Estimating Percentile Confidence Intervals with Samplics

[kevintcaron]

Hi @MamadouSDiallo,

I am attempting to use samplics to replicate results from the CDC's National Exposure Report Biomonitoring Tables. So far, I have been successful at reproducing geometric means, their 95% confidence intervals and percentile estimates. However, I have not been able to correctly reproduce the 95% confidence intervals for the percentile estimates. For example, the reported 2013 Blood Total Mercury for Age 6-11 years at 95th percentile is 1.62(1.38-2.19) µg/L. I can get 1.62 for the estimate, but I do not know how to get the confidence intervals for this in samplics (it appears samplics reports the CIs of the proportion, not the estimate when using PopParam.prop).

The Procedure for Estimating Percentiles is described on the CDC website, along with sample SAS code that employs SAS-callable SUDAAN. I get the correct confidence intervals when using SAS/SUDAAN, but was hoping I could replicate these results using samplics in order to dynamically generate future exposure reports without SAS or SUDAAN. The SAS/SUDAAN code is a bit difficult to read but is provided in the link above. Please let me know if you have any thoughts or recommendations on how to proceed. Thank you again for this tool!