Statistical validation of a commonly used method for personnel dosimetry issuance determinations.

摘要

In 1992, the federal requirements for personnel monitoring of occupational radiation doses were modified, making monitoring compulsory only for individuals "likely" to exceed 10% of the applicable annual dose limits. This regulatory change served as a catalyst for many radiation protection programs to re-evaluate their personnel monitoring issuance criteria to determine if some monitoring services might be reduced or eliminated. But in the absence of any clear definition of the term "likely," radiation protection programs relied on adjustments largely based on reviews of previous monitoring results and professional judgment. Although such semi-quantitative assessments may have been appropriate, the approach was not without inherent programmatic risks, as radiation dosimetry programs have been shown to be a consistent source of regulatory non-compliance upon inspection and serve as essential elements in defense against litigation efforts purporting workplace-related radiation injury. The objective of this study was to subject this commonly used method for post-regulatory change dosimetry issuance determinations to statistical validation. Personnel monitoring data from The University of Texas Health Science Center at Houston Radiation Safety Program pre- and post-1992 regulatory changes were accessed and descriptively analyzed. The dynamic nature of the institutional environment made direct comparisons of individuals between years impractical, so 1990 mean dose levels for various independent variables were compared using the Kruskal-Wallis non-parametric ANOVA test at a significance level of alpha = 0.05. The data from 1998 were then analyzed to ascertain if the groupings identified as statistically different in 1990 were still being monitored. The analyses statistically validated the monitoring program reductions made by the University of Texas Health Science Center at Houston radiation protection program. In addition, statistical support for further monitoring program reductions were identified, if such reductions were deemed to be necessary in the future.