COMBINING A REAL-WORLD OBSERVATIONAL STUDY AND A NETWORK META-ANALYSIS OF RANDOMISED CONTROLLED TRIALS TO DETERMINE RELATIVE TREATMENT EFFECTS IN TYPE 2 DIABETES
Author(s)
Reason T1, Hankins M1, Lovato E1, Pitcher A1, Schlueter M1, Abrams KR2
1IMS Health, London, UK, 2University of Leicester, Leicester, UK
OBJECTIVES: Since clinical trials are not always generalizable to a real-world population the aim of this study was to compare relative treatment effects from randomised controlled trials (RCTs) to relative treatment effects from a real-world data (RWD) source. First line type 2 diabetes mellitus was chosen as a case study due to the large number of trials conducted in this area and the importance of appropriate diabetes treatment to budget holders with aging populations. METHODS: A retrospective cohort study using UK electronic medical record data from The Health Improvement Network (THIN) was conducted in parallel with a network meta-analysis (NMA) of RCTs identified from a previously conducted systematic literature review. Patients and studies of patients with type 2 diabetes mellitus on first line monotherapy with either metformin or dipeptidyl peptidase-4 (DPP4) were included. The outcome studied was changed from baseline in HbA1c. Relative treatment effects were determined based on THIN using multi-level regression and propensity score matching. For the NMA, a Bayesian approach using baseline HbA1c as a covariate in meta-regression was employed. Relative treatment effects from both approaches were combined using the power prior method. RESULTS: The underlying populations from the two data sources were different in terms of baseline HbA1c and age. However, estimated relative treatment effects were consistent between the observational study and the NMA; both methods showed that metformin was generally superior to DPP4s in first line monotherapy. CONCLUSIONS: Due to the differences in underlying populations, relative treatment effects should ideally be estimated using both RCTs and RWD. Relative treatment effects from both sources can be combined using the power prior method. Health technology assessment agencies should consider combined evidence if available.
Conference/Value in Health Info
2016-10, ISPOR Europe 2016, Vienna, Austria
Value in Health, Vol. 19, No. 7 (November 2016)
Code
PDB7
Topic
Clinical Outcomes
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
Diabetes/Endocrine/Metabolic Disorders