Psychometric Properties of the Thai Health Literacy Assessment Using Word List with Extended Questions to Test Comprehension in Thai Muslim Patients

Abstract

Objectives: To test validity and reliability of the Thai Health Literacy Assessment Using Word List with Extended Questions to Test Comprehension (THLA-W+) in Thai Muslim patients and to determine its cut-off. Methods: This study used the THLA-W+ developed by Phadoong Chanchuto. The scale consisted of 48 words with 4 corresponding choices to test the comprehension of word meaning. Scoring of the scale was conducted in 3 methods: 1) calculating the scores from reading test (THLA-W+R), 2) calculating the scores from comprehension test (THLA-W+C) and 3) calculating the scores from both reading and comprehension test (THLA-W+RC). Subjects were 767 Muslim outpatients at Yaring hospital in Pattani province. Validity testing of THLA-W+ was conducted by determining its correlation with various indicators of health literacy (HL) and the Thai Health Literacy Assessment Using Nutrition Label (THLA-N8). Cut-off point and predictive ability were determined by the analysis of receiver operating curve. Results: Cronbach’s Alphas of THLA-W+R, THLA-W+C and THLA-W+RC were 0.98, 0.97 and 0.97, respectively. The THLA-W+C and THLA-W+RC could well discriminate the subjects with different levels of education, and better than the THLA-W+R did. Correlation coefficients (r) between all three THLA-W+ scores and HL indicators (e.g. reading ability, understanding of health documents) were positive and statistically significant (r=0.222-0.816; P<0.05). The THLA-W+C and THLA-W+RC showed a higher correlation coefficient than the THLA-W+R did.  The THLA-W+C and the THLA-W+RC highly correlated with r=0.987. When using reading ability, understanding of health documents and THLA-N8 as gold standards, THLA-W+R had an area under the curve (AUC) of 0.641-0.660 (depending on gold standards), which was less than those of the THLA-W+C and THLA-W+RC, which were 0.820-0.830 and 0.826-0.832, respectively. The cut-off values of the THLA-W+R, THLA-W+C and THLA-W+RC were 47, 37 and 37 respectively. Sensitivities of the scales were 41.90-46.52, 68.68-78.61 and 70.74-80.75, respectively. THLA-W+C and THLA-W+RC were more sensitive than the THLA-W+R did. Specificities of three measures were similar at 76.21-88.10, 67.96-83.33 and 66.02-80.95, respectively. Conclusion: The THLA-W+ was reliable and valid for using in Muslim patients. The scale should be administered as THLA-W+C because it provides the scores with a better psychometric property than THLA-W+R, and needs no staff to judge the pronunciation of the subjects. The THLA-W+C shows a satisfactory sensitivity and specificity with the score of 37 or lower indicating inadequate HL.