In years past, healthcare providers looking to use data analytics to predict volume and staffing needs in their emergency departments relied on an approach that blended experience, trial and error, and conventional wisdom. A full moon on a Friday night? Better staff up. Fourth of July weekend? Have trauma staff on hand. Mondays tend to be busy; Thursdays tend to be slow.
But that kind of analytics can’t predict unexpected spikes in volume or unforeseen disease outbreaks. Even the weather or changes in a bus schedule can affect staffing needs. And unexpected lulls can lead to wasted overtime.
Fortunately, the practice of data analytics has evolved significantly in recent years. New tools are available to enable more sophisticated analysis methods — including predictive analytics — that emergency department leaders can use to ensure they have the staff they need on hand.
“To the average layperson, it’s controlled chaos,” says Kirk Jensen, M.D., who has spent more than 20 years in emergency medicine management and clinical care. “And certainly, at times, it feels like that to the people working in the emergency room. But once you’ve spent some time in the ER you quickly learn it’s more nuanced than that.”
Finding the Necessary Resources at the Right Time
Resource allocation in the ED is a “rich and complex mix,” says Jensen, the chief innovation officer at Envision Physician Services, a national multispecialty physician group that has contracts in more than 800 emergency rooms and treats more than 18 million patients annually.
“You have multiple lines of dedicated professionals all focused around delivering patient care, and somehow you have to integrate that workforce. So, you have physicians, in some places residents, advanced practitioners, nurses, techs and unit coordinators. So how do you evolve and get the best mix of the right people to effectively manage care? Separate from that is, we may figure out how to manage our average workloads, but how do we respond to surge capacity as well?” says Jensen.
It’s important to balance the cost of running the ED in part by aligning staffing to patient volumes, says Bill Orrell, managing director at global consulting firm Berkeley Research Group.
“It’s getting everyone, including the physicians, the midlevels, the nursing staff and the ancillaries focused on that patient on the inpatient side,” he says. “What are we doing for today? What are we doing for the length of stay to be sure we’re working toward an estimated date of discharge based on the patients’ disease processes?”
Organizations must have the ability, the time and the data to align it all, Orrell says, adding that organizations should re-examine their numbers regularly: every four to six months plus any time something changes, such as a disease outbreak, unusual weather or changes in the market, such as a new hospital or urgent care center opening up.
“We’re Marrying Science and Analytics”
In most healthcare organizations, the emergency department is a loss leader: It’s an expensive setting in which to provide care that’s prone to inefficiencies around staffing, length of stay and throughput. But it’s also a source of downstream revenue: Of the 145.6 million annual ED visits in 2016, 12.6 million patients (8.7 percent) were admitted to the hospital, according to the Centers for Disease Control and Prevention.
Beyond finances, the ED is the first touchpoint for those 12.6 million patients and their families. Long door-to-treatment time, hallway boarding, poor bed management and other bottlenecks can ruin that first impression, Orrell says.
“If you don’t get length of stay right in the ED and patient satisfaction right in the ED, it will be hard to recover once patients are admitted to the inpatient side. Because the impression that the patients and the family have begins in the ED, and you’ll never recover once you get them to an inpatient bed,” Orrell says.
“It’s more and more important that organizations are focused on aligning staffing to the volume and focusing on that throughput — how we’re greeting them at the front door, how we’re managing through the whole length of stay in the ED.”
Historically, organizations called in the troops when staff and resources were overwhelmed. But “there’s a gap between the signal and the need,” Jensen says. Staffers usually only realize they’re in over their heads after additional resources would have helped. By the time extra staff comes in, it’s too late: Everyone’s struggling to catch up for the rest of the day.
“We’ve had to get more sophisticated,” Jensen says. “With analysis tools, we can shift some of that staffing further forward.”
Envision has introduced a web-based, data-driven demand capacity management tool that helps evaluate and define optimal staffing levels in the ED by modeling patient volume and acuity.
“We look at patient arrivals by hour of the day, day of the week and by season,” he says. “And we’ve also built in a methodology that allows us to look at length of stay and its impact.”
The tool drills down further by factoring in natural variances of physician and nurse productivity and mapping ED features such as triage and patient flow service solutions. Patients who spend more time in the ED take up more resources, so the tool helps identify treat-and-release patients to keep the lines moving.
“We’re marrying science and analytics with patient care and clinician needs, and there’s a tremendous opportunity here to optimize safety and make service improvements,” Jensen says. “Once we have all of this data in the system, the next iteration is adding the artificial intelligence component and data mining to see how much further we can optimize this.”
“It’s Not a Competition, It’s a Collaboration”
At Bergen New Bridge Medical Center in Paramus, N.J., the admission rate is much higher than the national average: About 4,000 of 14,000 patients treated in the ED are admitted to an inpatient bed. The community hospital has a diverse patient mix: Of its 1,079 beds, it has a 323-bed psychiatric unit, 523 beds dedicated to long-term care, and it also has a secure forensic unit servicing the local prison population.
With 40 percent of inpatient volume coming from the ED and a complex mix of patients who need medical attention and support for social, financial and other needs, “understanding who’s coming in not only affects how we staff our ED to make sure we‘re providing the best patient care that we can in that setting, but so does understanding our service mix and staffing mix on the units where our patients end up,” says Alex Filipiak, Bergen’s director of finance.
With an eye toward using analytics to improve quality, better coordinate care, ensure patients are transitioning to appropriate settings and to increase efficiency and optimize staffing, Bergen recently went through a major overhaul of its IT systems, allowing it to connect to other organizations though the New Jersey Health Information Network.
“The facility’s IT infrastructure was in desperate need of modernization, so we did a full infrastructure refresh of our hardware, we enhanced our Wi-Fi throughout the entire hospital, we launched email and other communication platforms for staff to be able to communicate properly, and we also upgraded our electronic medical records to the latest version to allow us to have this kind of interoperability and connectivity,” says Bergen CIO Jennifer D’Angelo.
“We’re trying to get facilities to streamline data so it’s meaningful and manageable to improve patient outcomes,” she adds. “It’s not a competition; it’s a collaboration of providers. The more we as providers collaborate with each other, with our community partners and physicians, our patients reap the benefits of that. It’s incredible what we can do if we partner together.”
“The More Data You Have, the Better”
Accurately predicting patient ebbs and flows and using that data to inform staffing has allowed Bergen to make some practical and fairly simple changes that affect the organization’s bottom line, says Thomas Amitrano, Bergen’s chief nurse executive. When the data revealed a recurring midday surge, for example, they added an 11 a.m. shift — an improvement over staffing equally for an average volume, Filipiak says.
“Certainly, that would reduce the need for overtime, and, from the employee satisfaction standpoint, it means someone’s not working a double, not getting called in, not getting sent home,” he says. “We always will be able to pull staff to meet volume. The question is whether we’re paying that staff overtime because we didn’t prepare for it or if we’re paying them regular time.”
Right-sizing staff is just the beginning of what predictive analytics can do.
“We ask who is showing up where, what kind of diagnoses are coming in and in what volume and acuity so that we can not only staff but staff appropriately,” Amitrano says. “The more data you have, the better, as long as you can put it into a framework that is actionable and that actually tells you something.”
Case in point: In March and May of 2019, New Jersey health officials reported a measles outbreak. Data allowed Bergen to ensure that everyone on staff was immune to the disease. Analytics also helps Bergen, which has the largest detox program in the state, to identify patients with substance abuse disorders and flag likely cases of doctor-shopping for opioid prescriptions. That data helps ensure patients are getting the help they need — the right level of care in the right setting, Amitrano says.
“Those are the type of analyses that this affords us that we didn’t have before,” he says. “For me, the excitement comes from how we look at delivering the care and what changes we can make to better the patient experience. This type of data helps us best prepare for who we care for.”
An Enormous Return on Investment (ROI)
Tools such as predictive analytics programs bring both soft and hard ROI — and can even pay for themselves.
“If you start applying a tool like this to the entire practice, the return on that investment in time, energy and critical thinking is enormous,” Jensen says.
But financial ROI isn’t necessarily the most important measurement. “If you can take an emergency department and move it from a staff feeling chaotic and overwhelmed to a staff that feels busy, efficient and confident that they’re delivering high-quality care, that’s a huge return on investment too,” he adds.
The data also improves quality of care, D’Angelo says. For example, it can alert staff to patients that meet high-risk criteria for hospital readmission based on the number of times they’ve visited the ED in a certain span of time.
“We’re able to log in to this portal and see a wealth of information immediately about that patient,” D’Angelo says. And data-sharing makes it easier to connect patients to community and other resources. “That’s really what the data should be used for,” D’Angelo says.