Use of evidence in cost-effectiveness decision modelling: synthesizing effectiveness evidence using individual participant data and aggregate data

Abstract: It is desirable to use all relevant sources to estimate effectiveness for use in a cost-effectiveness (CE) model.  Meta-analysis can be used to synthesise pair-wise comparisons of the same interventions. Mixed treatment comparisons (MTC) extends this framework allowing the simultaneous statistical synthesis of information on more than two interventions.  A proportion of this evidence, though, might be available at the individual participant level (IPD).

This paper describes existing evidence synthesis methods for CE assessment, and develops a series of novel statistical models aimed to fully exploit the existing data, regardless of whether available at aggregate - (e.g. study) or at IPD.   Bayesian statistical evidence synthesis models to estimate effectiveness parameters (from binary outcomes) were developed.  The proposed models allow the simultaneous analysis of IPD and aggregate data, together with study and individual level covariates.  The effectiveness of home safety education and the provision of safety equipments for the prevention of childhood injuries were used as motivating examples. 

Results show that including evidence at IPD level in the MTC synthesis is advisable; even when IPD are only available for a fraction of the studies forming the evidence base.  This will produce more precise estimates of patient-level covariate effects which may be valuable for estimating sub-group effects or adjusting for inconsistency in an MTC network.  Access to IPD when developing CE models is very valuable.