3.3 Approaches for Characterizing the Prevalence of Soil-Pica
In response to the first charge question--"What observational, questionnaire, and analytical studies are most valid for characterizing the extent of soil-pica behavior?"--the panelists discussed the different methods that have been used to evaluate the prevalence of soil-pica behavior, and identified strengths and weaknesses associated with each method. Given the limitations of individual methods, the panelists eventually agreed that a study that incorporates multiple methods would provide the best means for validating the prevalence of soil-pica behavior and the distribution of soil ingestion rates. The panelists thought that such validation was needed to develop a robust characterization of the prevalence of soil-pica behavior. Following is a summary of the panelists' discussion that led to this conclusion.
The utility of questionnaires. Two panelists noted that surveys on soil ingestion, which researchers typically administer to caretakers, are useful for getting meaningful insights on general issues (e.g., "does your child ever eat dirt?") but are not particularly useful for getting robust answers on specific issues (e.g., "how often does your child eat dirt?" or "how much dirt does your child consume?") (SD, NF). Expanding on this comment, one panelist indicated that parents typically do not observe their children constantly and therefore are not able to comment reliably on specific details of their soil ingestion behaviors (NF).
Other panelists listed reasons why administering surveys to parents about their children's behavior might lead to spurious results. First, parents might provide responses they think surveyors want to hear, rather than responses that accurately reflect their children's behavior (NF). Second, parents might provide inaccurate responses in efforts to conceal information that might reflect badly on them as parents (e.g., their children eating large amounts of dirt) (JM, DV). Third, because "unusual" behavior is more easily recognized in families with multiple children, parents' perceptions of "unusual" behavior might vary from one family to the next, thus complicating efforts to characterize the prevalence of soil-pica behavior with surveys (PS).
The utility of analytical studies. When discussing the different methods available to characterize the prevalence of soil-pica, several panelists noted that soil ingestion rates predicted by analytical methods (i.e., mass balance tracer research) have varied considerably from one study to the next (SD, BL, DM). One panelist suspected that these inconsistent findings might result from the difficulty short-term analytical studies have identifying rare events (DM). Another panelist agreed, explaining that the analytical studies he has conducted and reviewed characterize soil ingestion behavior for a small number of people over a very short time frame, typically 2 weeks or less (SD). With this study approach, he thought the analytical studies have a very small chance of identifying soil-pica events.
The panelists raised several other concerns about analytical studies. For instance, one panelist was not convinced that analytical studies can provide the most reliable account of soil ingestion, given his experience conducting two studies, both of which found that soil ingestion rates calculated from analytical approaches correlate very poorly with observational accounts of mouthing behavior and soil ingestion (SD). Another panelist added that inconsistent results might stem from the fact that studies are conducted in different regions and among children of various socio-economic status (DM). He was not surprised, for example, that soil ingestion rates observed among children in suburban communities in Massachusetts were different from those conducted on institutionalized children in Jamaica. One panelist did not think the prevalence of soil-pica behavior among the Jamaican children should be viewed as representative of that among children in the United States, given the small size of the study and the living conditions of the children considered (BL). Another panelist noted that mass balance studies assume an understanding of the digestive processes and degree of uptake, transformation, and excretion of tracers in young children (NF). She added that most tracer studies have been conducted on children, who have different gut permeability, metabolism rates, and excretion rates than adults. This reviewer commented that true mass balance studies would collect both feces and urine over a long period of time, because the temporal pattern of excretion in the two media will be different. She indicated that the mass balance studies included in the review materials were based only on fecal sampling over a limited time frame.
The utility of combining several methods. Given their concerns about the various individual methods for characterizing soil ingestion, most panelists advocated the use of multiple methods in one study to derive a robust, validated distribution of soil ingestion rates. Specifically, panelists highlighted the need for conducting an extensive study that integrates information on levels of soil contamination, biomarkers of exposure, and various metrics of soil ingestion (e.g., analytical, observational, and surveyed accounts) (BL, RW). Several panelists noted that such a study would be expensive, but thought the community at the Vasquez Boulevard/Interstate 70 Superfund site might be an excellent subject population for such research (NF, BL). Though not disagreeing with these suggestions, one panelist cautioned that using multiple methods to characterize soil ingestion has led to conflicting results in his previous research and that using multiple methods to derive a validated soil ingestion rate for pica children will be quite challenging, though worth pursuing (SD).
When commenting on the use of integrated approaches to characterize soil ingestion rates, two panelists identified existing data sets that ATSDR should review. One panelist indicated that one of his past studies on soil ingestion, which included observational and analytical components, also has biomarker data (i.e., blood lead levels) that have yet to be thoroughly examined (SD). Another panelist noted that biomarker data (i.e., levels of arsenic in urine) are available for a Superfund site in Washington state where ingestion of contaminated soils is of concern (JM).