Portland, OR – When Blake Lesselroth, MD, assistant professor of medicine and medical informatics at the Portland VAMC, and colleagues were asked to spearhead a hospital-wide implementation of computerized decision support for post-surgical deep vein thrombosis (DVT) prophylaxis, it seemed like a straightforward task that might require some technological tweaking and little else.
“Facility leadership recognized that we were not performing as well as we could in the domain of DVT prophylaxis,” Lesselroth said. “And with greater and greater visibility for DVT prophylaxis as an inpatient performance measure, combined with relatively sound evidence-based science, it seemed like something we could envision implementing without a lot of complete transformation of the culture.”
With the backing of the hospital’s chief of staff and the chief of the operative care division, Lesselroth and his hospital informatics department colleagues believed that their only challenge would be deciding how to program the electronic health record (EHR) to best help clinicians remember to include DVT prophylaxis as part of a patient’s surgical regimen.
Their assumptions turned out to be far off the mark, and technology was the least of their problems, Lesselroth said.
“Very quickly, we started realizing that we had a fairly naïve representation of all the stakeholders and agendas,” he explained. By the time the study was finished, Lesselroth and his colleagues found it necessary to develop an intimate understanding of how different clinical specialists use technology, how they communicate and how they think.
Venous thromboembolism, a dangerous post-operative complication affecting surgical patients, represents the most common preventable cause of in-hospital death. According to data cited in Lesselroth’s study, some 15-60 percent of patients are at risk of developing a DVT if prophylaxis is not employed. That risk can be significantly decreased by routine administration of low-dose anticoagulants or lower extremity automated compression devices, which are recommended in most published guidelines, according to the study. In 2007, the VA’s National Office of Quality and Performance established a deep vein thrombosis (DVT) prophylaxis performance target of 92% of candidate patients.
“Nationally conducted chart reviews showed that patients either did not receive prophylaxis or received non-standard prophylaxis 75% of the time,” Lesselroth and his co-authors wrote in the recent study. “Consequently, hospital executive leadership and the surgical department jointly sponsored the development and implementation of health record CDS and an educational campaign to encourage surgeon compliance with DVT prophylaxis recommendations.”
Lesselroth and his team developed a series of electronic menus with DVT prophylaxis recommendations and deployed it in the hospital’s EHR system. Performance compliance with DVT prophylaxis before implementation was 77%. After this first implementation, that number did not rise much. Initial use of the menu system was only 20%.
“We put a ton of effort into creating this decision support structure that was embedded in the EHR. Then, right after first implementation, the use wasn’t that great,” Lesselroth said.
Part of the problem was technical in nature. Not all surgeons used the same electronic pathways in going through medical procedures. A better understanding of how surgeons actually went about their tasks and what electronic tools they use in their environment allowed the researchers to redesign the menu system to better intersect with physicians’ workflow.
“It wasn’t revolutionary, but it was important,” Lesselroth said.
What was revolutionary was the realization that not all specialists think the same way. There was a cognitive dissonance between the DVT prophylaxis information that Lesselroth and his team were presenting in the EHR menus and the way that some surgical specialists envisioned their roles.
“Most clinicians aren’t trained in cognitive science. You don’t see a lot of these kinds of papers in the medical literature. But I’ve since discovered that it’s a hot topic in other industries, especially those that address safety,” Lesselroth explained. “In very complex environments, if you don’t understand all the cognitive things that are going on, you’re not in a position to build something that works and that fits into their mental framework.”
An internist himself, Lesselroth and his colleagues built all of their decision support so that it made perfect sense to an internist. However, surgeons and subspecialists had very different ideas of how to go about their work.
“I approach DVT prophylaxis in a way that looks at the variables in each patient, creates a risk model, then uses that model to make a risk assessment,” Lesselroth said. “It’s a very algorithmic way of thinking and emblematic of a way an internist thinks.”
An orthopedic surgeon, however, might see the situation very differently and want the menu presented in a way that speaks to which specific surgical procedure is being performed.
“The big picture is that we had written the menus in a way that didn’t register with the way a lot of surgeons thought,” Lesselroth said. “They opened it up and saw a lot of text that didn’t speak to them, and they did what any of us would do—they walked away.”
The team eventually spoke with each subspecialty, such as urology and orthopedic surgery, to get a better representation of their needs and how they think before going back and reengineering the entire decision support structure for each subspecialty.
“It made it a lot more work for us, but the take-home message was that not one size fits all,” Lesselroth said.
The third revelation, which was the most complicated and the greatest surprise to Lesselroth, involved the socio-behavioral organizational culture of the hospital and the barriers it produced.
“We talked to upper leadership who told us to do this, but what we hadn’t done very well was go around and meet with every surgical subspecialist, at the very least, and ask them what they think, in a way that was non-challenging,” Lesselroth said. “To transform a culture, you have to think about their perspectives.”
Once Lesselroth and his team started sitting down with specialists and asking what their concerns were, they found them to be legitimate, although not always predicated on medical literature.
Once they started addressing the socio-behavioral issues, as well as the cognitive and technical ones, Lesselroth began seeing a gradual adoption of the technology, with use improving 50% to 90%.
“We’ve managed to sustain a performance target and have seen this slow increase over time, which I think is the transformation of the culture, which is never fast,” Lesselroth said.
Asked how this project changed the way he and his team approach quality-improvement initiatives, Lesselroth said they have become more thoughtful in how they approach projects. “Right now, we’re trying to improve medication reconciliation at the Portland VAMC, and I’m using everything I learned in this particular study,” he said. “I’m trying to understand what are the belief systems people have about medication reconciliation, and, by extension, patient-centered care models. The goal is to identify what potential barriers—technical and cognitive—need to be overcome before beginning.”
A full description of the project and its results appeared in the January issue of the BMJ journal Quality & Safety. “If nothing else, I want people to see that we started out with a traditional approach to implementing DVT prophylaxis but, immediately after we started to encounter problems, we jumped out of our comfort zone.”