Objectives: To increase the credibility of retrospective exposure assessments. documentation of the procedures and presentation of measures on quality control is recommended. The aim of this study was to present and evaluate the procedures used in a case-control study on leukaemia.
Methods: A series of 1,087 cases and matched controls were interviewed. Exposure assessments were performed for 13 occupational agents and ten leisure-time activities and the confidence of the assessments was coded. The exposure assessments were performed blind by three occupational hygienists. Ten percent of the interviews were reassessed by two of the three occupational hygienists.
Results: The leisure-time activities contributed to a large extent to the overall prevalence of exposure. For organic solvents approximately 25% of the controls classified as exposed would be misclassified if leisure-time exposure were not considered. The proportions of subject assessments with low confidence were higher for next-of-kin than for in-person interviews. A negative correlation was seen between the proportions of assessments with low confidence and the reliability. A significant difference was seen in the inter-rater comparison between cases and controls when the reliability was calculated for each assessed period; no difference was seen for the subject assessments used for relative risk estimation.
Conclusion: When low-dose exposure in epidemiological studies are being assessed there is an obvious risk of misclassification if leisure-time activities are not included. Furthermore, the reliability of the assessments may suffer if next-of-kin interviews are used to a large extent. For cancers with poor prognoses, prospective studies are preferable to minimise possible information bias.