The factorial design of the RECORD trial attempted to determine the relative contributions of calcium versus vitamin D on fractures. In this secondary prevention trial, 5292 ambulatory patients were randomised to receive calcium alone (1000 mg daily), vitamin D3 (800 IU daily), a combination of the two, or placebo.

After follow-up of at least 24 months, there were no significant differences in fracture rates between the four groups. Interpretation of the study is limited by two main factors. First, compliance with medication was only moderate. It declined to 63% after 2 years and might have been as low as 45% when non-responders to the questionnaire about compliance were included. On this basis it is difficult to see how the authors can be sure there was "no evidence that true differences may have been obscured by poor compliance".

Although the analysis was appropriate by intention-to-treat, it is possible to correct for the effect of non-compliance, which will dilute any physiologically achievable treatment effect in inverse proportion to the degree of compliance and so widen the confidence intervals.4 Second, the study perpetuates the limitations of most previous studies by measuring 25-hydroxyvitamin D in only a small sample (n=60-ie, just 1-1% of the study population). Thus the vitamin D status of the trial population at baseline remains largely unknown, although, because the patients were younger than in other studies, ambulatory, and living in the community, they were less likely to have vitamin D deficiency.

There have been two subsequent studies of vitamin D in patients living at home. Trivedi et al5 reported a significant reduction in fractures after treatment with 100 000 IU cholecalciferol every 4 months compared with placebo in a randomised trial of 2686 patients over 5 years. Serum 25-hydroxyvitamin D was not assessed at baseline but was measured after 4 years of the trial in a subgroup (n=253) and was 53-3 nmol/L in the placebo group.