<img alt="" src="https://secure.lote1otto.com/219869.png" style="display:none;">
Skip to content

The Model-Based Approach: A Better Way to Forecast Enrollment

8.12_SWR_Banner

 

To continue our Summer Weekend Reads series, Cytel presents “The Model-Based Approach: A Better Way to Forecast Enrollment.” Please click below to access the full publication.

The clinical phases of drug development represent the eagerly awaited period during which, after years of research and development, promising treatments are now ready for volunteer patients. But these Phase 3 trials are not only the most costly and time-consuming period of drug development, they also have difficult odds to overcome – less than two-thirds are successful. And the most common cause for incomplete Phase 3 trials? Enrollment.1

Indeed, as many as 37% of trials miss discontinuity enrollment targets, and 11% fail to enroll a single patient.2

Conventional approaches to projecting patient enrollment, such as rudimentary forecasting through the use of spreadsheets, however, may rely on simplistic assumptions about trial performance, such as that all sites will be ready to recruit from the same start date or that patients will arrive in a linear fashion. With a conventional approach, it can be difficult to move beyond these assumptions. A model-based approach, in contrast, can allow trial sponsors to incorporate more complex assumptions into their projections.

The Model-Based Approach

The model-based approach differs in that it captures two realities of the enrollment process: nonlinearity and randomness. The process is nonlinear because, for example, sites may open at different times, and the rate at which patients enroll may accelerate as more sites open. It is random because different numbers of patients will be recruited at different times, simply due to chance.

A model-based approach can easily accommodate these two realities, and once the model is set up using relevant inputs (factors that will affect enrollment), trial sponsors can team up with statisticians to simulate virtual runs.

Monte Carlo Simulation

The Monte Carlo Simulation technique provides a range of possible outcomes of a number of virtual runs and their associated probabilities. Virtual runs can be repeated (such as, 1,000 times) with different values each time within the constraints specified by the previously determined relevant inputs. This range of outcomes can help make realistic forecasts about uncertainties in patient enrollment.

To learn more about the model-based approach and its advantages, simulation outputs, and acceleration strategies, click below to download the publication:

Download publication


Endnotes

1 Kasenda, B., Von Elm, E., You, J., et al. (2014). Prevalence, Characteristics, and Publication of Discontinued Randomized Trials. JAMA, 311(10), 1045-1052.

2 Tufts Center for the Study of Drug Development. (2013). Impact Report, 15(1).


 

Read more from the Summer Weekend Read series:

New call-to-action

New call-to-action

New call-to-action

New call-to-action

New call-to-action

New call-to-action

 

If you'd like updates on our Summer Weekend Read series, sign up for email updates below.

contact iconSubscribe back to top