การพัฒนาแบบประเมินความแตกฉานด้านสุขภาพสำหรับชาวไทยชนิดรายการคำที่มีคำถามทดสอบความเข้าใจ (THLA-W+) การทดสอบชุมชน

Abstract

This study aimed to test the validity and reliability of the Thai Health Literacy Assessment Questionnaire Using Word List with Extended Questions to Test Comprehension (THLA-W+) in the public and to determine the cut-off value of the scale to interpret the level of health literacy (HL). The researcher tested the THLA-W+ developed by Padong Chanchuto (2017) in 522 people living in Kor-en community within Phuket. There were 3 methods to calculate the scores for THLA-W +, 1. from the number of words with correct pronunciation from reading test, 2. from the number of words with correct choice from comprehension test, and 3. from the number of words with correct pronunciation and choices. The THLA-W+ test took an average of 6.69 +1.188 minutes. Reliabilities of the scores calculated from reading test, comprehension test and both tests were 0.801, 0.869 and 0.869, respectively, which was satisfactory. Average scores of the THLA-W+ were 45.57 ± 3.24, 31.58 ± 7.65 and 30.59 ± 7.82, respectively Addition of extended questions lowered average score and increased standard deviation. Subjects with a higher level of education got a higher score on the THLA-W+. Correlation coefficients between THLA-W+ score with different HL indicators (eg, understanding of health documents) were positive and statistically significant (r=0.129-0.369). However, the HL indicators measuring various dimensions of the construct beyond reading skill (eg, understanding of health documents) exhibited a lower correlation with THLA-W+ score calculated from reading test compared to that from the selection of correct choice. Correlation coefficient of THLA-W+ calculated from choice selection and that from reading test together with choice selection were as high as 0.98. THLA-W+ score based on the selection of correct choices and that on reading test together with choice selection showed the same result from the analysis of the Receiver Operating Characteristics (ROC) curve in terms of sensitivity, specificity, areas under the ROC curve, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR +) and negative likelihood ration (LR-). However, their profiles from analysis were better than THLA-W+ based on reading test. THLA-W+ based on selection of correct choice had a cut-off value at 37 indicating that those with the score 37 having inadequate HL. The score on THLA-W+ could well discriminate those with different levels of HL when using reading ability and understanding of health documents as gold standard. The scale showed sensitivity at 77-82%, specificity at 81-83%, areas under the ROC curve at 0.86-0.87, and PPV at 99%, but low NPV at 7%. LR + was 4.24-4.64, while LR- was 0.27-0.28, implying that the scale could provide important information for decision making in practice. The scale was appropriate for HL surveys in public. It is therefore concluded that use of the THLA-W+ for public survey of HL should calculate the score from choice selection because of its better psychometric properties compared to those calculated from reading test. Moreover, it simplifies testing process by not requiring to have the staff for testing reading skill of subjects.