16
rheumatoid arthritis and osteoarthritis. Since most of the current generation of mea-
sures were developed according to the principles of classical test theory with a fixed
number of items and a scoring rule based on combining individual items, results can
only be compared across studies that used the same PRO. One of the challenges
facing the field of PRO outcome research in rheumatology is therefore to facilitate
the comparability of outcomes across settings and diseases. As was demonstrated in
this chapter, PROs can assess health-related concepts reflecting different levels of
consequences of rheumatic diseases. At the most fundamental level are impairments
or symptoms of disease. In some cases this may require disease-specific instru-
ments. However, disability and particularly participation restrictions are universally
relevant outcomes across all rheumatic diseases, and indeed across all diseases.
Because of their indirect association with the disease process, these outcomes typi-
cally do not require disease- or population-specific measures. Particularly for these
higher level outcome domains, IRT-based concurrent calibration of existing mea-
sures may facilitate the development of a common currency of domain outcomes
across rheumatic diseases. For instance, a common metric of physical function may
be built in which all physical function instruments that are used across rheumatic
diseases are calibrated in one IRT-based item bank. Once this has been achieved,
fully comparable physical function estimates may be obtained from any subset of
calibrated items. This system can then be used to compare results across (com-
pleted) studies and across diseases, irrespective of which instrument has been used.
Furthermore, it would allow researchers to select the most relevant questionnaire or
even to construct study-specific forms based on what is known about the disability
levels of the target population. Another useful feature of IRT-based item banking is
that various linking procedures exist that can be used to add new items to the item
bank, without altering the item characteristics of preexisting items. This can be
utilized to improve the information richness of the item bank in poorly represented
areas of the measured trait. Finally, due to the fact that PROs are increasingly col-
lected in electronic format, we foresee an increasing role for computerized adaptive
testing as a means to resolve the trade-off between feasibility and reliability that is
associated with the use of traditional fixed-length questionnaires.
References
- US Department of Health and Human Services, Food and Drug Administration, Center for
Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research
(CBER), Center for Devices and Radiological Health (CDRH). Guidance for industry: patient-
reported outcome measures: use in medical product development to support labeling claims:
draft guidance. Rockville, MD: Food and Drug Administration. 2006. http://www.fda.gov/
ohrms/dockets/98fr/06d-0044-gdl0001.pdf. - van Tuyl LHD, Boers M. Patient-reported outcomes in core domain sets for rheumatic dis-
eases. Nat Rev Rheumatol. 2015;11:705–12. - Garratt A, Schmidt L, Mackintosh A, Fitzpatrick R. Quality of life measurement: bibliographic
study of patient assessed health outcome measures. BMJ. 2002;324:1417.
M.A.H.O. Voshaar and M.A.F.J. van de Laar