Discussion of Association between Soil Lead and Blood Lead
Generally, five study designs have been used to test the hypothesis of an association between soil lead and blood lead: ecological- temporal, ecological-spatial, cross-sectional, prospective soil removal, and descriptive. Each of the study designs has limitations.
The cross-sectional study design is the most frequently applied study design. This study design is severely limited due to exposure misclassification and outcome misclassification. It is well known that in urban areas, soil lead concentrations decay exponentially away from the roadside. It is also knownthat soil lead concentrations are elevated adjacent to homes due to the effect of past aerosols coming into contact with the side of buildings and dropping to the soil below. In addition, in homes with exterior lead paint, the soils adjacent to homes can become contaminated due to deterioration of leaded paint. Cross-sectional studies that are based on single sample exposure will result inexposure classification because point samples can be highly variable for the above-mentioned reasons. Outcome misclassification arises due to the well-known blood lead seasonality effect. Blood lead seasonality of groups of children has shown that blood lead concentrations can increase over 100% between winter and summer. The exposure and outcome misclassification makes it very difficult to correlate due to the high variability of exposure (soil lead) and outcome(blood lead). This likely explains the reason why cross-sectional studies sometime have historically low correlations, and this is due to the inadequacy of the study design. An example of a cross-sectional study was a study of the association between soil lead and blood lead at four superfund sites which was completed by the Agency for Toxic Substances and Disease Registry's (1999). This study found that the predicted blood lead level corresponding to a soil lead level of 500 mg/kg was 5.99 microg/kg with a 95% prediction interval of 2.08-17.29.
The ecological spatial study design is completed by taking the average blood lead of a small area for a given period of time, and correlating that blood lead with the average soil lead concentration for the same small unit, across all the small area units in a city. This study design has been successfully applied (R2 >0.65) in Syracuse (Johnson and Bretsch, 2002), New Orleans(Mielke et al. 1997), and Shenyang (Ren HM et al. 2005). This study design appears to be successful because the exposure (soil lead) is generally an average of a large number of soil samples in a small area, thus dampening out the exposure misclassification that occurs in the cross-sectional study design. Furthermore, since it is time invariant, the misclassification from seasonality effect that is found in the cross-sectional study design is not a factor. The limitations of this study design is that the confounding variable, homes with paint lead, are not accounted for. However, this variable can be adjusted for.In fact, Mielke found that the p value for spatial association between soil lead and blood lead was many orders of magnitude lower than the association between the percentage of home with paint lead with blood lead (Mielke et al.,1997). Another limitation is that in the ecological study design, the association is between the average blood lead of the group and not the individuals – thus the inference of a blood lead soil lead association atthe individual level should be made cautiously
Ecological – Temporal
The ecological temporal study design is conducted by selecting a city, then taking the average monthly blood lead concentration of all the children, for aspecific time period, then regressing against soil moisture and climate variables (Laidlaw et al. 2005). Of temporal variation in urban children'sBPb, 87% in Indianapolis (R2 = 0.87, p = 0.0004), 61% in Syracuse (R2 = 0.61, p= 0.0012), and 59% in New Orleans (R2 = 0.59, p = 0.0000078) are explained bythese variables (Laidlaw et al. 2005). This approach is believed to besuccessful because it is thought that soil moisture and climate variables are proxy indicators of urban lead contaminated soil re-suspension (Laidlaw and Filipelli, 2006). This approach eliminates spatial concerns, such as confounding by paint lead, and examines temporal fluctuations only. The limitation of the study design is that it does not account for variation inlead particles released from the opening and closing of windows. However, it is not believed that the opening and shutting of windows is related to temporal changes variables such as PM10 and soil moisture, thus it is believed that lead particles released from the opening and closing of windows is not related to a significant proportion of the variation in children's blood lead seasonality. In addition, ecological studies are based upon the group, rather than the individual, and inferences at the individual level must be made with caution.
Prospective Soil Removal
Prospective soil removal studies remove lead contaminated soil from yards of selected homes while leaving lead contaminated soil in yards of other homes. Blood lead concentrations of children in the homes are then compared between treatment and controls. These study designs have limitations. First, the samplesizes can be very low. Second, the blood lead samples must be collected at the same point in time to prevent outcome misclassification due to blood lead seasonality which could result in lower correlations and p values. Some of these studies have failed to account for differences in how the homes are sealed as well. These studies can also fail because they do not take into account the fact that dust that travels to a home does not originate from the soil from the home but the surrounding community (von Linden et al., 2003).This was illustrated by Sheldrake, S.; Stifelman, M. (2003) at the Near theBunker Hill Superfund site where they found that cleanup of residences has athree-fold greater reduction of children's blood lead levels compared withcleaning only those homes where children currently reside by reducing exposures attributable to neighboring properties. This approach has been successfully completed by Maisonet et al. who conducted a pair-matched, case-control study of yard soil remediation and found that yard soil removal was a strong protective factor for elevated blood lead levels in children (odds ratio, 0.28;confidence interval, 0.08-0.92).
Isotopic studies hold promise, however, in the United States some people suggest that there is an overlap in isotope ratios of lead in paint andgasoline, which limits any source apportionment. However the use of isotopic ratios have shown promise in Armenia (Kurkjian & Flegal, 2003).