63, respectively.

There is strong anthropological 81 and social psychology 82, 83 data indicating that different regional cultures have different value orientations that influence local attitudes and norms of behavior. This study indicates that although regional culture influences oncologists' attitudes toward disclosure and, indirectly, behaviors, the impact is powerfully influenced by other factors, particularly by perceived institutional professional norms, the degree of training in breaking bad news, and the frequency of exposure to requests by family members to withhold information from the patient.

The impact of local professional norms and education are particularly salient since they are amenable to change. The impact of the respondents' perception of local professional norms regarding disclosure on their attitudes and behaviors was substantial and highly significant (p<0.0001) and, importantly, it was independent of local culture. Irrespective as to where they worked, in Western on non-Western countries, physicians who reported a strong institutional culture of disclosure tended to have both attitudes and behaviors conducive of disclosive communication and shared decision-making. In non-Western countries, almost 50% of respondents either disagreed (38.6%) or disagreed strongly (10.8%) with the assertion ‘In my work setting, I am expected to discuss matters first with the family and only later with the patient’ and these norms powerfully influenced their attitudes and behaviors as compared with respondents who agreed (32.3%) or agreed strongly (10.8%). The impact of perceived professional norms has not previously been studied in this way. Interestingly, Surbone has reported that changes in professional norms and expectations have substantially influenced the changes disclosive behaviors by physicians in Italy over the past two decades 23.

The impact of professional norms is augmented by education. The data also suggest that younger physicians as well as physicians whose work involves a large proportion of patients with advanced incurable cancer are more likely to have positive attitudes and behaviors regarding disclosure and shared decision-making.

Another important finding of this survey relates to the prevalence of requests by family members to have information regarding either diagnosis or prognosis withheld from the patient. The responding physicians from non-Western countries reported this phenomenon either often (64.6 and 53.8%) or always (10.1 and 23.4%) regarding the issues of the diagnosis and prognosis, respectively. Responding physicians from Western countries also reported this phenomena, but with far lower frequencies; often (13.9 and 23.0%), always (0 and 1.6%). Universally, requests by patients to have information withheld from family members were distinctly uncommon. The frequency with which physicians were requested to withhold information was an important factor that impacted on both attitudes and practices regarding disclosure. The relationship has been previously described in a survey of Turkish oncologists 40, a qualitative study of Western oncologists 55 and reports from Iran 73. We hypothesize that the frequency of these requests is an indirect indicator of prevailing cultural norms and that these norms have a very strong impact on the behaviors and expectations of family members and a variable impact on patient and physician attitudes and behaviors.

These findings support the importance of regional culture as a factor that influences the attitudes and behaviors of individual oncologists. Among the responding physicians from non-Western countries, 20–25% were extremely non-disclosive and this phenomenon was not observed among the respondents from Western Europe, Australia and the United States. Indeed, the combination of working in a non-Western country, frequent exposure to requests from a family members to withhold diagnostic and prognostic information from patients, a lack of training in communication issues regarding breaking bad news and local professional norms that supported non-disclosure practices often went hand-in-hand.

Disclaimers and limitations

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

This survey of ESMO oncologists had a low response rate (<10%) and the responding clinicians are self-selected and may not be representative. The demographics of the respondents, however, are similar to those of previous ESMO surveys 84 with regard to age, sex, experience, the frequency with which respondents care for patients with advanced an incurable cancer and places of work. The age and sex distribution of the participants was representative of the ESMO membership overall. The low response rate precludes definitive analysis of the frequency of behaviors and attitudes or of comparisons between individual regions. The survey was presented in English and this may have compromised participation by clinicians who are not fluent in English.

As behaviors are self-reported, it is not possible to confirm that the reported frequencies regarding the use of the various communication strategies accurately reflect actual practice. Indeed, we are aware that physicians often report higher levels of disclosure than the patients actually perceive 18, 78, 85. In this survey, we attempted to capture the variability of physician responses by asking physicians to indicate the frequency with which they would use a range of common strategies rather than asking to select any one preferred strategy. We, as well as the expert reviewers of the communication questionnaire, felt that this approach was more likely to approximate the variability seen in real clinical practice.

The local professional norms and expectations were derived from the self-reported perceptions of the surveyed physicians. They were not independently verified. Although they may be accurate, it is also certainly possible that these self-reported perceptions may be at variance with actual local professional norms and that the reported expectations may be self-serving justifications for the physicians, prevailing attitudes and practices.

The association between geographic region and culture is imputed. Data were not collected regarding specific racial, religious, or cultural background of the responding clinicians.

The education parameters evaluated in this study are indirect and we have no data regarding the actual educational experiences in breaking bad news, in disclosure, shared decision-making and managing the emotional responses of patients to such discussions. Despite this limitation, physicians who felt that they had been well trained and well prepared for this sort of discussion clearly had much more positive attitudes toward disclosure and shared decision-making and this was also reflected in that their self-reported clinical behaviors.

Conclusions

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

These data demonstrate the profound variability of physician behaviors in disclosiveness between physicians, and for any one oncologist, between different patients. Despite the limitations of this study, our data add to the preexisting body of evidence that even when patients express a desire for information, many oncologists are often unwilling, unable or uncomfortable to honestly disclose and discuss information regarding diagnosis, likelihood of benefit from treatments and prognosis.

The findings of this study indicate that positive attitudes regarding disclosure of information to patients and disclosive behaviors can be encouraged, even in non-Western countries, by the development of strong professional norms and education in breaking bad news and in coping with the emotional responses of patients. This study underscores the experience that clear policy regarding expected norms of behavior are educative in themselves and that they can influence attitudes and behaviors.

Appendix A

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

ESMO survey of Medical Oncologist attitudes and practice in communication with patients suffering from advanced cancer

  • equation image
  • equation image
  • equation image
  • equation image

Acknowledgements

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References

We wish to acknowledge the president and executive of the European Society of Medical Oncology who supported this project, Jimmie Holland and James Tulsky who critically reviewed the draft of the survey: Robert Wesley from the dept of statistics in the Clinical Center of the United States National Institutes of Health who assisted in the statistical analysis of the data, and Moshe Spero, Robert Arnold, James Tulsky, Phyllis Butow and Antonella Surbone and Lesley Fallowfield who read and constructively criticized previous drafts of this paper. This survey did not involve subject experimentation. It was approved by the executive of the European Society for Medical Oncology.

References

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. Disclaimers and limitations
  7. Conclusions
  8. Appendix A
  9. Acknowledgements
  10. References