Calibration of a physical functioning item bank for measurement of health-related quality of life in Singapore

Purpose We aimed to develop and calibrate an item bank to measure physical function (PF) in Singapore, a multi-ethnic city in Southeast Asia. Methods We recruited participants from community and hospital settings stratified for age and gender, with and without medical conditions to administer candidate pool of 61-items derived from the people's perspectives. We calibrated their responses using Samejima's graded response model of item response theory (IRT), including model assumptions, model fit, differential item functioning (DIF), and concurrent and known-groups validity. Results 496 participants (50% male; 41% above 50 years old; 33.3% Chinese, 32.7% Malay and 34.1% Indian; 35% without chronic illness) were included in the calibration of item bank. 6 items were excluded due to mis-fit and local dependence. Redundancies in the response level was collapsed and re-scoring, while preserving the 5-level response structure. We found the final 55-item PF bank had adequate fit to IRT assumptions of unidimensionality, local independence and monotonicity. Items generally showed discernible ceiling effects with latent scores between - 3.5 to + 1.5. We found no DIF with gender, ethnicity or education. The PF scores correlated in the hypothesized direction with self-reported global health (Spearman's rho = - 0.35, 95% confidence intervals - 0.43 to - 0.27) and discriminated between groups stratified by age, gender and medical conditions. Conclusion The 55-item Singapore PF item bank provides an adequate tool for measuring the lower end of PF, with greatest potential utility in healthcare settings where restoration to normal physical functioning is the goal of intervention.