Home
LogXact
|
|
Example 2
Adenocarcinoma Tumors in the Small Intestine
Two hundred female rats were treated with varying doses of Acid Red 114 and observed for the presence of adenocarcinoma tumors in the small intestine, as part of the U.S. National Toxicology Program of the National Institute of Environmental Health Sciences (NIEHS), (1991). There were four dose groups, 0 ppm, 150 ppm, 300 ppm and 600 ppm. At the 0 ppm dose, 0/50 tumors were discovered. At the 150 ppm dose, 0/35 tumors were discovered. At the 300 ppm dose, 1/65 tumors were discovered. At the 600 ppm dose, 2/50 tumors were discovered. The objective was to test for a dose-response trend. The usual trend test (Armitage, 1955) cannot be applied directly however because of the need to adjust for animals that die before they have had a chance to develop a tumor. For incidental tumors the NIEHS conducts an adjusted test for trend by including both the dose and the survival time of each animal as covariates in a logistic regression model. (The survival times for the 200 animals are not shown here due to lack of space, but they varied from 14 to 24 months.) After adjusting for survival, the asymptotic p-value (one sided) for the dose effect is 0.087. This estimate is unreliable however, for there were only three responses in the entire data set of 200 animals. The exact p-value is calculated by LogXact as 0.192, more than twice as large as the asymptotic estimate.
|
|
|
|
