The decline in children's Blood Lead Levels (BLL) raises questions about the ability of current lead poisoning screening criteria to identify those children most exposed. The objectives of the study were to evaluate the performance of current screening criteria in identifying children with blood lead levels higher than 50 µg/L in France, and to propose new criteria. Data from a national French survey, conducted among 3831 children aged 6 months to 6 years in 2008-2009 were used. The sensitivity and specificity of the current criteria in predicting blood lead levels higher than or equal to 50 µg/L were evaluated. Two predictive models of BLL above 44 µg/L (for lack of sufficient sample size at 50 µg/L) were built: the first using current criteria, and the second using newly identified risk factors. For each model, performance was studied by calculating the area under the ROC (Receiver Operating Characteristic) curve. The sensitivity of current criteria for detecting BLL higher than or equal to 50 µg/L was 0.51 (0.26; 0.75) and specificity was 0.66 (0.62; 0.70). The new model included the following criteria: foreign child newly arrived in France, mother born abroad, consumption of tap water in the presence of lead pipes, pre-1949 housing, period of construction of housing unknown, presence of peeling paint, parental smoking at home, occupancy rates for housing and child's address in a cadastral municipality or census block comprising more than 6% of housing that is potentially unfit and built pre-1949. The area under the ROC curve was 0.86 for the new model, versus 0.76 for the current one. The lead poisoning screening criteria should be updated. The risk of industrial, occupational and hobby-related exposure could not be assessed in this study, but should be kept as screening criteria.
Keywords: child; exposure; health; lead poisoning; screening evaluation.