Impact of Remote Patient Monitoring on Length of Stay for Patients with COVID-19

Introduction

As the novel coronavirus disease (COVID-19) pandemic emerged, hospitals across the nation prepared for an influx of patients. Demand for acute care, including hospital medical/surgical beds and intensive care units (ICU) increased with the rising number of cases within the United States. Preparing for an increasing hospital surge capacity became paramount to offer the appropriate level of care to the maximum number of patients, maintain biocontainment of confirmed and suspected cases of COVID-19 while still providing safe care to non-COVID-19 patients, and ensuring a safe working environment for health care providers.¹

An increased duration of hospital stay not only makes less acute care beds available for other patients, but also increases use of personal protective equipment and prolonged COVID-19 exposure for health care providers. During the COVID-19 pandemic, home telemonitoring or remote patient monitoring (RPM) emerged as a new and powerful modality.^2–4 RPM can provide an opportunity to monitor patients in environments outside of the acute care setting and may reduce the number of hospital visits and admissions, thereby decreasing the usage of personal protective material, reducing the pressure on health care personnel, and minimizing the risk of viral transmission.

RPM may allow for optimizing care for patients with or suspected to have COVID-19 infection while ensuring the sustainability of health care capacity and resources for those who need it most urgently.^5–7 Furthermore, the promise of continued patient surveillance upon discharge may allow for earlier discharge of patients, opening up additional hospital capacity to patients who need acute care level of care.

With increasing surges of COVID-19 patients, hospitals can anticipate higher volumes of patient censuses and admission rates. RPM can be deployed as part of a tiered approach to open up bed availability in hospitals; this approach extends the walls of the hospital virtually and allows earlier discharge of patients with continued virtual monitoring. Question remains whether RPM can decrease length of stay (LOS) while keeping patients safe. We describe the impact of an RPM program on LOS and quality metrics including return to the emergency department (ED) and hospital re-admissions for patients with COVID-19.

Methods

STUDY SETTING AND PARTICIPANTS

We initially completed a feasibility pilot RPM program from March 2020 to June 2020 with the inpatient medicine services caring for COVID-19 patients at the University of Colorado Hospital (UCH), a 690-bed tertiary care academic hospital in Aurora, Colorado; part of UCHealth, a 12-hospital system in Colorado. This was during the first COVID-19 surge at UCH when little information was available on the expected clinical trajectory for this patient population. The focus during the feasibility pilot was to create infrastructure to deploy RPM: inpatient pathways for enrollment of patients, building electronic health record (EHR) processes including an RPM order and downstream processes such as informing the enrollment nurses, and finally creating the infrastructure at the virtual health center (VHC) to be able to continuously monitor patients including technology, staffing, and training, and escalation protocols for decompensating patients.

At the time of discharge, providers would place an order for RPM in the EHR, allowing the RPM nurses to consent and enroll patients into the RPM program. The device placed on the patient, Masimo Radius RPG, allowed monitoring of respiratory rate, heart rate, and pulse oximetry with real-time data transmission from the app to a Health Insurance Portability and Accountability Act (HIPAA)-compliant cloud accessed at the VHC. These data were monitored 24 h a day, 7 days a week over an 8-day period by a VHC technician trained for this role (not licensed) with built-in escalation protocols to a nurse and/or attending physician located within the VHC, if needed. This was paired with daily telephone check-ins with patients with a standardized script, including symptom tracking for the duration of the enrollment. To qualify, patients had to have decision-making capacity and a smart phone to transmit data from the smart phone application (app) to facilitate monitoring.

During our second surge, between October 2020 and February 2021, we redeployed the RPM program. Given the large volumes of patients, the only restriction we placed on the RPM program was that the patients were discharging home after an index admission for COVID-19.

DATA COLLECTION

We retrospectively queried all COVID-19 patients discharged from UCH with home RPM between April 2020 and February 2021 from the hospital EHR data warehouse. We recorded demographic data, oxygen therapy, intensive care unit (ICU) level of care, Charlson comorbidity index, length of hospitalization (LOS), readmission data, and 30-day ED visit.

We separated the data into two phases: phase 1 was between April and June 2020 during the feasibility pilot and phase 2 between October 2020 and February 2021. These data were separated to account for the differences in the operations of each of the phases to allow for appropriate case–control matching. This study was classified as exempt by a local institutional review board (COMIRB 19-2104). Patients were consented at time of registration.

DATA ANALYSIS

Using a two-to-one–matched case–control design, we compared patients who were discharged with RPM with those who were not discharged with RPM from March 2020 to September 2020 and October 2020 to February 2021, respectively. Patients discharged in each time period were matched on age, sex, Charlson comorbidity index, and limited English proficiency.

The primary outcome was hospital LOS. Secondary outcomes were (1) 7-, 14-, and 30-day readmission to the hospital and (2) return to the ED within 30 days. We estimated means and standard deviations for continuous variables when approximately normally distributed (as assessed by visual inspection of histograms), and medians and interquartile ranges when not, and frequencies for categorical variables. A chi-square test was used to evaluate associations between discharge with RPM and 7-, 14-, and 30-day readmission rates as well as return to the ED within 30 days. A Student’s t-test was used to compare hospital LOS for matched case–control patients who were and were not discharged with RPM.

Secondary analyses were performed to determine whether potential effect modification was supported by the data. We hypothesized that the association between hospital LOS and RPM in each time period may vary according to whether patients were discharged with home oxygen therapy or not. To test this hypothesis, we included an interaction term between discharge with home oxygen therapy and RPM, allowing for the intervention’s effect to depend on being discharged with home oxygen therapy. Using general linear regression, models were adjusted for (1) age, (2) sex, (3) Charlson comorbidity index, and (4) limited English proficiency. Given that hospital LOS is right skewed, this variable was log-transformed to facilitate regression analysis. We reported a relative difference in hospital LOS by exponentiating the coefficient, subtracting 1, and expressing the result as a percentage.⁸

Patients with missing data on any variables necessary for a specific analysis were excluded from that analysis. All data analyses were performed using SAS Enterprise Guide 8.2 (SAS Institute, Inc., Cary, NC).

Results

Overall, 203 patients were enrolled in RPM representing ∼1,293 patient-days of monitoring. We have summarized patient characteristics, English proficiency, financial class, discharge on home oxygen, ICU days, LACE score and Charlson comorbidity index median, LOS, readmission within 30 days and return to ED rates in Table 1.

Overall, there was a decrease in LOS for patients discharging with RPM without increase in 30-day ED revisits or hospital readmissions. The data are summarized in Tables 2 and 3. There was a strong association between patients being discharged with RPM and with home oxygen (p < 0.0001).

Although no significant association was observed between LOS, readmission to the hospital, or return to the ED between patients discharged with RPM and those who were discharged without RPM, we did observe an interaction effect between RPM and discharge with or without home oxygen therapy (p = 0.0075; Fig. 1) in the second time period. Specifically, the LOS decreased by an additional 36% for patients discharged without home oxygen therapy in the RPM group compared with patients not in the RPM group. No other significant interactions, either overall or within the first time period, were detected.