Patient Preferences for Adjuvant Chemotherapy of Early Breast Cancer: How Much Benefit Is Needed?

Transcription

1 Patient Preferences for Adjuvant Chemotherapy of Early Breast Cancer: How Much Benefit Is Needed? R. John Simes, Alan S. Coates Adjuvant chemotherapy for early-stage breast cancer has been shown to delay recurrence and improve survival. However, the benefits are modest and must be balanced against the adverse treatment effects. We assessed the size of the survival benefit needed to justify the toxicity of chemotherapy, based on the preferences of women who had previously received adjuvant cyclophosphamide, methotrexate, and 5-fluorouracil (CMF). We also attempted to identify circumstances in which larger survival gains would be needed. In semistructured interviews, 104 women who had received adjuvant CMF chemotherapy were asked to rate the survival benefit that would justify 6 months of such treatment, using a series of hypothetical trade-offs between shorter survival without treatment and longer survival with treatment. Similar preferences were sought for a greater probability of 5-year survival. Most patients considered 6 months of adjuvant CMF chemotherapy worthwhile for relatively modest survival gains: 77% considered an increase of from 5 to 6 years worthwhile, 74% thought an increase of from 15 to 17 years worthwhile, and more than 70% considered such treatment justified for a 5% greater chance of living 5 or more years. Smaller survival benefits were needed for women who had experienced less toxicity (P =.01), had not received initial radiotherapy (P =.01), had better social support (P =.02), and had others at home dependent on their support (P =.0001). Modest survival benefits are sufficient to justify adjuvant cytotoxic chemotherapy for most women with early-stage breast cancer. Individual preferences are important when weighing trade-offs between survival and adverse treatment effects. [J Natl Cancer Inst Monogr 2001; 30:146 52] Chemotherapy used as an adjuvant treatment for patients with operable breast cancer has been shown clearly to reduce the chance of the breast cancer recurrence and to improve overall survival (1). While the evidence of benefit is clear, the magnitude of survival benefit has been modest. Greater absolute benefits occur among younger patients and among those at greater risk of relapse, such as women younger than 50 years with positive axillary lymph nodes, for whom an increase in the chance of 10-year survival of approximately 11% might be expected. By comparison, for older women without lymph node involvement, the survival benefit is less clear and may be as little as 2% 3% (1). These modest improvements in survival rate might translate into gains in life expectancy of 1 3 years (depending on assumptions made about long-term effects). The benefit of improved survival and reduced risk of recurrence must be balanced against the side effects of chemotherapy, such as hair loss, nausea, tiredness, and risk of infection, with a resultant detriment to quality of life (2,3). Furthermore, these adverse effects usually occur early and are obvious to the patient, whereas the benefits of treatment may be delayed and less evident for the person. In this setting, preferences of individual patients on the relative importance of these outcomes may be crucial to optimal decision making about whether to give or withhold treatment. The views of patients who have actually experienced such treatment may be particularly helpful in deciding whether or not to treat future patients who are similar. Eliciting patient preferences to guide decision making is an important but complex process. Answers may vary according to how questions are asked and under what circumstances, as well as by whose views are sought (4,5). Surrogate decision makers may give answers that are systematically more conservative than those of patients (5 8). Healthy volunteers know neither the anxieties of having cancer nor the actual side effects of treatment. Even women with breast cancer who have not experienced chemotherapy may make judgments based on the worst possible side effects rather than on a reasonable average expectation (9). We therefore designed a study to ask women who had experienced adjuvant chemotherapy for early breast cancer what survival benefit would justify the treatment as they had experienced it (10). The primary aim was to assess the size of the survival benefit needed to justify the toxicity of adjuvant cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) chemotherapy. The study was also designed to identify possible patient and disease factors affecting whether larger survival gains were needed to justify adjuvant chemotherapy. PATIENTS AND METHODS Patients Women who had received at least three cycles of CMF chemotherapy as adjuvant treatment after local treatment for operable breast cancer and who were attending a clinic at the Royal Prince Alfred Hospital, Sydney, Australia, from November 1986 to December 1987 were approached about participating in the study. Patients who, having started such therapy, withdrew from it either by their own choice or by the decision of their doctor were also eligible to participate. Consent was obtained both from the patient and her doctor. Of 129 patients considered for participation in the study, nine were excluded because of insufficient comprehension of English, five were considered too ill to participate, three were geographically inaccessible, two died before the interview, and two were not asked. Thus, 108 patients Affiliations of authors: R. J. Simes, National Health and Medical Research Council Clinical Trials Centre, University of Sydney, Sydney, New South Wales, Australia; A. S. Coates, Department of Public Health and Community Medicine, University of Sydney, and Australian Cancer Society, Sydney, New South Wales, Australia. Correspondence to: John Simes, M.D., F.R.A.C.P., National Health and Medical Research Council Clinical Trials Centre, Mallett Street Campus, University of Sydney, New South Wales 2006, Australia ( enquiry@ctc. usyd.edu.au). See Notes following References. Oxford University Press 146 Journal of the National Cancer Institute Monographs No. 30, 2001

2 were asked to participate; 104 patients consented and completed the initial interview. Patient Interviews All 104 participating patients underwent a semistructured interview with one of two trained observers not involved in the patient s care at least 3 months after the completion of the adjuvant chemotherapy. Information was also obtained at this time on each patient s sociodemographic, disease, and treatment characteristics as well as on toxic effects experienced from chemotherapy. Patient preferences for adjuvant chemotherapy were elicited by using a series of hypothetical trade-offs between a lesser survival period without treatment and a greater survival period with adjuvant treatment plus its associated toxicity. Time Trade-off Questions Patients were presented with hypothetical scenarios of the general form: Suppose that without treatment you would live 5 years. Based on your own experience of chemotherapy, what period of survival would make 6 months of initial treatment worthwhile? Patients were then asked to express a preference between 5 years survival without treatment or a longer period of survival beginning with 6 months of adjuvant treatment, using a series of cards to represent each scenario. The second card with the longer period was then altered until the patient considered it to be roughly of equal value to the 5 years without treatment. This was referred to as the 5-year trade-off. A similar sequence was then followed to establish equivalence for a hypothetical patient at lower risk with an expectation of 15 years survival without treatment. This was referred to as the 15-year trade-off. Survival Rate Questions These questions were similar to the time trade-off approach but expressed the outcome of treatment in terms of percentage chance of remaining alive at 5 years. Patients were asked to express a preference between a 65% chance of 5-year survival without adjuvant chemotherapy and a higher chance of 5-year survival with 6 months of adjuvant chemotherapy. The higher chance of 5-year survival was again varied until it was considered roughly equal to the 65% chance of 5-year survival without treatment. Similarly, patients were asked to express a preference between 85% chance of 5-year survival without adjuvant chemotherapy and a higher chance with 6 months of adjuvant chemotherapy. Retest Interviews Where possible, patients were interviewed 3 6 months after the initial interview. Thirty-nine patients were not interviewed a second time: 12 patients refused a second interview, four had died in the interval, six were excluded because of poor comprehension or anxiety at the first interview, and 17 were lost to follow-up before the planned second interview. Results of the repeat interviews were used to assess the reliability of the measures used and to assess changes in preference over time. Interview Methods To assess possible framing effects of questions, the sequence in which alternatives were offered was randomly assigned. For the 5-year trade off questions, patients were initially offered a period of 6 or 10 years with treatment versus 5 years without treatment. In either case, the options were altered in response to the patient s reply until equivalence was established. For survival rate questions, framing effects were also assessed by randomly offering initial options at the high or low end of the expected response range, beginning with an extra 1% or with an extra 10%. Statistical Methods The results of the time trade-off and survival rate questions were skewed, and no effective normalizing transformation was available, so primary analyses were nonparametric. The cumulative proportion of those accepting chemotherapy as worthwhile was plotted for each size of survival benefit. The comparison of preferences for major groups was undertaken by using the Kruskal Wallis test. Patient and disease factors predicting individual preferences were assessed in multivariate linear regression analysis, in which the outcome used was a normal score associated with the rank of each individual s time trade-off responses (11). To avoid the problems of using multiple outcomes, we specified a priori the total score of the two time trade-off questions as the primary outcome in these analyses: the time trade-off total. All variables in Table 1, except those with low frequency, were included in the multivariate analyses. RESULTS Patient and Treatment Details The patient and disease characteristics of the 104 women interviewed are shown in Table 1. The median age was 49 years (range, years). Almost all of the women were treated with mastectomy and at least six cycles of CMF chemotherapy. CMF chemotherapy was associated with some severe nonhematologic toxicity in 15% of the patients and with moderate or severe nonhematologic toxicity in 62% of the patients (Table 2). Patient Interviews Fifty-five interviews were completed by the first interviewer; the remaining 49 were completed by the second interviewer. Reliability Test retest reliability was assessed by Spearman s rank correlation between first and second interview in 65 patients. The correlation coefficient was.68 for the 5-year time trade-off question and.64 for the 15-year time trade-off. For the 65% and 85% survival rate questions in interviews with 63 patients, the correlation coefficients were.63 and.69, respectively. These figures somewhat overestimate reliability, since some patients were excluded from a second interview because they had had problems at the first interview. There was no systematic change in the answers to time trade-off questions, but there was a statistically significant change to a larger increment in survival rate (by an extra 1% 2%) needed for adjuvant treatment in second interviews (P.003). Among patient responses to the 5-year trade-off, the correlation was higher when the retest interview was done by the same interviewer (0.75) than when it was done by the other interviewer (0.63). Framing and Other Effects No statistically significant difference was observed in any of the end points selected as a result of the sequence in which Journal of the National Cancer Institute Monographs No. 30,

3 Table 1, A. Baseline patient characteristics Patient characteristic Patients (n 104), % Age at interview, y < Married 73 Educational level Primary 12 Secondary 63 Tertiary 25 Employment Full 31 Partial 21 None 48 No. of others at home None Support needed by dependents Nil 43 Partial 37 Full 20 Support available to patient Nil 7 Partial 31 Full 31 Table 1, B. Disease and treatment details Disease characteristic Patients (n 104), % Tumor stage at diagnosis T1 16 T2 74 T3 8 Positive axillary lymph nodes Disease status at interview Disease free 81 Local relapse 5 Distant relapse 11 Both 4 Time from diagnosis to interview, y > Time from chemotherapy end to interview, y > Surgery Lumpectomy 9 Mastectomy 91 Radiation therapy 18 Adjuvant endocrine therapy 18 Oral CMF* 96 Intravenous CMF* 4 No. of cycles given < Percentage dose received of total planned >75 59 *CMF cyclophosphamide, methotrexate, and 5-fluorouracil combined. Table 2. Worst toxicity grade reached with adjuvant chemotherapy* Toxicity type None, % Mild, % Moderate, % Severe, % Alopecia Mucositis Nausea and vomiting Hematologic Other Worst nonhematologic *Based on World Health Organization criteria. alternatives were offered. Furthermore, in the multivariate analysis, preferences did not differ statistically significantly according to which interviewer was involved, how long after diagnosis or treatment the interview was undertaken, or whether the patient s breast cancer had recurred before the interview. Patient Preferences A large majority of the patients felt that relatively modest improvements in survival duration or in the percentage chance of 5-year survival would justify 6 months of the treatment they received. This was true both in the relatively optimistic scenarios, with an untreated survival duration of 15 years or 5-year survival rate of 85%, and in the less favorable scenarios, with an untreated survival expectation of 5 years or a 65% 5-year survival rate. Details of the percentage of responding patients accepting that treatment would be worthwhile at various trade-off points are displayed in Fig. 1 and in Tables 3 and 4. A majority of the patients considered relatively small survival gains to be sufficient to justify treatment, and a substantial minority of the patients considered only 6 months of extra survival enough. Thus, 46% of patients considered a survival period of 5.5 years with treatment equivalent to 5 years without such treatment, and 39% of patients would accept a similar increment even with an expected survival of 15 years (Fig. 1, A; Table 3). Furthermore, 77% considered an increase from 5 to 6 years worthwhile, while 74% thought an increase from 15 to 17 years worthwhile. Larger survival gains were needed for the scenario with the longer (15-year) survival, indicating that women were discounting benefits of treatment that were appreciably delayed. Results of the survival percentage trade-off were more extreme, with almost one-half of the women judging a 1% improvement in 5-year survival probability as justifying treatment, whether the expected 5-year survival without treatment was 65% or 85%. The minimum extra survival at which a majority of patients would accept adjuvant treatment was 1 additional year, whether the baseline was set at 5 years or at 15 years (Table 3), with an additional 2% for each of the survival rate trade-offs (Table 4). Importantly, for some women even very large survival benefits would be insufficient to justify the toxicity of treatment. Factors Affecting Patient Preferences In a multivariate analysis of all prespecified baseline factors, statistically significant predictors of stronger preferences for adjuvant treatment (with smaller survival benefits needed) were less toxicity from chemotherapy (P.01), not receiving radiotherapy as part of the initial treatment (P.01), full-dose chemotherapy (P.02), having better social support (P.02), and having others at home dependent on their support (P.0001). 148 Journal of the National Cancer Institute Monographs No. 30, 2001

4 Table 3. Time trade-off decision points: minimum expected additional survival needed for patients to accept adjuvant chemotherapy treatment Preference, compared with living without having adjuvant therapy for Expected additional survival needed for patient to accept adjuvant chemotherapy 5y* 15y No. Cumulative % No. Cumulative % 0 3 mo mo mo y y y y No increase enough *Preference of one patient not obtained. Preferences of two patients not obtained. 3 6 months indicates that more than 3 months but not more than 6 months additional survival is needed to justify adjuvant chemotherapy. Table 4. Survival percentage trade-off decision points: minimum additional chance of 5-year survival (%) needed for patients to accept adjuvant chemotherapy treatment Compared with a chance of living at least 5 y without having adjuvant chemotherapy Additional chance, %, of 5-y survival needed for patient to accept adjuvant chemotherapy 65% Chance* 85% Chance No. Cumulative % No. Cumulative % No increase enough *Preferences of five patients not obtained. Preferences of six patients not obtained. 1 2 months indicates that more than an additional 1% chance but not more than an additional 2% chance of 5-year survival is needed to justify adjuvant chemotherapy. Fig. 1. A) Proportion of patients who would consider the extra years in survival plotted sufficient to accept adjuvant chemotherapy compared with 1) 5 years or 2) 15 years of survival without such treatment. B) Proportion of patients who would consider the extra chance of 5-year survival plotted to be sufficient to accept adjuvant chemotherapy compared with 1) a 65% chance or 2) an 85% chance of living at least 5 years. Support Required by Dependents of the Patient Support required was categorized as none (46 patients), partial (37 patients), and full (20 patients). This was strongly associated with the time trade-off endpoints selected by the women. Patients whose dependents required full or partial support were more likely to judge treatment acceptable and needed smaller increments in survival to justify treatment (Fig. 2; P.002 for 5-year trade-off; P.0004 for 15-year trade-off, and P.0004 for time trade-off total). No such effect was seen in the survival rate questions, perhaps because most patients selected similar small-percentage increments. This factor remained statistically significant in the primary main multivariate analysis, based on trade-off total (P.0001). Support Available to the Patient Patients to whom full support was available from others accepted smaller increments in survival as justifying treatment than did those patients with partial or no support. This was statistically significant for the 15-year trade-off (P.01) and for trade-off total (P.04). It remained statistically significant in the multivariate analysis (P.02). Treatment-Related Toxicity Univariate analysis showed statistically significant associations between 5-year trade-off and mucositis (P.04), while for 15-year trade-off hematologic toxicity and mucositis were statistically significant (both P.04). In the multivariate analysis based on trade-off total, the summary factor describing any nonhematological toxicity remained independently statistically significant (P.01). As expected, patients experiencing worse toxicity demanded greater improvements in survival to justify treatment. Journal of the National Cancer Institute Monographs No. 30,

5 factors in the multivariate analysis based on trade-off total (P.01). Factors Not Predictive of Patient Preferences No association was observed between the trade-off total and patient age, educational level, or employment status; the time between treatment and interview; the use of concurrent adjuvant endocrine therapy; the occurrence of relapse; or the use of any particular modality (including further chemotherapy) for the treatment of relapse. DISCUSSION Adjuvant chemotherapy is now widely used to treat women with early breast cancer with the aim of preventing recurrence and improving overall survival. However, the benefits of adjuvant chemotherapy may be difficult to quantify and weigh against the adverse effects of treatment. For younger women at high risk of recurrence, such as those with axillary lymph node involvement, the gains in survival are larger, with an improvement of about 11% in the 10-year survival rate (see Table 5). For those at lower risk, the gains in survival are smaller: In women younger than 50 years, the improvement in 10-year survival is estimated at 7%, and for women older than 50 years the gain is in the range of 2% 3% (1). When considering benefits in terms of prolongation of time to relapse and overall survival, Cole et al. (12) estimate that for younger women (<50 years old), polychemotherapy adds an additional 5.4 months of survival and an additional 10.3 months of relapse-free survival within the first 10 years of follow-up. For older women (50 69 years old), the gains are smaller, with an estimated additional 2.9 months of overall survival and 6.8 months of disease-free survival. This would translate into an additional 6 8 months of quality-adjusted survival for younger women and an extra 3 5 months for older patients, accumulated within 10 years (Table 5). These estimates ignore longer term Table 5. Estimated benefits of adjuvant polychemotherapy for early breast cancer Fig. 2. Variation in patient preferences for adjuvant chemotherapy according to the amount of support required from the patient for dependents at home: 1) no support needed, 2) partial support needed, or 3) full support needed. Proportion of patients willing to accept adjuvant chemotherapy for a given increase in survival duration compared with 5 years (panel A) or 15 years of survival (panel B) without such treatment. Dosage Reduction During Chemotherapy Patients whose chemotherapy dosage was reduced to 75% or less of the total planned dose demanded longer survival increments than did those receiving a higher dosage (P.05 for 5-year trade-off, 15-year trade-off, and trade-off total). This factor was independent of recorded toxicity in a multivariate analysis (P.02). Initial Radiotherapy Patients whose initial adjuvant treatment included radiotherapy as well as chemotherapy required larger increments of survival to justify treatment. This was independent of other Outcome Subgroup, y Benefit Increase in 10-y survival rate* Women aged <50 Axillary lymph node positive 11.5% Axillary lymph node negative 7.1% Women aged Axillary lymph node positive 3.2% Axillary lymph node negative 2.4% Survival gain within 10 y < mo mo Disease-free survival gain within 10 y < mo mo Extra quality-adjusted survival within 10 y < mo mo Potential survival gain after 10 y < y y *Based on estimates provided by Early Breast Cancer Trialists Group overview (1). Assumes the same relative treatment effect in lymph node-positive and lymph node-negative groups. Based on estimates provided by Cole et al. (12). Assumes that the utility score associated with disease relapse is 0.5 to 0.9, compared with 1.0 for full health and 0.0 for death. Assumes that the same relative reduction on breast cancer mortality continues beyond 10 years with no effect on non-breast cancer deaths. 150 Journal of the National Cancer Institute Monographs No. 30, 2001

6 benefits beyond 10 years. If the earlier gains in survival from a reduced risk of breast cancer death were maintained after 10 years, then younger patients would gain an extra 2.9 life years and older patients an extra 0.5 year after this time. Consequently, for many women, especially younger women with lymph nodepositive disease, the size of the benefit will be large enough to justify adjuvant chemotherapy. However, the situation is less clear-cut for older patients and for those at low risk of recurrence, for whom individual preference over trade-offs between survival and adverse treatment effects will be more critical. In this setting, the individual preferences of women with early breast cancer become even more important in deciding whether such treatment is worthwhile. Preferences of women who have actually experienced the acute adverse effects of chemotherapy provide valuable information to guide decision making. Our study has shown that relatively modest survival gains would justify the adverse effects of treatment for many women. A gain in 5-year survival of at least 5% or a gain in survival by an extra year from 5 years would be sufficient to justify treatment for more than 70% of women. This means that, for most women at higher risk of recurrence, such as premenopausal women, the benefits of adjuvant CMF chemotherapy will be of sufficient size to warrant treatment. For women at lower risk, the benefit of treatment may be more questionable, and individual preferences will assume greater importance. Even for those at higher risk, individual preferences may be important. For example, 15% of women in this study would require more than 15% improvement in 5-year survival to justify therapy, an unlikely improvement in most clinical settings. Our study identified several factors that influenced the size of the survival benefit needed. Family-related factors, such as the amount of social support available to each woman and the amount of support needed by other family members from the patient, were each associated with stronger preferences for adjuvant treatment. This suggests that for these women, the adverse effects of treatment may be of secondary importance, provided that others are available to help them or that they judged their future ability to care for others in their family to be critical. Women experiencing greater toxicity also indicated that greater survival benefit was needed. This is expected, since each patient was asked to consider 6 months of chemotherapy as similar to the treatment she had already experienced. Patients who had had a reduction in the dose of their chemotherapy also indicated that larger survival gains would be needed even after allowance had been made for treatment toxicity in an adjusted analysis. This may have reflected unrecorded toxicity. Alternatively, physicians may have been more likely to reduce dosage in a group of patients whom they assessed as less willing to accept treatment toxicity. The association with radiotherapy is less clear. Whether it reflects patient selection, an increase in unrecorded toxicity of the overall treatment, or other factors remains unknown. This study has a number of limitations. In the patient interviews, we did not assess benefits other than survival gain associated with adjuvant chemotherapy, such as the delayed recurrence of breast cancer. The delay in recurrence will provide a small additional benefit in quality-adjusted survival and so means that the survival gains estimated in this study provide conservative estimates of the value of treatment. The study has also not taken into account psychological benefits of treatment that might possibly be gained by a sense of taking charge and feeling in control of future events. This has been identified in other settings as one reason that patients elect to have treatment (13). A broad cross section of patients who attended a follow-up clinic were selected for the study, and almost all eligible women participated. However, these women may not be representative of all those patients considering adjuvant chemotherapy, and their views may have changed since having this treatment. It is possible that women who had chosen adjuvant chemotherapy earlier overrated its value to justify this decision. However, patients who had relapsed and who may have been less positive about their decision to have adjuvant treatment expressed similar preferences. A further limitation is that the survival rate questions did not explicitly spell out what would happen to survivors beyond 5 years. Some women may have assumed that a chance of living at least 5 years was the same as a cure, whereas others may have assumed that there would be an ongoing risk of recurrence and shortened survival after this 5-year period. Despite its limitations, this study had some advantages over other possible designs. There was a consistency of preferences over time that was presumably related to the use of standardized interviews. The views of women who have actually experienced both the side effects of CMF chemotherapy and the concerns associated with their prognosis are also more likely to be of relevance. A number of studies using hypothetical scenarios have attempted to assess the preferences of women with early breast cancer. This method allows a range of scenarios to be considered but does not allow for the consideration of the individual woman s experience. In our study, the women s adverse experiences of adjuvant chemotherapy were used. Furthermore, the preferences of women with breast cancer may differ from others who cannot fully appreciate their unique perspective. Patient preferences for future survival may assume far greater importance once they are faced with the reality of a possible fatal outcome from cancer. Other studies support this view. For example, Galper et al. (14) found that women with invasive breast cancer considered the adverse effects of axillary lymph node dissection much less important than survival gains compared with a group of women who had had in situ, but not invasive, cancer. It is also possible that the women with invasive cancer have formed views that are biased in favor of the procedure so as to support their previous decision making. Another study comparing the preferences of patients with advanced cancer with those of their relatives showed that the patients considered the side effects of chemotherapy much less important than survival gains when compared with their relatives views (5,15). Since this study was performed, other groups have addressed similar questions. In a study of the adjuvant therapy of breast cancer, Lindley et al. (9) and Ravdin et al. (16) reported tradeoffs remarkably similar to those that we observed. In other diseases, a similar willingness for patients to accept moderately or even extremely toxic therapy in return for modest survival gains has been a consistent finding (7,17 20). The use of adjuvant chemotherapy in early breast cancer has changed substantially in the last 10 years since this study was performed. Adjuvant chemotherapy is increasingly used in women with lower risk tumours, such as those with small primary tumors and lymph node-negative tumors. It is also being used more often in older women. The nature and duration of treatment have also changed. The treatment and prevention of side effects have also improved. Anthracycline-containing chemotherapy regimens have been shown to provide additional modest benefits over CMF chemotherapy in disease-free sur- Journal of the National Cancer Institute Monographs No. 30,

7 vival (7% improvement at 5 years) and overall survival (3% improvement) (1). These additional gains come at the cost of more severe side effects than those of CMF chemotherapy. However, the side effects are shorter lived in the most commonly used anthracycline-based regimen of four cycles of treatment over the course of 3 months. Supportive therapy has also improved over the course of the last 10 years, reducing the frequency and severity of some important side effects. There are also new combinations of adjuvant chemotherapy undergoing evaluation including taxane-containing combinations and highdose chemotherapy regimens. Further studies are needed to assess patient preferences in these settings. The evaluation of these regimens in the context of randomized trials in representative samples of patients will be of particular value. Furthermore, consideration needs to be given to undertaking such studies of women both before and after they receive such chemotherapy. This would enable one to assess the extent that preferences are altered as a result of the decisions made by women to undertake chemotherapy. These studies will need to be designed carefully to ensure that the interviews have appropriate and not unintended effects on the patient s actual decisions. In conclusion, we believe that it is feasible to obtain individual patient preferences for adjuvant chemotherapy in early breast cancer based on an assessment of the trade-off between the adverse effects of treatment and survival gains. For younger women at high risk, the improvements in survival will usually be sufficient to justify therapy, whereas for older women and for those patients at low risk of recurrence, assessment of individual preferences may be critical to optimal decision making. Additional studies will be of considerable value in guiding the use of more recently developed adjuvant chemotherapy regimens. REFERENCES (1) Early Breast Cancer Trialists Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet 1998; 352: (2) Hurny C, Bernhard J, Coates AS, Castiglione M, Peterson HF, Gelber RD, et al. Impact of adjuvant therapy on quality of life in women with nodepositive operable breast cancer. Lancet 1996;347: (3) Palmer BV, Walsh GA, McKinna JA, Greening WP. Adjuvant chemotherapy for breast cancer: side effects and quality of life. Br Med J 1980; 281: (4) McNeil BJ, Weichselbaum R, Parker S. Speech and survival: trade offs between quality and quantity of life in laryngeal cancer. N Engl J Med 1981;305: (5) Simes RJ. Application of statistical decision theory to treatment choices: implications for design and analysis of clinical trials. Stat Med 1986;5: (6) Slevin ML, Stubbs L, Plant HJ, Wilson P, Gregory WM, Armes PJ, et al. Attitudes to chemotherapy: comparing views of patients with cancer with those of doctors, nurses, and general public. BMJ 1990;300: (7) Bremnes RM, Andersen K, Wist EA. Cancer patients, doctors and nurses vary in their willingness to undertake cancer chemotherapy. Eur J Cancer 1995;31A: (8) Coates AS. Who shall decide? Eur J Cancer 1995;31A: (9) Lindley C, Vasa S, Sawyer WT, Winer EP. Quality of life and preferences for treatment following systemic adjuvant therapy for early breast cancer. J Clin Oncol 1998;16: (10) Coates AS, Simes RJ. Patient assessment of adjuvant treatment in operable breast cancer. In: Williams CJ, editor. Introducing new treatments for cancer: practical, ethical and legal problems. London (U.K.): Wiley; p (11) Iman RL, Conover WJ. The use of rank transformation in regression. Technometrics 1979;21: (12) Cole BF, Gelber RD, Gelber S, Coates AS, Goldhirsch A. Polychemotherapy for early breast cancer: an overview of the randomised clinical trials with quality-adjusted survival analysis. Lancet 2001;358: (13) Butow PN, Maclean M, Dunn SM, Tattersall MH, Boyer MJ. The dynamics of change: cancer patients preference for information, involvement and support. Ann Oncol 1997;8: (14) Galper SR, Lee SJ, Tao ML, Troyan S, Kaelin CM, Harris JR, et al. Patient preferences for axillary dissection in the management of early-stage breast cancer. J Natl Cancer Inst 2000;92: (15) Simes RJ. Treatment selection for cancer patients: application of statistical decision theory to the treatment of advanced ovarian cancer. J Chronic Dis 1985;38: (16) Ravdin PM, Siminoff IA, Harvey JA. Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 1998;16: (17) Levine MN, Gafni A, Markham B, Mac Farlane D. A bedside decision instrument to elicit a patient s preference concerning adjuvant chemotherapy for breast cancer. Ann Intern Med 1992;117: (18) Silvestri G, Pritchard R, Welch HG. Preferences for chemotherapy in patients with advanced non-small cell lung cancer: descriptive study based on scripted interviews. BMJ 1998;317: (19) Slevin ML, Stubbs L, Plant HJ, Wilson P, Gregory WM, Armes PJ, et al. Attitudes to chemotherapy: comparing views of patients with cancer with those of doctors, nurses, and general public. BMJ 1990;300: (20) Hayman JA, Fairclough DL, Harris JR, Weeks JC. Patient preferences concerning the trade-off between the risks and benefits of routine radiation therapy after conservative surgery for early-stage breast cancer. J Clin Oncol 1997;15: NOTES Supported by a grant from the National Health and Medical Research Council, Australia. We gratefully acknowledge the work of Kate Cocker and Lexie Press who undertook each of the interviews and Simon Margrie for undertaking statistical analyses. 152 Journal of the National Cancer Institute Monographs No. 30, 2001