A Novel Large Scale Integrated Telemonitoring Program for COVID-19

Introduction

Coronavirus disease 2019 (COVID-19), first described in December 2019, ¹ has been particularly tough on the Washington, DC, and Baltimore metropolitan regions (Capital Region). ^2,3 Kaiser Permanente of the Mid-Atlantic States (KPMAS) is an integrated health care delivery system that provides care for >770,000 members in the Capital Region. ⁴ In October 2019, 5 months before the arrival of COVID-19 to the Capital Region, KPMAS established the virtual home care program (VHCP). The aim of this telemedicine-based program was to transfer a portion of lower acuity cases out of the hospital setting. For example, a hemodynamically stable patient requiring intravenous antibiotics for cellulitis could be referred to the VHCP, instead of being sent to the hospital. VHCP providers (MDs, NPs, and PAs) perform daily virtual rounds (telephone or video check-ins) with patients in the comfort of their own home, monitor disease progression, and make necessary adjustments in the treatment plan.

When the COVID-19 pandemic hit the Capital Region in early March 2020, KPMAS saw an opportunity to leverage the newly formed VHCP in the care of its COVID-19 patients, based on previous work suggesting decreased rates of emergency room (ER) and hospital utilization in communities with heavier use of telemedicine, ^5,6 as well as expert opinion that telemedicine may help to slow down disease transmission from COVID-19 by allowing for more adherence to social distancing. ^{7 –9}

From the outset, the goal of the VHCP was to perform daily check-ins on COVID-19 positive patients to help identify those patients requiring a higher level of care in the setting of an urgent care, emergency room, or hospital. Secondary goals included offering supportive care to COVID-19 patients, by prescribing medications to help with nausea and cough, providing guidance on home isolation and social distancing to limit disease spread, giving advice on when and how to come off isolation and safely return to work, and offering reassurance for anxious patients.

As of November 1, 2020, 13,508 COVID-19–related cases had been referred to the VHCP, resulting in a 25-fold increase in the average daily patient census managed by the VHCP. This article aims to describe the rapid upscaling of this program, and the unique set of problems that were created and addressed along the way, as well as describing some of the shortcomings of the program. In addition, this article shares some preliminary observational data that have been generated by the VHCP.

Methods

For KMPAS members tested in a Kaiser Permanente facility, patients testing positive for COVID-19 (defined as positive nasal or nasopharyngeal NAAT test for COVID-19) were automatically referred to the VHCP through an algorithm built into the KP electronic medical record (EMR). Patients testing positive for COVID-19 outside of the KPMAS system, as well as patients with suspected COVID-19, could be manually referred to the program by any KPMAS physician, including their primary care physician. Drawing from these two referral sources, the VHCP was ultimately involved in the care of 73% (13,508) of all KPMAS members testing positive for COVID-19, through November 1, 2020. The remaining 27% (4,972) of COVID-19 patients were not referred, largely as a result of having tested positive outside of the KPMAS system, and/or being admitted to the hospital at the time of their diagnosis.

Starting March 1, the VHCP census rose from an average daily census of 40 active patients and peaked at 1,470 patients in early May, dropping off to ∼300 patients during the summer months before rising again in early October. This necessitated a proportional increase in provider staffing, from 4 physicians (known as the “core” physicians), 1 nurse, and 1 program manager to >292 physicians, 70 nurses, 17 nurse practitioners, and 4 physicians assistants. This resulted in a daily average of 29 patients per physician.

To accommodate this rapid expansion, standardized onboarding tools and workflows were created by the VHCP physicians and program manager. All new providers entering the program receive an in-depth orientation through Microsoft Teams™ meetings on the morning of their initial start date. Orientation sessions are led by one of the four “core” VHCP physicians, and involve the use of the screen sharing feature in Teams™ to demonstrate the specific workflows unique to our EMR, including medication ordering, charting, and tracking of a patient's pulse oximeter readings. In the beginning, an average of two to four new providers were oriented during each workday. Orientation sessions last for 1 hour on average but can take longer depending on the number of participants and their degree of familiarity in managing acute respiratory illnesses. During orientation sessions, the core physician demonstrates the use of the group chat feature built into Microsoft Teams, into which a provider can post a clinical question that is viewable by the entire team. A typical question posted might include a clinical description about a given patient, followed by a request for advice on whether to send that patient to the hospital, to start the patient on oxygen, etc. The chat feature is also used to disseminate critical information throughout the day, such as reporting on temporary closures of medical centers and sharing of essential phone numbers (translator services, pharmacies, and hospitals). Starting in April, a dedicated pulmonologist was assigned to the VHCP team each day to field higher level clinical questions, for example, regarding the initiation of antibiotics, steroid therapy, or oxygen for a given patient.

In addition, two team-wide virtual meetings are held each day, one in the morning and the other in the afternoon. During these meetings (referred to as “huddles”), the team discusses any challenging cases that came up throughout the day, reviews important announcements, and holds refresher sessions on various aspects of the digital workflow. In particular, the afternoon huddle has proved invaluable for new providers who needed advice on how to handle a difficult case.

Before the morning huddle, each provider is assigned a list of patients to contact that day. Patient lists contain basic demographic information (name, age, gender, etc.), as well as a brief written sign out from any previous providers, allowing providers to easily identify the highest risk patients on their list, so that they can prioritize contacting those patients earlier in the day. After the morning huddle, providers are free to proceed through their lists in the order they deem to be most clinically appropriate.

For all new referrals, a dedicated team of nurses (RNs) was established to perform chart reviews and make the initial outreach phone call for each patient. These RNs were drawn from KPMAS clinical specialties that had shut down or cut back on operations due to the COVID-19 crisis. In some cases, patients were already aware of their COVID test result, but if not, the RN informs them of their test result and provides specific advice on isolating and preventing the spread of disease to other household members. RNs on this team can escalate any case to one of a core VHCP physician in real time for patients with urgent symptoms, such as chest pain or shortness of breath at rest. The patient is then assigned to a physician or mid-level provider (NP or PA), who reaches out to that patient the following day.

All patients enrolled in the program received daily outreach calls by a provider, after the initial outreach call by the RN. These providers represented a wide range of specialties, including fields as diverse as dermatology, pain medicine, plastic surgery, orthopedics, ophthalmology, and obstetrics & gynecology. Recognizing the varied clinical backgrounds of these physicians, standardized note templates were created to facilitate patient care and documentation, with a list of essential symptoms to track such as fever, cough, and shortness of breath. ¹⁰ These templates contained built-in algorithms to help providers decide when and how to escalate a given case. For example, if a patient answered “yes” to having shortness of breath at rest, a decision tree would appear instructing the provider to send that patient into an urgent care site or ER. Note templates used by RNs had a few more decision aids built into them compared with notes used by physicians, including instructions to inform a physician about any patient reporting any degree of dyspnea.

A phone line was established to receive incoming calls, which is covered by an RN or a provider 24 h a day, 7 days a week, and patients were encouraged to call in directly for any urgent concerns, such as for worsening symptoms. During the period from May 1 to November 1, we received an average of 117 incoming phone calls per day.

The team understood that there would be limitations with performing patient assessments over the phone, including lack of objective data such as vital signs to aid with clinical decision making. As a result, VHCP invested in the use of portable pulse oximeter devices for COVID-19 remote monitoring, purchasing 5,000 devices in March 2020. Standardized criteria were created to determine which patients would receive a pulse oximeter, including age >75 years, body mass index (BMI) >30 kg/m², and history of chronic lung disease. These criteria were adjusted as the team obtained more data and gained more experience. For example, with the help of OB/GYN physicians who joined the program, a less restrictive set of criteria was created to determine which OB patients should receive a device.

Pulse oximeter devices are provided to patients at the time of discharge from KPMAS urgent care or hospital. A courier service was established to deliver pulse oximetry kits directly to patients at their own home, with an average turnaround time of <4 h from the time of order placement to delivery to the patient's home. VHCP partnered with a third party who developed a “COVID-19 Monitoring” virtual dashboard (Validic, Inc.^©), to allow patients to upload oxygen saturation percentage (SpO₂), heart rate, temperature readings, and severity of cough and shortness of breath (on a scale of 1–10) into a portal accessed by all VHCP clinicians. Patients enter these values in a web browser on a computer or smartphone and receive twice daily text-message prompts reminding them to check and upload their data into Validic. If a patient enters an SpO₂ value of <93%, he of she receives an alert to call the VHCP on-call number directly. Simultaneously, an alert is also sent to the provider covering the COVID phone, who would then proactively reach out to the patient to check in.

Once a patient starts showing signs of improvement, including no fever for 48–72 h, improvement of cough, resolution of shortness of breath, and no other concerning symptoms, the patient is transitioned back to his or her KPMAS primary care provider for ongoing care, though he or she was encouraged to call us back directly for any worsening symptoms or pulse oximeter readings. This observational analysis met IRB waiver criteria.

Results

As of November 1, 2020, preliminary data are available on the demographics and outcomes of COVID-19 patients in VHCP. These are descriptive results only and are not representative of a planned statistical analysis. Eighty-seven percent of all KP members with COVID-19 were referred in the program. Of these patients, 43% were male, 42% were black, and 3% were Hispanic. Obesity (defined as BMI >30 kg/m²) was the most common comorbidity within this population at 35%, followed by hypertension at 32%, diabetes mellitus 19%, chronic kidney disease (estimated glomerular filtration rate <60 mL/min per 1.73 m²) 4%, and chronic obstructive pulmonary disease (forced expiratory volume in one second to forced vital capacity ratio <0.7) 3%. In total, 8.2% of patients ultimately required hospital admission (1,104 out of 13,508 total), and among those, a median of 9 days elapsed between symptom onset and date of hospital admission (Table 1).

Among KPMAS members outreached by VHCP, those who were successfully contacted and consented to enrollment in VHCP (2,751 out of 13,508, 20.4%) had fewer urgent care visits, ER visits, and admissions than those who were unsuccessfully contacted or declined enrollment (109 out of 477, 22.9%) (Table 2). The preliminary overall mortality rate for all KPMAS members with COVID-19 was 2.32%, whereas for those COVID-19 patients who were actively followed by VHCP, the mortality rate was 1.33%

Discussion

Kaiser Permanente's VHCP was able to rapidly establish a telemedicine-based program for the management of COVID-19 positive patients in the DC and Baltimore Metro regions, and preliminary data suggest that such a program may be effective in keeping patients out of the hospital and/or ER. In the process of rapidly establishing this program, the VHCP overcame several hurdles and learned several things along the way.

The constant influx and egress of providers through the program were a big challenge. Dynamic staffing needs and clinic openings/closings created a constant turnover in our provider staffing, making it more difficult for these individual providers to gain the necessary continuity of care experience to comfortably care for COVID-19 patients. On examining patients in our program who required more aggressive levels of care, a lack of provider experience (such as in case of a specialty outside of internal/family/emergency medicine) was an identified variable. This was mitigated to some extent by adopting a clear escalation system, where each physician and nurse was assigned to a core VHCP physician and could escalate cases to them directly at any time. In addition, all VHCP staff were encouraged to use the chat feature within Microsoft Teams, where they could post a clinical question to the entire group and get an answer from someone with more experience or expertise in that matter in real time.

VHCP also experienced challenges in the management of Spanish-speaking patients. Although a phone translator was always used during encounters with Spanish-speaking patients, and instructions for the pulse oximeter devices were made available in Spanish, written instructions for the Validic platform were only available in English, and thus fewer Spanish-speaking patients were able to take advantage of this benefit.

Lastly, with regard to our pulse oximetry program, we initially did not have an alert system set up for when patients registering low SpO₂ readings. For patients using Validic, we had to proactively review each patient's readings at multiple times during the day, calling any patient with an SpO₂ value of <93%. For patients not using Validic, there was no alert feature built into the pulse oximeter devices themselves to warn them of a low SpO₂ reading. Eventually, we worked with Validic to introduce a feature such that any patient who entered an SpO₂ value <93% would receive an alert to call our direct number so we could advise them on the most appropriate next step. Later, Validic also introduced a feature allowing us to receive an alert as well, in the form of a text message sent directly to our COVID phone.

Regarding our preliminary results showing lower health care utilization and overall mortality in patients followed by VHCP, there are many factors that influence these numbers, chief among them being that KPMAS members not diagnosed at a Kaiser facility were more likely to have been diagnosed upon admission to an external hospital, and thus were sicker at baseline than those patients referred to VHCP. Additional analysis is being conducted to understand this difference.