Published in the May 2017 issue of Today’s Hospitalist
WHAT DO YOU NEED to succeed at quality improvement? “Everybody is doing QI in the hospital,” said Scott Flanders, MD, one of hospital medicine’s foremost quality-improvement experts, speaking at last fall’s management of the hospitalized patient conference at the University of California, San Francisco. “But few are doing it well.”
Dr. Flanders, who recently became vice chair for external relations and quality for internal medicine at the University of Michigan Medical Center in Ann Arbor, said that one big reason why so many QI efforts falter is that they can fall prey to many potential pitfalls. Those can include a lack of passionate physician leaders and targeted data, as well as cultural barriers that can bring the best-intentioned QI project to a halt.
TO ILLUSTRATE the need for a physician champion, Dr. Flanders described one project to curb inappropriate antibiotic use that was launched at both his academic center and a partner community hospital. In both hospitals, the No. 1 condition driving inappropriate use was UTIs.
“If you don’t have good data, you’re dead in the water.”
“We found that 60% of orders for urine cultures lacked an indication,” Dr. Flanders said. Among positive cultures, 60% had asymptomatic bacteriuria, while 25% of patients with a UTI received antibiotics for too long. But the biggest problem: “A huge percentage of asymptomatic bacteriuria was being treated.”
As the head of that QI project at the University of Michigan, Sarah Hartley, MD, first brainstormed with her colleagues to come up with strategies that hospitalists could use, then helped implement them. Dr. Hartley helped standardize testing recommendations and treatment algorithms. She followed up on hospitalists’ suggestion to create and distribute laminated cards that physicians could carry. Those cards spelled out when to send cultures and what drug and duration to use.
Dr. Hartley encouraged one doctor to put together an antibiotics app that physicians used on their cell phones. She also helped launch an antibiotic timeout in which pharmacists and hospitalists met to revisit patients who had been taking antibiotics for 72 hours. When it became clear that it was hard to keep track of that 72-hour window, those timeouts were rescheduled for Monday, Wednesday and Friday at a set time.
Dr. Hartley helped create and display posters, and she held an ongoing round of educational sessions, which were videotaped so doctors could watch them remotely. She came back to the hospital at night with dinners to talk to the nocturnists while they ate.
But at the partner community hospital, the project rollout took a much less aggressive approach. The project lead at that hospital gave the hospitalists the laminated card that had been created, and he asked them to watch Dr. Hartley’s video. The lead physician also offered paper authorship to the hospitalist lead in exchange for making sure the physicians watched the video.
When project results came in, the difference in engagement between the two leaders became clear. At the academic center, the percentage of asymptomatic bacteriuria cases being treated with antibiotics fell by nearly one-third. But at the community hospital, the treatment rate barely budged, falling less than 5%. (Dr. Hartley was lead author of a study that appeared in the November 2013 Infection Control & Hospital Epidemiology with those results.)
The conclusion? Physician champions heading up QI projects must have a clear process of education and feedback, Dr. Flanders said, and they must provide consistent modeling. While prior experience is a plus, it’s not required.
“Having a good leader is critical,” said Dr. Flanders. “It’s not all you need. But if you don’t have one, don’t even bother.”
THE NEXT POTENTIAL pitfall that Dr. Flanders highlighted was data, or rather the lack of appropriate, actionable data. That problem was illustrated by the University of Michigan’s efforts to improve hand hygiene.
Based on results from unit-based, covert observers, the academic center reported that providers complied with handwashing requirements less than half the time. But the hospitalists didn’t begin to believe that those data really reflected their practice.
After all, the “secret shoppers” were unit-based, while hospitalists saw patients throughout the academic center. And with 125 providers in the hospitalist program—100 physicians, 25 advanced practitioners—how often were they each being observed?
Another detail that hospitalists found damning: While the academic center provided overall compliance rates, the data didn’t identify specific clinical situations where clinicians were consistently failing to wash their hands.
“My reaction was: I can’t go to my group and tell them we need to improve,” said Dr. Flanders. “For us to understand the problem, we need better data.”
He took those concerns to the center’s infection control team, which agreed to dedicate two observers to the hospitalists on a pilot basis. The observers received the hospitalists’ photos as well as their schedules so they could easily identify doctors.
Big handwashing fails
When Pilot results came back, it turned out the hospitalists were performing even worse than previously thought, slumping to 45% compliance. But importantly, the new data identified their major handwashing fails. For one, doctors were not washing up before putting on gloves and gowns for patients in isolation.
“To illustrate why this was a problem, the infection control team created a video in which they had chocolate syrup all over their hands, then put on gowns and gloves,” Dr. Flanders said. “When they were finished, everything was covered with chocolate. The video showed clearly why you needed to wash your hands even when using gowns and gloves.”
Another big problem: Doctors seeing patients would excuse themselves to answer their phone and leave the room, forgetting to wash their hands both when they walked out of the room while talking on the phone and when they walked back in.
The fix included educational sessions as well as compliance audit and feedback at both the service-line and physician levels. In addition, the hospitalists agreed to include hand hygiene in their annual incentive, setting a threshold of 90% compliance before they would earn that portion of their bonus.
In one year, their compliance rate rose from 45% to 90%. Dr. Flanders’ takeaway: “If you don’t have good data, you’re dead in the water, and you’re not going to make progress.”
For groups struggling to get the right data, he suggested tailoring QI efforts to take advantage of data already being collected within their hospital or health system. Chances are, that’s the case for not only hand hygiene but readmissions, ED wait times, HCAHPS scores or any existing EHR data field.
And if you do have to manufacture your own data, decide on how big of a sample you really need. “Rather than looking at all patients with a specific condition, pick 50,” he said. (If charts need to be checked, be sure that task falls to non-providers.) “Some of your physicians won’t be reflected in that sample, but you could still establish a group-based metric.”
It’s all about culture
ANOTHER BIG PITFALL is “a key element that we often forget when we’re doing QI,” said Dr. Flanders: culture. Not confronting cultural issues head-on can cause QI efforts to “fail disastrously,” which is how he characterized one initiative at his hospital designed to curb CAUTIs.
The project seemed to do everything right, assembling a multidisciplinary team with data support to track catheter-use indications. Backed by research on how to disrupt the life cycle of a urinary catheter, the project’s main goal was to ensure prompt catheter removal.
“The technical solutions for this are actually well-studied, and they work,” Dr. Flanders pointed out. Doctors would have to fill out a new CPOE order, listing an indication for the catheter. They also would put in stop orders allowing bedside nurses every day to assess patients’ need for a catheter and remove it when no longer needed.
The roll-out on four pilot units proceeded with plenty of nurse education. But as CAUTI rates were tracked over a year, they were still (slightly) on the rise, and urinary catheter use had not dropped significantly.
What went wrong? A lot of small issues, Dr. Flanders said. The indication order had an “Other” box that was checked too often, making it unclear whether a true indication existed. Project leaders hadn’t thought about what to do when patients requested a catheter, and it wasn’t always apparent what good alternatives there were to indwelling catheters.
But the big fail, he said, was not dealing with the cultural implications of the stop orders.
“The physicians didn’t really buy into the project,” he noted, “so nurses were uncomfortable removing the catheters. All it took was one doctor confronting a nurse over why a catheter had been discontinued, and none of the nurses would ever remove one again.”
The University of Michigan got a second chance at preventing CAUTIs, this time as a participant in an AHRQ-funded (to the tune of $20 million), national, four-year project. That project benefited from the expertise of extended faculty (including Dr. Flanders) and experts from key professional societies, including the Society of Hospital Medicine and the Society of Healthcare Epidemiology of America.
The project also set up the Interdisciplinary Academy for Coaching and Teaching (I-ACT), which was written up in the October 2014 issue of the American Journal of Infection Control, to train those faculty. Faculty members would in turn speak about CAUTI prevention at regional hospital meetings and assist multidisciplinary coaches at individual hospital sites.
But first, “we recognized that if you’re coming to someone with stop orders to use at a hospital, that’s a technical solution,” he explained. “If you give them only those, the project will fail, just like it did in the first pilot at our institution.” During a several-day meeting developed for faculty, I-ACT coached them—often through role-playing—in how to train hospital personnel to tackle cultural problems head-on.
“We walked through how to break down physician resistance, how to handle the urologist screaming at a nurse, what to say to patients asking for a catheter,” Dr. Flanders said. “We then coached people at individual sites on how to deal with all of those.”
The results of the project were published in the June 2, 2016, New England Journal of Medicine, by lead author Sanjay Saint, MD, a University of Michigan hospitalist-researcher. With the program implemented in more than 900 units across the country, CAUTI rates in those units fell more than 30%.
Phyllis Maguire is Executive Editor of Today’s Hospitalist.
Tips for successful QI
TO HELP HOSPITALISTS succeed with quality-improvement initiatives, Scott Flanders, MD, a QI expert who practices at the University of Michigan in Ann Arbor, offered the following pearls during a hospital medicine conference last fall:
• Less is not necessarily better. While examples of overtreatment abound, QI efforts should also target undertreatment. A fib is one example. “Some of these patients may have better long-term outcomes if you send them to an electrophysiologist after discharge, but very few patients are referred,” Dr. Flanders noted. “You may drive up costs for those patients, but it’s the right thing to do.”
• Don’t stick too rigorously to guidelines when creating local quality-improvement tools. For one, guidelines don’t let doctors know what they need to do.
“Telling someone that rhythm control is better than rate control for some patients with A fib doesn’t convey what dose to give and for how long,” said Dr. Flanders. “What we need instead are pathways.” And invariably, you need to modify guidelines and pathways to local conditions. “Bend them a little and get buy-in.”
• Carefully pick the scope of your project and start small. To curb inappropriate antibiotics for asymptomatic bacteriuria, Dr. Flanders pointed out that the initial project at his center targeted only hospitalists and not the ED, where a lot of problematic prescribing occurs.
“We eventually brought that project to the ED,” he said. “But by starting with the hospitalists, we proved that it worked. Once you can point to success in one department, you get much better buy-in when you expand.”
• Borrowing and stealing are fine when it comes to QI ideas and techniques. Dr. Flanders said he’s been amazed at how generous clinicians in other hospitals are in sharing pathways and other solutions, even when those doctors work for hospitals that compete locally with his.
One cardiologist in his health system, for instance, called another at a local heart hospital and asked if the second cardiologist had a good pathway for treating A fib patients with rapid ventricular response to help reduce admissions. “That pathway,” said Dr. Flanders, “jumpstarted our own project.”
• Fake it until you make it. In that particular A fib project, Dr. Flanders said his group employed what’s known in consumer marketing as a “vapor test”: advertising a product that doesn’t really exist to see what the demand for it might be. Based on results, companies then decide whether they want to manufacture that actual item.
The University of Michigan wanted to reduce admissions among A fib patients presenting to the ED, but emergency physicians maintained that they didn’t have any place to send patients other an inpatient bed. Project leaders then found a physician assistant who worked with the electrophysiologists and who was willing to slot in patients two days a week for rapid follow-up.
An ED doctor then agreed to refer appropriate A fib patients to the “post-discharge A fib clinic,” which really didn’t exist, rather than admit them. “We got 16 patients, and both the ED and the patients loved it,” said Dr. Flanders. Project leaders then approached cardiology and hospital leadership, who agreed to hire an additional PA. The hospital then launched an actual electrophysiology clinic last year.
• To avoid burning out from a constant barrage of QI efforts, “try projects that are tied together and build on one another,” he advised. One project at his center on VTE, for instance, made it easier to move into QI projects related to PICC lines and intravascular devices, which cause VTE. Because most PICC lines are used for IV antibiotics, those projects in turn informed others on antibiotic stewardship.
And bring in new people. Dr. Flanders said he’s been impressed by how eager newly trained doctors are to take on QI. “Senior people can provide the guidance, but we spread QI out over everybody. That way, we haven’t had a lot of people say, ‘Forget it, I’m done.’ ”
How to boost QI with infrastructure
IN COUNSELING DOCTORS how to improve quality improvement in hospitals, Scott Flanders, MD, vice chair for external relations and quality for internal medicine at the University of Michigan Medical Center in Ann Arbor, told hospitalists at a hospital medicine conference last fall to start small.
But then how do you scale up successful projects system-wide? One big potential pitfall for QI efforts, particularly at large institutions like his own, Dr. Flanders said, is the lack of QI infrastructure.
Without that infrastructure, one service won’t have any idea of what another may be doing. By way of an example, Dr. Flanders pointed out that a pulmonologist at his academic center had started a post-discharge clinic for COPD patients. The problem? None of the hospitalists or other front-line providers knew about it, so no patients were being referred.
The solution at University of Michigan was to establish a big-scale project called the Clinical Design Program. That program not only helps keep front-line providers informed about QI efforts system-wide, but it creates QI capacity and provides support and resources.
As Dr. Flanders explained, the program doesn’t have a huge staff: a physician, an administrator, a business manager and two project managers. One project manager is an expert Lean trainer, while the other is an industrial engineer who can do value-stream mapping, laying out all the dozens of steps in any clinical process so barriers can be identified.
Program personnel rely on front-line clinicians to help identify projects to tackle. Selection criteria for potential projects include identifying conditions or procedures that demonstrate a lot of variability in costs as well as in clinical processes and outcomes, as well as high-volume or high-impact patient populations. Projects selected have targeted tertiary or quaternary services that University of Michigan provides to referring hospitals, as well as conditions and procedures covered by bundled payments.
One project flagged was for A fib patients with rapid ventricular response. “We have a ton of those patients who would be cardioverted or not or anticoagulated or not, depending on the day of the week and the ED physician,” he said. Other big QI projects the program has tackled include colorectal surgery and hip and knee replacements.
In the program’s first workshop on any given project, the improvement specialists and project leads create a value-stream map of every step of the treatment or procedure being considered, “from the first phone call to the clinic to five-year outcomes,” Dr. Flanders said. That mapping identifies what works, what doesn’t and where the roadblocks are.
Front-line doctors then review those maps in a second workshop. “There are so many problems identified that you don’t know where to begin,” Dr. Flanders pointed out. “That’s when clinicians vote on what to prioritize.”
After dividing a whiteboard into four quadrants in terms of both high and low effort and impact, “people stick Post-its on problems they think they can work on,” he says. Areas of the board that get the most “votes”—particularly in the high-impact, low-effort section—are “the problems the QI project will focus on.’