There is no standard method for measuring multimorbidity. Choices that have to be made include the type of
dataset used, the morbidities to include, the definition of each morbidity in the dataset chosen, and whether or
not to weight the final multimorbidity measure in any way. The choices made should be driven by the purpose
of measurement, but will inevitably be at least partly subjective and partly pragmatic (for example, because of
limitations of the dataset chosen for measurement). All multimorbidity measures are therefore contestable, but
the choices made should be as explicit as possible.w1 There is only space to discuss this briefly in the main paper
methods, so this supplementary file documents in detail the rationale and limitations of the choices made in our
study