OBJECTIVES: Item response theory (IRT) combined with computer adaptive testing (CAT) can improve assessment of self-reported measures by reducing measurement error and respondent burden. We previously used IRT to develop an item bank of health impact on role functioning (RF) in 3 areas (family, social, occupational). We tested the validity of a simulated 4-item CAT based on 10 items from our previously-developed bank in acute coronary syndrome (ACS) patients and compared its performance with 2 established SF-36 scales - role physical (RP) and role emotional (RE). METHODS: Adults (n=759) in Worcester, MA and Macon, GA were interviewed during an ACS hospitalization for the TRACE-CORE (Transitions, Risk, and Actions in Coronary Events: Center for Outcomes Research and Education) study. Ten questions from the RF item bank were administered; we assessed their construct validity using factor analysis.  We produced a 4-item CAT using simulations with the 10-item data. We compared score accuracy and ability to discriminate between groups based on levels of self-reported general health (SF-36), depression (PHQ9) and anxiety (GAD7) for the 10 items, the CAT and the SF-36 role scales. RESULTS: Average patient age was 61.1±11.6 years, with 64.8% men, 76% non-Hispanic white, 13% African-American. We confirmed the construct validity of the 10-bank items. The CAT improved measurement precision and expanded measurement range by 0.5 SD versus the SF36 RP. All measures discriminated across levels of general health, depression, and anxiety (p <0.0001). The instruments showed high agreement on the mean score for each subgroup, but the CAT scores had lower SDs. Relative validity coefficients suggest that CAT discriminated better than the SF36 RP and RE scales. CONCLUSIONS: Use of CAT improved assessment accuracy and discrimination among pre-defined groups, suggesting modern psychometric approach leads to practical advantages in assessment. Results will further be validated with clinical criteria and predictive analysis.