Schreinemachers, DM. 2003. Birth Malformations and Other Adverse Perinatal Outcomes in Four U.S. Wheat - Producing States. Environmental Health Perspectives 111:1259-1264.

In a study comparing the rates of birth defects counties where wheat is grown abundantly to places where it is not, Schreinemachers concludes that herbicides used on wheat may be causing birth defects. Babies born in wheat growing areas of the US west are more likely to have several types of birth defects than babies born in the same region but where wheat is less common.

While the study falls short of proving causation, it is consistent with earlier research on animals and with people indicating that a family of herbicides commonly used on wheat, the chlorophenoxy herbicides such as 2,4-D, disrupt fetal development.

What did she do? Schreinemachers analyzed the incidence of birth defects in 147 rural counties in 4 states, comparing rural regions of high wheat production with rural regions of low wheat growing. Two chlorophenoxy herbicides, 2-4,D and MCPA, are used heavily on wheat, much less so on crops like soy that dominate the counties where wheat is less abundant. Hence by comparing high wheat and low wheat counties, Schreinemachers sets up a comparison of high chlorophenoxy exposure vs. low.

The study focused only on Caucasian singleton (not twin) children born to mothers at least 18 yrs old. Data for the comparisons were obtained from health records maintained in the National Center for Health Statistics. In all, records for 43,634 births in the counties of interest from 1995 to 1997 were analyzed.

What did she find? Birth outcomes differed significantly in several ways:

• Birth defects of the respiratory and circulatory systems were greater in high wheat areas, as were defects of the musculo-skeletal system like clubfoot, fused digits and extra digits. These birth defects were 60% to 90% more likely in high wheat counties. The odds-ratios for these increases were all significant.
• The chances of birth defects rose for babies conceived in the spring, when herbicide spraying was most intense. Boys born in April-May in high wheat growing counties were almost 5x more likely to have a birth defect than boys born in low wheat counties at other times of the year.
• Infant death due to congenital abnormalities was higher in boys born in wheat growing counties compared to boys in low wheat counties. Most of the infant boys' deaths were caused by heart and musculoskeletal birth defects. No comparable elevation in risk of infant death was seen for girls.

What does it mean? Epidemiologists in general agree that 'ecologic' studies like Schreinemachers--examining geographic patterns to highlight statistical associations--stop short of demonstrating causation. Instead they are one piece of evidence that can be considered along with parallel studies using other techniques, all approaching the question of human impact from different directions. Indirect work like this is necessary because purposeful experimentation on humans to determine, for example, whether an herbicide causes birth defects, would be unethical.

Schreinemachers work neatly sidesteps a common weakness of ecologic studies of the effects of agricultural pesticides by focusing all of her analysis on rural communities. Comparisons are often made between rural and urban residents; these studies unavoidably add a raft of confounding variables due to the differences in urban vs rural environments, health care, etc.

But by comparing people living in different rural counties, varying in the intensity of wheat growing, Schreinmacher doesn't avoid all confounding variables. Two stand out: first, some wheat is grown in the low wheat counties, and thus some, albeit reduced, exposure to chlorophenoxy herbicides will take place. Second, other types of herbicides can be used heavily on the crops grown instead of wheat, for example, atrazine on corn.

Neither of these considerations would bias Schreinemachers' analysis toward showing a link between chlorophenoxy herbicides and birth defects. If anything, they would have decreased her chances of finding a positive result, by increasing the numbers of birth defects in low wheat counties. The ideal comparison would have been rural counties with no agriculture vs. high wheat, but rural counties with no agriculture within the region are rare at best.

Hence the fact that Schreinemachers found heightened risk of birth defects for several comparisons needs to be taken seriously, even more so because the weight of evidence from other studies strongly links chlorophenoxy herbicides with birth defects. Notable amidst that earlier work are studies showing higher birth defect rates in Minnesota, New Zealand and Norway associated with chlorophenoxy exposures, and experimental studies in showing that chlorophenoxy herbicides cause birth defects in rodents.

References

Hanify JA, Metcalf P, Nobbs CL, Worsley KJ. 1981. Aerial spraying of 2,4,5-T and human birth malformations: an epidemiological investigation. Science 212:349–351.

Kristensen P, Irgens LM, Andersen A, Bye AS, Sundheim L. 1997. Birth defects among offspring of Norwegian farmers, 1967–1991. Epidemiology 8:537–544.

Schardein JL. 1993. Chemically Induced Birth Defects. 2nd ed.New York:Marcel Dekker, Inc.