Introduction

The majority of patients currently diagnosed with breast cancer (BC) expect a “patient-centered” process of treatment-related decision making. Several studies have suggested that patients who take a more active role in this process are better motivated and have better compliance with their care, and may even have a more favorable outcome [1–5]. The reported rates of patients with BC in different studies who were willing to delegate responsibility for their treatment selection completely to their physician vary between 20 and 38 %. In contrast, 40–60 % of patients wish to select their treatment collaboratively with their physician as equal partners, in accordance with the principle of shared decision-making (SDM) that includes patients’ perceptions, concerns, and expectations [6, 7]. The preference for taking part in medical decision-making is reported to depend on age, educational level, and current health status, with younger and better-educated patients preferring a more active role [6, 8]. Treatment decisions for or against systemic adjuvant therapy are based on estimates of the risk of recurrence and on predictions of the treatment response. Current clinical practice is essentially to compare the potential reduction in the risk of recurrence with expected adverse treatment effects [9, 36]. Patients with a perceived high risk of recurrence are treated with all the available treatment options, whereas low-risk patients mainly receive follow-up or treatments that have a favorable side effect profile. The problem is that the area of overlap between these extreme groups of patients with BC is only partly covered by clear guidelines, while each individual patient has different values and different perceptions of health, well-being, and future prospects. There has even been evidence suggesting that clinicians are mostly inaccurate in predicting patient values [10]. Comorbidity, the individual risk of side effects, life expectancy, the strength of evidence supporting the treatment, and in particular the patient’s preferences also affect the treatment decision. Cancer patients need to know the facts and circumstances of their disease and the possible benefits and adverse effects of all the treatment options to allow informed decision-making regarding treatment. In many cases, these facts are incompletely explained or patients are overwhelmed by the cancer diagnosis and are emotionally unable to cope with immediate complex decisions regarding adjuvant treatment [11–13]. Nevertheless, a recent study in Germany found a high rate of acceptance of SDM among physicians, with 68 % completely accepting this approach [14].

Previous studies suggested a pronounced discrepancy between the expectations of patients and those of their physicians with regard to utility assessment (preference values) and thresholds for accepting therapy [15, 16]. Mismatches between physician and patient viewpoints may result in detrimental treatment decisions. In the present study of patients with BC, patients and physicians were, therefore, asked about the desired treatment efficacy, using virtual BC cases and proposed therapy regimens with specific benefits and side effects.

Patients and methods

Study design

The study was divided into an initial stage, conducted from October to December 2005, and the subsequent main stage, from April 2006 to December 2007. The pilot phase was used to test the plausibility and applicability of the specific questionnaires by sending them to 50 patients with BC who were treated at our center (the patients’ questionnaire) and to 100 oncologists in private practice (the physicians’ questionnaire), along with an evaluation sheet. Patients and physicians assessed the comprehensibility of the questionnaire on a scale from 1 (incomprehensible) to 10 (coherent); the response rates were 96.0 % (n = 48) in the patient group and 23.0 % (n = 23) in the physician group. The median comprehensibility was slightly lower in the group of BC patients, at 8.28 (range: 7.73–8.95), in comparison with the physicians, at 8.72 (range: 8.10–9.48), but was acceptable in both groups. Targeted revision of the questionnaire was carried out on the basis of the data from the pilot phase and in collaboration with patient advocacy groups, to optimize intelligibility and to include further suggestions from patients and physicians.

Questionnaire

Patients and physicians received a letter informing them about the background and intention of the study, along with the respective questionnaire. The two types of questionnaire contained a common part, with six case descriptions in the adjuvant setting, and special sections for patients (asking for personal details and medical history) or for physicians (asking for details about their practice, qualification, specialty, etc.). All participants were asked to indicate (for each case and each treatment option) which additional percentage benefit on top of a given 5-year overall survival rate (60 or 80 %) would be necessary for the adjuvant treatment to be acceptable. The structure of the questions was as follows: “If a breast cancer patient would have an 80 % probability to survive the next 5 years without chemotherapy, which increase (in percent) would you demand to indicate chemotherapy in that case?” The six virtual cases are summarized in Table 1.

Table 1 Summary of the cases for which the patients and physicians were asked about the extent of a desired treatment effect required to allow them to decide in favor of the respective therapy (OS, overall survival)
Full size table

Data collection

The study was conducted in collaboration with the patient advocacy groups in Germany, “Frauenselbsthilfe nach Krebs e.V.,” “mamazone e.V.,” and “Brustkrebs Deutschland e.V.,” which provided contact addresses for a further 444 self-help groups. These groups, along with the patient advocacy groups, distributed the questionnaires to 6,460 patients with BC. In addition, 815 follow-up patients at our BC center and 1,725 patients from 26 other certified German breast centers also received a questionnaire during one of their follow-up visits. All of the patients received a prepaid envelope for returning the questionnaire. On the physicians’ side, 6,938 gynecologists, general practitioners with oncological expertise, and oncologists from all parts of Germany were contacted by mail. In addition, an Internet site was created to make further information about the study available for patients and physicians (www.gutinformieren.de). It was also possible to complete the whole questionnaire online on the Internet site. A total of 9,000 questionnaires were sent to patients and 6,938 to physicians. The response rate among patients was 23.9 % (n = 2,155), with 1 % (n = 90) using the Internet to answer the questions. The response rate among physicians was significantly lower, with a total of 6,938 questionnaires mailed and a return rate of 7.6 % (n = 527), 0.8 % (n = 55) of which were via the Internet. The response rate among patients was similar to that in previous reports in the literature, in which comparable studies achieved return rates of around 24 %, but the response rate among physicians was lower than in previous reports, in which rates of 16.7 % have been described [17–20].

Statistical analysis

The increases in survival offered in the questionnaire were regarded as ordinal categorical response options. The responses of the patients and physicians were compared using Wilcoxon rank sum tests. Within patients and physicians, characteristics were associated with responses using the appropriate statistical tests. Wilcoxon rank sum tests were used for categorical characteristics, and Spearman correlation tests were used for continuous or ordinal categorical characteristics. Unadjusted P values are shown. All of the tests were two-sided, and a P value <0.05 was regarded as statistically significant. Calculations were carried out using the R system for statistical computing (version 2.11.1; R Development Core Team, Vienna, Austria, 2010).

Results

Participants

At the time of the survey, the patients’ average age was 59.1 years. Most of the patients had unilateral BC (92.3 %) and an early tumor stage (31.4 % pT1) without lymph-node involvement (64.1 % pN0). Eleven percent (n = 238) of all participating patients with BC had already experienced a relapse following adjuvant treatment, and 8.4 % (n = 182) were in the metastatic stage. More than half of the patients had received chemotherapy (56.7 %); 71.1 % (n = 1,533) had received radiotherapy, 7.1 % (n = 152) had recieved antibody treatment, and 65.9 % (n = 1,420) of the patients had received or were still receiving endocrine therapy. Complementary treatment options were used by 60.1 % (n = 1,296) of all the participating patients. Tumor and patients’ characteristics are shown in Table 2.

Table 2 Patients’ characteristics
Full size table

The average age of the physicians (n = 462) was 49.7 years; 52.9 % were women and 93.9 % were gynecologists specializing in oncology. Only 3 % of the physicians were still in training, and 80.8 % were in private practice. The average number of patients with primary BC seen per week was 7.2; while the average number of patients with metastatic BC seen per week was significantly lower (2.8). In all, 88.6 % of the physicians stated that they were providing BC follow-up care (Table 3).

Table 3 Physicians’ characteristics
Full size table

Perception of treatment benefit and thresholds for treatment acceptability

The thresholds for accepting a specific treatment regimen differed significantly between patients and physicians for all six of the hypothetical cases and for both scenarios, one of which was based on a 60 % rate of overall survival and the other on an 80 % overall survival rate after 5 years without treatment (Figs. 1 and 2). In all cases, the physicians preferred a lower treatment effect than the patients. The P values for all comparisons in all six cases were below 10−10.

Fig. 1
figure 1

a Percentage distribution of the desired treatment effect for chemotherapy with a baseline 5-year overall survival (5y-OS) rate of 60 % for physicians and patients. (P value for comparison between patients and physicians: P < 10−10). b Percentage distribution of the desired treatment effect for endocrine therapy with a baseline 5-year overall survival (5y-OS) rate of 60 % for physicians and patients. (P value for the comparison between patients and physicians: P < 10−10). c Percentage distribution of the desired treatment effect for antibody therapy with a baseline 5-year overall survival (5y-OS) rate of 60 % for physicians and patients. (P value for the comparison between patients and physicians: P < 10−10)

Full size image
Fig. 2
figure 2

a Percentage distribution of the desired treatment effect for chemotherapy with a baseline 5-year overall survival (5y-OS) rate of 80 % for physicians and patients. (P value for the comparison between patients and physicians: P < 10−10). b Percentage distribution of the desired treatment effect for endocrine therapy with a baseline 5-year overall survival (5y-OS) rate of 80 % for physicians and patients. (P value for the comparison between patients and physicians: P < 10−10). c Percentage distribution of the desired treatment effect for antibody therapy with a baseline 5-year overall survival (5y-OS) rate of 80 % for physicians and patients. (P value for the comparison between patients and physicians: P < 10−10)

Full size image

Chemotherapy

Two findings were of particular interest in comparisons of the desired benefit between patients with BC and physicians. First, the requested treatment benefit was greater in the patients’ group in comparison with the physicians, with about one-third of patients demanding a 5-year overall survival benefit of 30 % or more in addition to a given 60 % 5-year overall survival rate without therapy to consent to treatment. In comparison, in the group of physicians the proportion demanding an additional benefit of 30 % or more was only about 5 %. More than half of the physicians (56.7 %) rated a 5–10 % benefit in the 5-year overall survival as acceptable for chemotherapy, based on a 5-year overall survival rate of 60 % without treatment, while only 18.2 % of patients shared this opinion.

Second, a substantial group of patients (15.9 %) were willing to accept the suggested treatment regimen even with a 5-year overall survival benefit of only 1 %. This group of patients appear to be motivated by a minimal benefit, regardless of possible side effects. In contrast, the number of physicians satisfied with such a minimal chemotherapy benefit rate was significantly lower (3.7 %). The results are shown in Fig. 1a. The case based on a 5-year overall survival rate of 80 % yielded very similar results, which were not significantly different from the scenario with 60 % (Fig. 2a).

Endocrine therapy

The evaluation of the expected benefit of endocrine therapy revealed similar results among the patients, but slightly different results among the physicians. Again, one group of patients (21.3 %) was satisfied with a very small increase in the benefit, from 60 to 61 % (“minimal benefit group”) (Fig. 1b). On the other hand, a large group of patients (33.2 %) did not consider the suggested endocrine treatment acceptable until a benefit increase from 60 to 90 % or more was present. In the group of physicians, a benefit of 5 % was sufficient for one-third of the respondents, while 10.2 % of the physicians were satisfied with a benefit of 3 %. In contrast to the chemotherapy cases, a substantial “minimal benefit group” was observed for endocrine therapy among the physicians. A total of 15.1 % of the physicians would start this form of treatment if a benefit of only a 1 % increase in the overall survival was obtained. The scenario with a primary 5-year overall survival rate of 80 % (without any treatment) showed very similar significant results (Fig. 2b).

Antibody therapy

With regard to the expected benefit associated with 1 year of antibody treatment in case of a 60 % 5-year overall survival rate without therapy, more than 22 % of the patients were willing to accept a benefit of only 1 %, while this was acceptable to only 6.1 % of the physicians. The majority of physicians (53.4 %) would suggest the antibody treatment only if an increase of 5–10 % in the 5-year overall survival was achieved (Fig. 1c). Again, very similar results were seen with a 5-year overall survival rate of 80 % as the base value (Fig. 2c).

Factors influencing physicians’ treatment decisions

In the group of physicians, several factors were identified that significantly influenced the desired treatment benefit. With increasing age and professional experience, the physicians expected greater treatment benefits. The medical specialty (gynecologists versus nongynecologists) and whether the physician was seeing the patient in follow-up care did not have any significant effects. There was a trend toward a lower expected benefit with chemotherapy in physicians who were working in hospitals in comparison with physicians in private practice, but this was not consistent in all cases or for both baseline 5-year overall survival assumptions (60 and 80 %).

If the physician stated that he or she made treatment decisions without the input of a multidisciplinary conference in a certified breast center, lower benefits were desired for the chemotherapy cases and endocrine therapy cases. Associations between physician characteristics and the desired treatment benefit for all six cases are summarized in Table 4.

Table 4 P values indicating characteristics that were significantly associated with lower requirements for treatment benefit among physicians; e.g., younger age in case 1 was significantly (P = 0.001) associated with lower requirements for treatment benefit
Full size table

Factors influencing patients’ treatment decisions

Several personal details given by patients with BC in the questionnaire were identified as factors influencing the desired treatment benefit (Table 5). There was a highly significant positive correlation between the expected treatment benefit and age for all the cases presented. There was also a strongly significant association with the age of a patient’s child. The younger the child was, the less treatment benefit the patient was willing to accept to decide in favor of therapy relative to all of the cases presented. A higher educational level and active membership of a self-help group were also associated with lower desired treatment benefits. Another factor that was strongly correlated with the desired treatment benefit was whether the patient herself had previously experienced a specific treatment and the side effects associated with it. Patients who had been through any type of therapy were more likely to accept lower treatment benefits. Similarly, patients who had not experienced severe side effects were more likely to accept lower treatment effects (Table 5).

Table 5 P values indicating characteristics associated with patients’ assessment of lower benefits of therapy
Full size table

Discussion

This study, including 2,682 participants, investigated how closely patients with BC and physicians agree in their preferences for different adjuvant treatment options with specified benefits. Significant differences were observed in the perception of the expected benefits of adjuvant BC treatment between patients and physicians using six constructed cases with given benefit rates for chemotherapy, endocrine therapy, and antibody treatment. Personal and work-related factors were also identified that significantly influenced the perception of thresholds for adjuvant BC treatment in the two groups.

While the responses of most of the physicians were within the range of a realistically expected treatment benefit, the patients appeared to divide into two groups. One large subgroup of patients expected unrealistic benefits before they were willing to accept treatment (e.g., 34.2 % of patients vs. 5.5 % of physicians expected a 5-year overall survival benefit of 30 % or more with chemotherapy in the scenario of a 60 % 5-year overall survival without any therapy). Another subgroup of patients only had a very low expectation of treatment benefit (e.g., 15.9 % of patients were satisfied with a 1 % benefit in the 5-year overall survival with chemotherapy, in comparison with 3.7 % of physicians). This group of patients with very low treatment benefit expectations has also been identified previously in considerably smaller studies. These studies suggested that a substantial proportion of patients with BC judge benefit rates with adjuvant treatment to be acceptable that health-care professionals feel are not meaningful [13, 21].

Interestingly, the group of patients in the present study who expected very low treatment benefits was considerably smaller than the proportions reported in previous studies. For example, Duric et al. [22] surveyed 46 women with adjuvant chemotherapy 3–34 months after the treatment and found that the majority of the patients (59–72 %) only expected a chemotherapy benefit of 0.1 % (in the 5-year overall survival rate). More than half of the patients in the study stated that a survival benefit of even only 1 day would be sufficient for chemotherapy to be acceptable. In a similar study by Simes and Coates, including 104 patients, 73 % of the patients rated a minimal clinically significant difference of 5 % or less in the 5-year overall survival as acceptable for consent to chemotherapy treatment [23, 24]. The reasons for these differences between smaller previous studies and the present study with regard to the required benefits that justify adjuvant treatment for women with early-stage BC are elusive. The studies were conducted in different countries, and the selection of certain patient groups could also be a strong influencing factor, especially in smaller studies on shared decision-making (SDM). The present study, therefore, attempted to include a large number of patients (9,000) and physicians (6,938) and used various contact routes to achieve a potential cross section of all patients with BC in the population in Germany. Nevertheless, the way of distribution of the questionnaires could be a potential limitation of this study. We attempted to use various contact routes but still most of the 9,000 questionnaires (6,460) were distributed by advocacy groups. The low response rate among patients and physicians constitutes a significant limitation of our study and may result in selection bias of a particular “type” of patients responding to the questionnaire. It is all the more surprising that a smaller proportion of patients with very low treatment thresholds were observed in comparison with previous studies.

The same trend was found with regard to adjuvant endocrine therapy, with 21.3 % (20.9 %) of patients with BC regarding a 5-year overall survival benefit of 1 % as an acceptable increase, while 33.2 % of patients expected a 5-year overall survival gain of 30 % or more, based on a 60 % (80 %) 5-year overall survival rate. In contrast, a previous study that included 102 patients aged 40 years or younger with early-stage BC found that half of the participants considered that endocrine therapy is worthwhile for an absolute survival gain of 2 % from a baseline of either 65 or 85 % [25]. Another study by Duric et al., including 85 women 6–30 months after the completion of adjuvant endocrine therapy, showed results closer to the present findings, with one-third of patients considering gains of 1 % in survival rates or 6 months in survival time sufficient to give their consent to adjuvant endocrine treatment. More than half of the women required gains of at least 5 % in the survival rate or 3 years in survival time, while 16 % of patients expected gains of at least 15 % (or 10 years) to make therapy worthwhile. The study by Duric et al. [26] also used hypothetical scenarios with specified survival times and rates at 5 years.

Previous side effects, social support, age, education, and income have been identified as factors influencing patients’ treatment preferences in previous studies [27, 28]. In line with these data, the present study also found significantly increasing treatment thresholds with increasing age, previous side effects, and decreasing levels of education. Another parameter associated with preferences in the SDM process was the age of the youngest child. Women with younger children tend to be willing to accept treatment that offers little benefit. A larger number of children was associated with significantly higher treatment thresholds in two of the six cases. This finding could be explained by concerns about childcare during therapy.

The strengths of the present study include the large cohort of patients and the extensive dataset, with six different virtual scenarios for treatment decisions. However, the findings should be interpreted in the light of several limitations. First, in contrast to previous studies, questionnaires were used instead of personal interviews [22–24, 26–31]. This might have led to misunderstandings and might prevent patients and physicians from completing the questionnaire correctly. An attempt was made to minimize this potential bias and make the questionnaire as understandable as possible through intensive preliminary testing and revision of the questionnaire during the initial stage of the study.

Second, patients may be likely to accept any hypothetical treatment that offers at least a little benefit. In addition, behavioral measures and patients’ perceptions of involvement and thresholds often differ significantly in the SDM process [32]. The hypothetical nature of the six cases and treatment decisions used in this study as well as the potential bias due to patients′ experiences with previous adjuvant therapies (e.g., being recurrence free even without adjuvant chemotherapy) allows only limited conclusions to be drawn in comparison with prospective studies evaluating a real clinical decision faced at the time. Unfortunately, such large-scale prospective studies are still lacking and are very difficult to conduct.

All of the patients surveyed in this study had experienced the situation of serious decision-making before, during their active cancer treatment. A quite realistic assessment of treatment thresholds could therefore be expected, even in the study’s hypothetical setting. Despite this, unrealistic treatment thresholds were observed in a large proportion of patients suggesting shortcomings in the SDM process. It has been shown previously that most physicians only gave their prognostic assessments in qualitative terms and that only 46 % of women receiving chemotherapy had been given precise data on outcome variables such as the 5-year overall survival or disease-free survival rates. The percentages were even lower for other treatment regimens such as endocrine or antibody therapy [13]. Recent research data suggest that the current rates may now be higher, due to a more active role of patients with BC in SDM, increased communication skills training of physicians and due to the availability of web-based programs such as “Adjuvant! Online” that calculate individualized probabilities of survival and benefit [33–35, 37]. A significantly different perception of treatment thresholds was observed in the present study among physicians who used web-based risk calculation programs, with a significantly lower expected benefit required for indicating a specific therapy in three of the six cases.

In conclusion, this large study noted significant differences between patients’ and physicians’ perceptions of acceptable levels of benefit from adjuvant treatment for BC and identified influencing factors on both sides, which should be taken into account during shared decision making.