Utilizing Multidimensional Computer Adaptive Testing to Mitigate Burden With Patient Reported Outcomes.

Area of Science:

• Health Informatics
• Psychometrics
• Medical Informatics

Background:

• Patient-reported outcome measures (PROs) are crucial for real-time health assessments.
• Current PRO systems, like the Patient Reported Outcomes Measurement Information System (PROMIS), use Item Response Theory (IRT) and Computer Adaptive Testing (CAT) for efficiency.
• Unidimensional CAT implementations administer separate question sets for each trait, potentially increasing patient burden due to correlations between traits.

Purpose of the Study:

• To develop and evaluate a Multidimensional Item Response Theory (MIRT) model and associated Multidimensional Computer Adaptive Testing (MCAT) software.
• To compare the efficiency of the developed MCAT approach against the standard unidimensional CAT approach for PROs.

Main Methods:

• Developed a MIRT model utilizing existing PROMIS item banks for depression and anxiety.
• Created MCAT software to administer items targeting multiple correlated traits concurrently.
• Compared the efficiency of the MCAT approach with the unidimensional CAT approach in simulated testing scenarios.

Main Results:

• The MCAT approach demonstrated potential for improved efficiency in assessing multiple correlated PRO traits.
• MCAT aims to reduce patient burden by selecting items that measure several traits simultaneously.
• The study successfully developed and tested an MCAT system using established PROMIS item banks.

Conclusions:

• Multidimensional CAT (MCAT) offers a promising advancement over unidimensional CAT for PRO assessment.
• MCAT has the potential to significantly reduce patient burden and improve the efficiency of collecting patient-reported outcomes.
• Further research and validation are warranted to fully integrate MCAT into clinical practice.