Association of Pesticide Exposure With Neurologic Dysfunction and Disease

Freya Kamel; Jane A. Hoppin


Environ Health Perspect. 2004;112(9) 

In This Article


Most studies of neurotoxicity have documented an increase in symptom prevalence and changes in neurobehavioral performance reflecting cognitive and psychomotor dysfunction, but many found little effect of pesticide exposure on sensory or motor function or direct measures of nerve function. There are several potential explanations for these findings. Except for vibrotactile sensitivity, information on sensory and motor function is limited, and further study may reveal associations with pesticide exposure. Another possibility is that the increase in symptom prevalence is due to bias: Most studies were cross-sectional in design, and individuals with greater exposure or a history of poisoning may have been more motivated to recall or report symptoms. Confounding by head injury or neurologic disease, either of which might be related to both pesticide exposure and increased symptom prevalence, could also create the appearance of an association. Consistency of findings across many studies argues against these explanations, as do the positive findings of some studies that used more quantitative exposure measures. Further, bias and confounding are less likely to account for changes in neurobehavioral performance, which is assessed using objective test batteries. Thus, moderate pesticide exposure may in fact have greater effects on symptom prevalence and neurobehavioral performance than on sensory or motor function. The lack of specificity of the symptomatic response is also interesting. It is possible that the earliest or most general response to pesticide neurotoxicity is a general malaise lacking in specificity and related to mild cognitive dysfunction, similar to that described for Gulf War syndrome (White et al. 2001).

Although the weight of the evidence suggests that pesticide use is associated with increased symptom prevalence and deficits in neurobehavioral performance, there were some inconsistencies that future studies should attempt to resolve. It may be that certain functional domains are more sensitive to pesticides than others, but the current literature is too limited to resolve this question. Some of the inconsistencies among studies are likely due to methodologic differences. A critical concern is exposure assessment. Qualitative and quantitative aspects of the exposure under consideration differed among studies, as did the ability of the studies to assess exposure. Exposure measures ranged from job title to detailed assessment of cumulative exposure based on work history. There was, however, no clear-cut relationship between the quality of exposure assessment and the results of the studies.

The choice of comparison group may also influence results. Responses to symptom questionnaires and neurobehavioral performance are influenced by age, education, and cultural background (Anger et al. 1997), so it is important for comparison groups to be demographically similar to exposed populations. However, using a comparison group from the same community or workplace as the exposed participants can create problems. Although the former may have no documented exposure, they may nevertheless not be truly unexposed, limiting the power of the study to detect effects. There may be no one best solution to this problem.

Other aspects of study design, such as size, neurologic end points considered, and data analytic strategies including control for confounding, are likely to influence results. More than half of the studies considered were small, with < 100 exposed participants, and therefore had limited power to detect associations. Poor response rates in some studies may have biased results. Symptom questionnaires, neurobehavioral test batteries, and other methods for evaluating neurologic outcomes also varied among studies. In particular, different neurobehavioral batteries employ different tests of cognitive and psychomotor function. However, results were variable even for tests used in many studies. Implementation of a given test may vary between batteries; for example, a computerized version may differ from a paper-and-pencil model, but even this consideration may not explain all differences. A study of styrene found that grouping results of neurobehavioral tests provided increased power to detect effects of exposure, compared to evaluating individual tests (Heyer et al 1996). Use of similar analytic strategies might reduce inconsistencies among studies of pesticides.

Pesticide exposure may be associated with increased risk of Parkinson disease. Inconsistencies among studies are again likely to be caused by variations in study methodology, particularly lack of detailed exposure assessment in some earlier studies. The positive results from recent studies with more comprehensive exposure assessment, together with support from animal models, reinforces the hypothesis of an association. Results for ALS and Alzheimer disease are suggestive but too sparse to support firm conclusions. Whether the subtle signs of neurotoxicity found in studies of poisoning and occupational exposure are related to the later development of neurodegenerative disease is a question not adequately addressed by the literature, although one study showed that short- and long-term responses to moderate exposure are not necessarily related (Stephens et al. 1996).

Historically, most studies have focused on OPs, first to document sequelae of acute poisoning and then to explore the effects of chronic moderate exposure. There is also evidence suggesting that other types of pesticides, including organochlorines, carbamates, fungicides, and fumigants, are neurotoxic. No study has evaluated the association of herbicides with symptom prevalence or neurobehavioral performance, but these chemicals have been implicated as risk factors for Parkinson disease. Although it is important to identify classes of pesticides and even specific chemicals associated with neurotoxicity, it is also important to recognize that most workers are exposed to complex mixtures of pesticides, which may contribute synergistically to neurotoxicity.

Other aspects of the relationship of pesticide exposure to neurotoxicity remain to be clarified. Participants in most studies have sustained both chronic and acute exposures; because these are often correlated, the studies have not been able to disentangle their effects. It is also possible that studies of chronic moderate exposure have been influenced by inclusion of individuals with a history of pesticide poisoning in the exposed population. Several studies in which such individuals were excluded found no relationship of chronic exposure to neurobehavioral performance or nerve function (Ames et al. 1995; Engel et al. 1998; Fiedler et al. 1997), but other studies of nonpoisoned individuals have found associations (Kamel et al. 2003; Stephens et al. 1995; van Wendel de Joode et al. 2001), suggesting that moderate as well as high-level pesticide exposure is neurotoxic. An issue receiving increasing attention is genetic susceptibility to pesticide neurotoxicity. In particular, genetic variation in paraoxonase has been related to OP neurotoxicity.

In conclusion, there is mounting evidence that chronic moderate pesticide exposure is neurotoxic and increases risk of Parkinson disease. To substantiate these findings, future studies must employ more detailed assessment of exposure in individuals and consider the role of genetic susceptibility. More studies of pesticides other than OPs and greater attention to disentangling the effects of different types of pesticides are also needed. Better information is required to clarify the relative importance of acute and chronic exposure and the role of moderate exposure in the absence of poisoning. Finally, it will be important to clarify the relationship of pesticide-related neurotoxicity to neurodegenerative disease.