Traditionally in national surveys of the United States, census region is used to stratify the frame and ensure that regions are covered in the sample. In this paper, we focus on a health survey that applies a state-level healthiness index instead of regions (or other geographic identifiers) for initial stratification purposes. The National Health and Nutrition Examination Survey (NHANES) is an ongoing multistage annual survey with first-stage primary sampling units (PSUs) selected for four years at a time. NHANES collects thousands of health-related variables for each sampled person from an interview, physical examination, and laboratory tests. In the 2011-2014 four-year sample design, NHANES initially stratified the PSU frame by a state-level healthiness index, resulting in PSUs within the same strata (with the same general levels of healthiness) coming from different regions. This paper reviews this stratification process and examines the improvement in the precision of the estimates.