MCC Metric of the Month Blog: Subject Retention


Applied Clinical Trials

This month, let's look at a quality metric that's useful for tracking both protocol and site performance: subject retention percentage.

MCC has 100+ standardized metrics spanning timeliness, cycle time, quality, efficiency, cost and risk.  This month, let’s look at a quality metric that’s useful for tracking both protocol and site performance: subject retention percentage.

Why this metric is important:  Keeping track of subject retention is, of course, critical to completing a study; enough subjects have to complete the protocol to achieve statistical significance. However, tracking retention percentage can also help to identify problems with the protocol or with specific sites. For instance, a high dropout percentage across all sites can point to onerous visit schedules, onerous procedures or concomitant medication problems. Retention problems at only certain sites can imply site-specific staff or training issues or even improper screening or randomization.

Definition:  The retention metric is calculated as the percent of enrolled subjects who remain in the study and did not voluntarily withdraw. Involuntary withdrawals and discontinuations are not counted as withdrawals for the purposes of this metric. The metric can be calculated at the site, country, region and study level, and can be rolled up to the indication, therapeutic area or portfolio level.

How to calculate this metric:  The formula is simply the (number of subjects enrolled) minus the (number of subjects who voluntarily withdraw) divided by the (number of subjects who enroll).

+2% of the planned retention percentage is a good target value.  Industry benchmark data suggest that on average, across all protocol phases and therapeutic areas, you can expect an 18% drop out percentage (1).

Example:  In the graph below, you can see the retention percentage for a protocol over time.  By June, retention is declining significantly, and an additional decline occurs in October.  A look at the graph on the right (retention by site for October) reveals the reason. 

Country A is doing well, and Country B is doing fairly well.  However, Country C sites are all having problems (perhaps a local standard of care issue?) and sites C3-C5 are faring particularly poorly compared to their counterparts. Investigation of Country C sites and a comparison to Country A sites is immediately in order to ascertain the root cause issues driving the high voluntary drop out percentage in Country C.

What you need in order to measure this:  You need the following two things for each site (or study) at the end of each month or quarter:

  • the number of subjects enrolled

  • the number of subjects who have voluntarily withdrawn from the study.

What makes performance on this metric hard to achieve:  Performance can be hard to achieve on this metric due to complexity of the protocol, procedures, etc., or problems at individual sites.

Things that you can do to improve performance:  Once you are tracking this metric, the appropriate improvements are dependent on the trends that you observe. 

  • If the value of the metric is lower than expected for most or all sites, the metric is likely pointing to a systemic problem, such as the protocol “burden” for subjects (e.g., visit schedule frequency, invasive procedures, pill burden and other study inconveniences). Protocol changes may be required to improve performance. In this situation, it’s wise to look at sites that are performing particularly well on this metric to see if they are employing strategies that keep their retention high. Perhaps protocol modifications can be avoided by sharing these best practices with all sites.

  • On the other hand, if the value of the metric is low for only some sites, the metric is likely pointing to problems at specific sites. This metric could be an indicator of site quality problems that need to be addressed immediately.

  • If the retention percentage is better than expected, two possibilities arise: First, your study will probably end sooner and with possibly lower total enrollment than planned, since more subjects than expected are completing.  Second, this could be a signal of potential coercion. If things appear too good to be true, it’s important to explore the causative factors to ensure subjects are not being indirectly or directly pressured into staying in the trial.

Companion metrics:  Other metrics that you should consider in tandem with this metric include: the MCC Protocol Quality metric and its related tracking tool, the MCC Site Quality metric and its related tracking tool, and planned vs. actual screen failure ratio. 

Dave Zuckerman, CEO, Metrics Champion Consortium, [email protected]

Linda Sullivan, COO, Metrics Champion Consortium, [email protected]



(1)  Benchmark data from Clinical Performance Partners, Inc. and PhESi – 2012.

© 2024 MJH Life Sciences

All rights reserved.