Patient participation in the treatment decision-making process is widely advocated and essential in the context of treatment decisions when there is no ‘best’ treatment option. A key requirement to achieve this goal is thorough and balanced information provision about the benefits and harms of the viable treatment options. There are many factors that can negatively influence information provision in clinical practice. Unfortunately, insights in information provision during real-time patient consultations involving preference-sensitive decisions is limited. The objective of the work presented in the thesis is to assess information provision about adjuvant systemic therapy during consultations between early-stage breast cancer patients and medical oncologists in general. In this era of personalized medicine, prediction tools (e.g., Adjuvant!) are becoming an integral part of information provision during patient consultations. However, evidence is lacking about a) the availability and the quality of prediction tools for the early-stage breast cancer setting, b) how prevalent the use of such tools is during patient consultations, and c) whether and how the use of such tools influences information provision.
In Part I, we investigate the availability and accuracy of risk prediction models for decision-making about adjuvant systemic therapy for early-stage breast cancer. An essential prerequisite for the use of such tools in clinical practice, is that their estimates have to be accurate. In this part, we present the results of a systematic review of published risk prediction models for adjuvant systemic therapy selection in early-stage breast cancer, that includes an assessment of the strengths and weaknesses of the identified models. Most prediction tools were developed to inform clinicians’ decisions, yet they are also used to inform patients. Therefore, in the review we also assessed the required literacy level to comprehend the content of the output provided by the tools we identified. Next, we assessed the prognostic accuracy of the 10-year all-cause mortality estimates in breast cancer patients aged <50 years at diagnosis for two commonly used prediction models, namely Adjuvant! (currently offline) and PREDICT (http://www.predict.nhs.uk/predict_v2.0.html). We focused on young patients as previous validation studies had too few young patients and/or the follow-up time was too brief to draw conclusions about the accuracy of these tools in this younger patient population.
In Part II, we assessed oncologists’ attitudes towards and self-reported use of tools to communicate the benefits of adjuvant systemic therapy for early-stage breast cancer. With a survey amongst oncologists, we assessed their perception of the minimal benefit that makes treatment worthwhile given the side-effects. Clinical guidelines indicate that 3-5% is the minimum benefit that makes treatment worth considering given its side-effects. We assessed whether oncologists’ minimally required benefit to tip the scale in favor of treatment is in line with the guidelines. These insights are relevant as oncologists’ preferences and beliefs can influence their information provision and treatment recommendations. In the same survey, we also examined oncologist perceptions of and reasons for using prediction tools, and views on communicating the uncertainty associated with prognostic estimates from such tools, as little is known about this.
In Part III, we assessed information provision about the benefits and harms of adjuvant systemic therapy for early-stage breast cancer during real-time patient consultations. From the results of Part II, we learned that Adjuvant! was the most frequently used prediction tool, therefore in this part we focused on Adjuvant! use. We investigated the frequency and the influence of the use of Adjuvant! on information provision about the benefits and harms of adjuvant systemic therapy, and whether the use of this tool is associated with the likelihood of reaching a decision during the consultation. Next, we zoomed in on a controversial element of risk communication, namely the communication of the uncertainty associated with the prognostic estimates provided by prediction tools. There currently are no generally accepted guidelines on whether and how to communicate uncertainty, and evidence on whether uncertainty is communicated in clinical practice is also lacking. We assessed whether and which type of uncertainty was communicated during patient consultations and evaluated to what extent patients perceived the uncertainty associated with the prognostic estimates communicated during the consultation. Finally, we explored whether the presentation of information about adjuvant systemic therapy during the consultation contained implicitly persuasive elements. Such behaviors could inadvertently steer patients facing preference-sensitive decisions towards a particular choice that might not be in line with the patients’ values and goals.
From the work contained in the thesis, seven key findings can be distilled, namely:
1. Many prediction tools are available to aid adjuvant systemic treatment decision-making, however, they require either further calibration and/or broad validation.
2. Medical oncologists set higher thresholds for considering adjuvant systemic treatment worthwhile compared to the thresholds used in clinical guidelines.
3. Adjuvant! is regularly used by medical oncologists to inform patients about their prognosis and the potential treatment benefit.
4. In spite of reservations about the robustness of Adjuvant!’s relapse estimates, medical oncologists usually only communicate the relapse probabilities to patients.
5. The uncertainty associated with Adjuvant!’s estimates is not always communicated, and patients struggle with the concept of epistemic uncertainty.
6. The suboptimal information provision about treatment side-effects during consultations, suggests that the potential treatment benefits mainly drive adjuvant systemic treatment decisions.
7. In spite of the lack of a ‘best’ treatment option, medical oncologists use implicit persuasion to steer patients towards the treatment option they deem in their patients’ best interest.
The thesis is available at: https://issuu.com/ellenengelhardt/docs/proefschrift_ellen_engelhardt_digit. For a hardcopy, please e-mail Ellen at e.engelhardtvumc.nl.