Using data to predict patient recruitment - How to stay on time & exceed goals

Winning in patient recruitment means finding the right patients (in terms of inclusion and exclusion criteria) AND finding the right amount of patients.

As you know, 80% of trials fail to meet enrollment timelines; and 30% of Phase III trial terminations are due to recruitment issues.

Using data, and therefore, a predictive analysis can help your research site or organisation to figure out how to best reach your recruitment goal on time. On top, it can also help you to better understand how to retain patients in your trial.

As a first step, it is important that you know how many sites are involved, in which regions these sites are and what your patient profile looks like. If you have all these information, let data be your friend!

Collect historical data

Try to retrieve information about trials with similar patient profiles. Through this, you can investigate on what worked best to recruit this target group and what did not.

Also, collect information about how long it took to meet the recruitment quota and find out the recruitment issues. Through this approach, you can learn enough to make your trial recruitment a success.

Identify key variables

In trial recruitment you’re dealing with human beings. Humans can be unpredictable sometimes. Once you have all historical data collected, select key variables in these data points. These key variables are likely to influence the outcome you want to predict. During the recruitment process, keep a close eye on these variables and adapt your predictions accordingly.

Find correlations

In order to predict your patient recruitment even better, try to find correlations between key variables. Through this, you can create a predictive model. To validate the model, you should test it against a previous patient recruitment project. If the result from the model doesn’t match the actual data well enough, there is room for improvement.

Create a model with different scenarios

To be prepared for every event, run different scenarios in your model. Have a pessimistic model, an optimistic model and a realistic/most likely model. Like this, you can react immediately, if the pessimistic model might be the one that is taking over your study.

CitrusLabs uses a predictive data model in every step of the patient recruitment process. Through this, the CitrusLabs team is always maximizing the clients’ spend - regardless of the budget. If you’re interested to support your next recruitment effort with a predictive data model, get in touch.