Page 136 - Recognizing axial spondyloarthritis - Janneke de Winter

CHAPTER EIGHT
mild to potential infections. For each scenario the participant could answer to what degree he or she wants to (hypothetically) initiate preventive treatment on a 5-point Likert scale (from ‘No’ to ‘Yes’). We also included a question to investigate the most important factor for participants to decline using preventive treatment (partly multiple choice, partly open). Furthermore, participants scored their perception of the severity of SpA, their own risk to develop SpA, and whether they are preoccupied with the thought of developing SpA on a visual analogue scale (VAS) from 0-100. We conducted think-aloud interviews with SpA patients, healthy volunteers with a wide variety of age and educational level, and medical doctors and nurses to verify realistic and clear scenarios and questions.
Statistical analyses
Baseline data are presented as numbers (%) (categorical data) or the mean/ median (SD/IQR/range) (continuous data) as appropriate. The data on treatment preference are shown as percentages and analyzed using the McNemar’s test, enabling to compare two scenarios as paired data. To prepare the data for that analysis, the ‘preference for treatment’ outcome variable was dichotomized. A preference for treatment (the answers ‘Yes’ and ‘I probably would’) was assigned a score of 1, and a neutral preference (‘I don’t know) or preference for non- treatment (‘I would probably not’ and ‘no’) was assigned a score of 0.
To test for possible interactions of willingness to use preventive medication with age, gender, HLA-B27 status and the presence of back pain throughout the different scenarios, we used a generalized estimating equations (GEE) model with a logit link, binomial distribution and an exchangeable correlation. GEE enables analysis of repeatedly assessed preference scenarios. It corrects for the fact that patients’ answers to each subsequent scenario are related to their answers in previous scenarios. Outcome measures of the GEE were odds ratios and 95% confidence intervals. We calculated whether age, gender, HLA-B27 status or the presence of back pain were significant predictors for (non)treatment preference.
We tested the correlation of disease perception (the own risk assessment of developing SpA) with the willingness to start using preventive medication in a linear regression model. We performed all analyses in SPSS version 24.0.
RESULTS
Study population and response
The study population has been described in detail earlier (7). Of all 130 Pre- SpA participants, 106 completed the survey (response rate 81.5%). Baseline characteristics are shown in Table 1. There were no missing values. Baseline characteristics between responders and non-responders did not differ (data not shown).
Evaluating participants’ believes and perceptions of SpA (VAS -100) showed that they were not occupied by the thought of developing SpA (median 23, range
134