A predictive analytics tool based on nursing assessments, combined with remote monitoring, leads to earlier intervention of patients in clinical decline, while alleviating nursing burden.
How does Froedtert Health accelerate intervention of patients in the early stages of clinical decline? By remotely monitoring every individual occupying a bed with a predictive analytics tool that pinpoints a patient's condition. The formula incorporates nursing assessments into its algorithm—something similar predictive tools neglect—and, rather than burden overworked nurses with yet another task, the offsite center notifies them when action may be warranted.
During an era when advanced technology guides more healthcare decisions, something essential is missing from the data mix: the assessments of the providers who work most closely with patients. While other predictive tools leave most of this information buried in the EMR, the Rothman Index used by Froedtert brings it to the forefront.
The Rothman formula quantifies the information nurses add to the chart, combines it with real-time data from vital signs and labs, and assigns a score indicative of the patient's acuity and risk. If the score drops into questionable territory, representatives communicate with the responsible nurse, resulting in earlier patient interventions 18% of the time.
Other health systems using this model have documented the ability to lower mortality rates by as much as 32%, as well as shorten length of stay, both of which contribute to lower costs. By moving monitoring from nursing stations to a remote facility that observes all patients at three hospitals, Milwaukee-based Froedtert has taken an approach that also alleviates nursing workloads, while simultaneously enabling them to focus on patients who need their care most.
Early treatment for these patients would otherwise have been delayed or missed entirely, says Andrew Resnick, MD, chief medical officer for Froedtert Health and Medical College [of Wisconsin] Physicians, and associate dean for clinical affairs at Froedtert Hospital.
A look at the way Froedtert is using this tool from Charlotte, N.C.-based PeraHealth provides insights into the value of incorporating nursing assessments into predictive analytics, the benefits of remote monitoring, how to measure success, and other potential uses for the tool to address utilization and staffing issues.
Missing in Action: The Value of Nursing Assessments
"For an organization that has a focus on quality at all levels—from senior management down to the front line—having an early warning system embedded within the electronic record is a must-have these days," says Resnick. "There are a million data points every day. The ability to figure out patterns from all the data coming at clinicians is next to impossible—other than when it's obvious or too late."
Recognizing the need for such information, multiple companies, particularly those that have developed EMRs, have created products that work within the medical record to help providers recognize signs of clinical deterioration earlier. Resnick says most of those solutions are flawed because while nurses chart into the EMR, the products don't incorporate this information into the algorithm when evaluating the patient's status.
"As a clinician who trained during the paper record era—before the electronic record—when you would open up a chart, there'd be a flow sheet that the nurse would have filled out," explains Resnick.
In addition to vital signs, the flow sheet also delivered a synopsis of what was going on with the patient, he says. For example, the document would note whether the patient had lost their appetite, or if they were experiencing palpitations, increased pain, abdominal tenderness, or difficulty walking. Most predictive analytics tools are missing this information because they have not figured out how to quantify it, Resnick says.
"The ability to instantly look at a flow sheet and see the nursing assessment is difficult now because there's so much data and so many pages to go through within the [EMR]," says Resnick. Yet the paper flow sheets were not ideal either because there was no consistent way to capture information, says Resnick.
"The Rothman Index not only brings that information back, but quantifies those nursing assessments … and puts it in the hands of the providers," says Resnick.
Offsite Monitoring Alleviates Burden; Prompts Nurses to Take Early Action
The PeraHealth tool works with all commercially available EMRs. Once installed, the Rothman Index appears as a tab within each patient's record. Most facilities also set up a kiosk or large screen at nurses' stations, which enables nurses to monitor and track the status of patients. If a patient's status changes significantly, alerts, calls, or texts may be issued to the responsible nurse, depending on the organization's protocols.
Froedtert took a different approach, and was perhaps the first health system to use an offsite facility to monitor all patients at three of its hospitals. The remote center, which also serves as the system's eICU monitoring facility, is operated by Inception Health, which serves a digital health service hub for Froedtert and the Medical College of Wisconsin.
"From the beginning we thought about how we could incorporate the remote monitoring center into our workflows," says Resnick. "They continuously monitor Rothman Index trends and scores for all patients across our health system. There are different algorithms [that determine] when they talk to the clinical team or make interventions. When [the remote team] get alerts, they call the nurse taking care of the patient to make sure they understand [the patient's status] or can evaluate if there's a clinical change."
By using an offsite monitoring facility, Resnick points out nurses are relieved of monitoring responsibilities, and the information enables them to focus on the patients who need them most.
Measuring Success
"Going into this, we knew one of the biggest challenges would be [measuring the impact of the tool]. It's hard to measure something that doesn't happen," says Resnick. In other words, since the Index accelerates patient intervention earlier the process, how do you determine whether the patient would have had a worse outcome or died without that intervention?
Studies done by other health systems have demonstrated mortality rates have dropped as much as 32% after the facility began using the Rothman Index along with associated treatment protocols. For example:
- Yale New Haven Health had a 30% reduction in mortality as reported in BMJ Quality & Safety during October 2015, which was noted in an article published by HealthLeaders in January 2016.
- Houston Methodist reported a 32% reduction in risk-adjusted mortality in the December 2016 issue of BMG Quality & Safety.
While mortality rates offer one form of measurement, it's difficult to make comparisons between institutions, Resnick says, because one organization may handle more complicated cases compared to another. And, while there are algorithms to account for such differences, the science is not exact because many factors contribute to mortality.
Michael Rothman, PhD, co-founder and chief science officer of PeraHealth—and the man who invented the Index with his brother Steven Rothman after their mother died from treatable complications—admits to being initially surprised at the Yale study findings. "I'm very skeptical researcher," he says. "You always question things." But when the Houston study emerged with similar results, he found himself "starting to think maybe we have something here."
Other peer-reviewed scientific studies have documented decreases in sepsis mortality, reduced length of stay and other measures. In-house assessments, conducted by hospitals using this solution, also demonstrate associated cost savings.
Froedtert is in the process of conducting its own assessment of observed mortality rates, but meanwhile has collected data related to communications between the Inception Health monitoring facility and the nursing team. Over a year, this process has prompted nurses to take action 18% of the time someone from the monitoring center reaches out. This number rose from 15% after the first five months of use.
"Without [this tool], there might not have been an intervention, or it might have been delayed," says Resnick. "The Rothman Index is helping our remote monitoring center identify patients in potential clinical decline, who then can get an earlier intervention, which has been shown to greatly improve outcomes."
Exploring Other Uses
Beyond its current use, Froedtert is also exploring other ways to use the tool to improve patient outcomes, as well as utilization.
"The idea is to make sure the right patient is in the right bed and at the right level of care," he says. "We're thinking about using the Rothman Index as a signal to measure acuity and make decisions on staffing and location of patients," says Resnick. "It's a bit beyond its original intent, but like many of large academic medical centers, we have capacity challenges, especially [during] flu season. It also allows us to prioritize which patients we see first."
Froedtert is considering using the Rothman Index for the following potential uses:
- To prevent inappropriate transfer out of ICU
- To facilitate transfer into the ICU if a patient's scores are declining
- To prohibit inappropriate patient discharges
- To ensure patients are safe for travel through the hospital for diagnostic tests and/or procedures
“If you just go by quantifiable numbers—vital signs and lab values—you miss a lot of opportunity.”
Andrew Resnick, MD, chief medical officer, Froedtert Health
Mandy Roth is the innovations editor at HealthLeaders.
KEY TAKEAWAYS
While nursing assessments are a vital component of patient care, most predictive analytics tools don't include such data in their formulas.
The Rothman Index quantifies information in nursing assessments, along with vitals and labs, and assigns a score indicating whether a patient's condition is changing.
To alleviate nursing burdens, Froedtert employs a remote facility to monitor each patient's condition.
The tool leads to accelerated patient interventions 18% of the time, spurring early treatment and enabling nurses to focus on the patients who need them most.