Maxime Verhoeven

206 Chapter 11 Patients Detailed information on patients and obtained data has been reported elsewhere. 8 For this study, a rheumatologist (AAAW) assessed the medical records for the treating rheumatologist’s clinical impressions on disease activity. Methods and results of the assessments for this clinical classification are provided in the Supplementary file, Data S1. In short, ‘active’ was characterised by initiation of a new disease modifying anti- rheumatic drug (DMARD), or an increase of the dose of current DMARD, or intramuscular glucocorticoid injection, or the treating rheumatologist’s notes indicating active disease, all at the consultation. ‘Inactive’ was characterised by tapering or stopping the current DMARD because of absence of disease activity (not because of adverse effects) or the treating rheumatologist’s notes indicating inactive disease. A randomly drawn subset of 39 clinic visits assessed by AAAW was blindly re-assessed by two other rheumatologists (JT and JWGJ); agreement on the clinical classification was overall 92.5%. Statistical analyses Continuous variables were described using means with standard deviations (SD) or medians with interquartile ranges (IQR), where appropriate. Frequencies and proportions were calculated for categorical variables. DAS28 was plotted against DAS-OST in a scatter plot, and agreement of DAS28 with DAS-OST was tested using a random one-way intra correlation coefficient (ICC). The mean of the differences between DAS28 and DAS-OST was illustrated in a Bland- Altman plot. The value DAS-OST and DAS28 to classify RA as active or inactive was established applying logistic regression analyses and calculating AU-ROC, using the rheumatologist’s clinical classification as reference standard. Similar analyses were performed with OST- scores only, to evaluate the additional value of OST in DAS-OST. AU-ROCs were tested for statistically significant differences. For remission (≤2.6) and low disease activity (LDA, ≤3.2), for both DAS28 and DAS-OST, sensitivity, specificity, positive- (PPV), negative predictive values (NPV) and accuracy were calculated. Youden’s index was used to define optimal cut-offs for OST-scores, for inactive disease for males and females separately. Agreement between remission and LDA according to DAS28 and DAS-OST and the rheumatologist’s clinical classification (reference standard) was tested using Cohen’s kappa and Gwet’s AC1. 10 All tests were two-sided, a p-value ≤0.05 was considered as statistically significant. Analyses were performed in SAS version 9.4, NCSS version12.0.12, R version 3.6.2 with the package pROC 1.16.1, and AgreeStat version 2015.6.1.