Abstract
The issue of whether existing computer programs can accurately allocate (stratify) person years of exposure to the appropriate categories of time related variables in epidemiological studies was discussed in a letter. A previously published article asserting that most computer programs for stratifying person years of exposure do so only approximately and that such programs cannot be easily run on a personal computer, was criticized. The NIOSH Life Table Analysis System (LTAS) which stratifies each person day of exposure to the time related variables, age, calendar time, duration of exposure, and time since first exposure, was cited as an example of a computer program that stratifies person years of exposure exactly. NIOSH was in the process of converting the LTAS program for use on a personal computer. Initial test runs have shown that a database of 2,500 workers containing 14,000 job histories, 500 deaths, and 67,000 person years of exposure can be run on a personal computer in about 1 minute. A response by the author of the original study was presented. The author asserted that he did not intend to imply that all computer programs that use approximate methods to stratify person years always produce inaccurate results. His article was intended to show that exact methods generally produce more accurate results than approximate methods.
