Effect of preventive supplementation on young children in Niger.

摘要

In Reply: Drs Ebbeling and Ludwig clarify the nature of the attention and contact provided to participants in the NSB reduction group but not provided to their study's control participants. However, while they maintain this was an unlikely confounder, earlier findings documenting beneficial effects of mail and telephone contacts on weight control challenge this assertion. Only good experimental design holding therapeutic contact constant across experimental groups can resolve this issue. We accept the statement by Ebbeling and Ludwig that their subgroup analysis in children with higher baseline BMI was preplanned. However, we remain concerned about other aspects of the study design, including that BMI status was not described as an eligibility criterion, power calculations did not take this into consideration, and the high-BMI cutoff used in the preplanned analysis differed from that used in the stratified randomization, contrary to standard procedure. It is surprising that, given this subgroup finding in 2006, the corresponding analysis was not also preplanned, executed, and reported in a later study for which Ebbeling was corresponding author. For articles by other authors, we have similar questions about subgroup analyses when there were no statistically significant overall treatment effects, emphasizing the importance of authors making study protocols publicly available in clinical trials registries before commencing a study.