Computer Adaptive Testing for Patient-Reported Outcomes: Resources for Clinical Research

Abstract

Background and Purpose. The physical therapy profession actively continues to develop systems of standardized outcomes to provide evidence of treatment efficacy and to improve rehabilitation practices. The emergence of patient-reported outcomes (PROs) of perceived health outcomes and health-related quality of life has gained increased importance in clinical research and practice. The objective of this report is to provide an introduction to the development of PROs using item response theory and computer adaptive testing, and to provide an overview of the National Institutes of Health's Patient-Reported Outcome Measurement Information System (PROMIS). PROs are increasingly relying on computer adaptive testing (CAT), a promising technology that helps resolve the classic conflict between practicality and precision faced by traditional outcome measures. CATs require a set of items developed using Item Response Theory in combination with CAT software. Method/Model Description and Evaluation. Item Response Theory (IRT) methods for instrument development in combination with computer adaptive testing approaches for administration are advancing PROs in health care. Outcomes. PROMIS is a system of PROs based on IRT/CAT that are available to clinical researchers for adoption. Discussion. CATs can be integrated seamlessly with electronic data capture systems and electronic health records to provide a means to implement standardized systems of PROs across multiple sites. Conclusion. Health care research and practice systems are using these emerging technologies for developing and administering standardized health outcomes, and physical therapists' professional education should begin to include the necessary knowledge to enable them to become competent users and developers of PROs using CATs. Key Words: Computer adaptive test, Patient-reported outcomes, Standardized assessment. BACKGROUND AND PURPOSE The physical therapy profession has a critical need for standardization of measures as it continues to build evidence to support practice. Standardization of health and rehabilitation outcomes in this profession is an active focus of efforts by the American Physical Therapy Association (APTA). From a clinical service perspective, processes to monitor interventions for individual patients have become seriously fragmented as different hospitals, clinics, and programs select outcome instruments that cannot easily be compared across systems. Using an assortment of instruments also complicates credentialing and training for a large clinical staff. In many instances, clinicians simply lack the time for administration of the best instrument due to time constraints or length of the instrument. Enhanced monitoring supported by standardized outcomes should result in better program decisions and improved service delivery for patients and help in the identification of evidence-based clinical pathways. From a clinical research perspective, a uniform approach to health outcomes measurement could enhance the comparability of measures across clinical research sites and projects and improve the generalizability of study results.1 Although efforts are under way to develop clinically meaningful outcomes with low administration burden, few instruments or batteries of instruments have been accepted or adopted by the physical therapy profession for clinical practice or research. Evaluations of the effectiveness of medical care have conventionally focused on clinical end-points, including laboratory tests, imaging, and clinical exams measuring changes at the body structure and function level. However, many outcomes associated with the use of medical and surgical devices, medications, or interventions are known only to the patient. Patient reported outcomes (PROs) enhance clinicians' understanding of the impact of illness, health, and interventions from the patient's perspective. For example, symptoms (eg, pain, fatigue), activity behaviors and participation, perceived health, and health-related quality of life are all experienced by the individual, and these patient perceptions are often determined using PROs. …