150 / 245
150
· DOS Abstracts
Appropriate Methods for Development, Validation,
and Use of Patient Reported Outcome Measures
Jonathan Comins, Michael Krogsgaard, Svend Kreiner, John Brodersen
Department of Rheumatology , Copenhagen University, Institute of Public
Health /University Hospital Zeeland; Copenhagen University Hospital Bispebjerg;
Copenhagen University, Institute of Public Health; Copenhagen University,
Institute of Public Health
Background:
Patient-Reported Outcome Measures (PROMs) in the form of
clinical questionnaires have become a fundamental component of healthcare
assessment today. Also in the realm of sports medicine and orthopedics, clini-
cians and researchers besiege their patients with PROMs, only to be burdened
with the extensive administration and interpretation of these measures. From a
clinical standpoint, PROMs are important because they measure health from the
perspective of the patient. However, in order to understand how a sum score
(a number), which has been derived from the responses to a group of questions
can be considered a measure of anything, it is necessary to consider the basic
principles of what measurement in fact is. Clinicians and clinical researchers us-
ing PROMs need to have a basic understanding of the purpose and application
of PROMs as measurement scales.
Purpose / Aim of Study:
The objective of this paper is to illustrate how Rasch
Item Response Theory (IRT) is the most appropriate method for constructing
and validating PROMs.
Materials and Methods:
We present an in-depth description of how questions
that are confirmed to be relevant and comprehensive for the targeted patient
group should be generated, and we show how the Rasch model is used to con-
firm statistically the measurement/scaling properties of these questions.
Findings / Results:
Rasch IRT is the only statistical method used to validate
PROMs, which satisfies the fundamental mathematical constraints of measure-
ment.
Conclusions:
If we are to use instruments to measure non-physical attributes
such as pain, self perceived function, or psychosocial consequences as primary
outcome measures in comparative studies, then the validation methods must be
as stringent as possible and should include Rasch IRT analyses.
No conflicts of interest reported
101.