Article Text
Abstract
Background Radiographic damage is one of the core outcomes in Ankylosing Spondylitis (AS) and it is usually assessed with the modified Stoke AS Spine Score (mSASSS). Another score has recently been proposed, the Radiographic AS Spinal Score (RASSS)[1], which includes an additional assessment of the lower thoracic vertebrae. However, so far no further application of this proposed score has been reported, neither with respect to its feasibility, nor confirming its additional value compared to the mSASSS.
Objectives To assess the feasibility of mSASSS and RASSS.
Methods Spinal radiographs from patients included in the Outcome in AS International Study (OASIS) were used. Cervical and lumbar spine radiographs were taken every 2 years during a period of 12 years. Two readers independently scored the x-rays, including all the vertebral corners (VCs) to be assessed according to both scoring methods. Scores per VC were averaged between the readers. A missing value for a VC was considered when both readers assessed it as missing; otherwise the score of one of the readers was used. Only radiographs from both cervical and lumbar spinal segments were included when available with ≤3 missing VCs per segment. In cases with ≤3 missing VCs, the missing VCs were recorded as needing to be imputed. The availability of the radiographs for the calculation of each of the scores was expressed in percentage of the total number of available radiographs. The availability of individual thoracic VCs added in the RASSS was also investigated.
Results A total of 925 radiographs from the 217 patients initially included in OASIS were collected during the 12-year follow-up period. From those, 814 (88%) could be used for the mSASSS calculation and 637 (69%) for the RASSS calculation (both ≤3 missing VCs per spinal segment). In 177 out of 814 (22%) evaluable radiographs, all 4 additional thoracic VCs in the RASSS were missing, in 64 (8%) radiographs only 1 additional thoracic VC was available, and in 301 (37%) radiographs 2 additional thoracic VCs were available. In 365 out of 637 (57%) radiographs where the RASSS could be calculated, this could only be done based on 1 or 2 real scores from the 4 additional thoracic VCs, with the remaining missing VCs being imputed. These VCs were therefore uninformative.
Conclusions The RASSS does not seem to be a suitable instrument to assess radiographic damage in AS. Its calculation is frequently impossible or strongly based on artificial scores. Even though, it might capture additional information not reflected in the mSASSS, it shows a poor feasibility, an important aspect of the OMERACT filter to be accounted for when assessing an outcome measure.
Baraliakos et al. Arthritis & Rheumatism, 2009: 61, 764-771
Disclosure of Interest None Declared