RESUMO
Objective: Little is known about the measurement invariance (MI) of resilience instruments in cancer care. This study was designed to examine MI of 10-Item Resilience Scale (RS-SC-10) in Americans and Chinese with cancer using propensity score-based multidimensional item response theory (MIRT) analysis. Methods: A sample of 924 patients were enrolled in the Be Resilient to Cancer trial involving 1 hospital in America and 3 hospitals in China. Data were collected from the RS-SC-10 and Hospital Anxiety and Depression Scale. Propensity score matching and MIRT were performed to evaluate Differential Item Function. Integrated Discrimination Improvement and Net Reclassification Improvement were used to indirectly estimate the MI through incremental prediction ability of MIRT-based score over total score. Results: RS-SC-10 retained 10 items with monotonous thresholds and its original two-factor structure. Nonuniform Differential Item Function was recognized in Item 4 (P â= â0.0011, Δ%ß1 â= â4.15%) and Item 8 (P â= â0.0017, Δ%ß1 â= â5.99%). Net Reclassification Improvement ranged from 9.04% to 35.01%, and Integrated Discrimination Improvement ranged from 8.82% to 20.60%. Conclusions: Although partial MI has been identified between Americans and Chinese, RS-SC-10 remains a critical indicator to emotional distress in cancer care.