Risk Factors for Cholinesterase and Non-cholinesterase Effects of Exposure to Organophosphate Insecticides in California Agricultural Workers: 1982-1990

Prior efforts to evaluate the comparative risk of organophosphate (OP) compounds have used numerator illness data from California's Pesticide Illness Surveillance Program (PISP) with corresponding pesticide use information to calculate illness/pesticide use ratios (IPURs). We present here a means of evaluating the comparative risk of organophosphate (OP) compounds, employing a standard epidemiologic technique known as a case-control study to calculate odds ratios (OR) associated with individual OP compounds. The study population consisted of 396 cases of illness related to cholinesterase (ChE) inhibition among California agricultural workers and 758 comparison subjects derived from PISP. The study subjects were similar to the California agricultural population as a whole in terms of age, gender, and ethnicity and the 30 Standard Industrial Classification (SIC) codes represented by study subjects accounted for 95% of the state's agricultural employment. The number of study subjects for each of the 36 compounds represented among study subjects also showed a significant correlation with the corresponding number of reported pesticide applications (Rank correlation=0.76, p<0.001). The cases and comparison subjects differed chiefly in respect to the presence or absence of cholinesterase inhibition (definitely or probably present among the case group and definitely absent among the controls). Approximately 72% of the controls had nonspecific symptoms that could conceivably have been due to exposure to a cholinesterase inhibiting compound but had cholinesterase values within the normal population range. The remaining cases involved respiratory and ocular irritation, and some were demonstrably related to other specific medical diagnoses. The individual compound most frequently associated with exposure to both case and control subjects was mevinphos (158 cases [39.9%] and 337 controls [43.9%]. Other compounds accounting for 10 or more case subjects included oxydemetonmethyl, parathion, phosalone, dimethoate, methamidophos, diazinon, chlorpyrifos, azinphos-methyl, methidathion, demeton, and (as a co-exposure) the carbamate compound methomyl. The exposure factors identified as significant in the crude analysis included application work and field residue exposure. Exposure to multiple category 1 OPs and multiple ChE inhibitors, and several individual compounds proved significant risk factors in stratified analysis. These included phosalone, methomyl, oxydemeton-methyl, and mevinphos. For the application associated ChE-illnesses exposures to mevinphos (OR=6.2) and multiple ChE-inhibitors (OR=2.9) remained significant in the multivariate analysis. Based on the limitations of the cholinesterase assays without baseline values, some misclassification of illness among control subjects actually related to ChE inhibition was possible. The study was also limited by lack of a population-based control sample. However, neither random nor nonrandom bias was likely to have accounted for the increased risk associated with application work and exposure to mevinphos observed in the study. This study demonstrated that registry based casecomparison analysis can provide a useful alternative to the IPUR as a means of evaluating the comparative risk of OP compounds, allowing simultaneous examination of demographic factors, work tasks, and multiple insecticide exposures. Cases are the ChE-related illnesses resulting from OP exposures and controls the non-ChE illnesses. a These are taken to be representative of the general population exposures to OPs. ChE-Ill non-ChE ill Exposed a b Nonexposed c d The relative risk, or rate ratio (RR) for an exposure factor is computed as RR=[a/(a+b)]/[c/(c+d)], where b and d represent the entire non-ill population in the group of interest (e.g. applicators) and the comparison group, respectively. When the entire non-ill population cannot be determined, a proportionate morbidity ratio (PMR) can be computed using the same formula as the rate ratio. In this case, a and c represent the groups to which an event of interest has occurred (e.g. ChE-illness), and b and d represent the groups that experienced an event (e.g. non-ChE illness) but not the event of interest. Miettinen (Miettinen and Wang. An Alternative to the Proportionate Mortality Ratio. American Journal of Epidemiology 1981 114:144-148) observed that in this situation, the odds ratio (OR)=(ad/bc) is a more reasonable approximation to the rate ratio than the PMR.