Of the three main phases of clinical trials, by far the most expensive part of the process is the Phase 3 trial, typically requiring from several hundred to several thousand participants and costing up to $20 million on average. Furthermore, the process is fraught with risks with up to 70% of Phase 2 trials and 50% of Phase 3 trials failing for various reasons (1). Thus, there is an imperative to find ways to reduce the risks and increase the success rate of trials.
In this post, our COO and Co-founder, Dr Laura Webber, discusses how microsimulation modelling can contribute to the critical decision on progressing from Phase 2 to Phase 3 of a clinical trial and help improve the chances of success.
Improved decision-making in the Phase 3 Clinical Trial process
Phase 3 clinical trials usually comprise two randomised, double-blinded, placebo-controlled trials in a large number of patients to measure statistically significant improvements due to a given product compared with a placebo. However, from the data collected in Phase 2 it is often far from clear on whether or not to proceed to Phase 3 – hence the high number of failures.
So what can be done to improve the likely success rate of a Phase 3 trial?
The role of modelling and simulation
Microsimulation models developed in conjunction with clinical trials are a valuable solution to assist decision-making because the model can use the detailed patient level data generated from the initial Phase 1 and 2 trials to make predictions about the efficacy and design of Phase 3.
By incorporating Phase 1 and 2 trial data along with epidemiological data within the model, the results provided can inform many aspects of the process. Microsimulation modelling can provide insights about whether the drug is economically viable by determining if there is sufficient clinical evidence for the drug under development to achieve cost-effectiveness, as well as optimising the actual design of the trial.
Specifically, we have undertaken modelling to answer questions such as:
A particularly strong capability of the microsimulation approach to modelling, of relevance to clinical trials, is the level of granularity that can be achieved, which enables any number of “health states” to be simulated e.g. no disease, stage 1 disease, stage 2 disease and so on, determined by the number of states captured in the primary individual-level data source. This level of detail greatly increases the reliability of the model and is a distinct advantage over models based on “aggregate methods” where the number of states is limited.
Another critical capability of the microsimulation approach is the ability to account for individual differences based on detailed epidemiological data such as dynamic risk factors and relative risks which, again, makes it much more granular than other modelling techniques based on averages. This is an important distinction in epidemiology, especially when considering non-communicable diseases which are strongly related to ageing and behavioural risk factors.
For example, the effectiveness and cost-effectiveness of a drug will manifest differently in a younger compared with an older population, because younger people will live longer with a disease, which will consequently increase costs to the health system (as they would need the drug for longer) but may save money to wider society by enabling that younger person to work for longer, thus reducing the amount of lost productivity. In an older population an individual may not benefit from the drug for so long. These various scenarios can be modelled to help inform decision-making.
Phase 3 and beyond: real world implementation
Modelling can also enable interpretation of the clinical trial data in the context of the wider population or other populations. Our model has been adapted to simulate the outcomes of clinical trials to appraise the impact (clinical, social and economic) of recommended prevention strategies or treatments before expensive, time-consuming real-world implementation.
For example, one of our studies compared the cost-effectiveness of 12- and 52-week interventions in patients in a randomised control for weight-loss techniques and modelled these over a 25-year time-period (2). Outputs of the model can also contribute to reimbursement dossiers and support marketing.
Microsimulation is highly flexible leaving open the opportunity for model extensions and adapting to new environments, such as adding additional diseases, risk factors or populations of interest. This is key if future disease prevalence is to be accurately predicted and if the medium- and long-term impact of new products is to be precisely assessed.
All of these capabilities of microsimulation modelling taken together can help to improve the design of clinical trials, run them more effectively, and focus the trial populations, saving time and money – with the ultimate goal of getting drugs to patients more quickly and at a lower cost.
1. Fogel DB. Factors associated with clinical trials that fail and opportunities for improving the likelihood of success: A review. Contemp Clin Trials Commun. 2018;11:156-64.
2. Ahern AL, Wheeler GM, Aveyard P, Boyland EJ, Halford JCG, Mander AP, et al. Extended and standard duration weight-loss programme referrals for adults in primary care (WRAP): a randomised controlled trial. Lancet. 2017;389(10085):2214-25.