Abstract
Objective
To evaluate care utilization, cost, and mortality among high-risk patients enrolled in a coronavirus disease 2019 (COVID-19) remote patient monitoring (RPM) program.
Methods
This retrospective analysis included patients diagnosed with COVID-19 at risk for severe disease who enrolled in the RPM program between March 2020 and October 2021. The program included in-home technology for symptom and physiologic data monitoring with centralized care management. Propensity score matching established matched cohorts of RPM-engaged (defined as ≥1 RPM technology interactions) and non-engaged patients using a logistic regression model of 59 baseline characteristics. Billing codes and the electronic death certificate system were used for data abstraction from the electronic health record and reporting of care utilization and mortality endpoints.
Results
Among 5796 RPM-enrolled patients, 80.0% engaged with the technology. Following matching, 1128 pairs of RPM-engaged and non-engaged patients comprised the analysis cohorts. Mean patient age was 63.3 years, 50.9% of patients were female, and 81.9% were non-Hispanic White. Patients who were RPM-engaged experienced significantly lower rates of 30-day, all-cause hospitalization (13.7% vs 18.0%, P=.01), prolonged hospitalization (3.5% vs 6.7%, P=.001), intensive care unit admission (2.3% vs 4.2%, P=.01), and mortality (0.5% vs 1.7%; odds ratio, 0.31; 95% CI, 0.12 to 0.78; P=.01), as well as cost of care ($2306.33 USD vs $3565.97 USD, P=0.04), than those enrolled in RPM but non-engaged.
Conclusion
High-risk COVID-19 patients enrolled and engaged in an RPM program experienced lower rates of hospitalization, intensive care unit admission, mortality, and cost than those enrolled and non-engaged. These findings translate to improved hospital bed access and patient outcomes.
Abbreviations and Acronyms: COVID-19, coronavirus disease 2019; ED, emergency department; EHR, electronic health record; ICU, intensive care unit; RPM, remote patient monitoring
In response to the coronavirus disease 2019 (COVID-19) pandemic, many health care organizations implemented remote patient monitoring (RPM) programs to support patients with suspected or confirmed COVID-19 infections following a confirmed diagnosis or upon hospital discharge for the disease.1, 2, 3, 4, 5, 6, 7, 8, 9 These programs were enabled by various technologies, including automated short message service,8 mobile applications,3 or preconfigured devices1 with questionnaires for symptom tracking; peripheral medical devices or wearables for monitoring physiologic data; and/or telephone and video telehealth visits.6 , 7 Most programs provided centralized clinical support, integration of the patient-generated health data with the electronic health record (EHR), and were associated with high patient satisfaction.4 , 5 , 8 , 10
At Mayo Clinic, our team developed a COVID-19 RPM care model to support ambulatory patients with COVID-19 through the acute phase of illness, as well as those discharging from a complex COVID-19 hospitalization.1 , 11 , 12 For the high-intensity RPM program, all patients were required to be at risk for severe COVID-19 disease13; the in-home technology package was provided at no cost to patients. The technology-enabled monitoring was comprised of at least twice-daily patient-reported symptom assessments and physiologic data obtained from connected devices. A centralized team of registered nurses responded to alerts and escalated care as needed to COVID-19 care team providers. This high-intensity program was designed to meet Centers for Medicare and Medicaid Services criteria requirements as a billable RPM program14 which, in the absence of a universally accepted standard, represents the highest standard for an RPM program definition.
We previously reported the development and implementation of this multisite, multiregional, interdisciplinary COVID-19 RPM program1 which included a descriptive analysis of a diverse cohort of 7074 patients served by the program (including both low- and high-intensity monitoring) across 41 US states with an age range of 17-101 years and 27.5% racial/ethnic minority representation. Among patients engaged in high-intensity monitoring, the RPM technology engagement rate was 78.4%. Emergency department visit, hospital admission, and mortality rates within 30 days of RPM enrollment were 11.4%, 9.4%, and 0.6%, respectively.
Herein we report the results of a retrospective, matched cohort analysis of the high-intensity COVID-19 RPM program. The primary objective of this study was to evaluate care utilization among all eligible and enrolled in the RPM program, comparing those who did or did not engage with the technology. Secondary objectives were to evaluate cost and mortality in these cohorts. We hypothesized that identification of adverse health trends by the RPM technology and centralized care team would be associated with a reduction in hospital use.
Methods
Patients
Patients were eligible for the RPM program if they had a positive severe acute respiratory syndrome coronavirus 2 test at a Mayo Clinic location and one or more risk factors for severe COVID-19 illness as defined by the US Centers for Disease Control and Prevention and expert consensus.13 Patients were required to reside within the United States; however, those living in a skilled nursing facility were not eligible. Patients were not required to speak the English language or have a primary care provider, and they could participate regardless of underlying diseases and conditions.
Setting
Mayo Clinic is a nonprofit, specialty group practice with integrated research, education, and clinical practice activities. Patients were included in this study if they were diagnosed with COVID-19 at a Mayo Clinic hospital or ambulatory clinic within the Midwest tertiary campus (Rochester, Minnesota) or the affiliated Mayo Clinic Health System, comprised of more than 70 Midwest community-based hospitals and clinics in southern Minnesota, northern Iowa, and western Wisconsin.
RPM Intervention
At the outset of the COVID-19 pandemic, Mayo Clinic adapted its chronic disease and post-surgical/procedural RPM program to meet the unique needs of patients with acute COVID-19. Patients who enrolled in the high-intensity COVID-19 RPM program received a technology package comprised of a cellular-enabled tablet; preconnected, Bluetooth-enabled, medical grade devices (blood pressure cuff, pulse oximeter, and scale); and a thermometer for self-reported temperature. Vital sign measurements and symptom assessment questions were completed two to four times daily (four times for those immunosuppressed or receiving cancer-directed therapy). All patient-generated health data were integrated into the EHR. Alerts were triggered based on predetermined parameters. A centralized team of registered nurses responded to technology-generated alerts and used standardized care pathways for clinical assessments and patient management, including escalation to a COVID-19 care team of General Internal Medicine and Infectious Disease physicians and advanced practice providers. Clinical support was provided 24 hours per day, 7 days a week, including weekends and holidays. Program eligibility criteria, technology solution, and clinical operational model have been previously described.1
Study Design and Endpoints
A retrospective cohort analysis was conducted evaluating patients enrolled in the COVID-19 RPM program. Study endpoints included all-cause health care use, total cost of care, and mortality outcomes within 30 days of the COVID-19 RPM program enrollment (index) date, regardless of attribution to COVID-19. Data were abstracted from the EHR (Epic Systems, version May 2020), using validated billing reports for utilization endpoints and electronic death certificate data for mortality.
All-cause costs during the 30-day follow-up were abstracted from the Mayo Clinic Cost Data Warehouse, which has been previously described.15 Charges for hospital-based services were costed using Medicare cost-to-charge ratios, whereas the professional services were costed using the Medicare reimbursement rates for the corresponding current procedural terminology/Healthcare Common Procedure Coding System codes. The total cost was defined as the sum of both these costs and reported in US dollars.
Analysis Plan
To control for possible confounding, a propensity score–matched cohort was constructed to compare patient outcomes among those enrolled in the RPM program and “engaged,” as defined by one or more sets of vitals/symptoms submitted through the supplied technology, with those “non-engaged,” who enrolled but did not engage with the RPM technology. Specifically, one-to-one nearest-neighbor caliper matching was used to match engaged and non-engaged patients using a caliper equal to 0.2 of the standard deviation of the logit of the propensity score.16 The propensity score was estimated using a logistic regression model based on 59 baseline characteristics including age, sex, race, and other demographics, comorbidities, prior health care use, primary care empanelment at Mayo Clinic, EHR portal account access (web or mobile-based), COVID-19 risk factors17, monoclonal allocation screening score18, and index date month/year. Standardized mean differences (SMDs) were used to assess the balance of covariates after matching, with an SMD less than or equal to 0.1 indicating covariate balance. Baseline Elixhauser comorbidity scores, COVID-19 risk factors, and use were calculated using international classification of diseases version 10 diagnosis codes within 1 year before index date. Logistic regression and t test were used to compare engaged with non-engaged patients for binary and continuous outcomes, respectively.
Two subgroup analyses were performed for those patients who enrolled in the RPM program: (1) at the time of hospital discharge (following acute illness), and (2) after diagnosis in the outpatient setting (during acute illness).
Propensity score modeling and analyses were performed using Stata 16.1 (StataCorp). This study was approved by the Mayo Clinic Institutional Review Board (#18-009605).
Results
Between March 16, 2020, and October 18, 2021, 9679 high-risk patients enrolled in the COVID-19 RPM program. Among these patients, 5796 were evaluable and comprised the analysis cohort (Figure ). Reasons for exclusion included lack of authorization for retrospective research, assignment to low-intensity monitoring, sex, and risk score missing (required for matching). Patients were also excluded if they died or were hospitalized within 1 day of enrollment, as the RPM technology package is typically delivered to the patient’s home the day after enrollment. Additionally, those managed in the Mayo Clinic Southwest (Scottsdale, Arizona) and Southeast (Jacksonville, Florida) regions (n=1950) were excluded from the analysis given most patients only receive specialty care, but not routine care (including emergency department visits and hospitalizations) at Mayo Clinic, and to mitigate the effect of regional variability on comparative outcomes assessment.
Figure.
Flow diagram for patients enrolled in the coronavirus disease 2019 (COVID-19) remote patient monitoring (RPM) program and evaluable for the analysis.
Of the evaluable patients, 1162 (20.0%) did not engage with the technology. Before matching, non-engaged patients were generally older, with more comorbidities, or diagnosed with COVID-19 in the hospital (Table 1 ); however, sex, race, ethnicity, and primary language were similar between engaged and non-engaged cohorts.
Table 1.
| Characteristicc | Pre-matched population |
Matched population |
||||
|---|---|---|---|---|---|---|
| Non-engaged (n=1162) | Engaged (n=4634) | SMD | Non-engaged (n=1128) | Engaged (n=1128) | SMD | |
| Age, y | ||||||
| Mean (SD) | 62.6 (18.4) | 57.1 (17.7) | 0.31 | 62.3 (18.4) | 64.3 (17.5) | 0.10 |
| Median | 66.0 | 59.0 | 66.0 | 68.0 | ||
| Age distribution, y | ||||||
| 18-49 | 278 (23.9) | 1521 (32.8) | 0.20 | 277 (24.5) | 230 (20.4) | 0.10 |
| 50-74 | 488 (42.0) | 2242 (48.4) | 0.13 | 476 (42.2) | 495 (43.8) | 0.03 |
| 74+ | 396 (34.1) | 871 (18.8) | 0.35 | 375 (33.2) | 403 (35.7) | 0.05 |
| Sex | ||||||
| Female | 592 (50.9) | 2443 (52.7) | 0.04 | 569 (50.4) | 580 (51.4) | 0.02 |
| Male | 570 (49.1) | 2191 (47.3) | 0.04 | 559 (49.6) | 548 (48.6) | 0.02 |
| Married | 638 (54.9) | 2925 (63.1) | 0.17 | 629 (55.8) | 607 (53.8) | 0.04 |
| Race/ethnicity | ||||||
| White, NH | 941 (81.0) | 3638 (78.5) | 0.07 | 914 (81.0) | 933 (82.7) | 0.04 |
| Hispanic (all races) | 109 (9.4) | 530 (11.4) | 0.07 | 108 (9.6) | 92 (8.2) | 0.05 |
| Black, NH | 38 (3.3) | 176 (3.8) | 0.03 | 36 (3.2) | 39 (3.5) | 0.02 |
| Asian, NH | 13 (1.1) | 99 (2.1) | 0.08 | 13 (1.2) | 9 (0.8) | 0.04 |
| All other, NH | 28 (2.4) | 86 (1.9) | 0.04 | 26 (2.3) | 25 (2.2) | 0.01 |
| Unknown/missing | 33 (2.8) | 105 (2.3) | 0.04 | 31 (2.7) | 30 (2.7) | 0.01 |
| Primary language | ||||||
| English | 1029 (88.6) | 4036 (87.1) | 0.05 | 997 (88.4) | 1013 (89.8) | 0.05 |
| Spanish | 90 (7.7) | 411 (8.9) | 0.04 | 90 (8.0) | 80 (7.1) | 0.03 |
| Other | 38 (3.3) | 184 (3.9) | 0.03 | 37 (3.3) | 32 (2.9) | 0.02 |
| Missing | 5 (0.4) | 3 (0.1) | 0.07 | 4 (0.4) | 3 (0.3) | 0.02 |
| Paneled to primary care | 769 (66.2) | 3266 (70.5) | 0.09 | 753 (66.8) | 770 (68.3) | 0.03 |
| Portal accountd | 793 (68.2) | 3768 (81.3) | 0.31 | 789 (69.9) | 746 (66.1) | 0.08 |
| Diagnosed in hospital | 255 (21.9) | 679 (14.7) | 0.19 | 240 (21.3) | 254 (22.5) | 0.03 |
| ED visitse | 196 (16.9) | 650 (14.0) | 0.08 | 188 (16.7) | 197 (17.5) | 0.02 |
| Office visitse | 667 (57.4) | 2826 (61.0) | 0.07 | 651 (57.7) | 652 (57.8) | 0.00 |
| Hospitalizationse | 264 (22.7) | 798 (17.2) | 0.14 | 251 (22.3) | 265 (23.5) | 0.03 |
| COVID-19 risk factors per patient (sum) | ||||||
| Mean (SD) | 3.5 (2.2) | 3.0 (2.1) | 0.17 | 3.5 (2.2) | 3.7 (2.2) | 0.09 |
| Median | 4.0 | 3.0 | 3.0 | 4.0 | ||
| IQR | 2.0-5.0 | 1.0-4.0 | 2.0-5.0 | 2.0-5.0 | ||
| Monoclonal allocation screening score | ||||||
| Mean (SD) | 4.4 (3.1) | 3.8 (2.9) | 0.19 | 4.3 (3.1) | 4.5 (2.9) | 0.06 |
| Median | 4.0 | 3.0 | 4.0 | 5.0 | ||
| IQR | 2.0-7.0 | 1.0-6.0 | 2.0-7.0 | 2.0-7.0 | ||
| Elixhauser score | ||||||
| Mean (SD) | 3.1 (3.2) | 2.6 (2.8) | 0.17 | 3.0 (3.2) | 3.3 (3.2) | 0.07 |
| Median | 2.0 | 2.0 | 2.0 | 2.5 | ||
| IQR | 0.0-5.0 | 0.0-4.0 | 0.0-5.0 | 0.0-5.0 | ||
| Risk factors for severe COVID-19 illness | ||||||
| Cancer patient | 150 (12.9) | 664 (14.3) | 0.04 | 149 (13.2) | 154 (13.7) | 0.01 |
| Congestive heart failure | 214 (18.4) | 569 (12.3) | 0.17 | 206 (18.2) | 231 (20.5) | 0.06 |
| Chronic lung disease | 276 (23.8) | 970 (20.9) | 0.07 | 267 (23.7) | 277 (24.6) | 0.02 |
| Coronary artery disease | 287 (24.7) | 845 (18.2) | 0.16 | 279 (24.7) | 294 (26.1) | 0.03 |
| Immune compromised | 147 (12.7) | 812 (17.5) | 0.13 | 146 (12.9) | 130 (11.5) | 0.04 |
| End-stage renal disease | 248 (21.3) | 703 (15.2) | 0.16 | 239 (21.2) | 258 (22.9) | 0.04 |
| Arrhythmia | 336 (28.9) | 976 (21.1) | 0.18 | 323 (28.6) | 354 (31.4) | 0.06 |
| Depression | 157 (13.5) | 597 (12.9) | 0.02 | 151 (13.4) | 169 (15.0) | 0.05 |
| Diabetes with chronic complications | 231 (19.9) | 777 (16.8) | 0.08 | 223 (19.8) | 237 (21.0) | 0.03 |
| Diabetes without chronic complications | 203 (17.5) | 812 (17.5) | 0.00 | 197 (17.5) | 201 (17.8) | 0.01 |
| Fluid electrolyte disorder | 176 (15.1) | 480 (10.4) | 0.14 | 166 (14.7) | 190 (16.8) | 0.06 |
| Hypertension, complicated | 260 (22.4) | 727 (15.7) | 0.17 | 251 (22.3) | 274 (24.3) | 0.05 |
| Hypertension, uncomplicated | 334 (28.7) | 1263 (27.3) | 0.03 | 322 (28.5) | 329 (29.2) | 0.02 |
| Hypothyroid | 150 (12.9) | 500 (10.8) | 0.07 | 145 (12.9) | 158 (14.0) | 0.03 |
| Obesity | 255 (21.9) | 998 (21.5) | 0.01 | 250 (22.2) | 268 (23.8) | 0.04 |
| Peripheral vascular disorders | 145 (12.5) | 424 (9.1) | 0.11 | 139 (12.3) | 166 (14.7) | 0.07 |
| Renal failure | 214 (18.4) | 588 (12.7) | 0.16 | 205 (18.2) | 221 (19.6) | 0.04 |
| Mood disorder | 405 (34.9) | 1448 (31.2) | 0.08 | 388 (34.4) | 409 (36.3) | 0.04 |
| Current smoker | 333 (28.7) | 1065 (23.0) | 0.13 | 318 (28.2) | 338 (30.0) | 0.04 |
COVID-19, coronavirus disease 2019; ED, emergency department; NH, non-Hispanic; SMD, standardized mean differences.
Values shown are n (%) unless otherwise stated.
Other factors not listed that were used for balancing cohorts included: month/year of COVID-19 index date, testing site, and any risk factor for severe COVID-19 occurring at a frequency of less than 10% (pregnancy, chronic liver disease, active chemotherapy, alcohol, blood loss anemia, coagulopathy, deficiency anemia, drug abuse, HIV/AIDS, liver disease, lymphoma, metastatic cancer, other neurological disorders, paralysis, peptic ulcer disease, psychosis, pulmonary circulation disorder, rheumatoid arthritis/collagen vascular disease, solid tumor metastasis, valvular disease, weight loss, and bone marrow/organ transplant).
Portal Account denotes Mayo Clinic’s electronic health record–integrated web or mobile patient online services platform that facilitates secure messaging, appointment scheduling, bill pay, et cetera.
Health care utilization in the 3 months before COVID-19 index date.
After matching (Table 2 ), when compared with the non-engaged patients, those engaged in the RPM program experienced a significantly lower rate of one or more hospitalization (13.7% vs 18.0%, P=0.01), prolonged hospitalization 7 or more days (3.5% vs 6.7%, P=0.001), and intensive care unit (ICU) admission (2.3% vs 4.2%, P=0.01), as well as a significantly lower average hospital length of stay (6.7 vs 8.2 days, P=0.04). Total ICU days were markedly less for those engaged relative to those who were non-engaged (119 vs 313 days, P=.21); however, the difference was not statistically significant. Rates of one or more emergency department (ED) visits were similar among groups; however, those engaged were more likely to experience two or more ED visits (4.3% vs 2.4%, P=0.01) than those non-engaged.
Table 2.
Care Utilization and Mortality Outcomes Between Non-engaged and Engaged Patients Enrolled in the COVID-19 RPM Programa,b
| Outcomesc | Non-engaged (n=1128) | Engaged (n=1128) | OR (95% CI) | P |
|---|---|---|---|---|
| ED visits | ||||
| ≥1 ED visit (unique patients) | 158 (14.0) | 171 (15.2) | 1.10 (0.87-1.39) | .44 |
| >1 ED visit (unique patients) | 27 (2.4) | 49 (4.3) | 1.85 (1.15-2.98) | .01 |
| ED visit converted to inpatient hospitalization hospital admissions | 87 (7.7) | 99 (8.8) | 1.15 (0.85-1.55) | .36 |
| ≥1 Admission (unique patients) | 203 (18.0) | 154 (13.7) | 0.72 (0.57-0.90) | .01 |
| >1 Admission (unique patients) | 41 (3.6) | 29 (2.6) | 0.70 (0.43-1.13) | .15 |
| Prolonged hospitalization (7 or more days) | 76 (6.7) | 39 (3.5) | 0.50 (0.33-0.74) | .001 |
| ICU admissions | 47 (4.2) | 26 (2.3) | 0.54 (0.33-0.88) | .01 |
| Mortality (30 day) | 19 (1.7) | 6 (0.5) | 0.31 (0.12-0.78) | .01 |
| Mean difference (95% CI) | ||||
|---|---|---|---|---|
| Average length of stayd | ||||
| Mean (SD) | 6.7 (6.0) | 5.4 (4.7) | -1.3 (-2.4 to -0.1) | .03 |
| Median (range) | 5 (1-30) | 5 (1-30) | ||
| Total hospital daysd | ||||
| Total | 1660 | 1026 | ||
| Mean (SD) | 8.2 (7.2) | 6.7 (6.0) | -1.5 (-2.9 to -0.1) | .04 |
| Median (range) | 6 (1-30) | 5 (1-30) | ||
| Total ICU dayse | ||||
| Total | 313 | 119 | ||
| Mean (SD) | 6.7 (7.6) | 4.6 (4.9) | -2.5 (-5.4 to 1.2) | .21 |
| Median (range) | 3 (1-30) | 3 (1-21) | ||
| Overall cost of care | ||||
| Mean (SE) | $3565.97 ($525.25) | $2306.33 ($325.22) | -$1259.64 | .04 |
COVID-19, coronavirus disease 2019; ED, emergency department; ICU, intensive care unit; OR, odds ratio; RPM, remote patient monitoring.
Values shown are n (%) unless otherwise stated.
All data are reported for events that occurred within 30 days of RPM program enrollment for outpatient diagnosis or 30 days from discharge of a hospitalized patient.
Data reported for those patients who were hospitalized.
Data reported for those patients who had an ICU admission.
Those who were engaged in RPM experienced a significantly lower overall 30-day cost of care than those non-engaged ($2306.33 USD vs $3565.97 USD, P=0.04). The average cost savings among engaged RPM patients was $1259 per patient during the 30-day follow-up period.
All-cause, 30-day mortality rates were significantly lower for those who engaged in the RPM program than those non-engaged (0.5% vs 1.7%, odds ratio, 0.31; 95% CI, 0.12 to 0.78; P=0.01].
In a subgroup analysis of patients diagnosed with COVID-19 while hospitalized and RPM-enrolled upon discharge to home (Table 3 ), the rates of subsequent ED visits and rehospitalizations were similar between groups. However, the rate of prolonged hospitalization and mean length of stay were significantly lower for those engaged than those non-engaged.
Table 3.
Care Utilization and Mortality Outcomes Among Those With COVID-19 Diagnosed While Hospitalized and RPM-enrolled Upon Discharge to Homea,b
| Outcomesc | Non-Engaged (n=240) | Engaged (n=254) | OR (95% CI) | P |
|---|---|---|---|---|
| ED visits | ||||
| ≥1 ED visit (unique patients) | 37 (15.4) | 37 (14.6) | 0.94 (0.57-1.53) | .79 |
| >1 ED visit (unique patients) | 10 (4.2) | 9 (3.5) | 0.84 (0.34-2.12) | .72 |
| ED visit converted to inpatient hospitalization | 39 (16.3) | 40 (15.8) | 0.96 (0.60-1.56) | .88 |
| Hospital admissions | ||||
| ≥1 Admission (unique patients) | 51 (21.3) | 40 (15.8) | 0.69 (0.44-1.09) | .12 |
| >1 Admission (unique patients) | 10 (4.2) | 10 (4.0) | 0.94 (0.39-2.31) | .90 |
| Prolonged hospitalization (7 or more days) | 23 (9.6) | 9 (3.5) | 0.35 (0.16-0.77) | .01 |
| ICU admissions | 14 (5.8) | 8 (3.2) | 0.52 (0.22-1.28) | .16 |
| Mortality (30 day) | 5 (2.1) | 2 (0.8) | 0.37 (0.07-1.94) | .24 |
| Mean difference (95% CI) | ||||
|---|---|---|---|---|
| Average length of stayd | ||||
| Mean (SD) | 7.5 (7.0) | 4.8 (3.5) | -2.7 (-5.1 to -0.3) | .03 |
| Median (range) | 6 (1-30) | 4 (1-18) | ||
| Total hospital daysd | ||||
| Total | 450 | 262 | ||
| Mean (SD) | 8.8 (7.8) | 6.6 (6.4) | -2.3 (-5.3 to 0.8) | .14 |
| Median (range) | 7 (1-30) | 5 (1-28) | ||
| Total ICU dayse | ||||
| Total | 68 | 27 | ||
| Mean (SD) | 4.9 (5.7) | 3.4 (3.3) | -1.5 (-6.1 to 3.1) | .51 |
| Median (range) | 3 (1-19) | 2 (1-10) |
COVID-19, coronavirus disease 2019; ED, emergency department; ICU, intensive care unit; OR, odds ratio; RPM, remote patient monitoring.
Values shown are n (%) unless otherwise stated.
All data are reported for events that occurred within 30 days of RPM program enrollment for outpatient diagnosis or 30 days from discharge of a hospitalized patient.
Data reported for those patients who had a hospital admission.
Data reported for those patients who had an ICU admission.
A separate subgroup analysis of patients with COVID-19 diagnosed and RPM-enrolled in the ambulatory setting (Table 4 ) revealed that when compared with non-engaged patients, engaged patients had higher rates of two or more ED visits, but lower rates of hospital admission, prolonged hospitalization, ICU admission, and mortality. These outcome trends were similar to those for the overall cohort.
Table 4.
Care Utilization and Mortality Outcomes Among Those Diagnosed With COVID-19 and RPM-Enrolled in the Ambulatory Settinga,b
| Outcomesc | Non-engaged (n=888) | Engaged (n=874) | OR (95% CI) | P |
|---|---|---|---|---|
| ED visits | ||||
| ≥1 ED visit (unique patients) | 121 (13.6) | 134 (15.3) | 1.15 (0.88-1.50) | .31 |
| >1 ED visit (unique patients) | 17 (1.9) | 40 (4.6) | 2.46 (1.38-4.37) | .002 |
| ED visit converted to inpatient hospitalization | 48 (5.4) | 59 (6.8) | 1.27 (0.86-1.88) | .24 |
| Hospital admissions | ||||
| ≥1 Admission (unique patients) | 152 (17.1) | 114 (13.0) | 0.73 (0.56-0.94) | .02 |
| >1 Admission (unique patients) | 31 (3.5) | 19 (2.2) | 0.61 (0.34-1.10) | .10 |
| Prolonged hospitalization (7 or more days) | 53 (6.0) | 30 (3.4) | 0.56 (0.35-0.89) | .01 |
| ICU admissions | 33 (3.7) | 18 (2.1) | 0.54 (0.30-0.98) | .04 |
| Mortality (30 day) | 14 (1.6) | 4 (0.5) | 0.29 (0.09-0.88) | .03 |
| Mean difference (95% CI) | ||||
|---|---|---|---|---|
| Average length of staye | ||||
| Mean (SD) | 6.4 (5.6) | 5.6 (5.0) | -0.8 (-2.1 to 0.5) | .23 |
| Median (range) | 5 (1-30) | 5 (1-30) | ||
| Total hospital daysd | ||||
| Total | 1210 | 764 | ||
| Mean (SD) | 8.0 (7.0) | 6.7 (5.8) | -1.3 (-2.9 to 0.3) | .12 |
| Median (range) | 5 (1-30) | 5 (1-30) | ||
| Total ICU dayse | ||||
| Total | 245 | 92 | ||
| Mean (SD) | 7.4 (8.2) | 5.1 (5.4) | -2.3 (-6.6 to 2.0) | .29 |
| Median (range) | 4 (1-30) | 3 (1-21) |
COVID-19, coronavirus disease 2019; ED, emergency department; ICU, intensive care unit; OR, odds ratio; RPM, remote patient monitoring.
Values shown are n (%) unless otherwise stated.
All data are reported for events that occurred within 30 days of RPM program enrollment for outpatient diagnosis or 30 days from discharge of a hospitalized patient.
Data reported for those patients who had a hospital admission.
Data reported for those patients who had an ICU admission.
Discussion
This study suggests that patients with COVID-19 at risk for severe disease who enrolled and engaged in the RPM program experienced significantly lower rates of 30-day, all-cause hospital utilization, total cost of care, and mortality when compared with those who were non-engaged, especially when diagnosed and managed in the ambulatory setting through the acute phase of illness. We postulate the RPM program facilitated detection of adverse health trends and enabled early supportive care interventions, which in turn favorably altered the COVID-19 disease trajectory. As hospital bed and ICU capacity have been severely strained during the pandemic,19 these findings build upon our prior observations1 , 11 and those of others,20 and they provide a potential strategy to improve hospital access.
Furthermore, as racial/ethnic minority populations were as likely to engage in the RPM program as non-Hispanic White patients, it is feasible that RPM programs could help improve outcomes in this cohort disproportionately impacted by COVID-19.21 This was an important and unexpected observation given that RPM program participation was declined by several racial/ethnic minority, migrant workers at a meatpacking plant in the early days of the pandemic.22
Among those diagnosed while hospitalized during the acute phase of COVID-19, we expected the RPM program to enable earlier discharge, as shown by others.3 , 5 Our findings suggest that postdischarge RPM engagement may not reduce subsequent ED visits and re-admissions during the recovery phase; however, it was associated with a reduced mean hospital length of stay, which still conveys an improvement in bed capacity.
Emergency department visit rates were similar between groups; however, the multiple ED visit rate for those engaged with RPM was higher than for those non-engaged. These findings were anticipated as deteriorating patients were sent to the ED per established workflows. Future program iterations incorporating tele-emergency medicine and community paramedics are being explored to enhance in-home diagnostics, triage, and supportive care treatment interventions, such as intravenous fluid administration and initiation of supplemental oxygen.
Compared to non-engaged patients, the total cost savings for 1128 matched patients that were engaged in the COVID-19 RPM program was approximately $1.4 million. Thus, a well-engaged RPM program not only results in lower hospital utilization and better patient outcomes, but it could potentially yield substantial health care cost savings for patients and health systems.
There are several strengths of this COVID-19 RPM program analysis: (1) large cohort size (1128 matched pairs); (2) matching on 59 confounding variables; (3) inclusive representation of elderly, rural, and underrepresented minority populations; and (4) RPM program technology and clinical operational model that qualifies as a billable service by Centers for Medicare and Medicaid Services criteria.14 Importantly, this is among the first known reports to show improved care utilization rates among high-risk patients diagnosed with COVID-19 when managed and engaged in an ambulatory RPM program. Additionally, program engagement was associated with improved mortality and total cost of care.
Study Limitations
Results must be interpreted within the limitations of retrospective study design. We did not compare outcomes of those enrolled in the RPM program with those not enrolled under the “intention-to-treat” principle, as we could not identify a comparable high-risk control group managed without RPM. This finding was not entirely surprising as our COVID-19 care team physicians depended on the RPM program for care delivery at scale, especially to support ambulatory management of patients with COVID-19 at risk for severe disease. For this reason, as well, they refused a prospective randomized trial of RPM vs usual care for COVID-19 management. Therefore, we used an "as-treated" analysis focused on eligible cases enrolled in the RPM program. We recognize that patients who do not engage with the RPM technology may have a general predisposition to not engage with health care and acknowledge that participation bias may exist with this analysis. That said, some patients were non-engaged simply by lack of timely receipt of the technology package, especially during surges which strained the supply chain. Finally, the propensity score matching adjusted only for the observed patient characteristics, not the unobserved ones. The latter may include subjective factors that are associated with patient engagement; however, the cohorts were matched by portal access, an indicator of cellular or broadband telehealth access and indirect measure of digital literacy. Inability to adjust for unobserved or unmeasured variables is a well-acknowledged limitation of any retrospective study, including ours. Such unobserved confounding can be eliminated only through a prospective randomized controlled trial, which was not feasible at our institution during the COVID-19 pandemic.
Conclusion
Our findings on well-engaged RPM have potential policy and reimbursement implications for extending this acute care delivery model beyond the pandemic while additional prospective, confirmatory studies are performed. Further mixed-methods research is also needed to understand why enrolled patients do not engage with the technology and to evaluate the cost effectiveness of the program.
Potential Competing Interests
Dr Haddad has received research grant funding from Takeda Oncology to Mayo Clinic to conduct a clinical trial unrelated to the research presented in this manuscript. Dr Borah has received consulting fees from Exact Sciences and Boehringer Ingelheim in the last 3 years on topics not related to the content of the manuscript. The remaining authors report no potential competing interests.
Footnotes
For editorial comment, see page 2184
Grant Support: This study was made possible in part by funding from the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery. The funders had no role in the study design, conduct, and preparation of the manuscript. The Mayo Clinic Remote Patient Monitoring program is an enterprise shared service supported by the Mayo Clinic practice. There were no funding sources external to Mayo Clinic associated with this research study.
References
- 1.Coffey J.D., Christopherson L.A., Glasgow A.E., et al. Implementation of a multisite, interdisciplinary remote patient monitoring program for ambulatory management of patients with COVID-19. NPJ Digit Med. 2021;4(1):123. doi: 10.1038/s41746-021-00490-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mantena S., Keshavjee S. Strengthening healthcare delivery with remote patient monitoring in the time of COVID-19. BMJ Health Care Inform. 2021;28(1) doi: 10.1136/bmjhci-2020-100302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gordon W.J., Henderson D., DeSharone A., et al. Remote patient monitoring program for hospital discharged COVID-19 patients. Appl Clin Inform. 2020;11(5):792–801. doi: 10.1055/s-0040-1721039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kodama R., Arora S., Anand S., et al. Reengineering the discharge transition process of COVID-19 patients using telemedicine, remote patient monitoring, and around-the-clock remote patient monitoring from the emergency department and inpatient units. Telemed J E Health. 2021;27(10):1188–1193. doi: 10.1089/tmj.2020.0459. [DOI] [PubMed] [Google Scholar]
- 5.Grutters L.A., Majoor K.I., Mattern E.S.K., et al. Home telemonitoring makes early hospital discharge of COVID-19 patients possible. J Am Med Inform Assoc. 2020;27(11):1825–1827. doi: 10.1093/jamia/ocaa168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bell L.C., Norris-Grey C., Luintel A., et al. Implementation and evaluation of a COVID-19 rapid follow-up service for patients discharged from the emergency department. Clin Med (Lond) 2021;21(1):e57–e62. doi: 10.7861/clinmed.2020-0816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Medina M., Babiuch C., Card M., et al. Home monitoring for COVID-19. Cleve Clin J Med. 2020 doi: 10.3949/ccjm.87a.ccc028. Online ahead of print. [DOI] [PubMed] [Google Scholar]
- 8.Morgan A.U., Balachandran M., Do D., et al. Remote Monitoring of Patients With Covid-19: Design, Implementation, and Outcomes of the First 3,000 Patients in COVID Watch. NEJM Catalyst, Innovations in Care Delivery. Published July 21, 2020. https://catalyst.nejm.org/doi/full/10.1056/CAT.20.0342
- 9.Annis T., Pleasants S., Hultman G., et al. Rapid implementation of a COVID-19 remote patient monitoring program. J Am Med Inform Assoc. 2020;27(8):1326–1330. doi: 10.1093/jamia/ocaa097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Daly B., Lauria T.S., Holland J.C., et al. Oncology patients' perspectives on remote patient monitoring for COVID-19. JCO Oncol Pract. 2021;17(9):e1278–e1285. doi: 10.1200/OP.21.00269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Pritchett J.C., Borah B.J., Desai A.P., et al. Association of a remote patient monitoring (RPM) program with reduced hospitalizations in cancer patients with COVID-19. JCO Oncol Pract. 2021;17(9):e1293–e1302. doi: 10.1200/OP.21.00307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Haddad T.C., Blegen R.N., Prigge J.E., et al. A scalable framework for telehealth: the Mayo Clinic Center for Connected Care response to the COVID-19 pandemic. Telemed Rep. 2021;2(1):78–87. doi: 10.1089/tmr.2020.0032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.US Centers for Disease Control and Prevention Science Brief: Evidence Used to Update the List of Underlying Medical Conditions Associated with Higher Risk for Severe COVID-19. June 15, 2022. https://www.cdc.gov/coronavirus/2019-ncov/science/science-briefs/underlying-evidence-table.html [PubMed]
- 14.US Centers for Medicare & Medicaid Services. Final Policy, Payment, and Quality Provisions Changes to the Medicare Physician Fee Schedule for Calendar Year 2021. December 1, 2020. https://www.cms.gov/newsroom/fact-sheets/final-policy-payment-and-quality-provisions-changes-medicare-physician-fee-schedule-calendar-year-1
- 15.Visscher S.L., Naessens J.M., Yawn B.P., et al. Developing a standardized healthcare cost data warehouse. BMC Health Serv Res. 2017;17(1):396. doi: 10.1186/s12913-017-2327-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Austin P.C. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10(2):150–161. doi: 10.1002/pst.433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nanda S., Chacin Suarez A.S., Toussaint L., et al. Body mass index, multi-morbidity, and COVID-19 risk factors as predictors of severe COVID-19 outcomes. J Prim Care Community Health. 2021;12 doi: 10.1177/21501327211018559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ganesh R., Philpot L.M., Bierle D.M., et al. Real-world clinical outcomes of bamlanivimab and casirivimab-imdevimab among high-risk patients with mild to moderate coronavirus disease 2019. J Infect Dis. 2021;224(8):1278–1286. doi: 10.1093/infdis/jiab377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Rosenbaum L. The untold toll — the pandemic's effects on patients without COVID-19. N Engl J Med. 2020;382(24):2368–2371. doi: 10.1056/NEJMms2009984. [DOI] [PubMed] [Google Scholar]
- 20.Delgado M.K., Morgan A.U., Asch D.A., et al. Comparative effectiveness of an automated text messaging service for monitoring COVID-19 at home. Ann Intern Med. 2022;175(2):179–190. doi: 10.7326/M21-2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Gross C.P., Essien U.R., Pasha S., et al. Racial and ethnic disparities in population-level COVID-19 mortality. J Gen Intern Med. 2020;35(10):3097–3099. doi: 10.1007/s11606-020-06081-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Pritchett J.C., Haddad T.C. Avoiding the "Faux Equalizer. Mayo Clin Proc. 2021;96(12):2963–2965. doi: 10.1016/j.mayocp.2021.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]