Assessing the implications of contradictory farm exposure data in an aging rural population.

Occupational exposure to agriculture can be measured in a variety of ways, including a simple yes/no classification, as well as continuous measures, such as years in farming. Regardless of the exposure measure, the possibility of mismeasurement exists. This can be the result of memory decay, differential interpretation, and overly broad definitions of farming. Data from a 20-year longitudinal study were used to measure the extent of this misclassification, and also to compare results obtained from yes/no measures (ever having farmed and current farmer) versus a more refined measure (years of farming). There were 1,811 households in a rural seven-county region of central New York responding to the survey. All subjects were ≥50 as of 2009. The subjects' 2009 responses relating to agricultural exposure were validated against those given in 1989 and 1999. Subjects giving contradictory responses were contacted for clarification. Of the 445 subjects indicating any kind of agricultural exposure, 186 (41.8%) provided contradictory information. Correction of these led to a lower estimate of farming prevalence, but an increase in the mean years of farming. Contradictory responses were more common for females, subjects with partners, and those with better financial situations. For both the corrected and uncorrected data, ever farming, current farming, and years of farming were all strongly related to both farmer's lung and farm machinery injury. Contradictory information is commonly encountered when measuring agricultural exposure. However, when studying measures that are strongly related to agriculture, the extra effort devoted to clarifying these contradictions may be unnecessary. For outcomes where these relationships are more subtle, correction may be more important.