Inter-rater agreement of assessed prenatal maternal occupational exposures to lead

BACKGROUND: Industrial hygienists' assessments of prenatal occupational exposures based on parental job histories is a promising approach for population-based case-control studies of birth defects and other perinatal outcomes. However, evaluations of inter-rater agreement of such assessments have been limited. METHODS: We examined inter-rater agreement of occupational lead exposure assessments of maternal job reports by industrial hygienists in a population-based case-control study of parental occupational lead exposure and low birth weight. A total of 178 jobs with potential exposure to lead during the 6 months before pregnancy to the end of pregnancy were examined. Three industrial hygienists evaluated these jobs independently for exposure to lead including probability of exposure, type of exposure, route of entry, exposure frequency, duration, and intensity. Inter-rater agreement of these assessments beyond chance was evaluated using the kappa statistic (kappa). RESULTS: In general, inter-rater agreement was greater for assessment of direct exposures than assessment of indirect exposures. However, inter-rater agreement varied with the lead exposure metric under consideration, being: 1) fair to good for type of direct exposure (i.e., inorganic or organic), respiratory exposure and frequency of exposure to direct inorganic lead, hours per day of direct (i.e., inorganic or organic), and intensity of direct inorganic exposure; 2) poor for probability and type of indirect exposure (inorganic or organic); and 3) indeterminate for frequency of direct organic exposure, frequency of indirect exposures (organic or inorganic), and intensity of direct exposures (organic or inorganic). CONCLUSION: Retrospective assessment of maternal prenatal exposures to lead by industrial hygienists can provide some reliable metrics of exposure for studies of perinatal outcomes. Reliability studies of such exposure assessments may be useful for: quantifying the reliability of derived exposure metrics; identifying exposure metrics for exposure-outcome analyses; and determining the reliability of prenatal occupational exposures to other agents of interest. Birth Defects Research (Part A) 76:811-824, 2006. (C) 2006 Wiley-Liss, Inc.(dagger)