Reconstructing past occupational exposures : how reliable are women's reports of their partner's occupation?

Article English OPEN
Tagiyeva, Nara ; Semple, Sean ; Devereux, Graham ; Sherriff, Andrea ; Henderson, John ; Elias, Peter ; Ayres, J. G. (2011)

Objectives Most of the evidence on agreement between\ud self- and proxy-reported occupational data comes from\ud interview-based studies. The authors aimed to examine\ud agreement between women’s reports of their partner’s\ud occupation and their partner’s own description using\ud questionnaire-based data collected as a part of the\ud prospective, population-based Avon Longitudinal Study\ud of Parents and Children.\ud Methods Information on present occupation was\ud self-reported by women’s partners and proxy-reported by\ud women through questionnaires administered at 8 and\ud 21 months after the birth of a child. Job titles were\ud coded to the Standard Occupational Classification\ud (SOC2000) using software developed by the University of\ud Warwick (Computer-Assisted Structured Coding Tool).\ud The accuracy of proxy-report was expressed as\ud percentage agreement and kappa coefficients for\ud four-, three- and two-digit SOC2000 codes obtained in\ud automatic and semiautomatic (manually improved)\ud coding modes. Data from 6016 couples at 8 months and\ud 5232 couples at 21 months postnatally were included in\ud the analyses.\ud Results The agreement between men’s self-reported\ud occupation and women’s report of their partner’s\ud occupation in fully automatic coding mode at four-, threeand\ud two-digit code level was 65%, 71% and 77% at\ud 8 months and 68%, 73% and 76% at 21 months. The\ud accuracy of agreement was slightly improved by\ud semiautomatic coding of occupations: 73%/73%, 78%/\ud 77% and 83%/80% at 8/21 months respectively. While this\ud suggests that women’s description of their partners’\ud occupation can be used as a valuable tool in\ud epidemiological research where data from partners are not\ud available, this study revealed no agreement between these\ud young women and their partners at the two-digit level of\ud SOC2000 coding in approximately one in five cases.\ud Conclusion Proxy reporting of occupation introduces\ud a statistically significant degree of error in classification.\ud The effects of occupational misclassification by proxy\ud reporting in retrospective occupational epidemiological\ud studies based on questionnaire data should be\ud considered.
Share - Bookmark