- Email: [email protected]
Patients' preferences for adjuvant chemotherapy in early breast cancer: what makes AC and CMF worthwhile now?
Annals of Oncology 16: 1786–1794, 2005 doi:10.1093/annonc/mdi370 Published online 26 August 2005
Original article
Patients’ preferences for adjuvant chemotherapy in early breast cancer: what makes AC and CMF worthwhile now? V. M. Duric1, M. R. Stockler1,2*, S. Heritier1, F. Boyle3, J. Beith2, A. Sullivan2, N. Wilcken4, A. S. Coates5,6 & R. J. Simes1 1
NHMRC Clinical Trials Centre, University of Sydney; 2Sydney Cancer Centre; 3Mater Hospital; 4Nepean Hospital; The Cancer Council Australia; 6School of Public Health, University of Sydney, Sydney, Australia
5
Received 8 May 2005; revised 7 July 2005; accepted 12 July 2005
Introduction Adjuvant chemotherapy for early breast cancer has evolved substantially over the last 30 years. Randomised trials over this time have established that it reduces relapse and improves survival in women under 70 years old with early breast cancer [1], but is inconvenient and has significant side-effects. Adjuvant chemotherapy is now considered in older women and in women with lower risk disease than it was 15 years ago. Regimens have been intensified by adding or substituting new drugs, and by cycling them more frequently. Supportive care for the prevention and treatment of side-effects has improved substantially, particularly for nausea, vomiting and myelosuppression. The benefits accruing on average to groups of such women are established beyond reasonable doubt, but the benefits for any particular woman cannot be known. For some women, chemotherapy will cause side-effects but not prevent or delay disease recurrence and *Correspondence to: Dr M. R. Stockler, NHMRC Clinical Trials Centre, University of Sydney, Locked Bag 77, Camperdown, NSW 2050, Australia. Tel: +61-02-9562-5000; Fax: +61-02-9565-1863; E-mail: [email protected] Ó 2005 European Society for Medical Oncology
death. Optimal decision-making for an individual requires that the possible benefits have been weighed against the possible harms and inconvenience. We previously reviewed studies that quantified the benefits considered necessary to make adjuvant chemotherapy worthwhile by asking women who had experienced it [2]. The findings were remarkably consistent despite major differences in methods [2–6]. Most women considered small improvements in survival sufficient to make adjuvant chemotherapy worthwhile. An additional 6 months in life expectancy was considered sufficient by half the women, and an additional 12 months was considered sufficient by three-quarters. An additional 1% in the chance of cure or 5-year survival was considered sufficient by half the women, and an additional 5% was considered sufficient by three-quarters. Women who had dependants, good support, involved lymph nodes and fewer side-effects considered smaller benefits to be sufficient than their counterparts; women who had adjuvant radiation therapy considered larger benefits necessary. The women in these studies were treated between 1985 and 1995, predominantly with AC (doxorubicin and cyclophosphamide) and CMF (cyclophosphamide, methotrexate and 5-fluorouracil).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Background: Studies of women who had adjuvant chemotherapy for early breast cancer 10–20 years ago showed that many judged small benefits sufficient to make it worthwhile. Indications, regimens and supportive care have changed. We sought the preferences of contemporary women who received similar chemotherapy. Patients and methods: Ninety-seven consecutive consenting women who completed adjuvant chemotherapy for early breast cancer 3–34 months previously were interviewed. Preferences were elicited with a structured, scripted interview using the trade-off method. Women were presented with four hypothetical scenarios based on known life expectancies (5 and 15 years) and survival rates (65% and 85% at 5 years) without adjuvant chemotherapy. Results: Improvements of an additional year in life expectancy or 3% in survival rates were judged sufficient to make adjuvant chemotherapy worthwhile by 68–84% of women. Half the women judged 1 day or 0.1% sufficient to make adjuvant chemotherapy worthwhile. Recollections of better wellbeing during adjuvant chemotherapy, having dependants and having a friend or relative who died from cancer were independently associated with judging smaller benefits sufficient to make adjuvant chemotherapy worthwhile (all P < 0.05). Conclusions: Preferences were highly variable, but the benefits judged sufficient to make adjuvant chemotherapy worthwhile were even smaller than those found in previous studies. Preferences were influenced by factors other than direct benefits and harms of chemotherapy. Key words: adjuvant chemotherapy, early breast cancer, patient preferences
1787 Given changes in indications, supportive care and attitudes to shared decision making, have women’s preferences for adjuvant chemotherapy also changed? The aim of this study was to determine the preferences of women treated with similar regimens for current indications by asking them the benefits they considered necessary to make their adjuvant chemotherapy worthwhile.
Patients and methods Patients
Preferences interview Patients’ preferences for adjuvant chemotherapy in early breast cancer were elicited with a structured and scripted interview using the time trade-off method based on our previous work [5]. The interview was piloted in 10 health professionals and 10 breast cancer consumers and then refined with their assistance. The interview took about 45 min to complete. All women were interviewed by the same clinical psychologist (V.M.D.). Participants were asked a series of questions based on hypothetical clinical scenarios. The scenarios were similar to those in our previous study [5]. There were two different types of question: one based on adding time to a given life expectancy, the other based on adding to the chance of surviving a fixed length of time (‘survival rate’). There were two scenarios for each type of question, with different baselines of survival chosen to reflect better and worse prognoses. In the life expectancy questions, the baseline for the worse prognosis scenario was 5 years and for the better prognosis scenario was 15 years. In the survival rate questions, the baseline for the worse prognosis scenario was 65% chance of 5 year survival, and for the better prognosis scenario was 85% chance of 5 year survival. Each of the four scenarios implies a different prognosis: the 85% survival rate scenario is the most favourable, followed by the 15-year life expectancy scenario, then the 65% survival rate scenario, and lastly, the 5-year life expectancy scenario, which is the least favourable. This ranking was not explained to women. In the life expectancy questions, participants were asked to imagine that they knew they would have a given life expectancy without chemotherapy (5 or 15 years), and a longer life expectancy with chemotherapy. They were then asked a series of questions to determine the minimum amount of extra time (beyond 5 or 15 years) that would be sufficient to make their chemotherapy worthwhile. In the survival rate questions, participants were asked to imagine that they knew their chance of surviving 5 years without chemotherapy (65% or 85%). They were then asked a series of questions to determine the minimum extra chance of surviving 5 years (beyond 65% or 85%) that would be sufficient to make their chemotherapy worthwhile.
Patient and treatment details Women completed a questionnaire asking a range of common demographic questions. Disease and treatment details were collected from participants’ medical records. Women’s recollections of their experience of treatment were self-rated with the Patient Disease and Treatment Assessment Form, which has 23 generic items related to cancer, including symptoms, functions and global ratings, and 18 specific items related to side-effects of chemotherapy and other problems.
Statistical methods Preferences are presented descriptively with figures and frequency tables. Similar methods were used as in our previous study [5]. Associations between patients’ characteristics and preferences were assessed with linear regression after normal score transformation. To minimise the problems of multiple comparisons, the sum of the responses to the two life expectancy questions was specified as the primary outcome variable. Multivariable models were used to adjust for potential confounding and to test the independent significance of potential predictors. All candidate variables were included in the starting pool for multivariable analyses. Variables were considered independently significant if they were retained in the multivariable models selected by both forward and backward selection using inclusion and removal criteria of P <0.05. All P values are two-sided, and there were no adjustments for multiple comparisons. P values were used to indicate levels of support for individual associations, with lower P values indicating stronger associations. Distributions of preferences were highly skewed. This was only partly corrected by normal score transformation. Analyses were repeated with logistic and ordinal regressions to test the validity of our reported conclusions based on linear regression. The effects of framing (better- or worse-prognosis scenario) and visual aids (with or without pie charts) were assessed with the Mann–Whitney– Wilcoxon test for independent samples. Test–retest reliability (agreement between the two interviews) was assessed with the intraclass correlation coefficient (ICC) using a two-way mixed-effects model with absolute agreement and a single rater (time as a fixed effect and individual as a random effect) [8, 9]. The size of the ICC was interpreted according to the guidelines of Landis and Koch: 0.2– 0.40 fair agreement; 0.41–0.60 moderate agreement; 0.61–0.80 good agreement; and 0.81–1.00 excellent agreement [10, 11]. Because the preferences data were highly skewed, the ICCs were also calculated using raw scores and ranks, equivalent to a quadratically weighted kappa (data not shown) [12]. Differences between women’s preferences at the first and second interview were tested with Wilcoxon’s signed rank test for paired samples. Data were analysed with SPSS for Windows, version 11.5.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Consecutive women who had adjuvant chemotherapy for early breast cancer at five tertiary referral cancer centres in metropolitan Sydney were interviewed 3–34 months after finishing their treatment. Separate consent was obtained from the woman and her treating oncologist. Of the 193 who were initially eligible, 29 (15%) were uncontactable or lost to follow-up, 12 (6%) were excluded because of inadequate English and 152 (79%) were invited to participate. Of the 152, 131 consented and were interviewed (response rate 86%). Of the 131 women, 29 (22%) were clinical trial participants and five (4%) received adjuvant taxanes outside of a clinical trial; their responses will be reported separately. This report includes the 97 women who had received either AC or CMF, or both, as part of routine clinical practice. The present study was performed within the context of a larger, continuing study of patients’ preferences for adjuvant chemotherapy in early breast cancer in an international randomised trial.
The increments in life expectancy and survival rate were identical to those in our previous study, except that we included scenarios with benefits of 1 day and 0.1% [5] The script emphasised that: the choices were hypothetical; the numbers did not apply to the individual personally; there were no right or wrong answers; and their answers would not affect their medical care in any way. The interview was designed to minimise the influence of framing and anchoring, with the sequence of better or worse prognosis scenarios randomly assigned. However, all interviews began with a life expectancy question. The benefits of visual aids are unclear, so we randomly allocated half the women to have the information also presented as pie charts. We aimed to re-interview 50 women to assess the reliability of our interview and the stability of their preferences [7]. Second interviews were sought from 63 women: two had died, 10 were lost to follow-up, two had moved interstate and 49 were interviewed 3–29 months after their first interview (average 17 months).
1788
Results Baseline characteristics Demographic characteristics are summarised in Table 1. The median age at the time of the interview was 55 years (range
Table 1. Baseline demographic characteristics (n = 97) Characteristic
%
Age at interview (years) 1
31–40
4
41–50
21
51–60
46
61–69
28
Characteristic
Marital status Married or de facto
68
Separated or divorced
18
Widowed
Table 2. Baseline disease, menopausal and treatment characteristics (n = 97)
3
Single
11
Children
79
%
Primary tumour size £2 cm
35
2.1–5 cm
49
>5 cm
8
Unknown
8
Dependent children
42
Dependent people
54
Involved
60
Support always available during ACT
65
Not involved
33
At home during ACT
93
Friend or relative died of cancer
86
Highest level of completed education
Axillary lymph nodes
Unknown
7
Hormone receptor status ER or PR positive
68
ER and PR negative
29
Primary school
2
High school before intermediate
4
Unknown
Intermediate or school certificate
31
Tumour grade
Trade or technical qualification
22
1
12
University or college degree
41
2
30
3
42
Employment status at time of interview Full-time employment
42
Part-time or casual employment
27
Unknown
3
16
Premenopausal at time of surgery
66
Unemployed
3
Premenopausal at time of interview
10
Student
3
Disease recurrence before interview
3
Retired
18
Home duties
7
Type of paid employment Administration
25
Professional
Chemotherapy regimens AC4 alone
54
AC4 followed by CMF3
27
CMF6
19
37
Adjuvant radiation therapy
72
Paraprofessional
4
Adjuvant endocrine therapy
74
Sales or personal service
3
Other
2
Not in paid employment
29
Member of community groups
63
ACT, adjuvant chemotherapy.
ER, estrogen receptor; PR, progesterone receptor; AC4, four cycles of doxorubicin (or epirubicin) and cyclophosphamide intravenously on day 1, repeated every 21 days; CMF, oral cyclophosphamide days 1–14, intravenous methotrexate and 5-fluorouracil days 1 and 8, repeated every 28 days for six cycles if used alone (CMF6); or for three cycles after AC (CMF3).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
25–30
25–69). More than half the women had a life partner, and most had children. The mean number of children 15 years or younger was one (range 0–3 children) and older than 15 years was two (range 0–5 children). Half the participants (54%) had at least one person dependent on them for support and 42% had at least one dependent child. A substantial proportion was educated at a post-secondary level and many were employed. During chemotherapy, most lived at home and 65% indicated that support was available throughout their chemotherapy if they needed it. The average time it took to travel to their chemotherapy centre was 30 min (range 5 min to 5 h). Most had a friend or relative who had died from cancer.
1789 Baseline disease characteristics, menopausal status, and treatment details are summarised in Table 2. Many had involved nodes. Two-thirds were premenopausal at the time of their breast cancer surgery, but only 10% were premenopausal when interviewed. The average number of months between finishing chemotherapy and the interview was 17 months (range 3–34). Most women had received 3–6 months of chemotherapy including an anthracycline. Three-quarters also had adjuvant endocrine therapy and adjuvant radiation therapy. Three women had recurrence of their breast cancer before the interview.
Patients’ preferences The benefits that the women judged necessary to make adjuvant chemotherapy worthwhile are summarised in Figure 3A and B. Plausible benefits (an extra 1 year in life expectancy or 3% in survival rate) were judged sufficient to make adjuvant chemotherapy worthwhile by 68–84% of the women. However, 52–55% of the women judged 1 day or 0.1% sufficient to make adjuvant chemotherapy worthwhile, regardless of the baseline survival. At the other extreme, 2–4% women judged the maximum benefit offered insufficient to make their adjuvant chemotherapy worthwhile.
Experience of treatment Figures 1 and 2 summarise the women’s retrospective recollections of their experience of chemotherapy. Hair loss, fatigue, the thought of having chemotherapy and nausea were rated the most troublesome symptoms. Energy, appetite and physical wellbeing were the rated most impaired functions.
Factors associated with preferences
hair loss fatigue thought of actually having the treatment nausea altered sense of taste trouble concentrating anxiety trouble sleeping problems with needles or injections 0
1
2
3
4
5
6
7
8
9
10
Worst I can imagine
No trouble at all
Figure 1. How much that aspect troubled you on average during your breast cancer chemotherapy (ranked from most troublesome to least troublesome) (n = 97).
energy appetite physical well-being emotional well-being overall well-being mood mobility 0
1
Worst possible
2
3
4
5
6
7
8
9
10
Best possible
Figure 2. How you would have rated yourself on that aspect on average during your breast cancer chemotherapy (ranked from most impaired to least impaired) (n = 97).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Women’s baseline characteristics were not strongly associated with their stated preferences for adjuvant chemotherapy (Table 3). We found weak associations between judging smaller benefits
1790
A
100
Cumulative % of women choosing treatment
90 80 70 60
Life expectancy of 5 years
50
Life expectancy of 15 years
40 30 20 10 0 1day
1mo
3mo
6mo
1yr
2yr
3yr
5yr
7yr
20yr
Extra time in life expectancy needed to make adjuvant chemotherapy worthwhile
B
100
80 70 60
5 year survival rate of 65%
50
5 year survival rate of 85%
40 30 20 10 0 0.1
1
2
3
5
10
15
20
25
35
Extra chance in 5 year survival needed to make adjuvant chemotherapy worthwhile Figure 3. (A) Cumulative proportions of women considering whether adjuvant chemotherapy would be worthwhile for various improvements in 5 and 15 year life expectancy (n = 97). (B) Cumulative proportions of women considering whether adjuvant chemotherapy would be worthwhile for various improvements in 65% and 85% chance of 5-year survival (n = 97).
sufficient and having dependants (P = 0.04), having support available during adjuvant chemotherapy (P = 0.05), living at home during adjuvant chemotherapy (P = 0.06) and having a friend or relative who died from cancer (P = 0.04). There were no obvious associations between characteristics of the disease or treatment and the women’s preferences for adjuvant chemotherapy. Women’s recollections of their experience of adjuvant chemotherapy were strongly associated with their preferences in univariable analyses. Women who were more troubled by adjuvant chemotherapy judged larger benefits necessary to make it worthwhile. The strongest associations were with fatigue, nausea, problems concentrating, thought of actually having chemotherapy, and problems with needles and injections. Anxiety, altered sense of taste and hair loss were less strongly associated with preferences. There was a non-significant trend suggesting that women who completed their treatment more recently judged smaller benefits sufficient (P = 0.06). These results were corroborated by logistic and ordinal regressions (see Appendix). The following factors were independently associated with judging smaller benefits sufficient in multivariable analyses:
having dependants, having a friend or relative who died from cancer and better well-being during adjuvant chemotherapy. Plots of preferences versus well-being during chemotherapy suggested that women fell into two groups of similar size: a group whose preferences were independent of their well-being during chemotherapy (small benefits sufficient regardless of how severely their well-being was impaired); and a group whose preferences were related to their well-being during chemotherapy (larger benefits required if their well-being was worse). This is illustrated in Figure 4.
Test–retest reliability, framing and other effects Agreement between women’s preferences at their first and second interview was moderate to good (Table 4) and comparable to our previous study [5]. There was no systematic change between preferences at time 1 and time 2, as in Jansen et al.’s study [13]. Preferences were not affected by whether the better or worse prognosis scenarios were presented first (P = 0.7–0.9), or by whether visual aids were used (P = 0.3–0.5).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Cumulative % of women choosing treatment
90
1791 Table 3. Characteristics associated with judging smaller benefits of adjuvant chemotherapy sufficient
Table 3. (Continued) Factor
Factor
Linear univariable models P
Best linear multivariable model P
Linear univariable models P
Best linear multivariable model P
If the following were rated better:
Characteristics of women
Appetite
0.01
Age
0.8
Emotional well-being
Menopausal status at surgery
0.8
Physical well-being
<0.0001
<0.00001a
Menopausal status at interview
0.5
Overall well-being
0.0001
<0.00001a
0.0003
If she: a
0.4
Was employed at the time of interview
0.6
Has a life partner
0.5
Lives with others
0.6
Has children
0.2
Has dependent children
0.3
Has dependants
0.04
Had support always available during ACT
0.05
Was a member of community groups
0.5
Lived at home during chemotherapy
0.06
Had a friend or relative who died from cancer
0.04
Had less distance to travel to chemotherapy centre
0.7
Either physical or overall well-being were highly significant in the multivariable model, but not both. ACT, adjuvant chemotherapy; AC, doxorubicin, cyclophosphamide; CMF, cyclophosphamide, methotrexate, 5-fluorouracil.
Table 4. Test–retest reliability, based on paired measurements for 49 women 0.03
0.04
Scenario
Intraclass correlation coefficient
P for difference between interviews 1 and 2
Life expectancy of 5 years
0.7
0.2
Life expectancy of 15 years
0.4
0.2
5-year survival rate of 65%
0.5
0.1
5-year survival rate of 85%
0.7
0.4
Discussion
Characteristics of disease and treatment Tumour size
0.9
Lymph nodes involved
0.6
Hormone receptor status
0.7
Adjuvant radiotherapy given
0.3
Adjuvant endocrine therapy given
0.4
Disease recurrence before interview
0.7
AC alone versus CMF with or without AC
0.4
Shorter interval since chemotherapy finished
0.06
Experience of treatment If she was less troubled by: Problems coping with treatment
<0.0001
Thought of actually having the treatment
0.002
Nausea
0.002
Problems concentrating
0.006
Problems with needles or injections
0.008
Fatigue
0.008
Altered sense of taste
0.02
Anxiety
0.03
Hair loss
0.05
Problems with sleeping
0.4
Our most striking finding was that benefits of 1 day or 0.1% were judged sufficient by 52–55% of the women. In previous studies, an extra 6 months in life expectancy or an extra 1% in survival rates were judged sufficient to make adjuvant chemotherapy worthwhile by 50–52% of women [2]. Benefits of this size were judged sufficient by 62–72% of women in the current study. Benefits consistent with the Early Breast Cancer Trialists’ overview data (an extra 1 year in life expectancy or an extra 3% in survival rates) were judged sufficient to make adjuvant chemotherapy worthwhile by 58–77% of women in previous studies but 68–85% of women in the current study [1, 2]. There are many possible explanations for these striking results. Better supportive care may mean that chemotherapy is better tolerated and that smaller benefits are needed to offset the side-effects and inconvenience. Attitudes to cancer and its treatment may have changed making chemotherapy seem less dramatic and more acceptable. The larger number of women judging small benefits sufficient might be due to our including negligible benefits amongst our possible responses in the preferences interview. Our data do not shed light on which, if any, of these explanations is true. One interpretation of the plots of preferences for chemotherapy versus experience of chemotherapy is that women fell into one of two equal-sized groups. For women in the first group, the benefits judged necessary to make adjuvant chemotherapy worthwhile were related to the severity of side-effects experienced: larger benefits were required if side-effects were worse. These women were trading length of life for quality of
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Had tertiary education
Benefit needed: from least to most (normal score transformed)
1792
Figure 4. Relationship between the size of benefit judged sufficient to make adjuvant chemotherapy worthwhile and women’s recollections of their physical well-being during adjuvant chemotherapy. Open circles represent women who judged small benefits sufficient regardless of their well-being; filled circles represent women for whom the benefit required was related to their well-being.
life. For women in the second group, small benefits were judged sufficient to make adjuvant chemotherapy worthwhile regardless of the side-effects experienced: small benefits were sufficient even if side-effects were severe. These women seemed unwilling to trade length of life for quality of life. Women’s preferences were associated with their social circumstances, but not with their age, marital status or disease characteristics. Women with dependants and women who felt well-supported judged smaller benefits sufficient to make adjuvant chemotherapy worthwhile. As previously found by Yellen and Cella [14], women in our study with someone to live for (or someone living for them) seemed less influenced by the side-effects and inconvenience of their chemotherapy. The criterion for rational decision making is to maximise quality-adjusted survival. Quality of life is reduced during adjuvant chemotherapy. Some benefit is needed to make up this deficit. For almost half the women in our study, an extra day or 0.1% seemed sufficient. This could be interpreted as meaning that they were trying to live at all costs or that every day counts. Perhaps it means that feeling they have done everything confers other benefits or that some did not make their decision by trading direct harms and benefits of chemotherapy. It is tempting to attribute preferences for treatment with such small benefits to misunderstanding. Problems understanding and incorporating small probabilities in decision making are important and almost certainly part of the explanation, but a benefit of one extra day on a baseline of 5 or 15 years is hard to misunderstand. At least for some people, it may be more accurate to say that they understood the issues differently, rather than that they did not understand them at all.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Physical well-being during adjuvant chemotherapy (from best to worst)
The practical questions for clinicians are: what benefit is sufficient for them to offer adjuvant chemotherapy as a real option, and is this different for different individuals; how to deal with requests for adjuvant chemotherapy that is unlikely to be of benefit; and how to deal with people reluctant to have adjuvant chemotherapy with substantial benefits. Clinical practice guidelines offer a framework for making clinical decisions, but not absolute recommendations. In Australia, national evidence-based clinical practice guidelines specify appropriate indications for adjuvant chemotherapy, emphasise the importance of preferences and recommend that they be taken into account [15]. Our data reinforce the importance of explicitly considering benefits and harms when discussing adjuvant chemotherapy with patients. Different groups of people have remarkably different preferences for chemotherapy [2]. Reconciling the range and diversity of perspectives needs further consideration: the patients’ perspective that often favours treatments with small benefits; the clinicians’ perspective that favours treatments doing more good than harm; and the community’s perspective that favours treatments that are more cost-effective. Anxiety, fear, loss of control and regret are all part of being affected by cancer. These factors may influence people making decisions about cancer treatment. It is important to know whether and how these factors might influence decision making about potentially life-saving treatments. For example, women with a more pessimistic view of the future and limited social supports may be less inclined to prefer adjuvant chemotherapy. The main limitations of this study are related to its sampling frame and generalisability. We included only women who had experienced adjuvant chemotherapy. Their preferences were based on their recollections of what chemotherapy was like, and were likely to be influenced by psychological mechanisms, such as memory, adaptation, coping and cognitive dissonance reduction. Cognitive dissonance reduction occurs when an individual reinterprets past actions in a way that minimises conflict with their current attitudes. For example, the reason why a woman judges small benefits sufficient now may partly reflect the need to see her past decision to have chemotherapy in a positive light. A shift in her values that reaffirms her original decision to have adjuvant chemotherapy may also reflect adaptation and coping. The response rate in our study was relatively high and the results should be generalisable to comparable centres and women. However, the higher response rates should not obscure the fact that only women who had adjuvant chemotherapy were recruited, and that they are not representative of women considering whether to have adjuvant chemotherapy, some of whom declined. The preferences of health professionals, lay people and partners can also provide useful information about attitudes toward adjuvant chemotherapy for early breast cancer. We are currently completing studies determining the preferences of patients’ partners [16] and health professionals involved in providing their treatment. The attitudes of recently diagnosed women who are considering adjuvant chemotherapy are also of interest, but eliciting preferences in these women poses ethical and logistic problems. Studying women considering adjuvant chemotherapy removes the bias of selecting only those who chose to have it, but replaces it with the bias of how information
1793
Acknowledgements V.M.D. was supported by a PhD Scholarship from the National Health and Medical Research Council of Australia. Our thanks go to the study participants for their time and effort and the Consumer Advisory Panel of the Australia New Zealand Breast Cancer Trials Group for their help in piloting the interview. This research was supported by grants from the National Health and Medical Research Council of Australia, the Cancer Council of Australia, and the Australia New Zealand Breast Cancer Trials Group.
References 1. Early Breast Cancer Trialists’ Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet 1998; 352: 930–942.
2. Duric V, Stockler M. Patients’ preferences for adjuvant chemotherapy in early breast cancer: a review of what makes it worthwhile. Lancet Oncol 2001; 2: 691–697. 3. Lindley C, Vasa S, Sawyer WT, Winer EP. Quality of life and preferences for treatment following systemic adjuvant therapy for early-stage breast cancer. J Clin Oncol 1998; 16: 1380–1387. 4. Ravdin PM, Siminoff IA, Harvey JA. Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 1998; 16: 515–521. 5. Simes RJ, Coates AS. Patient preferences for adjuvant chemotherapy of early breast cancer: how much benefit is needed? J Natl Cancer Inst Monogr 2001; 30: 146–152. 6. Zimmermann C, Baldo C, Molino A. Framing of outcome and probability of recurrence: breast cancer patients’ choice of adjuvant chemotherapy in hypothetical patient scenarios. Breast Cancer Res Treat 2000; 60: 9–14. 7. Walter SD, Eliasziw M, Donner A. Sample size and optimal designs for reliability studies. Stat Med 1998; 17: 101–110. 8. McGraw K, Wong S. Forming inferences about some intraclass correlations coefficients: Correction. Psychol Methods 1996; 1: 390. 9. McGraw K, Wong S. Forming inferences about some intraclass correlation coefficients. Psychol Methods 1996; 1: 30–46. 10. Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977; 33: 363–374. 11. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–174. 12. Streiner DL, Norman GR. Health Measurement Scales: A Practical Guide to their Development and Use, 3rd edition. Oxford: Oxford University Press 2003. 13. Jansen SJ, Kievit J, Nooij MA et al. Patients’ preferences for adjuvant chemotherapy in early-stage breast cancer: is treatment worthwhile? Br J Cancer 2001; 84: 1577–1585. 14. Yellen SB, Cella DF. Someone to live for: social well-being, parenthood status, and decision-making in oncology. J Clin Oncol 1995; 13: 1255–1264. 15. National Health and Medical Research Council of Australia. NHMRC clinical practice guidelines for the management of early breast cancer. http:// www.health.gov.au/nhmrc/publications/2001. Available from: http://www. health.gov.au/nhmrc/publications/pdf/cp74.pdf (17 August, 2005, date accessed). 16. Duric V, Stockler M, Butow P et al. Comparing patients’ and their partners’ preferences for adjuvant chemotherapy in early breast cancer. ASCO annual meeting proceedings. J Clin Oncol 2005; 24: 00 (Abstr 6074). 17. Slovic P, Finucane M, Peters E, MacGregor D. Heuristics and biases: The psychology of intuitive judgment. In Gilovich T, Griffin D et al. (eds): Heuristics and Biases: The Psychology of Intuitive Judgment. New York: Cambridge University Press 2002; 397–420. 18. http://www.jncicancerspectrum.oupjournals.org/cgi/pdq/jncipdq, CDR0000067653?version=professional (16 March 2005, date accessed).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
about adjuvant chemotherapy is presented, which may also influence their decisions [17]. Regimens including taxanes have become a standard part of adjuvant chemotherapy for women with higher risk disease, but were not considered in this study. The addition of taxanes increases effectiveness but also increases duration and toxicity. We are currently completing a study of women participating in an international randomised trial to determine the benefits they consider necessary to make the addition of a taxane worthwhile [18]. Preferences are remarkably diverse, even within groups of people with similar characteristics. In the current study, moderate but plausible benefits were sufficient for most women, but negligible benefits were sufficient for many, and implausibly large benefits were insufficient for others. Preferences appear to have been motivated by more than simple trade-offs between direct benefits and harms of chemotherapy. Future studies should seek to identify and measure these additional considerations so that they can be incorporated into models for decision making. Patients facing decisions about adjuvant chemotherapy have a lot to consider. So do their clinicians. Clinicians’ recommendations should be based on the best available evidence about disease and treatment, but they should also reflect the particular patient’s attitudes, values and priorities. People facing lifethreatening illnesses are the experts about their unique priorities, concerns, fears and regrets. Clinicians should ask about these things and try to understand their patients’ motivations for treatment so that they can better tailor their advice.
1794
Appendix P values from univariable linear, logistic and ordinal regression analyses assessing characteristics associated with judging smaller benefits of adjuvant chemotherapy sufficient Linear univariable models Pa
Ordinal univariable models Pa
Logistic univariable models Pa
0.82
0.51
0.32
Had tertiary education
0.42
0.76
0.90
Was employed at the time of interview
0.58
0.50
0.35
Has a life partner
0.50
0.62
0.72
Lives with others
0.55
0.46
0.72
Has children
0.23
0.18
0.05
Has dependent children
0.34
0.43
0.36
Has dependants
0.04
0.09
0.05
Characteristics of women Age If she:
0.05
0.04
0.07
Was a member of community groups
0.52
0.37
0.16
Was at home during chemotherapy
0.06
0.03
0.06
Had a friend or relative who died from cancer
0.04
0.08
0.05
Had less distance to travel to chemotherapy centre
0.70
0.23
0.08
Characteristics of disease and treatment Tumour size
0.93
0.64
0.48
Lymph nodes involved
0.58
0.55
0.38
Hormone receptor status
0.73
0.49
0.36
Adjuvant radiotherapy given
0.25
0.42
0.29
Adjuvant endocrine therapy given
0.38
0.29
0.45
Disease recurrence before interview
0.68
0.68
0.96
AC alone versus CMF with or without AC
0.43
0.53
0.77
Shorter interval since chemotherapy finished
0.06
0.03
0.10
Fatigue
0.008
0.003
0.01
Problems with sleeping
0.40
0.28
0.65
Nausea
0.002
0.002
0.005
Anxiety
0.03
0.01
0.03
Problems concentrating
0.006
0.002
0.002
Altered sense of taste
0.02
0.004
0.006
Hair loss
0.05
0.02
0.03
Experience of treatment If she was less troubled by:
Problems with needles or injections Problems coping with treatment Thought of actually having the treatment
0.008
0.005
0.02
<0.0001
0.001
0.003
0.002
0.004
0.006
If the following were rated better: Appetite
0.01
0.003
0.005
<0.0001
0.0002
0.004
Emotional well-being
0.0003
0.001
0.007
Overall well-being
0.0001
0.0002
0.002
Physical well-being
a
P values £0.1 are indicated in bold. AC, doxorubicin, cyclophosphamide; CMF, cyclophosphamide, methotrexate, 5-fluorouracil.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Had support available at all times during chemotherapy (if needed)
Original article
Patients’ preferences for adjuvant chemotherapy in early breast cancer: what makes AC and CMF worthwhile now? V. M. Duric1, M. R. Stockler1,2*, S. Heritier1, F. Boyle3, J. Beith2, A. Sullivan2, N. Wilcken4, A. S. Coates5,6 & R. J. Simes1 1
NHMRC Clinical Trials Centre, University of Sydney; 2Sydney Cancer Centre; 3Mater Hospital; 4Nepean Hospital; The Cancer Council Australia; 6School of Public Health, University of Sydney, Sydney, Australia
5
Received 8 May 2005; revised 7 July 2005; accepted 12 July 2005
Introduction Adjuvant chemotherapy for early breast cancer has evolved substantially over the last 30 years. Randomised trials over this time have established that it reduces relapse and improves survival in women under 70 years old with early breast cancer [1], but is inconvenient and has significant side-effects. Adjuvant chemotherapy is now considered in older women and in women with lower risk disease than it was 15 years ago. Regimens have been intensified by adding or substituting new drugs, and by cycling them more frequently. Supportive care for the prevention and treatment of side-effects has improved substantially, particularly for nausea, vomiting and myelosuppression. The benefits accruing on average to groups of such women are established beyond reasonable doubt, but the benefits for any particular woman cannot be known. For some women, chemotherapy will cause side-effects but not prevent or delay disease recurrence and *Correspondence to: Dr M. R. Stockler, NHMRC Clinical Trials Centre, University of Sydney, Locked Bag 77, Camperdown, NSW 2050, Australia. Tel: +61-02-9562-5000; Fax: +61-02-9565-1863; E-mail: [email protected] Ó 2005 European Society for Medical Oncology
death. Optimal decision-making for an individual requires that the possible benefits have been weighed against the possible harms and inconvenience. We previously reviewed studies that quantified the benefits considered necessary to make adjuvant chemotherapy worthwhile by asking women who had experienced it [2]. The findings were remarkably consistent despite major differences in methods [2–6]. Most women considered small improvements in survival sufficient to make adjuvant chemotherapy worthwhile. An additional 6 months in life expectancy was considered sufficient by half the women, and an additional 12 months was considered sufficient by three-quarters. An additional 1% in the chance of cure or 5-year survival was considered sufficient by half the women, and an additional 5% was considered sufficient by three-quarters. Women who had dependants, good support, involved lymph nodes and fewer side-effects considered smaller benefits to be sufficient than their counterparts; women who had adjuvant radiation therapy considered larger benefits necessary. The women in these studies were treated between 1985 and 1995, predominantly with AC (doxorubicin and cyclophosphamide) and CMF (cyclophosphamide, methotrexate and 5-fluorouracil).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Background: Studies of women who had adjuvant chemotherapy for early breast cancer 10–20 years ago showed that many judged small benefits sufficient to make it worthwhile. Indications, regimens and supportive care have changed. We sought the preferences of contemporary women who received similar chemotherapy. Patients and methods: Ninety-seven consecutive consenting women who completed adjuvant chemotherapy for early breast cancer 3–34 months previously were interviewed. Preferences were elicited with a structured, scripted interview using the trade-off method. Women were presented with four hypothetical scenarios based on known life expectancies (5 and 15 years) and survival rates (65% and 85% at 5 years) without adjuvant chemotherapy. Results: Improvements of an additional year in life expectancy or 3% in survival rates were judged sufficient to make adjuvant chemotherapy worthwhile by 68–84% of women. Half the women judged 1 day or 0.1% sufficient to make adjuvant chemotherapy worthwhile. Recollections of better wellbeing during adjuvant chemotherapy, having dependants and having a friend or relative who died from cancer were independently associated with judging smaller benefits sufficient to make adjuvant chemotherapy worthwhile (all P < 0.05). Conclusions: Preferences were highly variable, but the benefits judged sufficient to make adjuvant chemotherapy worthwhile were even smaller than those found in previous studies. Preferences were influenced by factors other than direct benefits and harms of chemotherapy. Key words: adjuvant chemotherapy, early breast cancer, patient preferences
1787 Given changes in indications, supportive care and attitudes to shared decision making, have women’s preferences for adjuvant chemotherapy also changed? The aim of this study was to determine the preferences of women treated with similar regimens for current indications by asking them the benefits they considered necessary to make their adjuvant chemotherapy worthwhile.
Patients and methods Patients
Preferences interview Patients’ preferences for adjuvant chemotherapy in early breast cancer were elicited with a structured and scripted interview using the time trade-off method based on our previous work [5]. The interview was piloted in 10 health professionals and 10 breast cancer consumers and then refined with their assistance. The interview took about 45 min to complete. All women were interviewed by the same clinical psychologist (V.M.D.). Participants were asked a series of questions based on hypothetical clinical scenarios. The scenarios were similar to those in our previous study [5]. There were two different types of question: one based on adding time to a given life expectancy, the other based on adding to the chance of surviving a fixed length of time (‘survival rate’). There were two scenarios for each type of question, with different baselines of survival chosen to reflect better and worse prognoses. In the life expectancy questions, the baseline for the worse prognosis scenario was 5 years and for the better prognosis scenario was 15 years. In the survival rate questions, the baseline for the worse prognosis scenario was 65% chance of 5 year survival, and for the better prognosis scenario was 85% chance of 5 year survival. Each of the four scenarios implies a different prognosis: the 85% survival rate scenario is the most favourable, followed by the 15-year life expectancy scenario, then the 65% survival rate scenario, and lastly, the 5-year life expectancy scenario, which is the least favourable. This ranking was not explained to women. In the life expectancy questions, participants were asked to imagine that they knew they would have a given life expectancy without chemotherapy (5 or 15 years), and a longer life expectancy with chemotherapy. They were then asked a series of questions to determine the minimum amount of extra time (beyond 5 or 15 years) that would be sufficient to make their chemotherapy worthwhile. In the survival rate questions, participants were asked to imagine that they knew their chance of surviving 5 years without chemotherapy (65% or 85%). They were then asked a series of questions to determine the minimum extra chance of surviving 5 years (beyond 65% or 85%) that would be sufficient to make their chemotherapy worthwhile.
Patient and treatment details Women completed a questionnaire asking a range of common demographic questions. Disease and treatment details were collected from participants’ medical records. Women’s recollections of their experience of treatment were self-rated with the Patient Disease and Treatment Assessment Form, which has 23 generic items related to cancer, including symptoms, functions and global ratings, and 18 specific items related to side-effects of chemotherapy and other problems.
Statistical methods Preferences are presented descriptively with figures and frequency tables. Similar methods were used as in our previous study [5]. Associations between patients’ characteristics and preferences were assessed with linear regression after normal score transformation. To minimise the problems of multiple comparisons, the sum of the responses to the two life expectancy questions was specified as the primary outcome variable. Multivariable models were used to adjust for potential confounding and to test the independent significance of potential predictors. All candidate variables were included in the starting pool for multivariable analyses. Variables were considered independently significant if they were retained in the multivariable models selected by both forward and backward selection using inclusion and removal criteria of P <0.05. All P values are two-sided, and there were no adjustments for multiple comparisons. P values were used to indicate levels of support for individual associations, with lower P values indicating stronger associations. Distributions of preferences were highly skewed. This was only partly corrected by normal score transformation. Analyses were repeated with logistic and ordinal regressions to test the validity of our reported conclusions based on linear regression. The effects of framing (better- or worse-prognosis scenario) and visual aids (with or without pie charts) were assessed with the Mann–Whitney– Wilcoxon test for independent samples. Test–retest reliability (agreement between the two interviews) was assessed with the intraclass correlation coefficient (ICC) using a two-way mixed-effects model with absolute agreement and a single rater (time as a fixed effect and individual as a random effect) [8, 9]. The size of the ICC was interpreted according to the guidelines of Landis and Koch: 0.2– 0.40 fair agreement; 0.41–0.60 moderate agreement; 0.61–0.80 good agreement; and 0.81–1.00 excellent agreement [10, 11]. Because the preferences data were highly skewed, the ICCs were also calculated using raw scores and ranks, equivalent to a quadratically weighted kappa (data not shown) [12]. Differences between women’s preferences at the first and second interview were tested with Wilcoxon’s signed rank test for paired samples. Data were analysed with SPSS for Windows, version 11.5.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Consecutive women who had adjuvant chemotherapy for early breast cancer at five tertiary referral cancer centres in metropolitan Sydney were interviewed 3–34 months after finishing their treatment. Separate consent was obtained from the woman and her treating oncologist. Of the 193 who were initially eligible, 29 (15%) were uncontactable or lost to follow-up, 12 (6%) were excluded because of inadequate English and 152 (79%) were invited to participate. Of the 152, 131 consented and were interviewed (response rate 86%). Of the 131 women, 29 (22%) were clinical trial participants and five (4%) received adjuvant taxanes outside of a clinical trial; their responses will be reported separately. This report includes the 97 women who had received either AC or CMF, or both, as part of routine clinical practice. The present study was performed within the context of a larger, continuing study of patients’ preferences for adjuvant chemotherapy in early breast cancer in an international randomised trial.
The increments in life expectancy and survival rate were identical to those in our previous study, except that we included scenarios with benefits of 1 day and 0.1% [5] The script emphasised that: the choices were hypothetical; the numbers did not apply to the individual personally; there were no right or wrong answers; and their answers would not affect their medical care in any way. The interview was designed to minimise the influence of framing and anchoring, with the sequence of better or worse prognosis scenarios randomly assigned. However, all interviews began with a life expectancy question. The benefits of visual aids are unclear, so we randomly allocated half the women to have the information also presented as pie charts. We aimed to re-interview 50 women to assess the reliability of our interview and the stability of their preferences [7]. Second interviews were sought from 63 women: two had died, 10 were lost to follow-up, two had moved interstate and 49 were interviewed 3–29 months after their first interview (average 17 months).
1788
Results Baseline characteristics Demographic characteristics are summarised in Table 1. The median age at the time of the interview was 55 years (range
Table 1. Baseline demographic characteristics (n = 97) Characteristic
%
Age at interview (years) 1
31–40
4
41–50
21
51–60
46
61–69
28
Characteristic
Marital status Married or de facto
68
Separated or divorced
18
Widowed
Table 2. Baseline disease, menopausal and treatment characteristics (n = 97)
3
Single
11
Children
79
%
Primary tumour size £2 cm
35
2.1–5 cm
49
>5 cm
8
Unknown
8
Dependent children
42
Dependent people
54
Involved
60
Support always available during ACT
65
Not involved
33
At home during ACT
93
Friend or relative died of cancer
86
Highest level of completed education
Axillary lymph nodes
Unknown
7
Hormone receptor status ER or PR positive
68
ER and PR negative
29
Primary school
2
High school before intermediate
4
Unknown
Intermediate or school certificate
31
Tumour grade
Trade or technical qualification
22
1
12
University or college degree
41
2
30
3
42
Employment status at time of interview Full-time employment
42
Part-time or casual employment
27
Unknown
3
16
Premenopausal at time of surgery
66
Unemployed
3
Premenopausal at time of interview
10
Student
3
Disease recurrence before interview
3
Retired
18
Home duties
7
Type of paid employment Administration
25
Professional
Chemotherapy regimens AC4 alone
54
AC4 followed by CMF3
27
CMF6
19
37
Adjuvant radiation therapy
72
Paraprofessional
4
Adjuvant endocrine therapy
74
Sales or personal service
3
Other
2
Not in paid employment
29
Member of community groups
63
ACT, adjuvant chemotherapy.
ER, estrogen receptor; PR, progesterone receptor; AC4, four cycles of doxorubicin (or epirubicin) and cyclophosphamide intravenously on day 1, repeated every 21 days; CMF, oral cyclophosphamide days 1–14, intravenous methotrexate and 5-fluorouracil days 1 and 8, repeated every 28 days for six cycles if used alone (CMF6); or for three cycles after AC (CMF3).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
25–30
25–69). More than half the women had a life partner, and most had children. The mean number of children 15 years or younger was one (range 0–3 children) and older than 15 years was two (range 0–5 children). Half the participants (54%) had at least one person dependent on them for support and 42% had at least one dependent child. A substantial proportion was educated at a post-secondary level and many were employed. During chemotherapy, most lived at home and 65% indicated that support was available throughout their chemotherapy if they needed it. The average time it took to travel to their chemotherapy centre was 30 min (range 5 min to 5 h). Most had a friend or relative who had died from cancer.
1789 Baseline disease characteristics, menopausal status, and treatment details are summarised in Table 2. Many had involved nodes. Two-thirds were premenopausal at the time of their breast cancer surgery, but only 10% were premenopausal when interviewed. The average number of months between finishing chemotherapy and the interview was 17 months (range 3–34). Most women had received 3–6 months of chemotherapy including an anthracycline. Three-quarters also had adjuvant endocrine therapy and adjuvant radiation therapy. Three women had recurrence of their breast cancer before the interview.
Patients’ preferences The benefits that the women judged necessary to make adjuvant chemotherapy worthwhile are summarised in Figure 3A and B. Plausible benefits (an extra 1 year in life expectancy or 3% in survival rate) were judged sufficient to make adjuvant chemotherapy worthwhile by 68–84% of the women. However, 52–55% of the women judged 1 day or 0.1% sufficient to make adjuvant chemotherapy worthwhile, regardless of the baseline survival. At the other extreme, 2–4% women judged the maximum benefit offered insufficient to make their adjuvant chemotherapy worthwhile.
Experience of treatment Figures 1 and 2 summarise the women’s retrospective recollections of their experience of chemotherapy. Hair loss, fatigue, the thought of having chemotherapy and nausea were rated the most troublesome symptoms. Energy, appetite and physical wellbeing were the rated most impaired functions.
Factors associated with preferences
hair loss fatigue thought of actually having the treatment nausea altered sense of taste trouble concentrating anxiety trouble sleeping problems with needles or injections 0
1
2
3
4
5
6
7
8
9
10
Worst I can imagine
No trouble at all
Figure 1. How much that aspect troubled you on average during your breast cancer chemotherapy (ranked from most troublesome to least troublesome) (n = 97).
energy appetite physical well-being emotional well-being overall well-being mood mobility 0
1
Worst possible
2
3
4
5
6
7
8
9
10
Best possible
Figure 2. How you would have rated yourself on that aspect on average during your breast cancer chemotherapy (ranked from most impaired to least impaired) (n = 97).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Women’s baseline characteristics were not strongly associated with their stated preferences for adjuvant chemotherapy (Table 3). We found weak associations between judging smaller benefits
1790
A
100
Cumulative % of women choosing treatment
90 80 70 60
Life expectancy of 5 years
50
Life expectancy of 15 years
40 30 20 10 0 1day
1mo
3mo
6mo
1yr
2yr
3yr
5yr
7yr
20yr
Extra time in life expectancy needed to make adjuvant chemotherapy worthwhile
B
100
80 70 60
5 year survival rate of 65%
50
5 year survival rate of 85%
40 30 20 10 0 0.1
1
2
3
5
10
15
20
25
35
Extra chance in 5 year survival needed to make adjuvant chemotherapy worthwhile Figure 3. (A) Cumulative proportions of women considering whether adjuvant chemotherapy would be worthwhile for various improvements in 5 and 15 year life expectancy (n = 97). (B) Cumulative proportions of women considering whether adjuvant chemotherapy would be worthwhile for various improvements in 65% and 85% chance of 5-year survival (n = 97).
sufficient and having dependants (P = 0.04), having support available during adjuvant chemotherapy (P = 0.05), living at home during adjuvant chemotherapy (P = 0.06) and having a friend or relative who died from cancer (P = 0.04). There were no obvious associations between characteristics of the disease or treatment and the women’s preferences for adjuvant chemotherapy. Women’s recollections of their experience of adjuvant chemotherapy were strongly associated with their preferences in univariable analyses. Women who were more troubled by adjuvant chemotherapy judged larger benefits necessary to make it worthwhile. The strongest associations were with fatigue, nausea, problems concentrating, thought of actually having chemotherapy, and problems with needles and injections. Anxiety, altered sense of taste and hair loss were less strongly associated with preferences. There was a non-significant trend suggesting that women who completed their treatment more recently judged smaller benefits sufficient (P = 0.06). These results were corroborated by logistic and ordinal regressions (see Appendix). The following factors were independently associated with judging smaller benefits sufficient in multivariable analyses:
having dependants, having a friend or relative who died from cancer and better well-being during adjuvant chemotherapy. Plots of preferences versus well-being during chemotherapy suggested that women fell into two groups of similar size: a group whose preferences were independent of their well-being during chemotherapy (small benefits sufficient regardless of how severely their well-being was impaired); and a group whose preferences were related to their well-being during chemotherapy (larger benefits required if their well-being was worse). This is illustrated in Figure 4.
Test–retest reliability, framing and other effects Agreement between women’s preferences at their first and second interview was moderate to good (Table 4) and comparable to our previous study [5]. There was no systematic change between preferences at time 1 and time 2, as in Jansen et al.’s study [13]. Preferences were not affected by whether the better or worse prognosis scenarios were presented first (P = 0.7–0.9), or by whether visual aids were used (P = 0.3–0.5).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Cumulative % of women choosing treatment
90
1791 Table 3. Characteristics associated with judging smaller benefits of adjuvant chemotherapy sufficient
Table 3. (Continued) Factor
Factor
Linear univariable models P
Best linear multivariable model P
Linear univariable models P
Best linear multivariable model P
If the following were rated better:
Characteristics of women
Appetite
0.01
Age
0.8
Emotional well-being
Menopausal status at surgery
0.8
Physical well-being
<0.0001
<0.00001a
Menopausal status at interview
0.5
Overall well-being
0.0001
<0.00001a
0.0003
If she: a
0.4
Was employed at the time of interview
0.6
Has a life partner
0.5
Lives with others
0.6
Has children
0.2
Has dependent children
0.3
Has dependants
0.04
Had support always available during ACT
0.05
Was a member of community groups
0.5
Lived at home during chemotherapy
0.06
Had a friend or relative who died from cancer
0.04
Had less distance to travel to chemotherapy centre
0.7
Either physical or overall well-being were highly significant in the multivariable model, but not both. ACT, adjuvant chemotherapy; AC, doxorubicin, cyclophosphamide; CMF, cyclophosphamide, methotrexate, 5-fluorouracil.
Table 4. Test–retest reliability, based on paired measurements for 49 women 0.03
0.04
Scenario
Intraclass correlation coefficient
P for difference between interviews 1 and 2
Life expectancy of 5 years
0.7
0.2
Life expectancy of 15 years
0.4
0.2
5-year survival rate of 65%
0.5
0.1
5-year survival rate of 85%
0.7
0.4
Discussion
Characteristics of disease and treatment Tumour size
0.9
Lymph nodes involved
0.6
Hormone receptor status
0.7
Adjuvant radiotherapy given
0.3
Adjuvant endocrine therapy given
0.4
Disease recurrence before interview
0.7
AC alone versus CMF with or without AC
0.4
Shorter interval since chemotherapy finished
0.06
Experience of treatment If she was less troubled by: Problems coping with treatment
<0.0001
Thought of actually having the treatment
0.002
Nausea
0.002
Problems concentrating
0.006
Problems with needles or injections
0.008
Fatigue
0.008
Altered sense of taste
0.02
Anxiety
0.03
Hair loss
0.05
Problems with sleeping
0.4
Our most striking finding was that benefits of 1 day or 0.1% were judged sufficient by 52–55% of the women. In previous studies, an extra 6 months in life expectancy or an extra 1% in survival rates were judged sufficient to make adjuvant chemotherapy worthwhile by 50–52% of women [2]. Benefits of this size were judged sufficient by 62–72% of women in the current study. Benefits consistent with the Early Breast Cancer Trialists’ overview data (an extra 1 year in life expectancy or an extra 3% in survival rates) were judged sufficient to make adjuvant chemotherapy worthwhile by 58–77% of women in previous studies but 68–85% of women in the current study [1, 2]. There are many possible explanations for these striking results. Better supportive care may mean that chemotherapy is better tolerated and that smaller benefits are needed to offset the side-effects and inconvenience. Attitudes to cancer and its treatment may have changed making chemotherapy seem less dramatic and more acceptable. The larger number of women judging small benefits sufficient might be due to our including negligible benefits amongst our possible responses in the preferences interview. Our data do not shed light on which, if any, of these explanations is true. One interpretation of the plots of preferences for chemotherapy versus experience of chemotherapy is that women fell into one of two equal-sized groups. For women in the first group, the benefits judged necessary to make adjuvant chemotherapy worthwhile were related to the severity of side-effects experienced: larger benefits were required if side-effects were worse. These women were trading length of life for quality of
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Had tertiary education
Benefit needed: from least to most (normal score transformed)
1792
Figure 4. Relationship between the size of benefit judged sufficient to make adjuvant chemotherapy worthwhile and women’s recollections of their physical well-being during adjuvant chemotherapy. Open circles represent women who judged small benefits sufficient regardless of their well-being; filled circles represent women for whom the benefit required was related to their well-being.
life. For women in the second group, small benefits were judged sufficient to make adjuvant chemotherapy worthwhile regardless of the side-effects experienced: small benefits were sufficient even if side-effects were severe. These women seemed unwilling to trade length of life for quality of life. Women’s preferences were associated with their social circumstances, but not with their age, marital status or disease characteristics. Women with dependants and women who felt well-supported judged smaller benefits sufficient to make adjuvant chemotherapy worthwhile. As previously found by Yellen and Cella [14], women in our study with someone to live for (or someone living for them) seemed less influenced by the side-effects and inconvenience of their chemotherapy. The criterion for rational decision making is to maximise quality-adjusted survival. Quality of life is reduced during adjuvant chemotherapy. Some benefit is needed to make up this deficit. For almost half the women in our study, an extra day or 0.1% seemed sufficient. This could be interpreted as meaning that they were trying to live at all costs or that every day counts. Perhaps it means that feeling they have done everything confers other benefits or that some did not make their decision by trading direct harms and benefits of chemotherapy. It is tempting to attribute preferences for treatment with such small benefits to misunderstanding. Problems understanding and incorporating small probabilities in decision making are important and almost certainly part of the explanation, but a benefit of one extra day on a baseline of 5 or 15 years is hard to misunderstand. At least for some people, it may be more accurate to say that they understood the issues differently, rather than that they did not understand them at all.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Physical well-being during adjuvant chemotherapy (from best to worst)
The practical questions for clinicians are: what benefit is sufficient for them to offer adjuvant chemotherapy as a real option, and is this different for different individuals; how to deal with requests for adjuvant chemotherapy that is unlikely to be of benefit; and how to deal with people reluctant to have adjuvant chemotherapy with substantial benefits. Clinical practice guidelines offer a framework for making clinical decisions, but not absolute recommendations. In Australia, national evidence-based clinical practice guidelines specify appropriate indications for adjuvant chemotherapy, emphasise the importance of preferences and recommend that they be taken into account [15]. Our data reinforce the importance of explicitly considering benefits and harms when discussing adjuvant chemotherapy with patients. Different groups of people have remarkably different preferences for chemotherapy [2]. Reconciling the range and diversity of perspectives needs further consideration: the patients’ perspective that often favours treatments with small benefits; the clinicians’ perspective that favours treatments doing more good than harm; and the community’s perspective that favours treatments that are more cost-effective. Anxiety, fear, loss of control and regret are all part of being affected by cancer. These factors may influence people making decisions about cancer treatment. It is important to know whether and how these factors might influence decision making about potentially life-saving treatments. For example, women with a more pessimistic view of the future and limited social supports may be less inclined to prefer adjuvant chemotherapy. The main limitations of this study are related to its sampling frame and generalisability. We included only women who had experienced adjuvant chemotherapy. Their preferences were based on their recollections of what chemotherapy was like, and were likely to be influenced by psychological mechanisms, such as memory, adaptation, coping and cognitive dissonance reduction. Cognitive dissonance reduction occurs when an individual reinterprets past actions in a way that minimises conflict with their current attitudes. For example, the reason why a woman judges small benefits sufficient now may partly reflect the need to see her past decision to have chemotherapy in a positive light. A shift in her values that reaffirms her original decision to have adjuvant chemotherapy may also reflect adaptation and coping. The response rate in our study was relatively high and the results should be generalisable to comparable centres and women. However, the higher response rates should not obscure the fact that only women who had adjuvant chemotherapy were recruited, and that they are not representative of women considering whether to have adjuvant chemotherapy, some of whom declined. The preferences of health professionals, lay people and partners can also provide useful information about attitudes toward adjuvant chemotherapy for early breast cancer. We are currently completing studies determining the preferences of patients’ partners [16] and health professionals involved in providing their treatment. The attitudes of recently diagnosed women who are considering adjuvant chemotherapy are also of interest, but eliciting preferences in these women poses ethical and logistic problems. Studying women considering adjuvant chemotherapy removes the bias of selecting only those who chose to have it, but replaces it with the bias of how information
1793
Acknowledgements V.M.D. was supported by a PhD Scholarship from the National Health and Medical Research Council of Australia. Our thanks go to the study participants for their time and effort and the Consumer Advisory Panel of the Australia New Zealand Breast Cancer Trials Group for their help in piloting the interview. This research was supported by grants from the National Health and Medical Research Council of Australia, the Cancer Council of Australia, and the Australia New Zealand Breast Cancer Trials Group.
References 1. Early Breast Cancer Trialists’ Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet 1998; 352: 930–942.
2. Duric V, Stockler M. Patients’ preferences for adjuvant chemotherapy in early breast cancer: a review of what makes it worthwhile. Lancet Oncol 2001; 2: 691–697. 3. Lindley C, Vasa S, Sawyer WT, Winer EP. Quality of life and preferences for treatment following systemic adjuvant therapy for early-stage breast cancer. J Clin Oncol 1998; 16: 1380–1387. 4. Ravdin PM, Siminoff IA, Harvey JA. Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 1998; 16: 515–521. 5. Simes RJ, Coates AS. Patient preferences for adjuvant chemotherapy of early breast cancer: how much benefit is needed? J Natl Cancer Inst Monogr 2001; 30: 146–152. 6. Zimmermann C, Baldo C, Molino A. Framing of outcome and probability of recurrence: breast cancer patients’ choice of adjuvant chemotherapy in hypothetical patient scenarios. Breast Cancer Res Treat 2000; 60: 9–14. 7. Walter SD, Eliasziw M, Donner A. Sample size and optimal designs for reliability studies. Stat Med 1998; 17: 101–110. 8. McGraw K, Wong S. Forming inferences about some intraclass correlations coefficients: Correction. Psychol Methods 1996; 1: 390. 9. McGraw K, Wong S. Forming inferences about some intraclass correlation coefficients. Psychol Methods 1996; 1: 30–46. 10. Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977; 33: 363–374. 11. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–174. 12. Streiner DL, Norman GR. Health Measurement Scales: A Practical Guide to their Development and Use, 3rd edition. Oxford: Oxford University Press 2003. 13. Jansen SJ, Kievit J, Nooij MA et al. Patients’ preferences for adjuvant chemotherapy in early-stage breast cancer: is treatment worthwhile? Br J Cancer 2001; 84: 1577–1585. 14. Yellen SB, Cella DF. Someone to live for: social well-being, parenthood status, and decision-making in oncology. J Clin Oncol 1995; 13: 1255–1264. 15. National Health and Medical Research Council of Australia. NHMRC clinical practice guidelines for the management of early breast cancer. http:// www.health.gov.au/nhmrc/publications/2001. Available from: http://www. health.gov.au/nhmrc/publications/pdf/cp74.pdf (17 August, 2005, date accessed). 16. Duric V, Stockler M, Butow P et al. Comparing patients’ and their partners’ preferences for adjuvant chemotherapy in early breast cancer. ASCO annual meeting proceedings. J Clin Oncol 2005; 24: 00 (Abstr 6074). 17. Slovic P, Finucane M, Peters E, MacGregor D. Heuristics and biases: The psychology of intuitive judgment. In Gilovich T, Griffin D et al. (eds): Heuristics and Biases: The Psychology of Intuitive Judgment. New York: Cambridge University Press 2002; 397–420. 18. http://www.jncicancerspectrum.oupjournals.org/cgi/pdq/jncipdq, CDR0000067653?version=professional (16 March 2005, date accessed).
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
about adjuvant chemotherapy is presented, which may also influence their decisions [17]. Regimens including taxanes have become a standard part of adjuvant chemotherapy for women with higher risk disease, but were not considered in this study. The addition of taxanes increases effectiveness but also increases duration and toxicity. We are currently completing a study of women participating in an international randomised trial to determine the benefits they consider necessary to make the addition of a taxane worthwhile [18]. Preferences are remarkably diverse, even within groups of people with similar characteristics. In the current study, moderate but plausible benefits were sufficient for most women, but negligible benefits were sufficient for many, and implausibly large benefits were insufficient for others. Preferences appear to have been motivated by more than simple trade-offs between direct benefits and harms of chemotherapy. Future studies should seek to identify and measure these additional considerations so that they can be incorporated into models for decision making. Patients facing decisions about adjuvant chemotherapy have a lot to consider. So do their clinicians. Clinicians’ recommendations should be based on the best available evidence about disease and treatment, but they should also reflect the particular patient’s attitudes, values and priorities. People facing lifethreatening illnesses are the experts about their unique priorities, concerns, fears and regrets. Clinicians should ask about these things and try to understand their patients’ motivations for treatment so that they can better tailor their advice.
1794
Appendix P values from univariable linear, logistic and ordinal regression analyses assessing characteristics associated with judging smaller benefits of adjuvant chemotherapy sufficient Linear univariable models Pa
Ordinal univariable models Pa
Logistic univariable models Pa
0.82
0.51
0.32
Had tertiary education
0.42
0.76
0.90
Was employed at the time of interview
0.58
0.50
0.35
Has a life partner
0.50
0.62
0.72
Lives with others
0.55
0.46
0.72
Has children
0.23
0.18
0.05
Has dependent children
0.34
0.43
0.36
Has dependants
0.04
0.09
0.05
Characteristics of women Age If she:
0.05
0.04
0.07
Was a member of community groups
0.52
0.37
0.16
Was at home during chemotherapy
0.06
0.03
0.06
Had a friend or relative who died from cancer
0.04
0.08
0.05
Had less distance to travel to chemotherapy centre
0.70
0.23
0.08
Characteristics of disease and treatment Tumour size
0.93
0.64
0.48
Lymph nodes involved
0.58
0.55
0.38
Hormone receptor status
0.73
0.49
0.36
Adjuvant radiotherapy given
0.25
0.42
0.29
Adjuvant endocrine therapy given
0.38
0.29
0.45
Disease recurrence before interview
0.68
0.68
0.96
AC alone versus CMF with or without AC
0.43
0.53
0.77
Shorter interval since chemotherapy finished
0.06
0.03
0.10
Fatigue
0.008
0.003
0.01
Problems with sleeping
0.40
0.28
0.65
Nausea
0.002
0.002
0.005
Anxiety
0.03
0.01
0.03
Problems concentrating
0.006
0.002
0.002
Altered sense of taste
0.02
0.004
0.006
Hair loss
0.05
0.02
0.03
Experience of treatment If she was less troubled by:
Problems with needles or injections Problems coping with treatment Thought of actually having the treatment
0.008
0.005
0.02
<0.0001
0.001
0.003
0.002
0.004
0.006
If the following were rated better: Appetite
0.01
0.003
0.005
<0.0001
0.0002
0.004
Emotional well-being
0.0003
0.001
0.007
Overall well-being
0.0001
0.0002
0.002
Physical well-being
a
P values £0.1 are indicated in bold. AC, doxorubicin, cyclophosphamide; CMF, cyclophosphamide, methotrexate, 5-fluorouracil.
Downloaded from http://annonc.oxfordjournals.org/ at NERL on May 23, 2015
Had support available at all times during chemotherapy (if needed)