Description

Background and rational:

Proton pump inhibitors (PPIs) are the most frequent treatment of gastric acid related disorders. The use of PPIs is increasing, as well as concerns about their inappropriate use. Long-term use of PPIs has been associated with adverse events (e.g., nutritional deficiencies, osteoporosis, infections). Therefore, deprescribing (stopping or reducing dose) PPIs when they are not or no longer indicated is expected to benefit the patients. While general practitioners (GPs) and patients may be reluctant to deprescribing, studies suggest that clinicians would be more comfortable with deprescribing after additional training, and that patients' acceptance increases following recommendations from their doctor. This highlights the importance of providing GPs and patients with the necessary tools for safe deprescribing. In Switzerland, there is evidence of PPI use without adequate indication or dose.

In the pilot survey conducted within the current project, which involved 48 GPs, 55% of them reported often encountering patients who were prescribed inappropriate PPIs, and 58% expressed a desire for a PPI deprescribing guideline. Optimizing PPI use in Switzerland is needed and of interest. Therefore, the investigators aim to conduct a clinical trial to evaluate an intervention to guide deprescribing of inappropriate PPIs. Additionally, alongside the trial, there will be an integrated process evaluation of the intervention, and a cost-effectiveness evaluation.

Study design:

This interventional, open-labelled, cluster-randomized controlled trial in Swiss primary care setting, aims to evaluate an intervention to guide deprescribing of inappropriate PPIs. In this trial, adult patients with inappropriate PPI prescription will be recruited by GPs in the German speaking part of Switzerland. Based on 1:1 cluster randomization of GPs, patients will be assigned to either the control group or to the group receiving an intervention to guide deprescribing of inappropriate PPIs. The control group will receive usual care. The investigators will compare the effectiveness and safety of the intervention with usual care, over a 12-month follow-up period. Alongside the trial, an integrated process evaluation of the intervention and a cost-effectiveness evaluation will be conducted.

Objectives

The main goal of this study is to evaluate an intervention to guide deprescribing of inappropriate PPIs. Therefore, the main trial aims to investigate if the study intervention leads to the deprescribing of inappropriate PPI prescription (i.e., effectiveness of the intervention), while ensuring noninferiority safety, in comparison to usual care. Additionally, the trial aims to investigate its impact on other clinical aspects, like the number of medications used, the health-related quality of life, and additional safety endpoints.

Alongside the main trial, an integrated process evaluation of the intervention aims to evaluate the quality of the implementation, as well as GPs' and patients' acceptance and fidelity to the intervention, patient typology, and the mechanisms supporting or hindering the success of the intervention.

Also, alongside the main trial, a cost-effectiveness evaluation aims to investigate the cost implications of the

Statistical considerations:

Statistical methods for the main trial The prescribed PPI dose will be quantified using the defined daily dose (DDD). To quantify its change the investigators will 1) estimate the average PPI dose over the 12 months of follow-up using the area under the curve divided by the time under observation and, 2) calculate the relative change from baseline as one minus the ratio between the average prescription and the baseline prescription level. The difference in the change in prescribed PPI dose between the two groups will be calculated using a linear model, adjusted for the baseline dose and including a random effect for the cluster. The difference in the change in upper gastrointestinal symptoms between the two groups will be calculated from a repeated-measures linear mixed-effects model, adjusted for the baseline value and including the intervention group, the timepoint, and the interaction between group and timepoint as fixed effects. Effects of the cluster and patient will be added as random effects. Both intention-to-treat (ITT) and per-protocol analyses need to meet non-inferiority to claim success for the co-primary safety endpoint.

Statistical analyses for other endpoints Repeatedly measured continuous secondary endpoints will be analyzed using the same model as for the safety co-primary endpoint. Count endpoints will be analyzed using a generalized linear mixed model and a negative binomial distribution, including the random effect of the cluster and the time of observation as an offset. Binary endpoints assessed at the end of the follow-up period will be analysed using mixed effects logistic regression, adjusted for the time under observation and including the random effect of the cluster.

Statistical methods for the integrated process evaluation of the intervention The process evaluation will be done based on the Medical Research Council's evaluation framework for complex interventions. Adherence to deprescribing decisions will be investigated using temporal dynamics modelling. Mechanisms of impact will be investigated based on the health action process approach, an established health behaviour change theory. The qualitative part will consist of semi-structured in-depth interviews with both patients and GPs.

Statistical methods for the cost-effectiveness evaluation Health economic analysis will be performed from a Swiss statutory health insurance perspective with a primary time horizon. QALYs will be estimated as the area under the survival curve resulting from utility estimates obtained for the different assessment timepoints during the trial. Utilities will be derived from EQ- 5D-5L questionnaire responses based on published valuation algorithms. A regression-based approach to assess intervention effects will be adopted due to the clustered nature of the trial data and possibility of residual baseline imbalances. The investigators will use Generalized Structural Equation Models (GSEMs), which will allow to simultaneously estimate incremental costs and QALYs (i.e., the intervention effects on costs and QALYs in the regression models), while accounting for the clustered nature of the data.