CitationDeLong, G. Robert & Leslie, Paul W. (1997). Iodination of Irrigation Water and Infant Mortality: Reply. Lancet, 350(9089), 1482.
AbstractSir—With respect to C Ronsmans and T F de C Marshall, analysis of each township separately sacrifices power relative to the full model used in our analysis. The effect of iodination on neonatal and infant mortality in Tusala is highly significant, as is that effect when data from all townships are combined. Tests of the other two townships separately do not show a significant effect of treatment. For Long Ru, this is not surprising; with the smallest population and so few deaths in any year, even a large true effect of treatment would probably be swamped by stochastic effects. As we noted, the apparent effect of iodination in Bakechi is smaller than in Tusala; indeed, it is not statistically significant. A smaller effect would be expected there, since Bakechi is the least iodine deficient of the towns studied. There may be a real but more modest effect, but with a sample size too small to show this clearly. Clearly, variations in early mortality over time and among the townships involve more than iodine availability. As we noted, mortality in Bakechi was much higher than in the other towns, and that difference cannot be attributed to iodine deficiency. Other determinants of mortality may have changed over the period examined, and may have changed differentially in the communities. The secular decline in mortality in Bakechi seems somewhat steeper than in Tusala, and this may help to obscure, statistically, the effect of iodination.
Reference TypeJournal Article
Author(s)DeLong, G. Robert
Leslie, Paul W.