Quantitative Assessment of Effect Modifiers in NMA - Leveraging Subgroup Data


Disher T
EVERSANA, West Porters Lake, NS, Canada

OBJECTIVES: Anchored indirect treatment comparisons provide valid estimates of comparative effectiveness in the absence of head to head trials under the assumption of no imbalance in effect modifiers across trials. Effect modifiers can be any aspect of a trial that would lead to changes in relative effects including differences in patient characteristics. Best practices suggest the need for robust evidence of effect modifier status prior to adjustment. We provide an overview of a quantitative approach to the assessment of effect modification that leverages sparsity assumptions to decrease false positives.

METHODS: Stochastic search variable selection is used in combination with weakly informative priors to assess for the presence of effect modification under either a strong or data-based shared effect modifier assumption. Under the shared effect modifier assumption the relevant summary is the pooled average value of the effect modifier on the relevant linear predictor scale. Network meta-analysis methods for contrast data are leveraged for assessment of the presence of relationships that violate the shared effect modifier assumption. Approaches are validated via simulation.

RESULTS: Both approaches are appropriately calibrated under their respective data generating model and lead to superior inference compared to either testing from a single trial or conducting comparisons using vague priors.

CONCLUSIONS: Evaluation of effect modifiers in anchored ITCs in disease areas where subgroup effects are commonly reported can benefit from meta-analyses leveraging sparsity inducing priors. These priors seem particularly relevant in this application since it is commonly assumed that there are only a small number of true effect modifiers across therapies and can help reduce issues related to multiple testing of under powered estimates. Areas for future development include use of shrunken effects in subsequent adjustments and leveraging multivariate models in areas where there is substantial missing subgroup data.

Conference/Value in Health Info

2024-05, ISPOR 2024, Atlanta, GA, USA

Value in Health, Volume 27, Issue 6, S1 (June 2024)




Clinical Outcomes, Methodological & Statistical Research

Topic Subcategory

Comparative Effectiveness or Efficacy, Confounding, Selection Bias Correction, Causal Inference


No Additional Disease & Conditions/Specialized Treatment Areas

Your browser is out-of-date

ISPOR recommends that you update your browser for more security, speed and the best experience on ispor.org. Update my browser now