First-order vs. second-order structural validity of the Health Literacy Scale in patients with diabetes

Abstract

OBJECTIVE: The purpose of this study was to determine the structural validity of the Health Literacy Scale (HLS) in Korean patients with diabetes based on the previously reported first-order three-factor and two-factor models and a newly proposed second-order model, to identify which model best represents the structure of the Korean version of the HLS (HLS-K). METHODS: The HLS was translated from Japanese into Korean using a translation and back-translation technique. A secondary data analysis was used to validate the structure of the HLS. Data were obtained from a cross-sectional survey involving 459 adults with diabetes recruited from outpatient clinics at two university hospitals in South Korea. The structural validity was examined using confirmatory factor analysis. Additionally, the known-groups validity by education level and internal consistency validity were assessed. RESULTS: The second-order three-factor model of the HLS-K exhibited a good fit to the data, as indicated by chi(2) /df = 3.891, SRMR = 0.042, GFI = 0.924, RMSEA = 0.079 (90% CI = 0.069-0.090), and CFI = 0.962. The second-order three-factor model empirically demonstrated that both communicative and critical factors explained the variance in the overall health literacy better than did the functional factor. The HLS-K mean score was significantly lower for patients with only elementary school education than for those with higher education levels, implying the presence of known-groups validity. Cronbach's alpha for the total scale was 0.90. CONCLUSIONS: This study found that the second-order three-factor model of the HLS-K is better than that the original first-order three-factor and first-order two-factor models. Further validation studies are needed to generalise the underlying structure of the instrument in diabetes populations across various cultures.