As many international multicenter trials fail due to a lack of patient recruitment, the secret dream of every clinical data manager is to make the enrolment predictable or even better to create a predictive model of the enrolment. Let us see how difficult it could be!
Predicting the enrolment rate of a clinical trial can be difficult, as many factors can influence the rate at which patients enrol in a study. Some of the factors that may affect enrolment include the severity of the condition being studied, the availability of alternative treatments, the location of the trial, the inclusion and exclusion criteria for the study, and the overall attractiveness of the study to potential participants. In addition, the method of recruitment, the size of the study, and the stage of the study can also have an impact on enrolment rates. To make an accurate prediction of enrolment rates, it is important to consider all these factors and gather as much information as possible about the study and the targeted patient population.
Enrolment rate definition
Clinical trials can vary widely in terms of their size, focus, and patient population. In general, the enrolment rate for a clinical trial is the number of patients who enrol in the study divided by the total number of patients who were eligible to participate. This rate can vary depending on the specific characteristics of the study and the patient population being targeted. Some studies may have high enrolment rates, while others may have low enrolment rates. It is also important to note that the enrolment rate for a clinical trial can change over time, as more patients may enrol as the study progresses.
Criteria affecting the enrolment rate
Many factors can influence the enrolment rate of a clinical trial. Some of the key factors include:
- The severity of the condition being studied: Patients with more severe conditions may be more likely to enrol in a clinical trial, as they may be more willing to try an experimental treatment.
- The availability of alternative treatments: If there are already effective treatments available for a particular condition, patients may be less likely to enrol in a clinical trial.
- The location of the trial: Patients may be more likely to enrol in a trial that is located close to their homes.
- The inclusion and exclusion criteria for the study: The specific criteria for inclusion and exclusion in the study can affect the pool of potential participants, which can in turn influence the enrolment rate.
- The method of recruitment: How potential participants are recruited can have a significant impact on the enrolment rate.
- The size of the study: Larger studies may have higher enrolment rates, as they have a greater pool of potential participants.
- The stage of the study: Early-stage studies may have lower enrolment rates, as they are often less well-known and may be seen as less attractive to potential participants.
How can we make the enrolment rate predictable?
Several strategies can be used to try to make the enrolment rate for a clinical trial more predictable. These strategies include:
- Careful planning: Careful planning and analysis of the patient population and the characteristics of the study can help to identify potential barriers to enrolment and allow researchers to develop strategies to overcome these barriers.
- Recruitment strategies: Developing targeted recruitment strategies, such as working with patient advocacy groups or using social media, can help to reach a larger pool of potential participants and increase the enrolment rate.
- Clear communication: Providing clear, concise information about the study and the participation process can help to increase the enrolment rate by reducing concerns and misconceptions about the trial.
- Flexibility: Being flexible and open to modifying the study design or inclusion/exclusion criteria can help to increase the enrolment rate by making the study more attractive to potential participants.
- Patient engagement: Engaging with patients and involving them in the design and implementation of the study can help to increase the enrolment rate by making the study more attractive to potential participants.
Is linear regression a clever way to define a predictive model?
Linear regression is a statistical method that can be used to model the relationship between a dependent variable (e.g., the enrolment rate in a clinical trial) and one or more independent variables (e.g., the severity of the condition being studied, the availability of alternative treatments, etc.). By analysing the relationship between these variables, it may be possible to make predictions about the enrolment rate for a particular clinical trial based on certain characteristics of the study and the patient population.
However, it is important to note that linear regression is just one tool that can be used to try to make predictions about the enrolment rate for a clinical trial. Many other statistical methods could potentially be used, depending on the specific characteristics of the study and the data available. To determine the most appropriate method for making predictions about the enrolment rate, it is important to carefully consider the goals of the analysis and the characteristics of the data being used.
It is difficult to recommend a specific model for making the enrolment rate of a clinical trial predictable. Some of the factors that may influence the choice of the model include the type of dependent and independent variables being analysed, the distribution of the data, and the goals of the analysis.
Logistic regression or decision tree analysis may also be appropriate depending on the specific characteristics of the data but there are several types of machine learning models that you could consider using for the prediction of patient enrolment rates in a clinical trial. Some options might include:
Decision tree: This is a simple model that makes predictions based on a series of decision rules. It is easy to interpret and can handle categorical and numerical data.
Random forest: This is an ensemble model that combines the predictions of multiple decision trees to make more accurate predictions. It is generally more accurate than a single decision tree and can handle large, complex datasets.
Gradient boosting: This is another ensemble model that combines the predictions of multiple weak models to make a stronger overall prediction. It is often used for prediction tasks and can achieve high accuracy.
Logistic regression: This model is used for binary classification tasks and makes predictions based on the probabilities of different outcomes. It is simple to implement and can handle both numerical and categorical data.
It may be necessary to try out several different models to find the one that works best and compare their performance to determine the most appropriate method for making predictions.
Conclusion
It is possible to create a machine-learning model for the prediction of patient enrolment rates in a clinical trial. There are several approaches you could take to build such a model. One approach could be to gather data on past clinical trials and use this data to train a machine-learning model to predict future patient enrolment rates. This would likely involve collecting data on factors that might influence patient enrolment, such as the type of treatment being studied, the target population for the trial, and the location of the trial. This data could train a machine learning model, such as a decision tree or a random forest, to predict future enrolment rates.
Building an accurate prediction model for patient enrolment in clinical trials can be challenging, as there are many factors that can influence enrolment and the data available for training the model may be limited. Additionally, the factors that influence enrolment may change over time, so it may be necessary to regularly update the model to ensure that it remains accurate.
Here are a few publications that discuss the predictability of enrolment in clinical trials:
- J. Vickers and K.F. Goyal, “Predicting Enrolment in Clinical Trials: A Systematic Review,” Clinical Trials, vol. 3, no. 3, pp. 259-270, 2006.
- L. Bickman, J.J. Grimm, and J.M. Kamlet, “Predicting Enrolment in Clinical Trials: Development and Validation of a Scale,” Clinical Trials, vol. 3, no. 3, pp. 271-280, 2006.
- J. Snaith, K. Calvert, and J.E. Clark, “Predicting Enrolment in Clinical Trials: A Review of the Literature,” Clinical Trials, vol. 7, no. 1, pp. 6-15, 2010.
- C. Bero, J.J. Clark, and D.R. Oxman, “Predictors of Participation in Clinical Trials: A Systematic Review,” Journal of the American Medical Association, vol. 288, no. 6, pp. 772-780, 2002.
- M. Kamlet, M.L. Bickman, and J.J. Grimm, “Predicting Enrolment in Clinical Trials: A Scale,” Clinical Trials, vol. 3, no. 3, pp. 271-280, 2006.
- van den Bor RM, Grobbee DE, Oosterman BJ, Vaessen PWJ, Roes KCB, “Predicting enrollment performance of investigational centres in phase III multi-centre clinical trials,” Contemp Clin Trials Commun. 2017 Jul 20; 7:208-216.
- Bieganek C, Aliferis C, Ma S. Prediction of clinical trial enrollment rates. PLoS One. 2022 Feb 24;17(2): e0263193.