Authors' reply

摘要

We agree with Nicholas Harvey and colleagues that multivariable regression analyses cannot determine causality. However, we have clearly shown that the previous ultraviolet B (an ecological proxy for vitamin D) association with bone-mineral content in ALSPAC was confounded. Notably, in the absence of age or any other confounder adjustment, there was no association of maternal circulating 25(OH)D (the real exposure) with offspring bone-mineral content. Based on regression analyses in a much smaller study than ours, Harvey and colleagues concluded that vitamin D supplementation of pregnant women, especially during winter months, could lead to longlasting reductions in the risk of osteoporotic fracture in their offspring, and this has affected policy.