Article Text

PDF

0411 Separating within- and between-group exposure effects in a panel study on pesticide use and early biological effects in the Corn Farmers study
  1. Lützen Portengen1,
  2. Anneclaire J De Roos2,
  3. Laura Beane Freeman3,
  4. Roel Vermeulen1,4
  1. 1Institute for Risk Assessment Sciences, Utrecht University, Utrecht, The Netherlands
  2. 2Department of Environmental and Occupational Health, School of Public Health, Drexel University, Philadelphia, PA, USA
  3. 3Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD, USA
  4. 4Julius Center, Utrecht Medical Center, Utrecht, The Netherlands

Abstract

Objectives We aimed to estimate the effect of pesticides on selected early biological effects among farmers, allowing for different effects of within-person and between-group (unexposed controls versus farmers) changes over time. Using a group-level estimate of exposure is a well-known approach to reduce impact of measurement error on estimated exposure-response relations. With only few exposure groups this results in an ecological study design, with potential for “aggregation” bias. By group-mean centering of individually assigned exposures it is possible to separately estimate within-individual and between-group exposure effects

Method Pesticide exposure information, blood and urine were collected throughout a growing season from male corn farmers (n = 30), and non-farming controls (n = 10). We used a hierarchical mixed model to relate change in cumulative exposure to atrazine and 2.4-D from before to during the spraying season to plasma levels of 22 immuno-modulatory cytokines and compared this to a more conventional model using individual exposures only.

Results Model fit for group-mean centred models averaged better than for non-centred models (lower AICs for 20/22 models for atrazine and 18/22 models for 2.4-D). Estimates for between-group differences in exposure were very similar to those from conventional (non-centred) models, while standard errors for estimates based on within-individual differences were relatively large (5x those for estimates based on between-group differences).

Conclusions Group-mean centering of exposures allowed us to estimate and contrast exposure-response relations based on differences between groups and within individuals. Comparison with a more conventional approach, ignoring the clustering of individuals, showed that effect estimates were dominated by differences between groups.

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.