Psychometric validation of ICF Rehabilitation Set-17 through item response theory and network analysis: implications for standardized functioning assessment in Chinese inpatient rehabilitation.

Abstract

Background: The International Classification of Functioning, Disability and Health (ICF) offers a comprehensive biopsychosocial model widely used to capture functioning and disability status. The development of the national assessment standard regarding the ICF-Rehabilitation Set (ICF-RS)-17 represents the first step in implementing value-based healthcare (VBHC) in China. However, studies relating the ICF to managing goal setting and guiding VBHC are scarce.

Aim: This study aims to investigate the psychometric properties of the ICF-17 and further tailor goal settings in inpatient rehabilitation settings.

Design: A multi-centered, cross-sectional study.

Setting: Three general hospitals and two rehabilitation hospitals in Jiangsu Province, China.

Population: Patients received inpatient rehabilitation treatment.

Methods: We administrated ICF-RS-17 and Barthel index to assess inpatient patients' function, and quality of life with the Short Form-12 (SF-12). We performed a non-parametric and parametric Item response modeling (IRM) pipeline as a psychometric measurement. We further constructed graphical modeling (GM) to visualize the network structure of functional categories and calculate the treatment benefit index (TBI).

Results: The 2-parametric logistic model (2PLM) emerged as the optimal model generating a unidimensional scale with 17 ICF items (9 'd - Activities and Participation', 8 'b - Body Functions'). The final scale presented strong reliability with a latent class reliability coefficient=0.963, indicating great internal reliability. The estimated individual functional competence by the 2PLM model was moderately related to the index score of the PCS (p = 8.68 × 10^-114, ȓPearson = 0.52) and MCS (p = 7.41 × 10^-45, ȓPearson = 0.34), as well as a strong positive correlation with the BI (p = 7.41 × 10^-45, ȓPearson = 0.81). The findings also demonstrated measurement equivalence of the final model for individuals with different gender and across age groups as well as different measurement points (e.g. admission and discharge). Moreover, the GM indicates "d510 Washing oneself" possesses the most potent alleviating effect in the functional network (perturbance power=1.019).

Conclusions: The IRM-verified ICF-RS-17 tool has good construct validity and internal consistency for assessing the inpatient functional level. Moreover, enhancing "d510 Washing oneself" ability demonstrates the highest benefit for overall functional performance.

Clinical rehabilitation impact: The IRM might be helpful in facilitating person ability-matched goal setting by comparing person abilities and item difficulties along the same logit scale. GM visualizes the spread and perturbance power of functional limitations within the network model, which may manifest process-oriented goals within a complex network of functional tasks. The combination of IRM and GM might aid in transitioning towards VBHC and optimize patient care.