Development and Validation of Thai Hand Function Assessment Questionnaire Using the Rasch Model

Abstract

Objective: To develop and validate the Thai Hand Function Assessment Questionnaire. Methods: The first draft of the Thai Hand Function Assessment Questionnaire was developed based upon Thai and foreign questionnaires and from an interview with physical therapists and occupational therapists. Its face validity was assessed by 3 experts’ opinion. Index of item-objective congruence ranged 0.67-1.00. The final version of the Thai Hand Function Assessment Questionnaire comprised 33 items with a 5-point scale. It was administered in to 180 patients in the physical therapy departments of 10 hospitals. Data were analyzed by the Rasch model. Results: The questionnaire showed unidimensionality from the variance explained by measures (54.96%) and the variance of residuals in the first contrast (3.61%) was greater than 40% and less than 15% respectively. All items displayed local independence. Person reliability and person separation index were 0.93 and 3.59 whereas item reliability and item separation were 0.95 and 4.29 respectively, which were regarded as acceptable. Test of item fit statistics revealed that 33 items had optimal infit mean-squares and outfit mean-squares ranging between 0.50-1.50 except for item 18 that outfit mean-squares was 1.53. After combining 5 response categories to 3 categories and excluding 29 mis-fitted persons from the Rasch model, the test of item fit statistics found that all 33 items had acceptable infit mean-squares and outfit mean-squares. Item difficulty was not well distributed when compared with the distribution of ability of subjects’ hand function. Some groups of subjects had their hand function ability lower than the easiest item while others had their hand function ability higher than the most difficult item. Item difficulty ranged between -1.19 to 1.55 logits while ability of hand function ranged between -6.62 to 5.12 logits. Conclusion: The Thai Hand Function Assessment Questionnaire has unidimensionality and local independence with high validity and reliability. However, further studies should include additional items with more difficult and easier items to capture all hand function levels of the subjects.