As with many constructs in healthcare (e.g., ‘evidence-based medicine’, ‘health-related quality of life’, ‘decision aid’) ‘shared decision-making’ is formative not reflective, that is, ‘it’ has no existence prior to its definition and measurement. Any particular formative construct is preference-sensitive, being based on the preferences of those who form it by their indicator selection and weighting. These preferences often reflect interests of various sorts, some material, many not (at least not directly), but often ones aligned with particular beliefs, ideologies or ideals. So cave litteras maiusculas - sdm not SDM. Since ‘shared’ is an adjectival qualifier of ‘decision-making’, fundamental preferences relevant to decision-making are relevant in any construction of sdm. We highlight two major preferences in relation to health decisions. One is for provider-controlled, direct-to-patient intermediation (inter) as contrasted with provider-independent, direct-to-person apomediation (apo). The second is for verbal deliberative reasoning (vdr) as contrasted with numerical analytical calculation (nac). From their cross-tabulation we can see that, within both practice and research - and in legal standards and ethical guidelines for both - sdm is currently being constructed exclusively within the intermediative verbal deliberative reasoning (‘inter-vdr’) frame. We compare and contrast inter-vdr with the three other possibilities - ‘inter-nac’, ‘apo-vdr’ and ‘apo-nac’. Dismissal or disregard of the latter, especially the last, on the grounds of credibility and trustworthiness, needs to be challenged by preference-based comparative evaluations, using unbiased measurement of costs and effectiveness, in order to optimise the development and delivery of personalised support for health and healthcare decisions.