4.1 Ingestion Rates for an Incident of Soil-pica
The first charge question pertaining to ingestion rates for soil-pica children was: "ATSDR currently assumes children who exhibit soil-pica behavior, on average, ingest 5,000 mg of soil per day. Based on your review of the literature on soil-pica behavior, is this soil ingestion rate scientifically defensible? Does this represent plausible exposures for children (i.e., is this soil ingestion rate unrealistically high or unrealistically low)? What soil ingestion rate would you recommend?" The panelists agreed that ATSDR's assumption that soil-pica children may ingest 5,000 mg of soil per day appears to be supported by only a few subjects in soil ingestion studies (i.e., two incidents in Massachusetts and several children in Jamaica). Referring to the distribution of soil ingestion rates published in the literature, some panelists thought ATSDR's assumed ingestion rate for soil-pica children seems high. Other panelists thought the available data might be inadequate for characterizing the distribution of soil ingestion rates among children. Tables 4-1 and 4-2, at the end of this section, present some of the findings published in the literature that formed the basis of the panelists' discussions.
The panelists also agreed that ATSDR should err on the side of being protective when examining exposures to soil contaminants. Following is a summary of the discussions that led to these key findings.
Review of data on maximum daily ingestion rates. The panelists first reviewed the maximum soil ingestion rates reported in the literature. Citing data reported in the background documents, one panelist commented that evidence of gram-level soil ingestion has only been observed in two incidents in Massachusetts and in six children in Jamaica, with the highest daily soil ingestion rate reported being 50,000 mg (JM). After someone else noted that the 50,000 mg ingestion rate was observed for a developmentally disabled child (DM), this panelist said the next highest daily soil ingestion rate reported is 20,000 mg. She questioned ATSDR's use of a soil ingestion rate (5,000 mg per day) that is consistent with the most extreme soil ingestion rate ever reported (JM). Another panelist agreed, noting that nearly every child that has been considered in analytical studies has had ingestion rates considerably lower than 5,000 mg per day, though he acknowledged that the design of these studies might not be able to capture rare peaks in soil ingestion (SD).
One panelist indicated that a statistical analysis of the existing soil ingestion data (Calabrese and Stanek, 1998) estimates that 42% of children will ingest 5,000 mg of soil on 1 or 2 days a year--a finding he thought supports ATSDR's use of this soil ingestion rate (PS). Other panelists argued, however, that this finding should be viewed as an estimate since it is based on a statistical analysis of very limited data; they suspected that the available soil ingestion data are not sufficient to characterize the distribution of soil ingestion rates among children (SD, NF).
Comments on the distribution of soil ingestion rates. Despite their concerns about whether the available data adequately characterize this distribution, two panelists noted that data presented in EPA's Exposure Factors Handbook suggest that children who ingest 5,000 mg of soil a day are well above the 95th percentile of soil ingestion rates (NF, RW). Citing Table 4-1 in the Exposure Factors Handbook, for example, one panelist indicated that a study conducted by Ed Calabrese and Ed Stanek reported a 95th percentile soil ingestion rate of 208 mg per day (RW). Commenting further on this table, another panelist noted that the 95th percentile soil ingestion rate calculated for seven of the eight tracers considered ranged from 117 to 518 mg/day; she added that only by considering manganese as a tracer did the study report 95th percentile soil ingestion rates greater than 5,000 mg per day (NF). Given the great differences among the tracers, this panelist questioned the validity of the analytical method for characterizing soil ingestion rates. Nonetheless, based on the presentation of these 95th percentile soil ingestion rates, two panelists thought a case could be made for using a soil ingestion rate lower than 5,000 mg per day for soil-pica children.
When discussing the implications of the analytical studies, some panelists noted that Ed Calabrese and Ed Stanek not only observed a 95th percentile soil ingestion rate of 208 mg/day based on the 64 children they studied, but also reported a 95th percentile soil ingestion rate of 1,751 mg/day based on a statistical analysis of this same data (see Table 4-2 in EPA's Exposure Factors Handbook) (DM). Another panelist stressed that this higher soil ingestion rate was not actually observed, but was based on various assumptions made when extrapolating data collected over less than 2 weeks to soil ingestion behavior over an entire year (JM). Nonetheless, some panelists noted that the more recent findings reported by Ed Calabrese and Ed Stanek are more consistent with ATSDR's estimated soil ingestion rate.
Suggested approaches to defending a soil ingestion rate for pica children. Without knowing how ATSDR derived its soil ingestion rate for pica children, the panelists ultimately found it difficult to critique the validity of this parameter. The panelists suggested that ATSDR should consider two approaches for deriving a defensible soil ingestion rate. First, ATSDR should conduct a comprehensive soil ingestion study that uses multiple methods to characterize the prevalence of pica behavior and to quantify the distribution of soil ingestion rates (NF, BL, RW). Second, until a more comprehensive study is performed, ATSDR should defend its soil ingestion rate by conducting statistical analyses of the various analytical studies reported in the literature (3) (DC, SD, NF). These panelists acknowledged, however, that such analyses will continue to be limited because the available data are sparse and possibly not representative of all geographic areas, cultures, and seasons. Given these uncertainties, the panelists stressed the importance of validating any conclusions drawn from the soil ingestion data currently available. Recommendations for validating these conclusions included measuring biomarkers, identifying adverse health effects, and conducting a multi-faceted study that relies on different approaches for characterizing soil ingestion.
3. Some panelists thought Ed Calabrese and Ed Stanek may have already performed such a statistical analysis, as documented in Table 4 of his publication in the Environmental Law Reporter (Calabrese and Stanek, 1998).