California surveillance for pesticide-related illness and injury: coverage, bias, and limitations.

The California Pesticide Illness Surveillance Program (PISP) is a major resource for pesticide illness epidemiology. This work attempts to improve characterization of pesticide illness in California, evaluate case ascertainment of the PISP and identify PISP's limitations and biases for studying the incidence and epidemiology of pesticide-related illness. Abstractors collected standardized information from 1994-1996 PISP files, poison control logs, hospital records, and death certificates. Linkage produced a merged file of all records that documented consideration of pesticide exposure as a cause of ill health. We identified 23 deaths and estimated 1,310 hospitalizations attributable to pesticide exposure, corresponding to 0.024 fatalities and 1.38 hospitalizations (95% confidence interval [CI] = 1.01-1.74) per 100,000 California population person-years. Most intoxications followed domestic exposures and were not reported to the surveillance program, which identified primarily occupational exposures. Excluding groups of five or more, we estimated 16% ascertainment (CI = 0.05-0.27) for non-agricultural occupational cases. Agricultural connections increased the probability of ascertainment to about 50%, cultural and economic barriers notwithstanding. Surveillance records existed for all identified episodes in which five or more people were exposed. California pesticide surveillance data appear complete for mass exposures, and adequate to characterize agricultural and occupational exposures. Work is needed to explore domestic exposures and occurrences beyond the reach of the health care system.