ADHD:Braun et al.Respond

摘要

We appreciate the comments of Brondum,and Konofal and Cortese,and the opportunity to clarify our results(Braun et al.2006).It is common practice to select variables with a p-value of 0.2 for inclusion in multivariable models(Katz 1999).Although the association of blood lead levels and ADHD appeared"tenuous"in bivariate analysis(i.e.,p = 0.19),this was largely an artifact of our decision to categorize blood lead levels.When we entered lead into our multivariable analysis as a continuous variable,we found a 1.2-fold oincreased odds [95% confidence interval(CI),1.0-1.4;p = 0.02] of ADHD for each 1.0-ug/dL increase in blood lead levels.The blood lead quintiles were not divided into exactly equal sample sizes because we used weighted percentages to categorize the data.