Rethinking Routine Mammography Among Older Women: Tackling Issues of Health Care Overuse Through Interdisciplinary De-Implementation Science

August 20, 2021

For most translational research, the final challenge is getting widespread adoption and routine delivery of an evidence-based practice, treatment, or guideline in the settings and populations that would most benefit from them. But what happens when research suggests that something widely embraced and accepted isn’t helpful—or may even be potentially harmful—for a specific population? How can researchers convince providers and patients that something they were taught as essential is now no longer necessary? How can we change healthcare systems and policies that facilitate or reinforce this behavior or practice?

That’s the challenge three researchers across Columbia School of Public Health and Columbia Medical School have embraced as part of their National Cancer Institute (NCI) R01 grant “De-implementation of Mammography Overuse in Older Racially and Ethnically Diverse Women.” Researchers Rachel Shelton, ScD, MPH, Nathalie Moise, MD, MS, and Parisa Tehranifar, DrPH, are collaborating to tackle this complicated issue.

Reconsidering the Status Quo

Implementation science focuses on methods and strategies that help put approved, evidence-based treatments, guidelines, and programs into real-world practice. Translating research into practice isn’t as simple as proving its merit and effectiveness in improving health outcomes. At every level—policy, institutional, provider, patient—there are challenges that are difficult to overcome. Education, provider recommendation, and systems automation (e.g. reminder letters) are tools that have helped establish some practices, like breast cancer screening, as part of the societal norm.

But de-implementation—the reduction, replacement, or discontinuing of ineffective or low-value clinical practices—is emerging as an important research area. As health care costs and health equity become a more urgent concern and new research highlights the potentially inequitable harmful impact of these practices on patients and health care systems, advancements in de-implementation science can help inform real-world practice efforts to improve outcomes and reduce unnecessary procedures and treatments that have health and psychological consequences.

“This area is really a priority now with the National Institutes of Health,” says Shelton, who helps lead an Implementation Science Initiative at the Irving Institute for Clinical and Translational Research. “They had a NOSI, a notice of special interest, on de-implementation across all institutes, and the NCI in particular has a huge interest in advancing science in this area to understand what strategies and approaches can be used to facilitate de-implementation.”

“Most of the work that has been done in implementation science has been focused on implementation—how do you actually get things widely delivered in real-world settings that are messy and complex?” she says. “But for de-implementation, the challenge is once things are already institutionalized and within these systems, and there's reinforcement at these different levels, how do we disrupt those institutional processes and norms and get something removed?”

Expanding the Scope

For Shelton and Tehranifar, this was a project years in the making. According to Tehranifar, “We’d been working for years on studies of women of screening ages that focused on state policies around mammography results disclosure and guideline changes in breast cancer screening. However, we noticed that many women 75 and older continued to get routine, often annual, mammograms even though guidelines suggested insufficient evidence for benefits and harms of mammography, and recommended that the decision to screen should be based on life expectancy, overall health, and personal preferences. While I’d worked for many years on underuse of screening among underserved communities, we started to recognize that overuse of screening may also be present in the populations we were conducting our studies.”

They had initially considered submitting a smaller formative grant application to NIH. “Through the Implementation Science Initiative at the Irving Institute, we have a working group where people present their ideas and get feedback, similar to a Work in Progress,” explains Shelton. “Parisa and I presented this idea to the group, and Nathalie was one of the members of the working group.”

Moise’s feedback and perspective as a primary care provider helped transform the project into something larger. “Nathalie thought the idea sounded great, and she gave us great feedback,” notes Shelton. “She felt strongly that we couldn’t do this in a two-year grant, that this idea was really much more ambitious.” So they decided to wait—and to bring Moise on as a multiple principal investigator (MPI).

With Moise on board, the scope of the project grew. “We pulled in eight clinics Nathalie was working in, and it became a much larger and ambitious grant where we weren’t only developing strategies to impact de-implementation, but actually testing them,” recalls Shelton. “Nathalie was the provider and a clinical trials and behavior change expert. I was the qualitative mixed methods expertise, with a focus on the theoretical frameworks. And Parisa brought experience with large scale surveys and epidemiology and has already done a lot of work with breast cancer screening. And all three of us bring different expertise related to implementation science methods and frameworks.”

The combination proved to be highly effective. “We have different approaches, different experiences,” explains Moise. “And that’s where our collaboration was key. When you have three very different researchers together, thinking about every single aspect of a grant, it does more than foster innovation. It makes you write specifically in a way that helps the reviewer really understand the message that you're trying to relay.”

The collaboration helped in smaller ways as well. “Working like this, you're catching things that you wouldn't have if you were just a co-investigator reading it once,” says Moise. “You’re very invested and encouraged to think outside the box. Somebody would say, ‘I've used discrete choice experiments before, but never in the context of having to stop behavior!’ We’re coming at things from very different perspectives, training, and experience, and it’s made all the difference.”

Rethinking Their Strategy

From early on in the process once they launched the grant, one thing became clear: They couldn’t simply approach this using traditional implementation science methods.

“I do a lot of intervention work—I'm constantly trying to get health care system providers and patients to adopt something new, to DO something,” explains Moise. “So the challenge here is how do you get a system that's on overdrive to stop? How do you work against something that is woven into a system?”

“We knew of promising work that used an innovation tournament, a crowdsourcing approach for getting suggestions and feedback for developing strategies in the context of health systems implementation. We decided to apply this method in the context of de-implementation”, says Tehranifar. But from the beginning, there were setbacks.

“We're in the scientific world that says, ‘This is what the guidelines are. This is what we should be doing,’” says Shelton. “But we have realized that not all providers are on board and not everyone is at the same starting place, so we’ve had to rethink some of our approaches.”

The team received some valuable feedback from an unexpected source. “I work with a creative director named Luis Blanco,” says Moise. “He used to work in private industry, and he's helped me create all of my materials for my implementation-focused interventions.”

The team invited Blanco to meet as part of their planning for the innovation tournament. At the time, they were approaching their work using familiar implementation science methods: Start with the guideline that everybody mostly agrees with and determine the best strategies from there.

But almost immediately, Blanco gave them feedback that made them rethink that stance. “Luis helped us see that this [de-implementation] is completely different, that you can't approach it the same way,” says Moise. “You have to meet people that may not even agree with this guideline. You have to slowly work your way into this conversation. Because if a person doesn't agree, it's already become a very different process.”

That discussion and others like it helped set the groundwork for a different approach. “We learned from our colleagues that it's not just about eliciting ideas—it's also a tool for engaging providers, getting them invested and starting a discussion, and to think about what would move them along that continuum of even considering reducing over screening in their clinical practice” says Moise.

It led to a completely new strategy. “Luis suggested we call it Re-Think,” she recalls. “Which is a great way of looking at it—let's rethink the entire concept of mammography screening for this population and come up with some strategies that would be specific to de-implementation.”

They began to develop a website-based platform called Re-Think that will allow providers and practice managers to make suggestions as well as rank, rate, and comment on them. “That way, it can serve as a crowdsourcing platform as well and we can evaluate it as a methodology as well. It's meant to encourage engagement and feedback and reflection, not just designed get information from them,” explains Shelton.

According to Tehranifar: “It's one thing when you're trying to get people to do something they already know they need to do. What's innovative here is that we're going to use a strategy they've chosen themselves,” she says. And by helping to get everyone more invested, the researchers are hoping it will be much more effective when it comes time to implement these methods.

The team feels optimistic that this can set a template for de-implementation that works across different scenarios. “The message can be, ‘You can be part of the discussion,’” says Moise. “De-implementation is not just do we actually reduce mammography? It’s about engagement and collaboration for systems change.”

Under the Wire

In February 2020, right before the height of the COVID-19 epidemic, the three were working day and night to meet the grant submission deadline. “We worked in night shifts, day shifts, passing it off to each other. It was definitely the most collaborative grant we've ever worked on,” remembers Shelton.

“We wondered if we would actually get this out the door,” says Moise, “and somehow, we did.”  The result? An immediate success. “It got funded the first time in, which almost never happens with grants,” she says. “It’s notoriously difficult. And we received a fourth percentile on the first try. It was amazing and testament to the importance of perseverance and collaboration.”

The team credits the collaboration and feedback made possible by connections and collaborations fostered through the Irving Institute. “It was definitely stronger because it had gotten all this feedback,” remembers Shelton. “It was three synergistic but distinct types of expertise that really complimented each other,” she says. “This is the type of collaboration that the Irving Institute enables and encourages. So many different perspectives helped shape this. This grant is just this really wonderful success story that we are proud of and excited to get started to contribute to advancing the science in this important area.”


mammograms, mammography, breast cancer, health screenings