[Validity of occupational histories from proxy respondents]

Occupational studies often use data obtained from proxy respondents. However, few investigations have been conducted on the validity of occupational data provided by workers' surrogates. This study aimed to compare self-reported occupational data to information obtained from next-of-kin, as a contribution to assess the validity of using proxy respondents to obtain information about workers.

Worker/next-of-kin pairs, residents in Southeastern Brazil, were interviewed separately in 1998 about worker's occupational past history. The concordance, sensitivity, and specificity of proxy reports about workers' occupations were examined comparing to self-reports.

A total of 2.163 worker/next-of-kin pairs were interviewed. The Kappa statistic for the agreement between worker's and next-of-kin's report about the worker's main occupation was 0.86 (CI95%; 0.85 - 0.88). The sensitivity of proxy reports on occupations ranged from 77.5% (64.6% - 90.4%) to 98.9% (97.3% - 100.0%), and specificity ranged from 96. 9% (96.0% - 97.7%) to 99.9% (99.7% - 100.0%).

These are encouraging findings for the use of proxy respondents in occupational studies when occupations are considered as categorical variables. However, caution is required when assessing occupational exposition by means of cumulative work time because next-of-kin may underestimates this information.