Psychometric Validation of the KDQoL-36 Using Item Response Theory and Computer Adaptive Testing among Patients Receiving Dialysis

Background: Health-Related Quality of Life (HRQoL) is a critical outcome in end-stage kidney disease care. The Kidney Disease Quality of Life-36 (KDQoL-36) survey is used widely to measure HRQoL but is limited by length and lack of personalization. We evaluated the psychometric performance of a multidimensional Item Response Theory (IRT) model for KDQoL-36 and simulated a Computerized Adaptive Testing (CAT) version in a large European dialysis cohort. Methods: We included 8,325 KDQoL-36 responses from dialysis patients in 170 NephroCare clinics across six European countries and used confirmatory factor analysis and bifactor IRT modeling to assess psychometric properties. We conducted CAT simulations to evaluate reliability and potential for burden reduction. Construct and criterion validity were assessed using hospitalization outcomes and known-group comparisons. Results: Bifactor models demonstrated superior fit for both Short Form-12 (SF-12) and disease-specific KDQoL-36 domains, with excellent reliability (ω > 0.90). IRT-based scores showed high agreement with classical test theory (R²: 0.77–0.94) and differentiated patient groups by hospitalization risk (logistic regressions: OR = 0.90-0.94; p < 0.05). CAT simulations showed strong agreement with full-length scores (R²: 0.93–0.99) and reduced item count by 25%. CCAT-based administrations averaged 7.40 ± 2.53 items per patient for the SF-12 domain and 18.75 ± 4.09 items per patient for the disease-specific domains while preserving measurement precision. Conclusion: Multidimensional IRT modeling supports the KDQoL-36 as a reliable, valid measure of HRQoL in dialysis care. CAT implementation may offer substantial efficiency gains with minimal loss of information, reducing patient burden, thus enhancing the meaningful integration of HRQoL into routine hemodialysis care.