Estimating site performance (ESP): can trial managers predict which trial sites will fail to recruit? Results from an exploratory study
Brighton, UK

Speaker: Hanne Bruhn

This is a talk at the International Clinical Trials Methodology Conference (ICTMC), Brighton, UK, 6th – 9th October 2019.  Venue is the Hilton Brighton Metropole.

Parallel Session 9A – Making Trials More Efficient


Hanne Bruhn, Shaun Treweek, Anne Duncan, Kirsty Shearer, Sarah Cameron, Karen Campbell, Karen Innes, Dawn McRae, Seonaidh Cotton.


Multicentre trials provide key evidence underpinning healthcare practice. They are also hard work, and generally expensive. Some work and expense is devoted to sites that fail to recruit. Methods to identify sites that will recruit to target would be helpful, especially simple ones.  The ESP project asked trial managers (TMs) to predict which sites would recruit to target, and why.


Trial managers at CHaRT (Aberdeen’s CTU) were asked to predict whether sites they were setting up would recruit to target. Predictions were made after site initiation and were collected as a ‘Yes/No’ prediction plus reasons for the prediction.

After ≥ eight months of recruitment, all CHaRT TMs (not just those making predictions) were invited to a ‘reveal’ where predictions were presented together with sites’ actual recruitment. Individual TMs reflected on their predictions and the performance of sites. The prediction reasons from the forms and TM reflections were used in qualitative work to identify ‘red flags’ linked to correct predictions of recruitment failure.


Ten TMs made predictions for 56 site visits recruiting to eight trials. Trial managers’ sensitivity was 82% and their specificity was 32%, correctly identifying 65% of sites that would hit their recruitment target and 54% of those that did not.  Eight ‘red flags’ for recruitment failure were identified: previous poor site performance; slow approvals process; strong staff/patient preferences; the site recruitment target; the trial protocol and its implementation at the site; lack of staff engagement; lack of research experience among site staff; busy site staff.


Using the unguided prediction model from ESP we have developed a modified guided prediction tool based around the red flags.  We expect this to improve predictions but this needs evaluation.  We encourage trial managers working in any country to contact us about contributing to this much larger evaluation (ESP2).

Start typing and press Enter to search