To measure the effects of questionnaire and imputation improvements in the Current Population Survey () on the estimated prevalence of high medical financial burden, that is, families spending more than 10 percent of income on medical care.
Matched longitudinal sample of data for 2013 and 2014 calendar years.
The used a split‐sample design to field traditional and redesigned questions about 2013 income, and old and new out‐of‐pocket premium imputation procedures, respectively. For both samples, data for 2014 were from the redesigned income questions and the new imputation procedures. We quantify the effects of the combined survey improvements using differences‐in‐differences methods.
The improvements were not associated with changes in the estimate of burden in the full sample. Estimated prevalence increased by 2.6 percentage points among nonelderly adults with private insurance, decreased by 6.6 percentage points among nonelderly adults with public coverage, and decreased by 5.8 percentage points among elderly adults with Medicare and no private coverage.
Improvements in the changed the estimated prevalence of high medical financial burden among key subgroups. Researchers should use caution when tracking burden across the time‐period in which these improvements were implemented.