Some considerations in the design of a nutrition intervention trial in Linxian, People's Republic of China.

Various issues pertaining to the design of a nutrition intervention trial in Linxian, a county in North Central China where esophageal cancer rates are extraordinarily high among a population with chronic deficiencies of multiple nutrients, are addressed. Two target populations are identified: One is a group of patients diagnosed with severe esophageal dysplasia, the other is the general population of the high-risk area. Interventions involving the supplementation of each group with vitamins and minerals are described, and a rationale is provided for the selection of nutrients and choice of dose levels. Several potential statistical designs are evaluated, with particular emphasis on the balancing of gains in specificity afforded by full and fractional factorial designs against their potential loss in sensitivity compared with the simplest design involving a placebo versus a multiple vitamin and mineral supplement.