Sir - The paper by Eskenazi and colleagues has raised concerns about the effect of in utero exposure to DDT and DDE on neurodevelopment in Mexican American children. The findings are of interest to environmentalists, policy-makers, child-health specialists and public- health specialists. However, the paper's conclusion that the study has 'important implications for countries that are reconsidering or continuing the use of DDT for malaria control..', assumes that the results of this study are conclusive, and should thus affect policy. I would like to raise the following concerns about the study to illustrate that more data are needed: (i) to provide conclusive evidence that in utero exposure to DDT and (to a lesser extent DDE) may be associated negatively with neurodevelopment and (ii) before policy-changes regarding DDT-use in resource-limited countries where controlled spraying of DDT best eradicates the malaria vector, can be recommended.
Sample: Although the data were gathered from participants in a longitudinal birth cohort study our sense is that the 6, 12 and 24 months data on development should be interpreted as assessments, at three time points, on groups that slightly differed in their characteristics i.e. the assessments do not represent developmental trajectories of the children assessed at 6 months. This assumption is supported by the fact that data are presented for 360 mothers (not 330). It would be important to see the relationship between DDT and neurodevelopment in exactly the same group of children at 6, 12 and 24 months. If the sample of children at 6, 12 and 24 months differ, one wonders whether the reasons for either missed visits or missed assessments at 12 and 24 months were in any way related to exposure or outcome.
Loss to follow-up/non-inclusion in analysis: Although the original cohort was 601 women, blood levels of DDT were obtained for only 526 women, and the association between DDT/DDE levels and BSID scores were only ascertained in 330, 327 and 309 children (6, 12 and 24 months respectively) i.e. in a little more than half the original sample. It is not clear how / whether the children / mothers excluded from the final analysis differed from children/mothers included in the analysis. Were the children included in the analysis at higher risk of obtaining lower scores, for reasons other than DDT exposure? Did mothers who were excluded have lower DDT/DDE levels, or would their infants have had poorer neurodevelopmental outcomes than those included in the sample due to reasons other than DDT?
Measuring exposure: Exposure was agricultural - It is not clear what levels of agricultural exposure had led to the high blood DDT/DDE levels, and whether controlled indoor household residual spraying of DDT for malaria control (using a low dose - approximately 1-2g/m2) would produce the same exposure level. Furthermore, mother's blood level was used as a proxy for infants exposure. Though this may be correct  , postnatal exposure to DDT/DDE was not known, and thus not accounted for.
Measuring and interpreting outcome: The Bayley scales of infant development (BSID) were used to measure mental and psychomotor development, using the US population as a standard. Several issues are of concern here: (i) The BSID may be used to describe the current developmental functioning of infants However, their stability (repeated measures using the same test) and predictive power are not high . The article, by stating that the study has implications for malaria control, suggests that the decrease in BSID scores are clinically significant. However, the original version of the BSID scale for infants and toddlers up to 24 mo of age has failed to show construct and predictive validity. Although BSIDII incorporated a new standardization of the mental scale it has shed no new light on its construct validity up to the ages of 18-24 months [2,3,4]. Studies have shown that the median correlation between the mental development scores obtained at some point between 7 and 12 months and an IQ obtained sometime between 5 to 7 years of age was about 0.20 [5,6].This means that the finding that a 10-fold increase in p,p'-DDT, o,p'-DDT, and p,p'-DDE levels were significantly associated with decreases in 12- and 24-month MDI scores are likely to not have any bearing / significance for later development and IQ. Furthermore, although the analysis controlled for psychometricians, it would have been useful to have conducted and reported on a preliminary pilot to determine the correlation coefficients for intra and inter- observer BSID scores.
Measuring confounders and other variables: Postnatal exposure to other neurotoxicants, and other confounders such as nutritional status (infants weight for age / weight for height at each assessment), diet, type of delivery, length of second stage of labour, birth trauma were not measured and controlled for. Did women with higher DDT exposure also have other such characteristics that increased their infant's risk of getting low scores on the BSID?. Furthermore although the study measured depressive symptomotology using a standardized scale it is not clear how relevant and reliable this scale was to identify depression in this newly immigrant population, and how this may have affected the relationship between exposure and outcome.
Statistical analysis: The statistical analysis treated DDT/DDE levels as a continuous variable - possibly because the researchers believe that any exposure to DDT leads to adverse outcome. It would be interesting to also see the analysis conducted using DDT/DDE levels as a dichotomous variable (equal to or above the geometric mean and below the geometric mean) to determine whether the scores are significantly different. Furthermore, the analysis in this study lumped all DDT exposures and outcomes, including the group with no known exposure into one group. Future research on DDT exposure and developmental outcome should, for public health purposes, look at exposure and outcome for three different groups of mother-infant pairs: (i) known agricultural exposure (ii) known exposure only through controlled indoor household residual spraying (iii) no known exposure to DDT to determine whether outcome differs significantly between groups.
Other issues of concern: The authors report on standardized scores, which, in this population, are lower than the expected mean standardized score for the US population. The authors also do not state whether the BSID were modified to make them more relevant to the newly immigrant population. It is not clear what each mother was told about her child's neurodevelopment, and future, in relation to the scores obtained. Caution is usually warranted in using the BSID for populations that differ from the standardization sample and some examiners who use the BSID for research purposes report raw scores, rather than relying on U.S. norms  Furthermore, the assumption after reading the article is that BSID II was used. BSIDII requires that examiners have training and experience in administering and interpreting standardized assessments with infants as test administration and interpretation is more complex than with other standardized assessments because the examiner alters the sequence of items in response to the infant's behavior and performance . Typically examiners have training at the master's or doctoral level and supervised experience, in accordance with guidelines from the American Psychological Association. The paper does not state the level of skill of the pschometricians.
Looking forward: As stated by the authors I agree that (i) the beneficial role of breastfeeding, in the context of DDT use should be explored further, particularly as breastfeeding is usually a norm in areas of high malaria endemnicity, and in areas where DDT is currently being used for malaria control (ii) the cohort assessed should be followed up and developmental trajectories developed for each child to ascertain the clinical significance of the results.
1. Waliszewski SM, Aguirre AA, Silva CS, Siliceo J. Organochlorine Pesticide Levels in Maternal Adipose Tissue, Maternal Blood Serum, Umbilical Blood Serum, and Milk from Inhabitants of Veracruz, Mexico. Arch. Environ. Contam. Toxico. 2001; 40: 32-438.
2. Pollitt E. Statistical and psychobiological significance in developmental research. American Journal of Clinical Nutrition. September 2001; 74:3, 281-282, .
3. McCall RB, Mash CW. Long-chain polyunsaturated fatty acids and the measurement and prediction of intelligence (IQ). In: Dobbing J, ed. Developing brain and behavior. San Diego: Academic Press, 1997:295-329.
4. Pollitt E, Triana N. Stability, predictive validity, and sensitivity of mental and motor development scales and pre-school cognitive tests among low-income children in developing countries. Food Nutr Bull. 1999;20:45-52.
5. McCall RB. A conceptual approach to early mental development. In: Lewis M, ed. Origins of intelligence. 2nd ed. New York: Plenum Press, 1983:255-301.
6. McCall RB. The development of intellectual functioning in infancy and the prediction of later IQ. In: Osofsky JD, ed. Handbook of infant development. New York: Wiley, 1979:704-41.
7. Black MM, Matula K. Essentials of Bayley scales of infant development. II. Assessment. New York, NY: Wiley, 2000.
*Ameena Goga is supported by CAPRISA which forms part of the Comprehensive International Program of Research on AIDS (CIPRA) funded by the National Institute of Allergy and infectious Disease (NIAID), National Institutes of Health (NIH) and the US Department of Health and Human Services (DHHS) (grant# 1 U19 AI51794), and the Columbia University- Southern African Fogarty AIDS International Training and Research Programme (AITRP) funded by the Fogarty International Center, National Institutes of Health (grant # D43TW00231.
Conflict of Interest: None declared
To read other critiques of the paper in question and the authors' response, please visit: http://pediatrics.aappublications.org/cgi/eletters/118/1/233#2193