Interitem Psychometric Validation of the Stanford Integrated Assessment for Transplant Scale Among Thoracic Transplant Candidates

Abstract

Background Psychosocial evaluations are mandatory for transplant listing, however the methodology for creating psychosocial risk stratifications is unclear. The Standford Psychosocial Integrated Psychosocial Assessment for Transplant Scale is the most commonly used instrument, however its interitem validity has never been examined. Objective To investigate the interitem validity of a psychosocial assessment tool for transplant candidates among a sample of thoracic transplant candidates. Methods Clinic data consisting of Stanford Integrated Psychosocial Assessment for Transplant administrations from 173 heart and lung transplant candidates were fit to a partial credit model. Data were subsequently fit to 4 separate partial credit models based on subscale categories, demonstrating the discrimination parameter estimate of each item. Differential item functioning analyses were conducted on the data within each subscale by sex to investigate potential bias produced by each item. Results The initial partial credit model using the full scale did not converge, indicating the subscales possibly did not measure the same underlying construct. Subscale discrimination parameter estimates demonstrated that most items were adequately or highly discriminative. The item measuring history of substance use demonstrated poor fit and differential item functioning. Conclusions While the Stanford Integrated Psychosocial Assessment for Transplant has demonstrated strong potential as a standardized framework for psychosocial assessments in transplant, this study identified some areas for improvement in the scoring system. The subscale scores appeared to show greater construct validity when utilized individually than when aggregated to form a total score. The substance use/abuse/dependence item did not fit well into its respective subscale. Future studies should aim to optimize the scoring system and re-asses its construct validity to improve its accuracy in discriminating between high-risk candidates and those needing psychosocial assistance.