Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix A Power Calculations The Columbia University researchers have performed power calcula- tions to estimate the number of subjects that would be needed to conduct an informative study of health outcomes in US veterans of Vietnam. Such calculations comprise many variables, including the incidence rate of the disease being studied, the age of the subjects, the proportion of exposed subjects in the cohort, the statistical confidence levels considered accept- able, and the magnitude of the difference in incidence rates that could be detected. Illustrative calculations were made for non-Hodgkin's lymphoma (NHL).1 The sample sizes were estimated on the basis of a 1-tailed chi- square test with alpha = 0.05 and beta = 0.80, an unequal sample size for the exposed and comparison groups (1:4 ratio), and a 587.75/100,000 cu- mulative NHL incidence in the comparison group. They are given below in terms of the relative risk (RR, also known as the risk ratio) that could be detected. 1 The committees responsible for the Veterans and Agent Orange series of reports have found sufficient evidence of an association between exposure to the herbicides used in Vietnam and NHL. 17
18 EXPOSURE OF VETERANS TO AGENT ORANGE AND OTHER HERBICIDES RR Number Number Sample exposed comparison size needed 1.25 23,310 93,240 116,550 1.50 6,356 25,422 31,778 1.75 3,057 12,227 15,283 2.00 1,849 7,395 9,243 2.50 936 3,742 4,677 3.00 590 2,359 2,948 4.00 319 1,273 1,591 5.00 211 842 1,052 All else equal, diseases that are more common than NHL among individuals in the same age bracket as US veterans of Vietnam would require smaller sample sizes to detect the same RR; less common diseases would require larger sample sizes. Using a larger percentage of exposed individuals would also decrease the sample size needed to detect the same RR.