fetching data ... 
Background: ASDAS is a composite measure of disease activity for ankylosing spondylitis (AS). Its calculation includes three BASDAI questions, BASDAI Q2 (back pain), Q3 (peripheral pain/swelling), and Q6 (morning stiffness), plus ESR or CRP, depending on the version. In many patients' clinical records, BASDAI total score is included but the scores of its individual components are missing.
Objectives: Our aim was to assess the validity of imputing missing BASDAI items from the overall BASDAI score in order to calculate the patient's ASDAS in such a clinical situation.
Methods: Patients with complete baseline data for ASDAS and BASDAI components were identified from our local AS database. Original ASDAS calculations were repeated by replacing each or all of the BASDAI items included in the formula with the overall BASDAI score. Kappa statistics was used to assess the agreement between the original and the imputed ASDAS scores, in classifying patients into different disease activity states (inactive, moderate, high and very high). The mean difference between scores was calculated. The analysis was completed with the Bland–Altman method.
Results: 403 patients with AS (age 44.4±11.8 years; 73.7% males) with a mean disease duration of 9.7±8.1 years were included in this analysis. Their mean (SD) BASDAI, ASDAS-CRP and ASDAS-ESR scores were 3.6 (2.3), 3.1 (1.1), and 3.1 (1.1), respectively. Kappa values ranged from 0.80 to 0.89 (weighted kappa values from 0.84 to 0.96), with lower limits of the 95%CI all over 0.75 in the different imputation schemes (Table 1). The largest mean difference between the original and the imputed scores was 0.17.
Table 1. Agreement between the original ASDAS and imputed scores after replacing individual BASDAI item values with overall BASDAI scores *Mean difference and limits of agreement were calculated using Bland-Altman analysis.
Agreement level between the methods Limits of agreement between the methods* Kappa value 95% CI Mean difference Lower limit Upper limit ASDAS CRP vs ASDAS CRP imputed for three BASDAI items 0.8 0.74 to 0.85 0.07 −0.4 0.53 ASDAS CRP vs ASDAS CRP imputed for BASDAI item 2 0.8 0.75 to 0.85 0.17 −0.25 0.59 ASDAS CRP vs ASDAS CRP imputed for BASDAI item 3 0.88 0.84 to 0.92 −0.09 −0.38 0.2 ASDAS CRP vs ASDAS CRP imputed for BASDAI item 6 0.83 0.78 to 0.88 −0.02 −0.02 0.26 ASDAS ESR vs ASDAS ESR imputed for three BASDAI items 0.88 0.84 to 0.92 −0.02 −0.47 0.43 ASDAS ESR vs ASDAS ESR imputed for BASDAI item 2 0.89 0.85 to 0.93 0.12 −0.16 0.4 ASDAS ESR vs ASDAS ESR imputed for BASDAI item 3 0.82 0.78 to 0.88 −0.12 −0.49 0.26 ASDAS ESR vs ASDAS ESR imputed for BASDAI item 6 0.88 0.84 to 0.93 −0.02 −0.34 0.31
Conclusions: The close agreement between the original and the imputed ASDAS scores observed in this analysis support the validity of imputing missing BASDAI items from the overall BASDAI, when the score(s) of its individual component(s) are lacking.
Disclosure of Interest: None declared
DOI: 10.1136/annrheumdis-2015-eular.2504