Development of a Short Version of the Health Literacy Scale Based on Classical Test Theory and Item Response Theory

Abstract

Background Health literacy is closely associated with health status, most domestic studies tend to use multidimensional and multi-item tools for assessing health literacy, lacking simple and effective assessment methods. Objective To simplify health literacy scale and conduct psychometric test within the Chinese population. Methods Adults aged 18 years and above were selected from the "China Family Health Index Survey (2021) " for this investigation. Based on inclusion and exclusion criteria, a total of 7 449 participants were selected and randomly divided into two sample sets, including 3 680 cases in sample set 1 and 3 769 cases in sample set 2. The general information questionnaire, Short-form Health Literacy Questionnaire (HLS-SF12), Perception Social Support Scale (PSSS), and Family Health Scale Short-form (FHS-SF) were administered to the respondents. Classical test theory (CTT) and the Mokken model in item response theory (IRT) were used to screen the original items, and validation analyses such as reliability and validity of the simplified scales were also conducted. Results A 9-item version (HLS-SF9) and a 4-item version (HLS-SF4) were simplified by using CTT and the Mokken model, respectively. Both the HLS-SF9 and the HLS-SF4 had no ceiling effect or floor effect, and the Cronbach's α coefficients for both were 0.913 and 0.842, with split-half reliabilities of 0.871 and 0.815, respectively. The exploratory factor analysis of HLS-SF4 revealed one common factor, accounting for 67.813% of the cumulative variance, with factor loadings exceeding 0.81 for each item. The confirmatory factor analysis of HLS-SF9 showed that χ2/df was 10.844, goodness of fit index (GFI) was 0.985, adjusted goodness of fit index (AGFI) was 0.971, normative fit index (NFI) was 0.986, comparative fit index (CFI) was 0.987, and root mean squared error of approximation (RMSEA) was 0.051. The correlation analysis demonstrated positive correlation of HLS-SF9 and HLS-SF4 with PSSS (r=0.367, 0.292, P<0.001), as well as FHS-SF (r=0.340, 0.237, P<0.001), respectively. The intraclass correlation coefficients (ICC) (95%CI) for the criterion validity of HLS-SF9 against HLS-SF12 was 0.989 (0.988-0.999), while for HLS-SF4 against HLS-SF12 was 0.892 (0.886-0.899) . Conclusion The simplified health literacy scales have good reliability and validity, which are reliable and effective tools for assessing the health literacy of the Chinese population. Researchers can choose the scale according to the needs of research precision (HLS-SF9) or assessment time (HLS-SF4) .

Keywords

• health literacy|item selection|classical test theory|mokken model|short-form health literacy survey questionnaire