Enrollment as a Learning Problem

The Tufts Center for the Study of Drug Development’s January/February 2013 Impact Report has stirred comment and criticism. Paul Ivsin of Placebo Control dismisses the significance of learning that 48% of sites miss their enrollment targets; Ed Silverman at Pharmalot does the reverse.  The biggest takeaway from my perspective is this:

“Study timelines typically extended to nearly double their original duration to meet desired enrollment levels for all therapeutic areas.”

Nearly doubling your timelines is dismal enrollment performance regardless of details in the report.  I can’t say I’m surprised to learn that industry enrollment performance remains dismal. There are a hundred whispered anecdotes about enrollment struggles for every one about enrolling on time.  People think poor enrollment is a fact of life because for them it has been.

One reason that enrollment problems persist for most in the industry is that everybody focuses on what happens before enrollment starts. Thorough research and planning are essential but only take you so far. Site and patient databases are important resources and site selection and training are important preliminaries. But improvements in these areas won’t solve most real-world enrollment problems. The starting point for enrollment is only that: a starting point.

Every slow-enrolling study is like one of the unhappy families that Tolstoy famously talked about in Anna Kerenina – it is unhappy in its own way.  It has its own unique enrollment problems depending on dozens of factors that preexisting information can’t address.  Regardless of historical performance, circumstances change at each site with each study. The standard of care evolves.  The lineup of competing studies changes.  Things keep changing during the study. For example, news items about problems with marketed drugs in the same class may scare subjects away.

That’s why 90% of the work that makes the difference between enrollment success and failure happens after the study is underway and patients are enrolling.  Systems and processes help, but above all, you have to realize that enrolling successfully requires learning fast, not following a rote process.  If your current study is missing enrollment targets, you may not be learning enough fast enough.


Share Button

Leave a Reply

  • (will not be published)