Stepped Wedge Designs

Stepped wedge designs, on the surface, seem ideally suited to implementation because the intervention is eventually turned “on” at each of the sites. However, implementation is influenced by many internal and external factors, and is decidedly complex and timing of the decision to fully implement is an issue. We describe the utility of stepped wedge designs in the chapters Experimental Designs and Randomization Schemes and Designing With Implementation and Dissemination in Mind. Below, we describe the diffusion, dissemination, and implementation plan of the Lumbar Imaging with Reporting of Epidemiology (LIRE) trial, which uses a stepped wedge design, as well as a few of the unexpected complications that arose.

Case Example: Lumbar Imaging with Reporting of Epidemiology (LIRE)

The goal of LIRE: Determine if inserting epidemiological benchmarks (essentially representing the normal range) into lumbar spine imaging reports reduces subsequent tests and treatments.

The issues:

  • At one health system, despite buy-in from leadership, several individuals within certain clinics did not want the intervention. This had two effects. First, an individual radiologist could remove the intervention from a report during its creation. Since the dictation system defaulted to including the intervention information, and it required an extra step to delete the intervention text, this removal at the individual radiologist level happened relatively infrequently. Second, the leadership of a few clinics within the same health system wanted the text slightly modified and would not allow the intervention to be used until it was modified. Because this change required IRB approval, it took several months and resulted in a lack of adherence to random assignment during that time.
Strategy Details
Diffusion Any site with a radiology information system (RIS), radiology dictation system or EMR can automate the insertion of the prevalence information that constitutes our intervention. In fact, the text had already diffused to a limited extent at two health systems due to a publication in Radiology (McCullough et al. 2012). Investigators needed to ask these sites to stop using the intervention text prior to the start of the trial.
Dissemination Investigators worked closely with site PIs to individualize approaches as how to best introduce radiologists and primary care providers (PCPs) to the intervention. Since the intervention was “turned on” centrally, neither radiologists nor PCPs needed to alter their workflow. However, investigators needed to inform both groups prior to roll-out to garner their acceptance. They informed them through a variety of approaches including meetings with clinic leadership, staff meetings, and email notifications.
Implementation The stepped wedge design facilitates both implementation and sustainability because by the end of the study, the intervention is “on” at all sites.
Sustainability The decision whether to sustain the intervention varied by site. Two of the health systems do not have active plans to de-implement the intervention. Again, the stepped wedge design makes it easier for sites to sustain the intervention since this requires that they simply accept the status quo. One of the health systems is prudently waiting for the study results before deciding to keep the intervention. The final health system is in a unique position since they are changing their EMR vendor. After they switch systems, they would have to actively re-implement the intervention. This will require evidence that the intervention brings value, which in turn will require study results.




back to top

McCullough BJ, Johnson GR, Martin BI, Jarvik JG. 2012. Lumbar MR imaging and reporting epidemiology: do epidemiologic data in reports affect clinical management? Radiology. 262:941–946. doi:10.1148/radiol.11110618. PMID:22357893.


Zatzick D, Tuzzio L, Chambers D, et al. Dissemination and Implementation: Stepped Wedge Designs. In: Rethinking Clinical Trials: A Living Textbook of Pragmatic Clinical Trials. Bethesda, MD: NIH Health Care Systems Research Collaboratory. Available at: Updated August 22, 2017.