Statistical Design Aspects of the Ntp/Hei Collaborative Study on the Health Effects of Chronic Ozone Inhalation

Abstract

The purpose of the NTP/HEI Collaborative Study was to assess exposure- and concentration-related health effects associated with chronic exposure to ozone. Data were obtained from 164 animals specially dedicated to HEI from a standard ozone inhalation study conducted by Battelle Pacific Northwest Laboratories for the National Toxicology Program. The study involved a number of investigators, each interested in assessing a different type of ozone-related health effect, including respiratory function, as well as structural, cellular and biochemical changes in the nose, lungs, and airways. Designing and analyzing a study with multiple investigators raises many statistical challenges. The highest design priority was that each investigator's data be individually interpretable as an independent study. This means that each investigator had to be assigned an adequate number of animals, balanced with respect to concentration level and other factors such as gender and time of sacrifice. An additional feature of the collaborative study was the opportunity it provided to assess and quantify the effect of ozone exposure on a broad spectrum of outcomes, and to explore the relationship between the different types of effect. For example, the data allowed an assessment of whether the animals with the greatest degree of structural damage were also the ones with altered biochemistry. Maximizing the potential to assess these types of correlations required that investigators overlap as much as possible on measurements in individual animals. This aspect of the statistical design requires careful consideration of the compatibility between various investigators. Fortunately, the degree of compatibility was high. In most cases, for example, it was possible to assess respiratory function in the animals prior to their sacrifice, and then to divide the tissue between several different investigators. This article focuses on the statistical design of the collaborative project. Brief descriptions of each component investigation are given along with a discussion of logistical and other practical constraints. The allocation design of matching investigator groups to animal blocks is then presented, followed by a discussion of the impact of sample loss. The article concludes with a discussion of the features of the design that could be applied to future studies.