Background Computerized adaptive testing (CAT) provides the possibility to measure patient reported outcomes precisely at an individual level. As items are selected based on the previous answer of a patient, the number of items needed to achieve the same level of precision as obtained in a traditional questionnaire can be up to 40% less. Multidimensional CAT has the further advantage that simultaneous measurement of multiple dimensions increases the efficiency of the adaptive item selection procedure. Thereby comprehensive measurement across different aspects of fatigue becomes possible.

Objectives Aim of the study was the development of a multidimensional CAT for fatigue in rheumatoid arthritis (RA) whereby the perspectives of patients as well as modern psychometrics were included. Reporting on its construction is informative for health professionals since multidimensional CATs are still rare in health care and clear guidelines for their development are lacking.

Methods For the construction of a CAT, an item pool has to be developed and calibrated according to item response theory (IRT). Our item pool was based on interviews with patients and existing fatigue questionnaires. It was examined for content validity by a previous Delphi study with patients and professionals. The preliminary pool of 245 was examined with IRT and factor analysis to explore its statistical dimensionality structure and to fit a multidimensional IRT model. It contained 196 items and three dimensions of fatigue: severity, impact and variability of fatigue. Based on a functional script, and the item information gained by IRT analyses, software was constructed and provided as online application. Responses to the CAT were simulated for about 1000 virtual persons to determine start- and stopping rules.

Results Simulations showed that the standard error for each of the dimensions was acceptably low under the following conditions: test length of maximal 20 items, at least two random start items per dimension and at least five items per dimension. A low standard error is desirable as it indicates high measurement precision. The first version of the CAT was checked for programming problems. Shortcomings as non-functioning items with a “not applicable” category had to be adapted by the software specialist.

Conclusions This study provided the first version of a multidimensional CAT for fatigue in RA. In this presentation, further important issues for the construction of a multidimensional CAT will be discussed since the thorough development of measurement instruments is essential for precise and reliable assessment of patient reported outcomes. Only then, an instrument will be able to provide benefits for further research, possible interventions and/or the enhancement of the communication between patients and health professionals. Subsequently, a validation study will take place to examine the psychometric properties of the multidimensional CAT in more detail.

Disclosure of Interest None Declared