What a real sIRB pilot found about speed and cost

Single IRB review for multi-site trials is usually pitched on speed: one review instead of many, less duplicated effort, faster approval across every site. A pilot study of the NIH's single IRB model, using the SMART IRB reliance framework, actually measured that speed claim against a real comparison, and found it holds up. It also found the fuller picture is more complicated than the speed argument alone suggests.

The speed gains were real and measurable

Sites relying on the single IRB model reached approval in a mean of 81 days, compared to 121 days for a site that didn't use the reliance model, a 40-day difference on a process that already takes months. That result also beat the published benchmark for academic medical centers generally, where average time to approval runs closer to 103 days. The relying sites weren't just faster than the specific comparison site in this study; they were faster than the broader field's typical performance.

That speed advantage carried through to a more consequential downstream measure: time to first participant enrolled. Relying sites reached first enrollment in 126 days, compared to 149 days for the non-relying site and 169 days against published averages. In a trial where every week before first enrollment is a week the whole study timeline shifts later, a gap of this size is operationally significant, not a marginal efficiency gain.

The costs didn't move the same direction

Here's where the pilot's findings get more interesting than a simple speed win. Actual costs of the reliance model turned out to exceed the projected estimates for traditional, local IRB review at each site. That's a genuinely counterintuitive result for a model whose central selling point is efficiency, and it's worth taking seriously rather than explaining away. Single IRB review consolidates the review itself, but it doesn't automatically eliminate the coordination overhead of managing reliance agreements, communicating between the central IRB and each local site, and handling the administrative machinery that a reliance model still requires. That overhead, on this pilot's evidence, was underestimated at the budgeting stage.

Satisfaction was mixed, and culture was named as the real obstacle

Nearly half, 47%, of stakeholders reported being very satisfied or satisfied with the reliance experience. That's a genuinely mixed result for a process that delivered a clear, measurable speed benefit. The gap between "objectively faster" and "only about half satisfied" points at something the speed numbers alone don't capture.

The researchers' own conclusion identifies what that something is: transitioning to single IRB oversight represents a shift in practice and culture for many institutions, not just a procedural change. Local IRB review, whatever its inefficiencies, is a familiar process that institutions have built their own internal expectations and relationships around. Ceding that review to an external central IRB, even one that demonstrably works faster, asks institutions to give up a degree of local control and familiarity that a purely time-to-approval metric doesn't account for.

Why both findings matter for planning a multi-site study

A sponsor or research network reading only the speed and enrollment figures would reasonably conclude single IRB review is an unambiguous improvement worth adopting everywhere. The fuller set of findings suggests something more calibrated:

• Budget for coordination overhead explicitly, rather than assuming centralised review is automatically cheaper. The gap between projected and actual costs in this pilot is a specific, documented warning against under-budgeting the administrative work reliance still requires.
• Plan for a cultural transition, not just a procedural one. The 47% satisfaction figure, alongside clear speed gains, suggests the barrier to a smooth rollout isn't really about whether the model works. It's about whether institutions and staff are prepared for what it changes about their role in the process.
• Communicate the reliance agreement's practical mechanics clearly to every site up front. A model that removes local IRB review but adds a different kind of coordination burden needs each site to understand what they're actually taking on before enrolment begins, not discover it as friction partway through.
• Track both timeline and satisfaction as separate success metrics. A model that hits every speed target while leaving half of stakeholders lukewarm has succeeded on one dimension and has real, addressable work left on another.

The honest takeaway

Single IRB review, on this pilot's evidence, is genuinely faster, meaningfully so, for both approval and first enrollment. It is not automatically cheaper, and it is not automatically embraced by the people who have to work within it. Neither of those complications is a reason to abandon the model. They're a reason to implement it with realistic expectations about where the friction will actually show up, budget and culture, rather than assuming a faster approval time settles the whole question.