We performed a retrospective cohort study, including all consecutive COVID-19-confirmed patients with nasal oxygen therapy who consulted at our hospital (emergency department) and managed with remote monitoring (Terr-eSanté^©), from October 2020 to April 2021. Patients could be included at hospital discharge or just after an emergency department (ED) consultation.

The web app (Terr-eSanté^©) is an easy-to-use and free web application for healthcare workers (HCWs), which was designed to be straightforward and intuitive to use for patients and a panel of multidisciplinary healthcare professionals (infectious diseases specialists, emergency physicians, nurses, and primary care physicians). It allows several physicians (primary care, emergency, infectious disease, and other specialists) to exchange and communicate information on patients’ care and conditions (Figure 1).

Patients provided electronic consent for this remote monitoring program, but they also had access to the hospital medical ward outside of this web app.

To be included in the remote monitoring process, patients had to present all the following inclusion criteria after evaluation at our center (ED or in an acute medical ward): age ≥ 18 years old; oxygen flow requirement <5 L/min to obtain a saturation of at least 90%, temperature < 40°C, respiratory rate < 30c/min, arterial systolic pressure > 100 mmHg, and positive SARS-CoV-2 PCR results. Furthermore, the patients should not be isolated at home, should live within less than 1 h of our hospital, and should agree to participate in the process with nurse visits and compliance with remote monitoring. Finally, the patients should give their consent to be included in this cohort. Each patient could only be included once.

Exclusion criteria were: patient with already telemonitoring for any other medical cause, no oxygen requirement, and negative SARS-CoV-2 PCR results. Delay from symptoms onset and comorbidities was not an exclusion criterion.

Inclusion in the process of remote monitoring was performed by physicians of our hospital after consultation at the ED or at hospital discharge.

The following data were registered by physicians: baseline characteristics, including age, gender, phone number or email address, date of first symptoms, the presence of cardiovascular disease, diabetes, chronic lung disease, and immunodeficiency (transplant, active cancer treatment, and uncontrolled HIV infection).

Treatment was prescribed according to guidelines (corticosteroids, prophylaxis of thrombosis, etc.). Then, patients were monitored at home by a nurse once or twice a day, according to their clinical status and severity, based on a medical prescription.

The monitoring period was at least 14 days but could be prolonged based on physician advice, especially if patients still required oxygen therapy.

During monitoring, in-home nurses collected the following data on the web app: blood pressure, temperature, oxygen saturation, and respiratory rate. In the web application, HCWs had access to the nurse’s transmissions, which were checked every day by the medical hospital COVID-19 team.

If the patient’s conditions worsened or if the nurse needed advice, an alert was created on the app or by phone call to the dedicated phone line of the medical hospital COVID-19 team.

During this period, the patients could have an in-site evaluation at our hospital on demand or after an alert. Consequently, patients could be hospitalized or discharged according to their conditions.

In case of rapid worsening, patients were advised to directly contact the national emergency number (Service d’Aide Médicale Urgente [SAMU]).

Registration, active monitoring, answers to alerts, coordination of care, and prescriptions were performed by a multidisciplinary medical team, with a coordinator nurse and a dedicated emergency department physician with a 24/7 service hotline for information between every HCW.

The primary endpoint was failure of remote monitoring, defined by unplanned hospitalization (excluding ED reevaluation) within the time of oxygen requirement.

Quantitative variables are presented as mean ± standard deviation (SD), or median and interquartile range (IQR). Qualitative variables are presented as the number of occurrences and relative frequencies.

We used χ2 tests to compare the distributions of categorical variables, whereas two-tailed, unpaired t-tests were used to compare the distributions of quantitative continuous variables. All reported p-values were based on two-sided tests, and a p < 0.05 was considered statistically significant.

To identify risk factors associated with failure, a univariate analysis by logistic regression was performed, using demographic and medical characteristics as well as all clinical and biological data. A multivariable analysis by logistic regression was then performed using all variables from the univariate analysis that had a p ≤ 0.05. The final model was obtained using backward stepwise regression with 0.10 thresholds.

Odds ratios (ORs) were calculated from the univariate and multivariable analyses to quantify association with failure during oxygen requirement with 95% confidence intervals (CIs). Analyses were performed with the use of the Statistical Package for Social Science (SPSS) version 26.0 (SPSS, Chicago, IL, United States).

The research was conducted in accordance with the Declaration of Helsinki and national and institutional standards. Patients were informed that their clinical data could be used, after anonymization, for research purposes. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

Second, we performed a literature review of all articles on the literature review of studies on COVID-19 patients requiring oxygen with remote monitoring from 2020 to 2022.

We searched the Medline/PubMed and Web of Science databases with the following keywords: “remote monitoring,” “telemonitoring,” “COVID-19 monitoring,” and “oxygen requiring.”

During the study period, 386 COVID-19 patients were admitted to the ED or medical wards in our hospital. We excluded from our study 86 patients: 30 were already under monitoring at the beginning of the study, 20 were in remote monitoring for another disease (5 asthma, 1 pericarditis, 1 community-acquired pneumonia, 2 for pancreatitis, 1 for osteomyelitis, 1 for dehydration, 3 for acidocetosis, 1 for sigmoiditis, 1 for Horton disease, 1 for pulmonary embolism, 1 for severe dizziness, and 1 for appendicitis), 5 COVID-19 patients did not require oxygen, 21 medical charts were duplication, 6 patients were not able to benefit from remote monitoring, and 4 patients had an incomplete medical chart. The complete study flow chart is presented in Figure 2.

In our study, we analyzed a cohort of 300 distinct patients who had been discharged with oxygen therapy following COVID-19 pneumonia. Among these individuals, 72 (24.0%) were discharged from the emergency department, while 229 (76.0%) were released from inpatient admissions. The median age of the cohort was 51 years, with an interquartile range of 45 to 61 years, and of these patients, 184 (61.3%) were male patients. Detailed patient characteristics can be found in Table 1.

The median duration of follow-up for these patients was 26 days, with an interquartile range of 25 to 27 days. A total of 265 patients (88%) underwent successful oxygen weaning, with an average delay of 7.9 ± 7.3 days (ranging from 1 to 34 days) without any complications.

Notably, 28 out of the 300 patients (9.3%) required readmission to the hospital due to respiratory failure caused by COVID-19. Among these cases, six patients (2.0%) required intensive care unit (ICU) admission. Among the ICU admissions, four patients were directly admitted upon hospitalization, while two were transferred to the ICU on days 1 and 6 of their hospital stay. Of these readmissions, 26 were facilitated by nursing staff, while 2 patients sought consultation independently.

Within the subgroup of hospitalized patients, 19 out of the 28 (67.9%) who were readmitted for respiratory failure were eventually discharged home after successful oxygen weaning. Additionally, 36 out of the 300 (12.0%) medical charts indicated that monitoring was performed by a general practitioner.

Among the patients who were admitted to the ICU, five individuals required such intensive care. Their mean age was 60.2 ± 11.8 years. Among these patients, two out of five had a history of chronic respiratory failure, while one out of five was immunocompromised. Worsening in their condition occurred at an average of 11.4 ± 6.5 days.

Remarkably, the study documented a total of 3 out of 300 (1.0%) patient fatalities. None of these deaths occurred within the ICU, during the transportation back to acute care, or at home. Importantly, there were no instances of missed follow-up for any of the patients.

These unfortunate fatalities were observed in patients who had been admitted to acute care settings and were not transferred to the ICU due to underlying comorbidities. Specifically, one patient at the age of 48 with glioblastoma and a severe disability experienced respiratory failure on day 5. Another patient, aged 73, presented severe cardiovascular comorbidities (including stroke and ischemic cardiomyopathy) and experienced worsening on day 14. Finally, an 81-year-old patient with prostatic neoplasia deteriorated on day 13.

We performed a univariable analysis to identify factors associated with unplanned rehospitalization. In this analysis, factors significantly associated with rehospitalization were age, chronic respiratory failure, delay from symptoms onset, and telemonitoring.

Thus, we performed a multivariate analysis using these parameters. In this analysis, factors associated with unplanned rehospitalization were immunosuppression (OR = 4.476), chronic respiratory failure (OR = 3.736), and a short delay between symptoms onset and the start of telemonitoring (OR = 1.344).

Results from the univariable and multivariable analyses are presented in Table 2.

We performed a literature review on COVID-19 patients requiring oxygen managed at home. We identified seven studies, including two in France, one in the US, two in the Netherlands, one in Türkiye, and one in India (18–24). The main study by Banerjee et al. was a large cohort study performed in the United States, which showed that patients discharged home with oxygen also had low rates of mortality and rehospitalization (18). In a French solution, the patients entered their own data via a web app, and no nurse was required (19). Moreover, a regional remote center was organized, and, depending on specific alerts, patients were contacted. Patients with oxygen were included at hospital discharge and required 3 L/min or less of oxygen therapy. The majority of patients were included after day 10 from symptoms onset and most of them experienced intensive care unit hospitalization prior to discharge under oxygen therapy. The duration of at-home oxygen therapy was longer.

In this present study, we also included patients at hospital discharge, but most of our patients were included straight after their consultation at the ED. They probably had fewer respiratory sequalae than patients included after ICU management, which could explain the short duration with oxygen therapy.

In our literature review, we identified seven other experiences of patients with COVID-19 requiring oxygen and managed at home (Table 3).

Included patients could be at an early or late stage of the disease, with a median age comprised between 51 and 62 years, with a majority of male patients. Patients at high risk were included (37.8 to 67%). Home monitoring was performed via a web app in five of seven studies, phone calls in three of seven studies, and web apps and phone calls were associated in two studies. The monitoring team included nurses in one study, a medical team in four of seven studies, and medical students in one study. Finally, the mortality rate was low (maximum 1.93%), as was unplanned hospitalization (max 12%).

Our study has several limitations. First, it is an observational study with potential bias, considering the indication and selection of patients. Moreover, no control group is available.

Second, the study population is heterogeneous: oxygen therapy at the early stage of the disease vs. oxygen therapy at discharge (late stage of the disease). Moreover, the healthcare systems and organizations differ between countries. All of these factors limit the generalizability of these study findings.

Finally, we could question the best primary endpoint for these types of process evaluations. As the primary goal is to ease overburdened hospitals, the best criteria seem, in our opinion, to be unplanned hospitalization, especially in the ICU, independent of patients’ mortality rates. Thus, our process seems promising for relieving hospitals overwhelmed with COVID-19 patients requiring oxygen.