Last year, Tampa General Hospital (TGH), a 1,011-bed teaching hospital in Florida, initiated a clinical process improvement (CPI) program aimed at enhancing standardization and efficacy of treatment for conditions identified as high priority. At the time, officials at the patient care organization noted that the core goal was to develop scalable and repeatable models to tackle difficult clinical initiatives and improve patient outcomes—an objective that Peter Chang, M.D., CMIO at Tampa General Hospital, confirmed in a recent interview.
“As a doctor, naturally your inclination is to want to make the lives’ of your patients better and make them live longer with less comorbidities. So this initiative was born out of [simply] providing better care for our patients. And when we looked across the organization, we lacked a lot of standardized methodologies for how we care for them,” Chang says. He gives an example of a patient coming in for a disease process, but one group of physicians might manage that differently—not incorrectly, but differently—than another group. “So with variability comes variability in outcomes,” says Chang. “We have known this problem exists in the U.S., but also in our organization, so we set out on a path to try to attain standardization for how we care for our patients.”
Early on in the CPI initiative, a top priority was the early diagnosis and treatment of sepsis in the emergency department (ED), where 80 percent of TGH’s sepsis cases originate. TGH tested several known methodologies, yet they lacked scalability and self-service access to the information that stakeholders needed to make quick improvements, according to officials. These methods did not provide insight into the actual process or delivery of care (rather only focusing on outcomes data), which is required for sustainable and scalable process improvement, TGH officials noted.
So clinical leaders at TGH opted for another approach, which involved electronic health record (EHR)-generated alerts and treatment advice, and a clinical process improvement platform from Minneapolis-based LogicStream, which enabled clinical leaders to review clinician response times and treatment decisions in real time.
Chang notes that with the advent of EHRs, other avenues have opened up. Before, when everything was on paper, he says, there would be a paper order set that would have to be located for different things to be enacted. But now, with clinical alerts living inside the EHR, they can kick off different processes to ask the provider, “Do you think this is what’s going on, and if so, please follow the accompanying treatment track.” And Chang adds that much of this is driven by TGH’s quality reporting for CMS (Centers for Medicare & Medicaid Services) star ratings, value-based purchasing, and U.S. News and World Report hospital scores, all of which take some account of quality and outcomes into their calculations. “So there is a secondary gain for wanting the hospital to perform at a higher level. We want to show folks that we treat our patients well here. So that led us onto our path for process improvement and bettering clinical outcomes,” he says.
Regarding sepsis, Chang says that by looking at the Sequential Organ Failure Assessment (SOFA) score (used to track a person's status during the stay in an ICU to determine the extent of his or her organ function or rate of failure) or examining the “Surviving Sepsis” guidelines, one would get differing opinions on what doctors believe is the correct way to manage sepsis. But nevertheless, CMS has adopted a 3-and 6- hour rule for a core quality measure—the Sepsis CMS Core (SEP-1) Measure.
For TGH clinical leaders, being able to leverage its Epic EHR to allow these processes to get kicked off became a key priority. “We created a sepsis alert order that goes into the EHR and is an overhead page to everyone in the ED to say that there’s a sepsis alert in X room [for example],” Chang explains. “So when that happens, it initiates a chain reaction of events from nurses to physicians to pharmacists to make sure that the correct protocols are being followed. And the use of LogicStream in that sense allows us to analyze the compliance and usage of the elements that we have built into the EHR,” he says.
Chang says that Epic has two major tracks in its system to help treat patients: best practice advisories and order sets. The best practice advisory, he continues, looks at information that lives within the chart, be it labs, documentations, vitals, or discrete data inside the EHR, and will kick off an alert to a nurse or physician, leading that person to an order set. The nurse or physician can then see that even though the patient meets the criteria for labs and vitals, maybe he or she just has the flu and is not septic.
“It gives you a way to ‘opt out;’ the computer just knows the patient from what’s inside the EHR. So we use best practice advisories to alert nursing in triage and physicians to say that their patient could potentially have sepsis based on multiple criteria, do you want to go ahead and initiate that sepsis order alert, and do you want to continue down the path of using the order set we have outlined to achieve the 3- and 6- hour timeframe from the SEP-1 measure?” says Chang.
Currently, only physician leaders are using the software since it’s their way of trying to manage their doctors, so TGH is just starting to publicly display order set utilization scores, says Chang. He adds that the LogicStream platform doesn’t adjust for volume, so if one doctor is working five times the number of shifts of another doctor, the former’s compliance will appear higher because he or she is using the order sets and alerts more often. But, Chang notes, LogicStream does give good insight into how people are managing the alerts. “They are sometimes dismissing the alerts, as they should, but are they doing that too much?” he asks.
To this end, when more data comes in, an organization can come up with a range based on what it knows. So if 10,000 patients come in, 5,000 of whom are final coded for patients, in that instance, TGH practitioners should be in the range of 50 percent compliance with this advisory, Chang explains. “We can come up with those broad stroke numbers to get the picture and follow the best pathway when appropriate. So it’s a ‘scorecard’ to see how this individual is using the system, and how the person is prompting the alert. If we reverse engineer the data, we can find specific things, such as OK, provider X had five patients that didn’t meet the measure, but why? And then we drill through and if we find that the doctor is not using the order set or dismissing the alert, that provides an education opportunity to see why the doctor is getting the alerts but not responding to them,” he says.
As a result, Chang says, clinicians in the ED have become more engaged in their performance data as it relates to sepsis outcomes. “No one wants to cause harm to a patient,” says Chang when asked if doctors are OK with these new processes. But, he adds, “You have to paint that picture to them explaining why it’s important rather than tell them ‘you have to do this.’ The alerts can be firing too many times; it has to make sense.”
And while it’s still early on in the process, initial results suggest that TGH doctors will quickly understand the initiative’s importance. In one instance, TGH increased the use of IV lactated ringers for sepsis patients by 50 percent within one week, according to officials.
Chang says the biggest challenge with an initiative like this one is change management. “You can use technology, tools and tricks to present people with data and provide explanations, and allow them to drill down into it, but getting people to realize the importance of it and then start to shift in how they’re thinking about septic patients is the challenge,” he says.
“A lot of light bulbs have gone up in regards to our process,” he adds, noting that TGH also has processes outlined for conditions more well-defined, such as stroke. But stroke is a small bucket and a small group of patients, so it’s a much more focused effort on getting the patient what he or she needs. Sepsis, on the other hand, is a very broad category—it could be your 10-year-old kid or a 95-year-old grandmother, or anyone in between, Chang says. “So change management is opening up the window. Could that person be septic or could he or she be developing sepsis, and why? I think that’s the big issue—getting people to see the importance of using the order sets and complying with the alerts.”