The decision about major amputation takes place inside a clinical context in which the stakes are already measured in mortality. Zeng et al.(4) reported from three international cohorts that long-term mortality in diabetic foot disease is higher than that of overall cancer, a finding that reframes amputation as one turn in a longer illness rather than as its endpoint. Schaper et al.(1) supplied the classification that stages foot risk and guides prevention, and Xiang et al.(2) quantified the incidence of diabetes-related foot disease in two prospective cohorts and a meta-analysis. Zhu et al.(3) described how hospital burden, amputation risk and mortality have evolved over time, and Chuter et al.(5) observed that the declines in amputation are real but locally uneven, so the probability of limb loss is not distributed uniformly across the population that faces it.

Against that background clinicians attempt to predict who will lose the limb and who will survive its loss. Lu et al.(21) developed a risk prediction model for major amputation in patients with diabetic foot, Tillqvist et al.(22) showed that the clinical frailty scale predicts outcomes after lower extremity amputation among people with diabetes, and Trivedi et al.(23) mapped the amputation levels available in high-risk chronic limb-threatening ischemia. These instruments describe the probability space, but none of them encodes what the patient values, which is why the same prediction can be used to justify aggressive salvage in one consultation and early amputation in another. A model that predicts a high probability of amputation can, in one service, trigger an earlier referral for salvage and, in another, a faster acceptance of limb loss, because the instrument carries no instruction about which response respects the patient's priorities. The tools therefore sharpen the clinical question without narrowing the ethical one, and their developers report discrimination but not deliberation.

The treatment options themselves are described by authors who reach compatible conclusions about the trade-off. Nypaver(6) reviewed chronic limb-threatening ischemia and positioned revascularization against primary amputation, and Radomski et al.(7) examined how older adults face that same choice, where functional recovery and time at home matter as much as limb retention. Hinojosa et al.(24) added the trajectory view by showing how minor amputations progress to major ones in patients with diabetes and peripheral arterial disease, which means the decision is not a single event but a series of escalations. Each escalation is an opportunity for a shared decision that clinical practice does not consistently take, because the minor amputation that precedes the major one is not discussed in the same ethical terms as the major one that follows. The salvage option is summarized by Chiu et al.,(8) who found that multidisciplinary management reduces readmission, major amputation and mortality, by Fereydooni et al.,(9) who observed that an extremity preservation program improves quality of life, and by Leo et al.,(10) who reported that podiatric care is associated with reduced mortality and enhanced amputation-free survival. Chávez Moya et al.(11) summarized the podiatric and technological strategies that move a diabetic foot from ulcer to remission, and Ruemenapf et al.(25) described the therapeutic alternatives for patients without an option for revascularization, a group for whom primary amputation can be the humane rather than the failed choice. Unlike the findings of Chiu et al.(8) and Leo et al.,(10) who emphasized the gains of organized salvage, Ruemenapf et al.(25) described a subgroup in whom salvage is not available and for whom the clinical question is not whether to save the limb but how to time its loss. The gap that remains after these contributions is not clinical but deliberative: none of the authors evaluated whether their prognostic instruments changed the quality of the decision conversation itself, which is the step the next three axes examine.

The ethical reading of the amputation decision has its clearest statement in the analysis of Xu et al.,(14) who applied the four principles of biomedical ethics to the choice between amputation and limb salvage in limb-threatening lower extremity conditions and concluded that autonomy, beneficence, non-maleficence and justice pull in different directions in almost every case. Autonomy defers to the patient's expressed wish; beneficence points toward the intervention most likely to prolong a functional life; non-maleficence cautions against the morbidity of either path; and justice asks how scarce revascularization and prosthetic resources are distributed. The consequence of Xu et al.(14) is that a purely principle-based analysis clarifies the conflict but does not resolve it, because the principles must be weighted rather than applied. The fourth principle is also the least developed in this literature. None of the ethical analyses collected here quantifies how revascularization capacity, operating time and prosthetic funding are rationed across the populations in which, as the fourth axis shows, the amputation burden concentrates, which leaves the justice dimension of the decision asserted rather than measured.

What those principles mean on the ground is supplied by Uhrenfeldt et al.,(19) who described the ethical dimensions of living with one leg in a qualitative study and showed that amputation is experienced as an existential and moral disruption that goes beyond functional loss. Just as Xu et al.(14) found the principles in tension, Uhrenfeldt et al.(19) found that the meaning of benefit is reconstructed by patients after the fact, which weakens any clinician's confidence in predicting it beforehand. Amputation is experienced as an existential and moral disruption that reaches beyond functional loss, and the analysis of Uhrenfeldt et al.(19) implies that a limb a surgeon judges non-viable may still be the limb a patient needs for identity, while a life that looks shortened on a survival curve may still feel worth living to the person who owns it.

A second conflict concerns the point at which continued salvage becomes harm. Zhou et al.(26) argued that deferring amputation in diabetic foot osteomyelitis can do more harm than good, because prolonged infection and repeated procedures accumulate morbidity that a timely amputation would have spared. de Mestral et al.(27) described end-of-life care following leg amputation in patients with peripheral artery disease or diabetes, positioning amputation within palliative rather than curative trajectories, and Cattermole et al.(28) reported that integrating palliative care consultation into the management of chronic limb-threatening ischemia changes the conversation from limb to life. Martelli et al.(29) framed the decision to stop as the most difficult and painful one clinicians face, and Nash et al.(30) argued that amputation above the ankle can be the intervention that achieves the best outcome for the patient, reversing the assumption that salvage is always the virtuous choice. Guo et al.(31) showed that a statewide one-year mortality estimate can be generated prospectively and used to mark futility, although they also demonstrated that a probability is not a judgment. The unresolved question is empirical: the normative analyses of Xu et al.(14) have not been tested against the accounts that patients such as those interviewed by Uhrenfeldt et al.(19) actually give. That testing is precisely what the shared decision-making axis describes, and its results suggest that the principles are easier to assert than to operationalize.

The largest recent empirical contribution comes from Prout et al.,(16) whose PERCEIVE qualitative study documented the challenges of shared decision-making about major lower limb amputation and found that patients, families and clinicians struggle with timing, information and the weight of a decision that feels irreversible. The distance between the shared decision-making model set out in the introduction and the practice that Prout et al.(16) observed is the central finding of this axis, and the studies that follow locate the gap at each step of the conversation.

Ong et al.(15) showed in a scoping review that decision-making processes for non-emergency diabetes-related lower extremity amputations are heterogeneous, under-documented and not made explicit, which means the field cannot yet say what a typical decision looks like. Leonard et al.(32) found that patients and their care teams perceive shared decision-making differently, with patients reporting less involvement than clinicians believe they offer. Støvring(33) reported from a national cross-sectional survey that patient experience of pre-amputation communication is variable, and Støvring(34) described the facilitators and barriers that surgeons themselves perceive when deciding the level of amputation. The affective dimension is not incidental to these gaps. Wang et al.(35) found that anxiety and depression mediate the relationship between decisional conflict and decision regret in diabetic foot amputation patients, and Cach et al.(36) identified the factors that influence acceptance of lower extremity amputation. The timing of the conversation is part of the difficulty. Prout et al.(16) found that patients are approached about amputation at moments of crisis, when pain, sepsis or altered cognition limit the very capacity that shared decision-making presupposes, and Støvring(33) showed that the quality of pre-amputation communication improves when the conversation begins before that crisis.

Preferences are measurable and differ between the parties. O'Hara et al.(37) quantified patient preferences regarding limb salvage for diabetic foot ulcers in a discrete choice experiment and showed that patients assign measurable value to avoiding major amputation. Poehler et al.(38) developed a multiple criteria decision analysis to compare patient and provider priorities around amputation-level outcomes, and Poehler et al.(39) found that those priorities diverge, with patients and providers ranking the same amputation-level outcomes in different orders. Günbaş et al.(40) described how the fear of amputation shapes the lives of patients with diabetic foot ulcers even before any decision is made, which helps explain why the offer of information alone does not produce a shared decision.

Tools exist but remain at the edge of practice. Czerniecki et al.(17) developed the AMPDECIDE decision aid for patients facing amputation secondary to chronic limb-threatening ischemia, Henderson et al.(41) tested its feasibility, and Smolderen et al.(18) described the development and alpha testing of the SHOW-ME PAD decision aid for treatment decisions in peripheral artery disease. Laor et al.(42) assessed the online patient education materials available for major lower extremity amputation, and Anderson et al.(43) showed that shared decision-making for prosthetic care after amputation remains similarly underdeveloped. The digital layer is arriving without a settled role. ten Klooster et al.(44) interviewed stakeholders about machine learning models that predict diabetic foot ulcers and amputations, Shan et al.(45) traced how non-professional health information delays standard care, and Anil et al.(46) showed that discomfort avoidance and a desire for normality drive footwear choices in underserved communities at risk of ulceration. Goodney et al.(47) reviewed patient-reported outcome measures in chronic limb-threatening ischemia, and Alabdullah et al.(48) synthesized the lived experience of phantom limb after lower extremity amputation; both matter because the consequences of the decision are measured in the patient's own terms. No published trial yet links a decision aid to decisional conflict, regret or satisfaction after major amputation. The endpoints that such a trial would need are already in use: the conflict and regret scales employed by Wang et al.(35) supply the primary outcome, and the discrete choice methods of O'Hara et al.(37) supply the preference weights against which the decision could be audited. The decision also does not end with the incision. Alabdullah et al.(48) synthesized the lived experience of phantom limb after lower extremity amputation, and Anderson et al.(43) showed that the shared decisions that follow amputation, about prosthesis and mobility, are as under-supported as the decision that preceded it. A patient who consents to amputation is therefore consenting to a rehabilitation that the system may not deliver, and consent that stops at the operation overstates what the patient has actually agreed to.

Who reaches the amputation decision, and who is offered salvage, is shaped by factors that ethics and shared decision-making do not control. Hamtzany et al.(49) reported ethnic disparities in mortality following hospitalization for diabetic foot ulcer, Eakers et al.(50) found that patient insurance status and race modify outcomes for diabetic patients undergoing care for foot ulcers and infections, and Zhang et al.(51) used a large electronic health record database to show how social determinants of health mark the diabetic population that receives major lower extremity amputation. Lee et al.(52) described income-based disparities in diabetic limb preservation in South Korea, Hart et al.(53) linked the social vulnerability index to outcomes in chronic limb-threatening ischemia, and Daly et al.(54) showed that marital status, a proxy for caregiving support, predicts clinical outcomes in the same population. The consistency of these gradients across countries and designs suggests that the amputation decision is partly determined before any consent conversation begins.

The meaning of the foot and of its loss is culturally loaded. Otene et al.(55) described the cultural and spiritual impact of diabetes-related lower limb amputation among Māori, for whom the feet carry a sacred connection that standard consent forms do not capture. Kalhan et al.(56) quantified the caregiver burden of diabetes-related foot complications in multi-ethnic Asian families, and Turner et al.(57) reviewed multidisciplinary management of diabetic foot ulcers impacted by social determinants of health. These accounts converge on the same point: a shared decision is impossible when the decision is framed only in biomedical terms, because the values that should be elicited are social, spiritual and relational before they are clinical. The implication for consent is direct. A process can be formally valid and substantively empty when the risks it lists omit the loss of a spiritual relationship to the ground or the transfer of a caregiving burden, and both omissions appear in the accounts collected here.

The health system sets the outer boundary of what any conversation can achieve. Al Ghafri et al.(58) used a COM-B framework to describe how healthcare professionals in Oman perceive the management of diabetic foot disease and where their capability, opportunity and motivation fall short of what shared decision-making requires. Chuter et al.(5) documented the global declines in amputation alongside local inequities, Roberts et al.(59) described the structure and initial outcomes of a multi-specialist limb preservation clinic, and Nomali et al.(60) compared the 2019 and 2023 versions of the IWGDF guidelines to show how rapidly the normative frame is moving. Chuter et al.(5) tied the clinical axis to the equity axis at the level of the health system by showing that declining amputation rates have not reached all regions equally, so the decision is constrained before it begins. The consequence for practice is that equity cannot be added after the conversation. If the gradients of Hamtzany et al.(49) and Hart et al.(53) are reproduced in part because some patients arrive at the decision already disadvantaged, then the remedy lies upstream, in access to foot screening, podiatric care and revascularization, and downstream, in auditing each service's amputation decisions against the social profile of the patients who receive them. What none of these studies does is connect the documented disparity to the documented decision gap: the gradients in who is amputated are measured in one literature, and the failures of shared decision-making are measured in another, with almost no study that joins them.

Taken together, the four axes describe a decision that is prognostic in its inputs, ethical in its conflicts, preference-sensitive in its resolution and inequitable in its distribution. The synthesis is condensed in Table 1 (Table 1), which lists for each axis the key findings, the representative studies and the gap that remains.