Shared decision‐making and personalised risk: a reply

We thank Williams and Day 1 for their thoughtful reflections on our consensus recommendations 2. Opening avenues of dialogue within the peri-operative community was one of the central aims of our publication, and we welcome the opportunity to further consider how peri-operative shared decision-making can be delivered in practice and not merely endorsed. We acknowledge that the scope of our published consensus recommendations cannot comprehensively capture the complexities of delivering this in real-world situations. We therefore thank the correspondents for expanding on the important challenges of communicating individualised risk, recognising clinician action bias and involving non-surgical specialties in a timely and meaningful manner. Our intention in formulating these consensus recommendations was to provide an operational framework that brings greater structure, consistency and visibility to the shared decision-making already occurring in clinical practice. We recognise that much of this patient-centred care is already being undertaken effectively by committed clinicians. However, as the correspondents allude to, these services often remain person-dependent, are under-resourced and reliant on improvised local solutions. We hope that the model of care described in these recommendations will support clinicians in advocating within their institutions for the personnel, time, training and infrastructure required to move from informal practice to formal, funded models of care. Without such structures, peri-operative shared decision-making will continue to vary in both reach and consistency. With the advent of more patient-centred outcome measures, we also need more patient-centred ways to communicate risk. The need for this is only accelerating with the development of data-driven personalised peri-operative risk tools powered by large language models. Current tools remain valuable adjuncts, but as the correspondents point out, most rely on datasets that inadequately capture outcomes that patients identify as important (e.g. loss of independence, prolonged recovery and unacceptable quality of life). However, even if we had the perfect tool, communicating probabilistic risk in a way that is unbiased, accessible, transparent about uncertainty and attentive to patient priorities may be an unobtainable ideal 3. A degree of discussion based on statistical risk and uncertainty is required, and training clinicians in both experiential communication skills and evidence-based risk communication strategies will be an important step. However, a shift towards the additional inclusion of non-probabilistic ways of presenting risk, focusing instead on qualitative and narrative risk 4, may also offer a parallel pathway forward. We agree also with the correspondents that the implicit bias towards action should be acknowledged explicitly. Providing meaningful treatment choice (including watchful waiting and best supportive care) requires accessible pathways to early multidisciplinary input from teams that deliver non-operative care. Developing new ways of working to embed these shared decision-making processes within peri-operative pathways will necessarily be gradual and incremental, to balance carefully patient needs, staff needs and operational efficiency. Now that we have a proposed model of care for peri-operative shared decision-making, the next substantive phase is implementation. This is no small challenge. Successful implementation will require co-design with patients, clinicians, managers and health services as end-users. Such collaborative efforts will be essential if shared decision-making is to move from a conceptual model of care to a sustainable component of routine perioperative practice.