We investigated whether oncologists currently have unmet decision support needs in the context of palliative treatment decision-making for mCRC. We found that oncologists that participated in our inventory questionnaire only knew one DSS for decision-making about palliative care for mCRC, namely the Fendler et al. [4] 4 nomogram. However, currently good quality DSS to guide palliative treatment decision-making for mCRC are lacking [12]. Available DSS have several shortcomings that limit their relevance for clinical practice. The methodology used for the development of DSS is not always optimal and often they have not been adequately externally validated or have only undergone narrow validations (e.g., in a single setting or ethnic population) [2, 5]. Additionally, available tools do not meet oncologists’ need for a comprehensive tool. Available DSS do not compare both the pros and cons of multiple treatment options, nor do they present relevant ongoing trials, and additional non-clinical outcomes such as cost-effectiveness of treatment options. Another shortcoming of available DSS is that they do not incorporate all clinically relevant predictors, such as the treatment’s impact on quality of life, the side-effects of treatment, performance status and prognosis. Finally, using informatics solutions facilitating communication between intra- and extramural specialists could also facilitate decision-making. Better exchange of information between healthcare providers contributes to the quality of care in general. However, better information exchange could for example, also improve the accuracy of the assessment of patients’ performance status, which in turn would improve estimates from prediction tools incorporating performance status as a predictor. Given the meager offerings in terms of available DSS, it is perhaps not surprising that respondents have many unmet decision support needs.
Respondents indicated that they currently mainly used the national CRC treatment guidelines [18] and/or the national palliative care guidelines [19] to support their clinical decisions. Even though respondents are confident in the quality of the content of the clinical guidelines, their general nature means they cannot be used to personalize treatment advice. One of the needs expressed was the incorporation of clear decision trees and good quality prediction tools in the online platform where guidelines are located. This would facilitate the embedding of DSS in the clinical decision-making process.
Further, in spite of the growing evidence that some oncogenes (e.g., KRAS, BRAF or PIK3CA), and tumor suppressor genes (e.g., APC, TP53 or PTEN) are relevant to prognosis, and could help to personalize treatment selection [13,14,15,16,17], respondents indicated that they currently do not consider genetic markers when pondering about their treatment advice. This is perhaps not surprising as for the mCRC setting more evidence is needed for such markers to be used in daily clinical practice. Given the rapid progress being made in the field of genetic markers and their potential to aid (further) personalization of treatment selection in the future, it is important that such factors are also incorporated in future DSS (once their prognostic value has been established).
If good quality DSS were available, they could help oncologists weigh the benefits and harms of treatment, when deliberating about their mCRC treatment advice. DSS, if they have a suitable patient interface, could also help oncologists to inform their patients about the benefits and harms of treatment. Given these potential benefits, it is pivotal to identify the mCRC treatment decisions for which oncologists have a need for decision support. In our study we focused on this element. A strength of our study is that we had access to all oncologists who are a member of the DCCG. Although the number of respondents is modest, it is likely that the specialists who participated are very experienced oncologists who use and/or feel a need for DSS in their daily practice. As in this study we aimed to inventory the requirements and characteristics of new DSS they need, this makes our sample of oncologists valuable.
Given the complex and multi-faceted nature of palliative treatment decision-making, and the need for a personalized and patient-centered approach, it is somewhat surprising that there are so few good quality DSS available for the mCRC setting. The development of new DSS for the mCRC setting could be of great value to clinical practice. However, development of new DSS meeting specialists’ needs is not straightforward. More research is needed to identify relevant predictors. This is particularly the case for genetic markers to help in the selection of patients for treatment. It is thus unlikely that in the short-term a good quality DSS incorporating genetic markers will be developed. In the meantime, DSS estimating prognosis based on patient and disease characteristics are more feasible and could nonetheless be of added value to clinical practice (for example the mortality calculator proposed by Refro et al. [20]). Further, making DSS available that remain relevant and reliable in the long-term is a key consideration from a methodological standpoint. For any (newly developed) DSS to be useful in clinical practice in the long-term, it is important that the tool is continuously updated. We propose for example, embedding DSS in national patient registries, to facilitate updating as new insights and treatments become available. Finally, although DSS can be of great value in clinical practice and oncologists expressed a clear need for such tools, they also strongly felt that DSS are nor should be a replacement for their clinical judgement or for deliberation with patients.