With pragmatic research, there is no specific event at the end of a trial, such as regulatory approval, that signals the most appropriate dissemination and implementation pathway. Rather, when an intervention from a pragmatic clinical trial (PCT), such as those conducted by the NIH Pragmatic Trials Collaboratory, is shown to be beneficial, there are a multitude of different mechanisms for enabling uptake of the interventions in different healthcare settings.

“Sustaining any innovation is a complex undertaking—complex because implementation strategies (a) are typically multi-component and (b) must adapt to local contexts (Bauer et al. 2015).”

Health system leaders make decisions about adopting the new intervention and consider a multitude of factors, including current priorities of the healthcare system, how the intervention was received by staff, cost, and related (possibly competing) initiatives. In other words, health system leaders’ decision to adopt or sustain the intervention are impacted by factors other than trial results (Green et al. 2024).

In a series of case studies on implementation from NIH Pragmatic Trials Collaboratory trials, authors found that, in some instances, health system leaders did not adopt an intervention that had been shown to be effective by the trial’s criteria (e.g., improvement in primary outcome), citing changed clinical circumstances, prohibitive costs (Green et al. 2024), or the complexity of the intervention (Staman et al. 2023). In other instances, interventions were adopted as standard practice despite nonsignificant or null findings of trial’s primary outcome. Instead, adoption was motived by benefits in a subgroup, secondary or post hoc analyses or because of a perception that the intervention improved patient or staff satisfaction (Green et al. 2024; Platt et al. 2024). Additionally, changes to policy and guidelines have major impacts on what health system leaders prioritize for change.

Based on the discordance between the results of PCTs and implementation of those results in healthcare settings, even in the settings that championed the work, researchers should consider:

1. Including secondary outcome measures that are salient to health system partners
2. Collecting all appropriate data to allow for post hoc analysis of subgroups
3. Collecting experience data from clinicians and staff
4. Engaging policymakers before starting the trial (Green et al. 2024)

Platt and colleagues posit that to provide evidence that health system leaders will actually use, other changes to the way we design trials are necessary:

• Trials need to be faster (2-3 years)
• Trials should consider outcomes that health system leaders care about (such as cost and subgroup analyses)
• The evidence required for change should be at the level that health system leaders use (not necessarily rejecting the null hypothesis at P<0.05 but also considering results of Bayesian methods and interim analyses)