Patient level pooled analysis of 68 500 patients from seven major vitamin D fracture trials in US and EuropeBMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d5245 (Published 30 August 2011) Cite this as: BMJ 2011;343:d5245
The authors of this research paper—the DIPART (vitamin D Individual Patient Analysis of Randomized Trials) Group—have become aware of a coding error relating to treatment allocation in one of the studies (the Porthouse study) included in their pooled analysis (BMJ 2010;340:b5463, doi:10.1136/bmj.b5463). This error affects the data relating to studies of vitamin D plus calcium reported in table 2 and figure 4. The authors have recalculated their data to take account of the error. They state that in table 2 the hazard ratios (and confidence intervals) shown in the “vitamin D 20 µg plus calcium” column should be 0.99 (0.84 to 1.19) for any fracture, 1.19 (0.81 to 1.75) for hip fracture, and 0.99 (0.49 to 2.04) for vertebral fracture [not 0.95 (0.80 to 1.14), 1.30 (0.88 to 1.92), 0.97 (0.48 to 1.98) respectively, as was published]. In figure 4, the hazard ratio for hip fracture in CaD [calcium and vitamin D] trials should be 0.83 (0.69-0.99) [not 0.84 (0.70 to 1.01) as published], which is now statistically significant, and the hazard ratio for all trials should be 0.96 (0.85-1.09) [not 0.97 (0.86 to 1.10)]. The authors state that the error led to a modest underestimation in their paper of the reduction in hip fracture risk for vitamin D supplements given with calcium, saying that “the results have been strengthened slightly by correcting this error and the reduction in hip fracture risk is now statistically significant.” After correction, the absolute hip fracture risk reduction remained at 0.4% for participants aged over 70 and 0.2% in participants with previous fractures, but the numbers needed to treat (NNT) are now 250 and 524 (not 255 and 548). The NNT for all fractures was correct as reported. For more information on this, see the authors’ Rapid Response (www.bmj.com/content/340/bmj.b5463.full/reply#bmj_el_268507).
Cite this as: BMJ 2011;343:d5245