| published 06-10-2007
|
Views : 454 |
Times marked as favorite : 53 |
This editorial review looks at the evidence that zinc reduces child mortality, including two recent articles in The Lancet. Both studies showed a reduction in mortality, at least in children older than one year, but the reductions were "statistically non-significant." The article makes the extremely important, often-overlooked point, though, that "In studies with results that do not reach statistical significance, if the upper boundary of the confidence interval includes an important benefit, the possibility that the treatment still might be worthwhile has not been ruled out." In plain English, the failure to find a "statistically significant effect" does not necessarily rule out a real, clinically important effect.
For example, in the Zanzibar study quoted, there was an 18% reduction in child mortality in the zinc supplemented group. This was labeled a "marginally significant" reduction (relative risk 0·82, 95% CI 0·68–1·00) because its statistical p value was 0.05 or, in confidence interval terms, the 95% confidence interval includes the ratio 1.0, i.e. no effect. However, the confidence interval includes the whole range from 0.68 through 1.0. This means that even if the true effect of zinc in this study were a 32% reduction of mortality, we could expect to get the observed results in at least 5% of similar studies. We have not ruled out an effect even as large as a one-third reduction in mortality, let alone a reduction of smaller magnitude.
The article continues to discuss the difficulties of finding more precise answers, mainly (1) the large number of subjects needed in a study looking at a moderate effect on an uncommon event (death) and (2) the ethical issues in doing more placebo-controlled trials of zinc.
Effect of zinc supplementation on child mortality. The Lancet 2007; 370:1194-1195
DOI:10.1016/S0140-6736(07)61524-4. 6 Oct 2007.
|
|
|
Readers' Comments  |
|
Average user rating
 (0 vote)
|
|
Add your comment
|
