2. Quality assessment items and possible scores.

Items	Scores
Item B Were the outcomes of participants who withdrew or were excluded after allocation described and included in an "intention‐to‐treat" analysis?	Score 2 if adequate detail of withdrawals and exclusions after randomisation exists, and an intention‐to‐treat analysis has been, or can be carried out Score 1 if number and reasons for withdrawal are mentioned but intention to treat analysis is not possible Score 0 if inadequate detail exists to allow the author to check or carry out an intention to treat analysis, or obvious differences with no adjustment
Item C Were the outcome assessors blind to assignment status?	Score 2 if blinding of all possible outcome assessors is clearly established Score 1 if there is a small or moderate chance of unblinding of assessors, or some but not other assessors who could have been blinded were blinded Score 0 if no attempt to blind assessors to the assignment of treatment is reported
Item D Were the treatment and control group comparable at entry?	Score 2 if groups are demonstrably comparable in respect of potential confounding factors on inspection of the characteristics on entry (means with some expression of the variation e.g. SD, SE, confidence intervals are required), or differences between groups adjusted for in the analysis (stratification, Mantel‐Haenszel technique, logistic regression, multiple regression, multivariable techniques) Score 1 if confounding appears small: although noted, adjustment has not been made Score 0 if description of the treatment groups at baseline, either in text or table, is inadequate to confirm comparability for all plausibly important confounders, or statistically significant differences between the groups are present but no adjustment has been made in the analysis
Item E Were the subjects blind to assignment status following allocation?	Score 2 if effective action has been taken to blind participants to assignment Score 1 if in a drug study, or in a study comparing a physical modality with a control, it is unclear whether participants were made aware, or could have become aware, of their assignment prior to measurement of outcomes, or the nature of the trial intervention is such that it is unlikely that they will have effects which allow identification of assignment (e.g. calcium supplements versus placebo) Score 0 if in a drug study, no treatment rather than a placebo is used, or in a placebo‐controlled drug study or in a study of comparable physical modalities, participants became aware of their allocation before outcome assessment and analysis
Item F Were the providers of care blind to assignment status?	Score 2 if the study is clearly double or triple blind Score 1 if it is unclear whether the treatment providers were blinded to the allocation Score 0 if in a placebo controlled drug trial, the providers of care were informed of the treatment allocation before outcome assessment and analysis, or a physical modality was used in one or more arms of the trial
Item G Were the care programmes, other than the trial options, identical?	Score 2 if it is clear that the care programmes other than the trial interventions were identical Score 1 if differences between the programmes are trivial Score 0 if the nature of the care programmes other than the trial interventions is unclear, or there are important differences between the programmes offered, other than the trial interventions
Item H Were the inclusion and exclusion criteria for entry clearly defined?	Score 2 if the inclusion and exclusion criteria are clearly defined and indicate that individuals currently exposed to a trial intervention were excluded e.g. vitamin D analogue, hormone replacement therapy Score 1 if the inclusion and exclusion criteria as described allow the possibility that individuals may have entered the study currently exposed to a trial intervention, or description of the inclusion and exclusion criteria is inadequate to determine how the sample was made up Score 0 if no description, other than age and gender, of inclusion and exclusion criteria was provided
Item J Was the ascertainment of fractures and other outcomes active and of clinically appropriate duration?	Score 2 if some form of concurrent collection of data about fracture e.g. subjects given postcards to mail back etc., with confirmation by interview, and by radiograph if positive, or, for vertebral fracture, routine confirmation by radiograph Score 1 if contact was made on a regular basis e.g. 6‐monthly phone call to establish if fracture had occurred or not, with confirmation by radiograph if positive Score 0 if fracture was registered as an outcome without confirmation by radiograph

Item A, which considered allocation concealment, was removed as this now forms part of the risk of bias assessment