Advancing the efficiency and efficacy of patient reported outcomes with multivariate computer adaptive testing

Area of Science:

• Health Outcomes Research
• Psychometrics
• Health Services Research

Background:

• The Patient Reported Outcomes Measurement Information System (PROMIS) developed patient-reported outcome (PRO) measures using unidimensional computer adaptive testing (UCAT).
• UCAT administers separate question sets for each trait, potentially increasing patient burden.
• Multidimensional computer adaptive testing (MCAT) can simultaneously assess correlated traits, offering a potential reduction in patient burden.

Purpose of the Study:

• To compare the effectiveness of MCAT versus UCAT in reducing patient burden for PRO measures.
• To evaluate the accuracy of trait estimates generated by MCAT and UCAT.

Main Methods:

• Developed one multidimensional item response theory (MIRT) model and three unidimensional item response theory models for anxiety, depression, and anger.
• Compared MCAT and UCAT performance using simulated individuals based on these models.

Main Results:

• Both MCAT and UCAT showed improved accuracy with more items for individuals aligned with item content.
• MCAT reduced the number of administered items by 50% compared to UCAT.
• Neither UCAT nor MCAT provided accurate estimates for individuals with low trait levels in the PROMIS Emotional Distress banks.

Conclusions:

• MCAT significantly reduces patient burden by administering fewer questions.
• For individuals targeted by specific PRO measures, MCAT and UCAT provide accurate trait estimates.
• MCAT holds promise for assessing a wider range of PROs with reduced patient burden.