Mercy Virtual Care Center: A deep dive into a virtual hospital

Mercy Virtual Care Center: A deep dive into a virtual hospital Bill Siwicki

With nearly a decade of successfully running a virtual ICU and a virtual care hospital known today as Mercy Virtual Care Center, Mercy in St. Louis, Missouri, embarked on amplifying a new digital-first frontier in 2015: A new virtual care model of remote patient monitoring for patients living with complex conditions.

THE PROBLEM

"In our ambulatory virtual programs, we manage complex chronically ill patients," said Annie Bannister, RN, executive director of vEngagement at Mercy. "We follow these patients longitudinally, meaning when someone enrolls with us, we will manage them alongside their primary care physician and specialists throughout their entire care journey.

"Some of the chronic diseases being managed are CHF, COPD, diabetes and chronic kidney disease, with many patients living with two or more of those conditions," she continued. "The patients are enrolled in RPM through Mercy Virtual's patient engagement program, vEngagement – "v" for virtual."

The program was born out of a need to better manage the sickest tier of Mercy's patients. The sickest 5% of this cohort are responsible for 50-60% of expenditures. Despite this high-cost care, these patients experienced fragmented care, frequently cycling in and out of inpatient and outpatient settings.

Historically, the traditional healthcare system was not equipped to deliver the high-intensity care required to manage this group of complex patients, she noted.

"Patients with chronic diseases often deal with daily symptom burden," Bannister explained. "Before vEngagement, the system was not set up to offer on-demand daily symptom triage and management. Sadly, after attempting to navigate this messy world of chronic disease alone, many patients stop trying to figure out what symptoms to be most worried about.

"Also, many patients with chronic disease do not want to be a burden to family members," she continued. "Because of this, many patients wait until symptoms are so bad that they have no choice but to call 911. The need for daily communication to help a patient navigate "Is what I'm feeling normal today" and other questions that vEngagement enables was a wholly new concept in care service."

Bannister and other providers came to vEngagement from internal medicine. They noticed a gap in care for this particular group of patients and believed that virtual care could bridge this gap between ambulatory and inpatient. They hoped to replace fragmentation with a cohesive, integrated, patient centered, data-driven care delivery.

"We started with a small pilot of 50 patients in Washington, Missouri," Bannister recalled. "We were determined to answer the question, 'Can we apply higher intensity care to this cohort and improve outcomes?' It didn't take long to determine that the answer was 'Yes.'

"In a few months, we began to see decreases in hospital utilization, patient satisfaction scores at 98% favorable and closing of gaps in care," she continued. "Our pilot program became the added layer of support this group of patients desperately needed."

"We recognized that we couldn't keep relying on hiring more labor and that personalized, individualized and proactive triage care was fast evolving."

Dr. Gavin Helton, Mercy

With the early introduction of home biometric monitoring, daily surveys and real-time access to providers, the virtual team was able to deliver high-intensity, comprehensive, individualized care that was previously not possible.

Many patients in RPM reside either in extremely rural or urban areas. Yet their obstacles to accessing care are the same regardless of geography.

"When talking about their specific diseases, we at Mercy Virtual intentionally take a holistic approach to patient care," said Dr. Gavin Helton, senior vice president, population health, at Mercy. "These individuals can have any number of diseases and chronic conditions that are incurable; oftentimes, some of their complaints or symptoms might be related to a secondary diagnosis or perhaps a behavioral health concern or unmet social needs, for example, loneliness.

"We realized the need to have a dedicated care team who very intentionally leverages technology to scale and provides the highest quality of care services to those who most need it, all in a proactive manner," he continued. "But we also very intentionally develop relationships. vEngagement is all about our singular focus on patient engagement. Our goal is to show respect for the dignity of the individual and treat them in a holistic manner."

As part of treating their chronic illnesses, staff help them to become comfortable maintaining their own self-care in the home. These patients still have continual needs requiring traditional healthcare episodically. When that occurs, they fall in the gaps in care and end up receiving acute care in a facility as opposed to care being brought to them proactively in the home.

Bannister keeps remembering the landscape when she started with this program and the COVID pandemic hadn't yet hit, virtual visits weren't a thing that were happening with patients.

"But we knew it was something our patients needed," she stated. "We needed to bring the care to them. For many of our elderly chronically ill patients, the burden of getting into the car and out of the car and down the office hallway into an exam room often prevented them from getting the care they need in the traditional office setting.

"These patients are short of breath and often don't move around well," she continued. "It's rewarding to just be able to bring care to them. I don't know that we highlighted how important that was to patients being able to connect them with a physician or a nurse practitioner or a nurse without having to have them leave their home."

It seems normal to Bannister now after the COVID pandemic.

"COVID has really normalized these virtual visits," she said. "But when we started, one of the greatest things we could accomplish was just bringing care to patients."

PROPOSAL

During the RPM pilot's first few years, patients' days in hospital decreased and satisfaction shot up. vEngagement was closing important gaps delivering the type of quality care so desperately needed, Bannister said.

"Patients increasingly started to enroll, and our healthcare communities called for expansion," she recalled. "We began scaling vEngagement for growth quickly from Washington to other Missouri communities, such as St. Louis and Springfield, and then to Northwest Arkansas and Oklahoma.

"We had also introduced biometric monitoring and daily surveys early to the program," she added. "For the first time, these patients with complex, chronic illnesses had real-time access to providers ─ something unheard of in their world. Like everyone else, they're used to calling their primary care physician or specialist's office and hearing a 'Press one' or 'Press two' triage or getting a voicemail."

Consequently, with the pilot's success, the objective was to scale this intervention to reach as many high-risk lives as possible.

"As we scaled the program from one community to the next, we also began increasing our team to take care of this ever-growing patient cohort," Bannister said. "By 2018, our team of registered nurses, advanced practice providers, providers and physicians grew nearly five-fold and continued to increase every time more patients were added.

"When our vEngagement team grew to more than 100 members, we shifted to looking at leveraging software-enabled productivity," she continued. "We knew we would be unable to continue this growth rate or we would risk encountering a shortage of nurses and advanced practitioners in the St. Louis area."

Bottom line, staff reasons for considering technology adoption were twofold: One was that once a care team becomes larger, the management infrastructure required serving a team that size becomes unwieldy.

The second reason is that they would soon run out of healthcare workers to hire and from a fiscal and operational standpoint, program expansion no longer made sense to continue without automation. Further, staff were sitting on a mountain of productivity data just waiting to be unlocked.

"Five years later, in 2020, we turned to Myia Health's cloud-based RPM platform to help us accomplish four RPM strategies reaching more patients without having to add more team members," she explained.

Risk stratify the patient population. This strategy was necessary to deliver the right intensity of care to every patient based on their individual needs at a moment in time. Staff were still applying a "One size fits all." If they wanted to add more patients, they needed to know which patients were stable and could have their care de-escalated and which patients needed frequent touchpoints.

Myia Health had the promise of machine learning models to help staff achieve risk stratification and deliver the right intensity of care to patients: An acuity model and time-to-event model. The acuity model would break down the population into high, medium and low acuity levels. The time-to-event model would notify staff when patients appear to be nearing hospitalization.

Leverage asynchronous messaging. Then, patient communication was happening over the phone and through video and staff knew these modes were not the fastest and most convenient for patients to access care.

Myia Health held the promise of "closed loop" asynchronous messaging for patients, which was really desirable. Closed loop means a care team member receives "read" and "unread" messaging receipts to let them know if a patient received the intended message.

"For instance, a clinician sends the following e-message to a patient: 'Hey, Bob, you started taking your antibiotics yesterday. Are you feeling better?'" Bannister explained. "The clinician can move on attending the next patient knowing that in 30 minutes, if Bob doesn't read their message, an alert will inform the care team, 'Bob's message is unread.'

"This closed loop gives clinicians the security of trusting the messaging system knowing all important messages will be received," she added. "So they can intervene."

Reduce alert volume. Chasing alerts was by far how many of the Mercy caregivers spent most of their day. Complex, chronically ill patients take their vital signs daily, which are automatically transmitted to the care team.

Certain alerts trigger if vital signs are outside the individual patient's threshold. For example, if blood pressure is greater than a certain number, an alert flags the provider. But if thousands of patients are entering daily vitals, the number of alerts being generated skyrockets.

"We needed a way to dial down the volume of false alert noise," Bannister said. "Myia Health would apply a data-driven approach to ensure clinicians weren't chasing unnecessary alerts and therefore focused on the patients who they needed to care about."

Panel views to allow providers (not alerts) to drive care. Mercy's former software platform was alert-driven. Providers would jump from alert to alert, bypassing assigned patients and their potential care needs.

Instead, the alerts were driving the care.

"In 2023, we plan to launch a co-developed standard deviation weight model that will reduce the weight alerts by greater than 60% and overall alert volume by 23%."

Annie Bannister, RN, Mercy

"We wanted to change to a panel view of patients sorted by tiers based on acuity levels and put the provider in the driver's seat administering care," Bannister said.

Myia Health's promise of a human-centered design approach to development was especially attractive, Helton noted.

"A couple years before the pandemic, we were transitioning to virtual care at a time when few health systems were serious about telehealth," he recalled. "We did not want a turnkey solution, believing that solution type would not actually meet our needs. As our patient populations changed and as our vEngagement program changes, the RPM program and platform must change with us.

"We were growing and changing and the appeal of iteratively evolving Myia's solution with clinician feedback was definitely well received," he continued. "Mercy Virtual's RPM program is all about serving more patients more effectively. We recognized that we couldn't keep relying on hiring more labor and that personalized, individualized and proactive triage care was fast evolving."

There was an intentional dependency on technologies for asynchronous communication, or messaging, that superseded one-on-one scheduling of staff time, was reactive in nature, provided educational and other resources outside the traditional in-person visit, and was more aligned with the patient's convenience and schedule, he added.

MARKETPLACE

There are many vendors of telemedicine technology and services on the health IT market today. Healthcare IT News published a special report highlighting many of these vendors with detailed descriptions of their products. Click here to read the special report.

MEETING THE CHALLENGE

Myia Health equips Mercy Virtual patients with RPM kits containing a tablet to monitor their vital signs and online capabilities to call, text or conduct a video consult with their virtual care team; biometric monitoring devices such as a blood pressure device, scale, pulse oximeter and thermometer.

Mercy Virtual also sends a "rescue kit" akin to a pencil bag affixed with a precautionary message: "Please do not use these medications without talking to your providers first" and has the appropriate phone number. When the vEngagement provider meets the patient, they determine which health conditions typically send them to the hospital.

They then send medications to their local pharmacy to keep in the rescue kit. For example, if they frequently have COPD exacerbations, staff send steroids and antibiotics to the pharmacy to pick up and put in the rescue kit. The rescue kit comes in handy when, for example, a patient with frequent urinary tract infections needs an antibiotic at 2 a.m. Christmas morning when the pharmacy is closed.

Mercy Virtual implemented four remote patient monitor program strategies.

"First, we risk-stratified the complex, chronically ill population with the machine learning-based acuity and time to-event models," Bannister said. "The acuity model, which utilizes electronic health record and zip code social determinants of health data, sets appointment cadence to match patients' right level of acuity assigning a risk status of high, medium or low.

"With that, we scheduled intentionally for the first time each participant's risk status at weekly (high), bi-weekly (medium) or monthly (low) cadence to match their acuity," she continued. "Previously, our team saw patients round the clock. The time-to-event model assisted how clinicians monitor and triage through alerts activating at the same time, prioritizing the identified high-risk patients."

Mercy Virtual implemented its second strategy developing a closed loop asynchronous system for messaging patients. With this system, providers are alerted to any message that went "unread." This closed loop allowed providers to push much of the conversation previously occurring over the phone to messaging.

"Our third strategy, reducing alert volume, presented one of the greatest value adds to our alerting through Myia," Bannister said. "A single provider feedback feature attached to every alert poses the question, 'Did this alert cause a change in care? Yes or no.'

"Every time a clinician clears an alert, this question is a hard stop," she continued. "The clinician's response allowed us to gain rich data around which alerts were not yielding changes in care and could be eliminated. While this question seems so simple, its inclusion saved much time prioritizing alerts that mattered and eliminating those that didn't."

To date, staff have gained a deeper understanding of the different types of alerts and found that the weight alerts were the highest volume of alerts, accounting for 44%. These alerts, volume-wise, are in the thousands of numbers that staff are clearing on a monthly basis. Low weight alerts result in a change in care only 1.7% of the time. High weight alerts resulted in a change in care only 9.8% of the time.

But when combining a symptom-based patient survey response with vitals, for example, if their weight is abnormal and the patient says they're short of breath or experiencing swelling, that would increase the likelihood that staff would change care by two to three times. About 71% of all alerts result in care changes less than 10% of the time.

"This closed loop feedback helped us visibly account for our time and acquire an understanding of which levers to pull to reduce alert volume," Bannister said. "In 2023, we plan to launch a co-developed standard deviation weight model that will reduce the weight alerts by greater than 60% and overall alert volume by 23%.

"For the fourth strategy of implementing panel view, Myia Health created team views that allow a clinician to review a whole team's work priorities or only their assigned patients," she continued. "This panel view features a 'watch list' empowering providers to add patients to the 'priority patients' list for their full focus at any time and drive care, not alerts."

If the patient answers a survey that they are feeling worse than usual or has vital sign readings out of range, they move into this bucket of priority patients. That gives the provider a chance to see their entire group including priority patients at a glance.

The "watch list" used to be a sticky note on every provider's desk listing the names of patients who they worried about. Having a list of all patients about whom they are concerned is not only a time saver, but increases the quality and safety of the care provided, Bannister said.

"Now, if a provider is unexpectedly absent, all the other team members know exactly who that provider was concerned about," she said. "Other software enabled productivity gains through Myia Health are: Calling patients directly to tablet, consumer view ability to call patients, and bundling multiple alerts to one patient for a more efficient, patient-centric workflow."

The Myia Health platform integrates through an API to Mercy's Epic EHR. Myia pulls patient demographic data on a nightly basis.

"What is really nice is when entering a patient referral for vEngagement into Epic, a new patient profile is built automatically," Helton explained. "This demographic data is now accessible in Myia until that patient is assigned to a care team. Once the patient is assigned, Epic will send different types of notifications such as a hospitalization if they are admitted, discharged, scheduled for appointments and so forth visible on the clinician's tablet.

"Epic EHR is our source of truth and so it was really important that Myia be directly connected bi-directionally" he continued. "What we worked hard to avoid was multiple, siloed data sources that lead to fragmented approaches to care."

RESULTS

From the initial pilot of 50 patients in 2015 to today, Mercy Virtual's vEngagement remote patient monitoring program has grown and scaled exponentially. As of October 2022, it has 4,474 active lives under management and 207 patients waiting to onboard.

Of this population, 60.5% is between the ages of 70-89 and 84% is between the ages of 60-89, which represents the largest number of enrollees; and overall, 56% are female and 42% are male.

Bannister detailed some of the hard results achieved so far:

Doubled lives under management without increasing FTEs. One of the biggest problems staff tried to resolve was increasing the number of lives under management without increasing the number of team members. Myia's approach to RPM technology enabled staff to double the number of patient lives without adding any FTEs. Staff scaled from around 2,300 patients to the current 4,600, with plans to enroll 5,000 more patients on the same staff structure.

Reduced cost total operating expense/UOS (patient) from > $750/pmpm to <$300/pmpm. Scaling has reduced the total operating expense per patient per month from more than $750 per patient per month to less than $300.

Established an NPS patient satisfaction score of +86. Bannister and staff are proud of their high patient satisfaction rating. Understandably, this group of patients who struggle with daily symptom burden and have experienced prior consumption of fragmented care are not satisfied historically. Their enrollment in the RPM was a turning point. Staff received more than 3,000 patient responses in the last round of quarterly patient surveys as of Nov. 14, 2022. Staff currently are maintaining an NPS 86 score with the latest unique survey capture of 3,128 patient responses.

Program evaluations over the years show a 50% reduction in inpatient utilization. Over the years, evaluations have examined the program's influence on inpatient utilization. That is a metric that staff always are trying to impact; for example, striving to increase the number of healthy days at home and decrease hospital utilization. Internal and external evaluations have hovered around a 50% reduction in inpatient utilization or 50% less times a patient is admitted. An average hospitalization for a CHF admission is approximately $14,000.

"First and foremost, we have an obligation to optimize care for the patients that we're privileged to serve," Helton stated. "Secondly, our technology investments must all be sustainable solutions proving cost avoidance and measurable savings, especially as Mercy is committed to moving up the risk curve in terms of our value-based reimbursement model. That's our financial justification for not only continuing RPM but continuing to grow the program.

"We don't ask the patients to pay for this service, which is most likely a barrier to adoption," he continued. "The program's financial support is derived from our participation in the Mercy Health ACO LLC, with upside and downside risks. We also have multiple Medicare Advantage contracts that place us at either complete or near complete financial risk for the care of these patients."

While talk of financing seemingly separates Mercy Virtual from its primary goal, both complement one another, he added. It's important to represent the sustainability of high-quality care, he said.

ADVICE FOR OTHERS

"We knew going into this program that we had years of clinical and operational experience of caring for complex, chronically ill patients in the home," Helton said. "That's what led us to seek a better alternative in terms of a platform allowing us to data drive different clinical behaviors and individually triage patients as well as seek automation and asynchronous messaging.

"But we also recognized there was no solution available on the market to meet those needs," he continued. "Therefore, we committed to a strategic relationship knowing we had years of experience in this space. We were confident that Myia Health was the right team to evolve their solution to iteratively solve more problems and unlock scale."

Years later, Helton would look back and say, "Gosh, we knew it would be difficult. We knew it would be time-consuming. We didn't know it would be this difficult or this time-consuming."

"I think that's part of the reality check for both Myia and Mercy," he said. "Yet looking back, would we do it again? The answer is absolutely. This realization should not be sugar coated. Quite the opposite. We should embrace that operational experience we started with and have gained since forming our relationship with Myia.

"It's that relationship of a tech software analytics driven platform plus our clinical team that together is surfacing and guiding new requirements, and being willing to change clinical behaviors based on the data," he continued. "I'm so proud of our progress, because it does allow us to establish a new standard of care for these patients."

Staff follow evidence-based guidelines to direct medical therapy with this program; and all the while they are practicing a standard of care far superior to what's available on the market, he added.

"And that's only because we've taken this collaborative approach that's been very patient-centered," he said. "Myia and Mercy have done what's right for the individual patient with an eye toward sustainability from the start. That makes it sound easier than it has been, but we should not hide from sharing our experience. If it were easy, someone else would have already done this and we would have bought it off the shelf."

For clinician Bannister, projects take longer than desired to be delivered in the tech space.

"Expect to plan for far more development time than is projected," she cautioned. "We were surprised by the whole body of work required to analyze our volumes of data. Our data is rich from the collection of vital signs entered and daily surveys entered and messages sent by thousands of patients.

"You must be able to digest and make sense of the data ─ a body of work that we didn't appreciate initially," she continued. "While it seems easy to be labeled 'data driven,' the actual interpretation of the data was far more difficult and time-consuming than we anticipated."

Make sure the clinical care delivery model is prepared for incorporating virtual care ─ especially for large integrated healthcare delivery systems, Helton advised.

"In fact, as opposed to thinking in terms of a fee-for-service or a volume-type of mindset, I recommend instead focusing on strong alignment preparing both your care delivery model for digital transformation in tandem with your reimbursement model," he said. "The two models should focus on the same goals: maintaining the health of the patient, preventing unnecessary utilization, and improving the quality of care and quality of life."

Bannister recommends reducing as many barriers as possible for patient entry to an RPM program.

"Mercy does not bill our chronically ill patients' insurance for interactions with the nurses and advanced practitioners or bill for the cost or use of the vital sign equipment," she explained. "Our type of RPM program is unique and takes getting used to.

"We build patients' trust so that they learn to love it; however, adoption is not instant," she continued. "Even adoption from community providers was not instant. Acceptance and buy-in from our community partners and patients took a long time to earn."

It took a couple of years for community partners to appreciate and endorse the high intensity delivered by the vEngagement team.

"Patient by patient, community by community, we were able to gain their trust delivering on-demand, individualized patient experience that had not been replicated previously," Bannister said. "Lastly, patient adoption of your RPM technologies must be prioritized. Our acceptance rate is still not 100% despite the fact that our program has measurable, proven results.

"Our acceptance rate is still only about 60%, meaning 40% of patients whom we solicit in outreach to join ultimately decline – 'Do you want this free program with access to nurse practitioners and physicians whenever you want and all of this free equipment?'" she continued. "Here is where we lean on our community partners to endorse us."

When a primary care physician advises a patient to join vEngagement, the success rate is threefold higher. Securing buy-in from the community partners is key for RPM enrollment to be successful, she concluded.

Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email the writer: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.