Background By modern treatment regimens, increasing numbers of patients achieve a state of sustained low disease activity or remission. Once disease control has been achieved, rheumatologists may decide to taper down or discontinue treatment. As no current guidelines exist with respect to de-escalation of treatment, large differences are expected to occur as to whether and in which patients they taper down medication in RA patients.
Objectives Objective of this study was 1) to identify patient characteristics that play a role in rheumatologists' decision to de-escalate treatment and 2) to assess whether and how rheumatologists differ with respect to the weights they attach to these factors using a discrete choice experiment (DCE).
Methods To construct the DCE questionnaire, first the relevant patient characteristics were identified first by interviewing 12 randomly selected rheumatologists practicing in The Netherlands. Subsequently, a DCE questionnaire with a highly efficient design was developed, consisting of a series of 16 patient profiles presented in pairs. All presented patients were assumed to be in a state of low disease activity or remission, using both a TNF-inhibitor and methotrexate. This survey was administered by 156 rheumatologist. From each pair, they were required to pick the profile they deemed most suitable for tapering medication, or choose not to taper at all (opt out). A conditional logit model was estimated to reveal how the likelihood that rheumatologists decided to taper medication depended on the patients' characteristics. A latent class analysis using a latent class conditional logit model was performed to identify subgroups of rheumatologists with respect to decision making for tapering.
Results Complete questionnaires were obtained from 156 rheumatologists. Overall, the opt out (taper neither patient of a pair) was chosen in 26% of cases. Compared to a patient having the more favourable characteristics (no swollen joints, longer remission duration, uncomplicated history, etc.), the presence of 2 swollen joints and a history of erosive disease combined with difficulties achieving remission were the most important factors influencing decision to taper. Latent class analysis revealed that willingness to taper and relative importance attributed to the various patient characteristics varied among rheumatologists. Willingness to taper was low in class 5, and high in classes 3 and 4. Swollen joints were important in class 1, 2 and 5, whereby class 1 also considered patient history, and class 5 remission duration. Decisions in class 3 are mostly driven by DAS and history. In class 4, none of the patient characteristics dominated the tapering decision.
Conclusions While the willingness to taper is generally high, doctors seemed to weigh patient characteristics differently in their tapering decisions. Swollen joint count and patient history are most important overall, but their weight into the tapering decisions varies strongly across latent classes, reflecting the lack of evidence. Policies could be directed towards promotion of tapering in patients where agreement across doctors is high, combined with evidence development to enable better decisions about the other groups.
Disclosure of Interest None declared