Systematic Literature Reviews
Although correct analytical methods are important, an evidence synthesis only as good as the evidence on which it is based. This evidence must be gathered by a process that is transparent, reproducible, unbiased, and comprehensive. A systematic literature review uses a pre-specified protocol to search all relevant sources of literature and appraise the information by a process that meets these criteria.
It is the first key step in evidence synthesis to acquire essential information for assessing relative effectiveness of interventions, rates of outcomes for cost-effectiveness models, or indeed any health economic question. We have extensive expertise in the use of systematic literature reviews incorporating meta-analyses, network meta-analyses, cost-effectiveness models, population adjusted indirect comparisons, and value of information analyses for your needs.
Meta-Analysis & Network Meta-Analysis
Meta-analysis is a method to synthesise multiple estimates of a quantity of interest (e.g. a treatment effect) into a single estimate. Network Meta-Analysis (NMA) is a method to indirectly compare two medical interventions that have not been compared directly in head-to-head trials; this is achieved by meta-analysing all available trials in a network of interventions.
Our expertise with meta-analysis and NMA is deep and wide ranging, covering disconnected evidence networks, supplementation of randomized controlled trial evidence with real world evidence, shared parameter models to combine different data types, survival outcomes network meta-analysis, and multiple outcomes network meta-analysis.
Population Adjusted Indirect Comparison
Population-Adjusted Indirect Comparison is a method to adjust for differences in patient populations when indirectly comparing interventions. These differences can bias other methods, such as network meta-analysis, if they pertain to treatment effect modifiers. We have extensive experience with both matching adjusted indirect comparison (MAIC) and simulated treatment comparison (STC) for adjusting for differences across populations. We also have experience using MAIC and STC to overcome disconnected evidence networks through so-called unanchored population-adjusted indirect comparison.
Real World Evidence
Randomized controlled trials (RCTs) are the gold standard of unbiased evidence on the efficacy of interventions in targeted populations. However, their results are not always generalisable to the real world population of general practice or to longer time frames. They also have limited use in estimating absolute, rather than relative, efficacy of interventions; absolute efficacy is necessary for cost-effectiveness modelling.
We have expertise in the use of registry, claims and cohort studies to strengthen RCT evidence, or networks of RCT evidence in network meta-analysis, when estimating relative and absolute treatment effects.
Cost-Effective Modelling
Cost-Effectiveness modelling, or more general health economic modelling, is the use of mathematical models to extrapolate and generalise evidence to wider populations and longer timeframes. We have expertise with all types of cost-effectiveness models, including decision tree, Markov, semi-Markov, and individual level simulation models, We have particular methodological expertise in structural uncertainty and Bayesian methods of multiparameter evidence synthesis.
Our models are developed in Excel, BUGS or R, depending on the model requirements, and we can support the latter with graphical user interfaces implemented in RShiny.
Value of Information Analysis
Value of information analysis is a method to quantify healthcare decision uncertainty from the payer perspective. Although its primary application is to estimate the value of further research, its use as a measure of payer uncertainty burden and in negotiating managed access agreements has been increasing. We have unique expertise in value of information analysis and have contributed to the development of methods to efficiently calculate it for complex cost-effectiveness models.
