Make evidence-based medicine individualized: synthesizing value-weighted composite endpoints in clinical trials

ID: 

309

Session: 

Poster session 3

Date: 

Tuesday 18 September 2018 - 12:30 to 14:00

All authors in correct order:

Huang CJ1, Chiang CE2, Chen CH3, Cheng HM1
1 Center for Evidence-Based Medicine, Department of Medical Education, Taipei Veterans General Hospital, Taiwan
2 General Clinical Research Center, Taipei Veterans General Hospital, Taiwan
3 Department of Medicine, National Yang-Ming University, Taiwan
Presenting author and contact person

Presenting author:

Chi-Jung Huang

Contact person:

Chi-Jung Huang
Abstract text
Background: Many researchers have recognized the importance of assigning distinct weights to individual adverse events in a composite endpoint. However, an adequate weighting method has not been developed for meta-analysis.
Objectives: We developed an approach by utilizing two metrics to calculate weighting values for deriving the overall effect on a composite endpoint, which was used further in the modeling of meta-analysis.
Methods: We conducted a meta-analysis of the aggregative data from nine randomized controlled trials comparing vitamin K antagonists (VKAs) with non-VKA oral anticoagulants (NOACs) in patients with nonvalvular atrial fibrillation. We adopted two weighting metrics including the disability-adjusted life-years (DALYs) developed by the WHO Global Burden of Disease Project and clinical endpoints preferences surveyed by questionnaire. We calculated the value-weighted composite endpoint consisting of myocardial infarction (MI), ischemic stroke (IS), intracranial hemorrhage (ICH), gastrointestinal bleeding (GIB), and all-cause death with corresponding weights. We used random-effects meta-analyses for pooling effect measures, expressed as relative risk (RR) with 95% confidence interval (CI). We addressed the uncertainty in the assignment of weighting values using the bootstrap estimates of treatment effects with 1000 resamplings.
Results: For DALYs metric, GIB had the smallest weight relative to death, followed by MI, IS, and ICH. While subjects assigned approximately equal weights to MI, IS, and ICH, the lesser weights were assigned to death and GIB. After performing our weighting procedure using DALYs, NOACs were significantly associated with a reduced risk for composite adverse events (RR 0.87, 95% CI 0.83 to 0.91), showing a slightly larger effect of treatment compared with that without using weights (RR 0.90, 95% CI 0.85 to 0.94). However, the risks were comparable between procedure with or without using the weights of subjects preferences.
Conclusions: By giving different weights to the individual endpoints, we can evaluate the pooled effect of different NOACs on composite endpoints and also take into considerations of patients’ value.
Patient or healthcare consumer involvement: Patients’ viewpoints on different endpoints could have substantial influences on the meta-analysis through our present proposed new method of meta-analysis.

Relevance to patients and consumers: 

Patients’ viewpoints on different endpoints could have substantial influences on the meta-analysis through our present proposed new method of meta-analysis. By giving different weights to the individual endpoints, we can evaluate the pooled effect of different treatment strategies on composite endpoints and also take into considerations of patients’ value.