Readability and Missing Data Rates in CAHPS 2.0 Medicare Survey in African American and White Medicare Respondents

Abstract

Purpose To examine associations between readability of survey items and missing data rates in a sample of white and African-American Medicare enrollees in managed care plans. Methods Consumer Assessment of Healthcare Provider and Systems (CAHPS) 2.0 health plan survey data collected from 139,284 respondents (127,524 whites and 11,760 African Americans) in 321 health plans. Product-moment correlations were computed between Flesch-Kincaid (F-K) readability estimates and the CAHPS item-missing data rates. Results F-K reading levels for items ranged from 4.8 to 17.7 with a mean of 8.9 across items. Missing data rates ranged from 1% to 10%, with African Americans having significantly higher missing data rates. Correlations between missing data rates and item-level readability were statistically significant for whites (r = 0.33, P = .0515) and African Americans (r = 0.37, P = .0284). Conclusions The significant associations between missing data rates and item-level readability estimates indicate that the completion of survey items varies by their readability. Enhancing the readability of survey items can improve the inclusion of survey data collected from different respondents.