Pesticides hit home: rating the risks for kids in California. (Science Selections).

Pesticides have long been suspected of causing childhood cancer, but establishing a cause-and-effect relationship is difficult. Children are exposed to unknown varieties and quantities of pesticides. Parents seldom know which compounds are being used in nearby agricultural fields. Epidemiologic studies often suffer from case-response bias: parents of sick children are more likely to remember using pesticides. And when cancer clusters are investigated, the cases are often too few to prove an association with pesticides.

Robert Gunier and colleagues at the California Department of Health Services describe a methodology to address these limitations [EHP 109:1071-1078]. The methodology compares pesticide use to location, and weights both hazard and usage to give an effective measure of the actual danger of the chemicals. Using data from California's Pesticide Use Report, a mandatory statewide pesticide-application reporting system begun in 1990, they determined which census block groups (subdivisions of census tracts) had received which pesticides from 1991 to 1994. To increase statistical power, they focused on the 38 most used insecticides, fungicides, and herbicides in the state and grouped the pesticides in four categories: genotoxicants, reproductive and developmental toxicants, probable carcinogens, and possible carcinogens (California has banned known human carcinogens from agricultural use). More than 36 million pounds of genotoxic active ingredients were used in California during the average year.

Pesticides were assigned a numerical hazard factor, based on toxicity and exposure factors:

* cancer class (probable or possible human carcinogen), determined by the U.S. Environmental Protection Agency;

* cancer potency, or how much the incidence of cancer increases with dose (usually determined by laboratory animal studies);

* volatilization flux rate, or how rapidly the compound enters the atmosphere (important because inhalation is a major exposure route); and

* field half-life, or how long the compound lingers in the field.

Among probable human carcinogens, the hazard factor varied by a factor of seven or eight between the least and most hazardous compounds. Among reproductive and genotoxic compounds, the maximum disparity was a factor of four. The hazard factor was multiplied by usage rates to find "hazard-adjusted use"--the overall hazard level for each pesticide. So, for example, while 1.6 million pounds of the insecticide propargite were actually used in the state, the compound's hazard factor of 1.4 gave it a hazard-adjusted use rating of 2.2 million pounds. On the other hand, the 2.4 million pounds of chlorpyrifos actually used had a hazard factor of .096, giving it a hazard-adjusted use rating of only about 233,000 pounds.

Pesticide use also varied widely by location. Seventy-seven percent of census block groups received less than 1 pound of active ingredient per square mile. But 493 block groups (home to about 170,000 children) received more than 569 pounds per square mile.

The study could not evaluate some widely used pesticides for which toxicologic data are lacking, and California's reporting system ignores home and garden pesticides. However, the hazard-adjusted use rating could help solve an old question in epidemiology: What is the connection between childhood cancer and pesticides? The researchers plan to use the results to compare childhood cancer rates in census block groups with intense use of hazardous pesticides to block groups without such use.