Every clinical trial requires some manner of trial forecasting, normally for feasibility and patient enrollment. However, studies reveal that more than 50% of clinical trials fail to meet enrollment targets, and that enrollment is the most commonly cited reason for Phase 3 trial discontinuation .
Thankfully, the methods used to generate accurate forecasting tools are advancing. Improvements in simulation, prediction and modeling techniques mean that trial forecasting is becoming even more powerful. Developing a trial strategy that takes into account these high-precision forecasting tools means keeping the following three trends in mind:
- Data-Driven Strategies: It may seem obvious that historical data needs to play some role in trial forecasting, but if someone were to ask you how confident you are in your enrollment strategy, could you provide a specific number? Forecasting tools can now tell you the probability that you will achieve a certain enrollment milestone within a certain timeline, based on data gathered at particular sites in the countries and regions in which you are running a trial. This can help you spot potentially problematic sites early, and prepare for foreseen challenges.
- Model-Based Simulations: Model-based approaches to forecasting consider various relevant parameters of feasibility, (e.g. regions and countries of various enrollment sites, number of patients screened per week, etc.) Then, they set up a model of the trial based on how these inputs relate to each other. Powerful computer simulations take into account possible fluctuations at each site to calculate how likely it is that a study will meet the demands of the overall timeline. Strategically, this helps determine where to focus attention and resources. Whether or not to worry about a 70% chance of meeting enrollment at Site A may depend significnatly on whether Site B has a 20% of meeting its enrollment or a 90% chance.
- Accounting for Variability: Many forecasting tools give you a single-point probability estimate for the likelihood of success of an enrollment milestone. However, if there are many factors influencing enrollment which you either do not know or on which there is disagreement, a probability range may offer more strategic insight. For example, even if you can't be sure that you will reach a milestone with 40% probability, it may be useful to know that the chances are 35% - 45%. This will also appease members of your trial team, who may disagree on interpretations of data that are being put into the enrollment forecasting tool.
Related Items of Interest