Reliability of environmental and occupational exposure data provided by surrogate respondents in a case-control study...

This study used data provided by 40 non-demented Parkinson's disease patients and 101 community controls, and by their 110 spouses and 31 adult children to assess the reliability of surrogate-provided rural environmental and occupational exposure information on the index subjects. The level of overall raw agreement between the index subjects and the spouse or adult child surrogates varied from 50.0 to 100.0% for the case-surrogate group and from 80.6 to 96.0% for the control-surrogate group. We did not detect significant differences in overall raw agreement between the case-surrogate and control-surrogate groups or between the spouse-surrogate and adult child-surrogate groups, for any of the variables studied. Considering all index subjects and their surrogates, the level of overall raw agreement was 80.3% for well water consumption, 82.3% for farm living, 85.8% for agricultural work, 87.1% for use of pesticides, 87.9% for field crop farming and 91.9% for use of fertilizers. However, the kappa estimates were lower, varying from 0.48 (SE = 0.20) for fertilizer use to 0.66 (SE = 0.11) for crop farming. The level of specific agreement was 52.2% for fertilizer use, 64.0% for pesticide use, 71.4% for agricultural work, 73.9% for crop farming, 80.9% for farm living, and 83.6% for well water consumption. The overall findings of this study support the use, if necessary, of spouses and adult children of index subjects as surrogate respondents in case-control studies of rural environment and occupational exposures and Parkinson's disease and, possibly, other neurologic diseases. Specific agreement seems to be a better index of reliability than overall agreement in studies where exposure is rare.