To develop a new index for disease activity in ankylosing spondylitis (ASDAS) that is truthful, discriminative and feasible, and includes domains/items that are considered relevant by patients and doctors.

Eleven candidate variables covering six domains of disease activity, selected by ASAS experts in a Delphi exercise, were tested in a three-step approach, similar to the methodology used for the disease activity score in rheumatoid arthritis. Data on 708 patients included in ISSAS (International Study on Starting tumour necrosis factor blocking agents in Ankylosing Spondylitis) were used. Cross validation was carried out in the OASIS cohort (Outcome in Ankylosing Spondylitis International Study).

Principal component analysis disclosed three factors with eigenvalues >0.75: patient assessments, peripheral joint assessments and acute phase reactants. Discriminant function analysis resulted in a correct classification in approximately 72% of the cases (prior probability approximately 50%). Regression analysis resulted in an index with five variables (total back pain, patient global assessment, duration of morning stiffness, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR)). Three additional candidate indices were designed using similar methodology while omitting either ESR or CRP or patient global assessment. All four scores correlated with the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI; r = 0.67-0.80), patient (0.58-0.75) and physician's global assessment (0.41-0.48) of disease activity. All four candidate ASDAS indices performed better than BASDAI or single-item variables in discriminating between high and low disease activity state, according to doctors as well as patients in the OASIS cohort.

The first steps in the development of a new assessment tool of disease activity in AS derived four candidate indices with good face and construct validity, and high discriminant capacity.