USING RANDOMISED CONTROLLED TRIAL (RCT) DATA AS EXTERNAL CONTROL TO OBTAIN COMPARATIVE EFFECTIVENESS IN A BAYESIAN NETWORK META-ANALYSIS MODEL
Author(s)
Martina R1, Prisco L2, Van Nooten F2
1University of Liverpool, Liverpool, UK, 2Dompé farmaceutici S.p.A, Milan, Italy
OBJECTIVES: In many disease areas there are no approved therapies. As a result, it is difficult to obtain comparative effectiveness results when developing first in class or novel compounds in these indications. The aim of this abstract is to describe comparative effectiveness results in an indication with limited publicly available comparative data. METHODS: A systematic review was performed to describe the treatment options available to patients. Treatments were included in fixed and random effects Bayesian network-meta-analysis models, using the vehicle response of two randomized clinical trials as external control to link the RCT results with those from normal clinical practice, which included RCTs, comparative studies and single arm trials. The magnitude of the vehicle response was varied to investigate the robustness of the findings. Additional analyses were performed using a single and two-way Bayesian power prior approach to simultaneously vary the size of the assumed vehicle response and the variability of the observed vehicle from the RCTs. Odds ratios for treatment effects, standard errors with credible intervals and treatment rankings were evaluated throughout. The Deviance Information Criterion (DIC) was used to evaluate the models. RESULTS: The fixed effects model performed the best throughout. The odds ratio (95% credible interval) for experimental therapy vs vehicle was 4.1 (2.0 7.7). Results of the comparison using RCT data only, all comparative studies and those from all available evidence, including data from single arm studies showed consistent results across all groups. The results of the power models confirmed that the fixed effects Bayesian model perfromed the best throughout. The treatment rankings remained consistent across all analyses. CONCLUSIONS: Using the RCT data and vehicle response to bridge the RCT results and those from external sources allowed for the indirect comparisons of treatments that would have otherwise not been possible. This provided comparative effectiveness results useful for health technology assessments.
Conference/Value in Health Info
2018-11, ISPOR Europe 2018, Barcelona, Spain
Value in Health, Vol. 21, S3 (October 2018)
Code
PSS6
Topic
Clinical Outcomes
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
Sensory System Disorders