Site selection visits - what still matters in a hybrid world
As decentralised and hybrid trials become more common, site selection has had to evolve. Traditional visits focused on physical infrastructure: sample storage, exam rooms, IT equipment. Today, that baseline is still necessary but no longer sufficient. In studies with remote monitoring, digital data entry, and fewer in-person visits, what matters most is how a site operates digitally and how responsive its team is.
Recent research backs up that shift with real data rather than intuition. A machine learning model trained on 460 sites across 42 trials for a rare cardiac disease, and externally validated on 761 sites across 89 trials for an entirely different condition, achieved strong predictive accuracy for site risk using factors far removed from a physical equipment checklist. The strongest predictors were data quality risk, screen failure rate, and enrollment risk, in that order, and the ranking of what mattered held up consistently across two very different disease areas. None of those three predictors are things a walk-through of exam rooms and sample fridges would ever surface.
Beyond the equipment checklist
Modern site assessment (whether in-person or virtual) should go further than confirming facilities. It should explore how well a site can handle the demands of a digital-first study:
- Are staff comfortable using electronic case report forms, or do they rely on paper first and enter data later?
- Is internet access stable enough for video visits or live screen-sharing sessions?
- How does the site support participants who struggle with eConsent or app-based tasks?
- Are coordinators familiar with resolving queries electronically rather than waiting for a monitor visit?
These questions reveal whether a site is likely to operate smoothly once the study starts, not just whether it looks good during the selection process, and they map closely onto the data quality and enrollment risk factors the research above found most predictive.
What virtual walkthroughs can reveal
Remote site selection visits using screen sharing, guided Q&A, or video tours are now common. When done well, they can surface:
- How responsibilities are divided between staff members
- Whether there is a documented process for uploading and verifying data
- How the team has handled remote-first protocols before
- What kind of ongoing participant support they are prepared to offer
Asking a site to demonstrate how they would complete an eCRF entry or resolve a flagged query can be more informative than a standard checklist.
The human factors still count
Technology is only half the equation. A site can have good connectivity and all the right systems but still underperform if the team is stretched, poorly organised, or resistant to digital processes.
Signs to look for:
- Do they ask specific, informed questions about remote workflows?
- Are they willing to be transparent about past delays or challenges?
- Is there clear ownership of the study among named individuals, rather than diffuse responsibility?
Good site selection is about fit: not just capability but culture. You want teams who see value in the remote model and are motivated to make it work, not ones who are tolerating it.
What still matters from traditional selection
Despite the digital shifts, fundamentals remain:
- Can they recruit from the intended population, through channels that work for the study?
- Are they current with ethics and regulatory submissions?
- Is the principal investigator genuinely engaged, or primarily a figurehead?
These have not changed. What has changed is how they interact with the delivery model. A site with a strong track record of walk-in recruitment may struggle if participants in this study are enrolled online and screened remotely. That mismatch is worth exploring during selection rather than discovering three months in, and it's precisely the kind of enrollment-related risk the predictive model above flagged as one of its three strongest signals.
Where the data points selection efforts
If a team's time for site qualification is limited, the research above suggests a useful order of priority: dig hardest into a site's historical data quality patterns and screen failure rates first, since those carried the most predictive weight across two very different disease areas. Equipment and infrastructure checks still matter, but they're the easier, more visible part of the assessment. The harder, more consequential part is asking a site to show its actual track record on the things that don't show up on a facilities tour.
Site selection is where you begin to identify real operational partners. Making that visit count, whether it happens in a clinic or over a video call, tends to pay off through the rest of the study.