The chart your link goes to shows that 4.25% of all US children who were actually tested for lead in the US in 2014 tested as high or higher than the kids who tested high in Flint.
You have to take the numbers provided and do the calculation, but that's what that chart's data shows. That chart was compiled by someone in the CDC who had their previous reference level, then known as the level of concern, in mind. That level was 10 ug/dL up to 2012. Note the chart shows the percentage of US kids under 72 months who tested higher than 10 ug/dL, but it does not give a % of kids testing higher than their current reference, i.e. 5 ug/dL.
The CDC would no doubt say that the 4.2% figure I calculated from their data doesn't accurately describe the situation for children in the US, i.e. that the lower figure the CDC gives, i.e. "only" 2.5% of US children would test as high as the Flint kids, is accurate, because, and they are no doubt right about this, testing tends to be more concentrated in populations of children authorities feel are more exposed to lead.
(My calculation: At the bottom of the chart , for 2014, it gives a total for the entire US, of children who tested higher than 5 ug/dL in 2014, i.e. 105,966, and it gives the total number of tests done that year, 2,496,140. I divided 105,966 by 2,496,140 to get 4.25%)
The CDC redefined what a significant blood lead level is in 2012 when they declared that no level of lead in blood is safe. They felt the evidence was so solid against lead they didn't want to be seen saying any level of it was safe in blood of children, so they give a "reference" level now, which they want authorities to use as a guide - if you see kids test higher than their reference, do something about it.
The CDCs own description of how they calculated the current 5 ug/dL reference level is here The 97.5 percentile of the NHANES blood lead distirbution they describe, means 2.5% of US children are expected to test higher than the value selected. NHANES data is the most authoritative snapshot of blood lead values at the time it is compiled that is available for US children.
A blog post more fully explaining what I'm talking about that I have so far not been able to get posted on TEC is here.