Sepsis management has long been a challenge for hospitals throughout the U.S., and severe sepsis carries high mortality rates and costs the healthcare system billions of dollars each year. Indeed, in America, sepsis kills more people than AIDS, breast cancer, and stroke—combined.
To help combat this national healthcare epidemic, more and more hospitals and health IT companies are teaming up to deploy technology that can help detect sepsis early on, as research has proven that early intervention will lead to higher survival rates. In New Jersey, for instance, the Cape May-based Cape Regional Medical Center (CRMC) has linked up with Dascena, a California-based healthcare startup that develops predictive algorithms which have the potential to facilitate the timely and accurate diagnosis of complex conditions.
In particular, Dascena’s sepsis detection algorithm, InSight, has been shown to significantly improve mortality rate, average length of stay, and readmission rate among several patient populations, including those at CRMC. Andrea McCoy, M.D., chief medical officer at Cape Regional Medical Center, says that the prime motivation behind the collaboration was the recognition that the most important factor in surviving sepsis is the early detection and early intervention for the patient. “We needed to find a way to identify patients sooner, compared to our existing process [for sepsis screening],” she admits.
Prior to implementing InSight, CRMC was using a twice-a-day manual process that occurred only when the nurses did their screenings. That included looking for certain criteria that suggested the patient might be septic, and beyond that, he or she had to have evidence of organ dysfunction as well, McCoy explains. But the issue with that process was that twice-daily screenings meant there were 11 hours when the patient wasn’t being specifically screened, unless there were dramatic changes in his or her health.
As such, subtle findings that were not present until later in the illness were being missed, leading to patients being diagnosed much later into their illness. Further, McCoy explains, previously, there was not any process in the ER to formally screen for sepsis. But now, InSight “enables us to screen those patients almost at their point of entry into the healthcare system, as well as have a continual evolution for the earlier signs of sepsis on the inpatient unit,” she offers.
At a high level, the algorithm is able to look at trends and correlations around the variables that might contribute to sepsis, explains Ritankar Das, who is the CEO of Dascena. While the tool only requires vital signs, it can incorporate other measurements as they become available. And it does it all of that automatically; it picks everything up from the EHR (electronic health record), meaning there is no extra work for clinicians, Das says.
The tool is also trying to help clinicians understand how the pattern looks relative to the millions of patients it has seen before. So for instance, if that pattern is a high-risk one for developing sepsis or one of its downstream complications, such as severe sepsis or septic shock, then if there is a high enough risk, a notification gets sent to the provider. In essence, it is learning what has happened with lots of other patients in the past, using just a small amount of information that’s available in the patient chart, and doing that continuously—in turn, updating the risk profile as time evolves, Das elucidates.
The most important findings from the CRMC study, which looked at how InSight could help with sepsis-related patient outcomes, were a reduction in length-of-stay for patients (9.55 percent), a reduction in the hospital’s mortality rate (60 percent); who presented with sepsis, and a reduction in sepsis-related 30-day readmission rate (50 percent).
McCoy notes that one of the keys for success was finding the “sweet spot” for the tool’s sensitivity and specificity, “so that you don’t have too many false positives, but also so that you are not missing patients who might have sepsis—which is what happens with lots of the other screening models that are out there.” Indeed, this is what happened with CRMC’s manual screening process, McCoy acknowledges. She says that even with the tool, in the early stages they did experience some alert fatigue because CRMC’s team was initially “generous” in how it was setting limits. “But we found out that doctors were getting lots of calls for patients who didn’t have sepsis. So we had to find the right spot to identify patients who had the early signs of sepsis and not some other disease process. We still will get the occasional patient who doesn’t have sepsis—but it’s nothing like before,” she says.
McCoy says that her team has been working with others in the state on early identification and to develop sepsis protocols. Since 2015, the Centers for Medicare & Medicaid Services (CMS) has had a national sepsis measure that assesses how well hospitals follow evidence-based protocol care, and just very recently in New Jersey, a law was passed that mandated education and sepsis screening protocols, says McCoy.
But she adds that implementing InSight has helped CRMC “stay ahead of the game,” since many screening protocols start looking later down the line. “Many patients will often present with vague symptoms, and sepsis is always out there hovering in the back of everyone’s minds. So having the ability to identify the early markers is what’s impactful in regard to patients’ outcomes,” McCoy says.