Ma, X, PA Buffler, RB Gunier, G Dahl, MT Smith, K Reinier and P Reynolds. 2002. Critical Windows of Exposure to Household Pesticides and Risk of Childhood Leukemia. Environmental Health Perspectives 110:955-960.

Ma et al. present results demonstrating a strong relationship between exposure to indoor pesticides and the risk of childhood leukemia. The risk of childhood leukemia is higher with greater exposures, and it is also highest for exposures that occur during pregnancy, compared to exposures before pregnancy and after birth. These results should caution families to be extremely careful about using indoor pesticides, especially during childbearing years.

This study is not the first to find a link between childhood leukemia and in utero pesticide exposure. In fact, most studies of this sort find the same basic result. Ma et al. avoid some of the weaknesses of earlier work, however, making their results that much more compelling.

What did they do? This case-control study is one of a series of investigations carried out by The Northern California Childhood Leukemia Study (NCCLS).

Between 1995 and 1999, NCCLS developed a registry of recently diagnosed leukemia patients between the ages of 0 and 14 yrs-old in northern California. Most patients were added to the registry within 24 hours of diagnosis. Based on characteristics of the patient (date of birth, sex, mother’s race (white, black, or other), Hispanicity (either parent is Hispanic), and mother’s county of residence at the time of child’s birth). Once the control was selected, commercially-available searching methods were used to locate the control and invite their participation in the study. If the initial choice declined, the search was continued using the same process until a control subject agreed to participate. Participation was strong: 83% of the cases and 69% of the eligible controls. A total of 162 matched case-control pairs formed the basis for the study.

The team interviewed the primary care giver shortly after diagnosis of the cases and controls to establish the pattern of household pesticide use for different periods: a total of 4 years: 3 months prior to pregancy, pregnancy, and each of the first 3 years after birth. The interviews ascertained the types and locations of pesticide use, as well as the frequency. Ma et al. then used statistical analyses to evaluate the impact of pesticide use on the likelihood that a child would develop leukemia, and in particular on how leukemia risk was affected by timing of pesticide use relative to birth or the type of pesticide application.

What did they find?

Overall, pesticide use was quite common. For example, during the first year after birth, more than half of households used insecticides or indoor pesticides.

Families of the children with leukemia used professional pest control services more commonly than the families of matched controls. This pattern held for all time periods studied. In fact, the odds-ratio for professional pesticide applicants exceeded 2 for all of the time periods except for the 3 months prior to pregnancy. This suggests that a child in a home using professional pest control services is more than twice as likely to develop leukemia. And the odds ratio for a child living in a home that was professionally treated at any time during the 4-yr period of concern was 2.8.

Because frequency of application was recorded in the interviews, Ma et al. were able to ask whether the odds ratios were affected by frequency of exposure. This is a very rough surrogate for amount of exposure, or dose. Comparing odds-ratios of children not exposed during the 4-year period to those with frequency indices of 1-5 and those with >5, Ma et al. found that increased frequency was associated with increased risk: ORs, respectively, were 1.5 (95% CI 0.6-3.6) and 2.4 (1.2-5.1). Thus, increased use was associated with increased risk.

A striking temporal pattern also emerged from the analysis. The risk was particularly high for application during pregnancy compared to other periods, as shown by the following figure.

Comparing odds-ratios at different times before, during and after pregnancy, for indoor vs. outdoor pesticides. Figure adapted from Ma et al.

In contrast to the links between indoor use and leukemia, according to Ma et al. "the magnitude of association between exposure to outdoor pesticides and the risk of leukemia was small." This difference could result from many factors, including the fact that the suite of pesticides used outdoors (more herbicides) differed from those indoors (more insecticides). The data set was not large enough to begin to tease these factors apart.

What does it mean? Ma et al.'s work supports suggestions from previous studies indicating an association between pesticide use and childhood leukemia. Three patterns stand out:

• the association between risk and the use of professional applicators in the home;
• the increase in risk associated with increased frequency of exposure,
• and the enhanced risk during pregnancy.

As always, epidemiological studies offer very blunt instruments to explore the link between contamination and health. Given all the variables, the relatively limited sample size, and the relative crudeness of exposure measurements, the magnitude of effects they observe are quite striking. These factors often obscure real relationships between exposure and impact in epidemiological studies (that is, they cause what statisticians call "false negatives"). The fact that Ma et al. detected such a strong signal amidst all the noise heightens concerns about the role of pesticides in causing childhood leukemia.

On the basis of these results, cautious parents should choose to avoid household use of pesticides prior to, during, and after pregnancy.