Hospitals today face a myriad of challenges as they continue to evolve their delivery models toward value-based care. They are tasked with increasing operational efficiency, improving margins and enhancing the patient experience — all in a way that preserves the cognitive capabilities of their care teams.
The promise of big data and analytics initially excited many hospital leaders, but over time reality has dampened that enthusiasm. Too many hospitals today operate in a deluge of data that fails to offer insights and actionability in real time.
Clinical staff should not have to "drown in dashboards and reports," said Mudit Garg, co-founder and CEO of Qventus, during a workshop at Becker's Hospital Review's 8th Annual Meeting in Chicago. More sophisticated than analytics, artificial intelligence software can process massive amounts of data, anticipate problems and communicate specific course corrections to frontline workers in real time.
During the workshop, executives of three leading healthcare institutions discussed how they prioritized a focus on operational excellence, supported by the use of Qventus, to improve efficiencies, achieve better outcomes and enhance the care experience for patients and clinicians.
Case One: How AI Supports Perioperative Care at Lucile Packard Children's Hospital Stanford
Kristin Petersen, vice president procedures and diagnostics and centers of excellence with Stanford Children's Health, Lucile Packard Children's Hospital in Palo Alto, Calif., said advancements in technology and pharmaceuticals have made pediatric care more effective, but also dramatically more complicated and expensive.
"The coordination of those children has become incredibly complex," said Ms. Petersen. Lucile Packard Children's Hospital has the highest case-mix index in the nation, with each pediatric visit requiring an average of 40 consultants. "Think about the communication — all of the data going into the EMR. It was leading to cognitive overload for our staff," she added.
The hospital identified a particular need to improve coordination of care and enhance efficiencies in the operating room — not only for scheduling purposes, but to provide the best possible patient experience.
"Like most hospitals, we have patient navigators to support the patient experience," said Ms. Petersen. These navigators provide updates to the anxious parents whose children are in surgery. This task means navigators must anticipate when a surgery might run over. "If a case goes over, the parents are just waiting and don't know what's going on. Meanwhile, the navigators are running back and forth, trying to keep them in the loop."
The process was inefficient and provided little comfort to parents, who were usually informed of overrun cases after the fact. After implementing Qventus' real-time decision-making platform, patient navigators were armed with real-time information to act proactively versus reactively. A navigator can now stay in the waiting room with parents and receive a text message regarding a surgery delay or a predicted run over.
The ability for patient navigators to remain by the parents' side and communicate with them about the status of the surgery in real-time was immensely comforting for parents. According to Ms. Petersen, patient satisfaction scores shot up 18 percent just two months after implementing the software.
Lucile Packard Children's Hospital also saw a reduction in case delays and improved start times, room turnover. Ms. Petersen said case time estimation was previously difficult due to the many stakeholders involved and numerous opportunities for breakdowns in communication.
The system enabled Lucile Packard OR staff to predict the likelihood of delays by analyzing the type of procedure, the surgeon's historical record, the patient's medical history and other factors, then prompt a decision-maker with options to adjust the schedule accordingly. Perioperative efficiency increased dramatically. Ms. Petersen cited a 10 percent reduction in case overruns and 15 percent decrease in case under-runs within four months of implementing the software. The percentage of cases delayed by more than 10 minutes decreased by 11 percent, cumulatively resulting in 520 hours of delays prevented since the launch, while same-day cancellations decreased by more than 25 percent.
Case Two: AI in the ED
Vance Moore, president of business integration at Chesterfield-based Mercy, said that in 2015 the health system identified a need to improve operational efficiency in the emergency department, where long wait times and inefficient workflows prompted too many patients to leave without being seen.
"We had some horrible situations in some of our EDs," said Mr. Moore. "Roughly 5 percent left without treatment. We were diverting patients all the time. "
Despite overall satisfaction with Epic as the health system's main medical records system, Mr. Moore likened the EMR to "writing books and then putting them away on the shelf" — a system of record versus a system of action.
"We need these systems of record, but we also need systems that can display and link data to solutions to prompt action," said Mr. Moore. Analytics systems that can do this, he added, will help health systems achieve a true return on investment on the EMR.
Through artificial intelligence, Mercy improved patient throughput and satisfaction in the ED. The Qventus system reviews EMR data and a range of demographic data to identify ED patients likely to leave the ED before being seen. If necessary, the system will provide care providers with recommended actions so they can speak to patients and intervene before they walk out.
This has yielded dramatic results, according to Mr. Moore. Since implementing the solution in February 2016, Mercy Fort Smith has seen a 30 percent reduction in patients who leave without being seen and a 24 percent reduction in door-to-doc time. It helped expand ED capacity by 15 percent using only existing resources.
"Our industry's traditional solution for ED issues was to be build, build, build," said Mr. Moore. "Now it's improve, improve, improve."
Case Three: Reducing Inefficiencies in Outpatient Care With AI
Karim Botros, chief strategy officer at MetroHealth, said the safety-net health system has a payer mix of 50 percent Medicaid, 25 percent Medicare and 25 percent commercially covered patients. MetroHealth doesn't have much financial leeway when it comes to operational inefficiency and waste, yet Mr. Botros said analysis revealed 15 to 20 percent of its primary care physician base was under-utilized, primarily due to a high rate of no-shows.
At the same time, many patients were waiting weeks and sometimes months to see a physician. These prolonged wait times became barriers to care and impeded population health management efforts that are becoming more important to the health system's success amid the transition to value-based care, he explained.
"If we have 300 primary care physicians and we only utilize 80 percent of their time, that's 40 to 50 FTEs worth of doc time that could be made available for better patient access and population health management," he said.
Of the 600,000 annual patient visits, MetroHealth had a no-show rate that ranged from 5 percent to 35 percent, depending on the location. Instead of adding more physicians or attempting to change behaviors, Mr. Botros and his team instead decided to change the health system's scheduling system. Instead of taking the "simplistic approach" —overbooking physician schedules — Mr. Botros and his colleagues sought an analytics system that could precisely predict the likelihood of a patient missing an appointment.
Because MetroHealth was an early adopter of the Epic EMR, Mr. Botros said the health system had no shortage of data. However, "we realized that with all of that data it becomes very hard to make decisions in real time," he said. Mr. Botros acknowledged some benefit to predictive analytics, but even those insights tend to be too broad. "We had to become more sophisticated — to get to the individual physician and patient level," he added.
MetroHealth is still in the process of implementing Qventus, but the solution has already made a tangible difference. The system's scheduling team is now equipped with real-time information to determine "this doctor can handle one more visit today or this doctor has too many patients scheduled," Mr. Botros said. "If that simplicity can be delivered behind the scenes, we will have unbelievable success."
In Conclusion
EMRs have helped standardize data entry and record-keeping in hospitals, but they also contribute to "death by data." Analytics systems that only offer the opportunity for a retrospective review or predictive features based on broad historical data are not enough to help clinicians make better decisions in real time. However, with AI, hospital staff are empowered to make accurate, timely decisions based on individual patient and provider data. This can result in millions of dollars of savings.
"The beauty of this is that your team becomes like a highly-coordinated football team," said Mr. Garg. "If you have a team of the strongest athletes and they just have a pre-planned set of moves and run really hard after the ball, they are not going to be very successful compared to a team that assesses and evaluates the situation and is able to call the right 'play' in the moment. When your team and operations adapt like that, you have a huge competitive edge. You can handle the complexity and variability of healthcare at a much lower cost."