Estimating the Mean, Variance, and Confidence Limits from Censored (Limit of Detection), Lognormally-Distributed Exposure Data

Academic Article

Abstract

  • A published statistical technique is applied to an exposure distribution having measurements that are less than the analytical limit of detection (LOD). The method uses the mean and standard deviation of known or quantitated values and the number of censored or nonquantitated (< LOD) values. These data are used to enter a table, which is presented. A parameter from the table is then used to calculate the overall mean and standard deviation of the censored distribution. In the case of a lognormal distribution, the logarithms of the exposure values are used. The technique is applied to 268 measurements of exposure to asbestos from an office building. Approximately 70% of this distribution was censored (< LOD). The results are compared to the more common technique of assigning the value of 0.5 LOD to all measurements < LOD, and the technique is also applied to the data under the assumption that they are normally distributed (a common, often inappropriate assumption). Lastly, another published technique is used to calculate confidence limits for the mean of a lognormal distribution. These two statistical techniques have not been previously applied to published industrial hygiene data. © 1990, Taylor & Francis Group, LLC. All rights reserved.
  • Authors

    Digital Object Identifier (doi)

    Author List

  • Perkins JL; Cutter GN; Cleveland MS
  • Start Page

  • 416
  • End Page

  • 419
  • Volume

  • 51
  • Issue

  • 8