Results
Patient Characteristics
All 98 patients completed all items of the SCQ. Characteristics of the patients are presented in Table 1 . The sample was similar to the 217 patients included in OASIS at baseline in terms of age, gender, symptom duration, disease activity and function (data not shown). The mean SCQ-score was 5.4 (S.D. 4.1) and the mean score for the mSCQ was 2.9 (S.D. 3.0). Table 2 shows the patients' responses on the SCQ and the frequency of comorbidities extracted from the medical records. Eighty-seven (88.8%) patients reported at least one comorbidity, of which 64 (65.3%) patients reported at least one non-rheumatic condition. With respect to the rheumatic conditions, 67 (68.4%) patients reported to have 'back pain', 19 (19.4%) patients reported 'OA' and 15 (15.3%) patients reported 'chronic rheumatic disease'. Within the last group, 11 (73.3%) patients had a history of peripheral arthritis, compared to 48 of 83 (57.8%) patients who indicated not to suffer from a chronic rheumatic disease (P = 0.26). At least one additional condition was reported by 43 (43.9%) patients and included, among others, IBD (n = 10), uveitis (n = 5), neurologic diseases (n = 7) and psoriasis (n = 1).
Criterion Validity
Table 2 also shows the agreement (κ) between patients' answers on the SCQ and medical records data. The agreement varied from slight (depression, ulcer or stomach disease and OA) to substantial or perfect agreement (heart disease, hypertension, diabetes, kidney-disease, anaemia and cancer).
The reliability analyses of the data acquisition from the medical records showed that both the intra- and inter-observer reliability for the specific items in the SCQ was substantial to almost perfect (κ 0.64–1.00), except for OA (κ 0.26 and 0.49 for intra- and inter-rater reliability, respectively).
Construct Validity
Table 3 shows the correlations of the SCQ and mSCQ with the Charlson index and Michaud–Wolfe index. The correlations between the instruments were low to moderate (0.24–0.57), but better for the mSCQ compared with the SCQ and better with the Michaud–Wolfe index compared with the Charlson index. The intra- and inter-observer reliability of data acquisition to calculate the Charlson and Michaud–Wolfe indices was very good (0.85, 95% CI 0.65–0.94) to excellent (0.91, 95% CI 0.78–0.96).
Table 3 also shows the correlations with the other constructs. All comorbidity indices correlated with age. The SCQ and mSCQ correlated both moderately with HRQoL and physical function, whereas for the Charlson and Michaud–Wolfe indices only weak correlations were found. The SCQ correlated weakly with some measures of disease activity, whereas the mSCQ did not. Strong associations were found between SCQ, mSCQ and the Michaud–Wolfe index with work disability.
The results of the regression analyses are shown in Table 4, Table 5 and Table 6. Assumptions of homoscedasticity and linearity were checked and were met. In multivariable analysis, the SCQ and mSCQ were significantly associated with the BASFI and the SF-36 PCS, whereas the Charlson and Michaud–Wolfe indices were not. A significant interaction between disease activity and both SCQ and mSCQ with work disability was found; therefore the models were stratified for level of disease activity in multivariable analysis. In patients with low disease activity (BASDAI <4, n = 46), comorbidity (SCQ or mSCQ) was the only variable which was significantly associated with work disability. In patients with high disease activity (BASDAI ≥ 4, n = 37), work disability was not associated with comorbidity. The Michaud–Wolfe index and Charlson index were not associated with work disability in multivariable analysis.