COVID-19 Clinical Predictors in Patients Treated via a Telemedicine Platform in 2022

1. Introduction

COVID-19 is a viral illness caused by SARS-CoV-2 (Coronaviridae family). In March 2020, the World Health Organization characterized the COVID-19 outbreak as a pandemic [1]. The most common clinical manifestations in COVID-19 patients include fever or chills, fatigue, headache, muscle or body aches, dry cough, pneumonia, and dyspnea [2]. Comparison studies between COVID-19 cases and other common colds highlighted their different symptomatology, which can aid in the diagnosis of this disease [3,4,5].

During the pandemic, circulation-restricting and social distancing measures were recommended to decrease viral transmission in the population [1]. Early diagnosis allows immediate isolation to be indicated, decreasing the COVID-19 transmission time [6]. The high specificity of RT-qPCR (i.e., the detection of nucleic acids in nasopharyngeal and oropharyngeal samples by reverse transcription quantitative polymerase chain reaction) rendered it the universal diagnostic method [2]. Nevertheless, should this test be unavailable, a combination of clinical and laboratory features can be applied to aid in the diagnosis [3,5].

Restriction measures have changed healthcare delivery, with telehealth providing a viable alternative modality of care during the pandemic [7]. In 2020, elective outpatient care for stable patients was suspended in public health units in the state of Rio de Janeiro, Brazil, according to SES Resolution No. 2004/2020 of March 18th 2020 [8].

By definition, telemedicine encompasses the remote diagnosis and treatment of patients via telecommunication infrastructures, whereas telehealth encompasses any service used to provide healthcare remotely [9]. The characteristic pattern of patients using telehealth/telemedicine services is similar to that of other healthcare and digital health services [7]. Telehealth is comparable to in-person care for several clinical and process outcomes, [7] and can be associated with cost savings for both patients and the broader healthcare sector [10].

Conexa Saúde is a telemedicine platform that connects patients and healthcare professionals through technological means. During the COVID-19 pandemic, the tool consolidated an extensive clinical database, used herein to evaluate telemedicine as a predictive diagnosis resource.

Therefore, this study aims to analyze the database of a telemedicine platform with the goal of developing a COVID-19 diagnosis prediction model.

2. Materials and Methods

This is a longitudinal study of patients seen via the Conexa Saúde Telemedicine platform in 2022. In this study, a database section comprising 88,287 records was used. Inclusion criteria included patients whose data indicated residence in the state of Rio de Janeiro, and confirmation of COVID-19 or flu-like syndromes (i.e., due to other viral infections yielding similar symptoms). Exclusion criteria included duplicate cases, unconfirmed suspected cases, post-COVID-19 cases, diagnoses unrelated to flu-like syndromes, and records with no description of signs or symptoms. Records that had been reported and confirmed via a diagnostic test were considered as cases. All patients with negative confirmation for COVID-19, or confirmation of any other flu-like syndrome were considered as controls.

This study was approved by the Ethics and Research Committee of the Evandro Chagas National Institute of Infectious Diseases of the Oswaldo Cruz Foundation (Fiocruz), under opinion no. 5.001.681.

Clinical data was collected from Conexa Saúde's telemedicine platform database. Analyses were conducted using Statistical Package for Social Sciences for Windows v. 16.0 (SPSS Inc., Chicago, IL, USA). Descriptive statistics were used to summarize the findings. The simple frequencies of the main signs and symptoms reported among cases confirmed as COVID-19 and controls were described. The association between categorical variables was verified by Pearson's χ2 test of proportions, or Fisher's exact test. P-values <0.05 indicated statistically significant tests. Statistically significant variables were included in the logistic regression analysis. A binary logistic regression analysis was applied to develop the predictive model for the outcome of interest in this study. A multiple binary logistic regression model, controls v. COVID-19, was developed to obtain an odds ratio (OR) and a 95% confidence interval (CI). Backward elimination with the likelihood ratio test was applied to obtain statistically significant variables. The calibration of each final logistic model was evaluated via the Hosmer-Lemeshow goodness-of-fit test and a p-value > 0.05 indicated good agreement between observed and predicted disease.

3. Results

Data from the telemedicine platform identified 4,600 records from the state of Rio de Janeiro in 2022, of which 1910 (41.5%) were excluded: 390 duplicates, 383 suspected cases, 70 post-covid cases, 100 diagnoses not related to flu-like syndromes, and 967 entries without reported data on signs and symptoms. A total of 2690 patients were included, of which 2559 cases and 131 controls. The presence of rhinorrhea, sneezing and burning nose, ocular symptoms, abdominal pain, fatigue/tiredness, rhinosinusopathy, and wheezing/asthma and bronchospasm were more frequent in controls, showing a negative association with COVID-19. Fatigue/tiredness showed a positive association with COVID-19 cases (Table 1).

Seven independent variables were entered into a multiple binary logistic regression analysis to identify independent predictors that distinguished COVID-19 from controls. Based on backward elimination, six factors were considered significant: rhinorrhea, ocular symptoms, abdominal pain, rhinosinusopathy, and wheezing/asthma and bronchospasm were more frequent in controls, with their presence indicating a greater probability of a flu-like illness than COVID-19. Tiredness and fatigue were 3 times more prevalent in COVID-19 cases. The Hosmer-Lemeshow goodness-of-fit test showed a good fit of the model to the data (p = 0.978). The results of the binary logistic regression analysis are summarized in Table 2.

4. Discussion

Data from the telemedicine evaluation allowed the identification of factors that can aid in the early diagnosis of COVID-19 by differentiating it from other flu-like syndromes – a challenge arising from non-specific clinical characteristics, and frequently from limited access to specific diagnostics for each virus [5,11,12,13,14].

The use of telemedicine data is fundamental as it becomes increasingly ingrained in patient care provisioning. Previous studies have highlighted the high effectiveness of telehealth service data in anticipating new COVID-19 waves, which can both help understand the dynamics of the epidemic and support the surveillance of different diseases [15,16,17].

The present study revealed that fatigue/tiredness was a characteristic COVID-19 symptom, corroborating studies that indicated fatigue/tiredness as one of its main symptoms [2]. The presence of rhinorrhea, ocular symptoms, abdominal pain, rhinosinusopathy, and wheezing/asthma/bronchospasm were characteristic symptoms of the different flu-like syndromes. Sirijatuphat et al. [5] also observed a difference in rhinorrhea prevalence between COVID-19 and Influenza cases, being more frequent in the latter. Iyadorai et al. [3] observed that patients with influenza virus infection had a higher prevalence of fever (83.0%) and a similar prevalence of cough compared to patients with COVID-19 infection, whereas other clinical symptoms (sore throat, hoarseness, nasal congestion, rhinorrhea, sneezing, headache, and myalgia) were more common in COVID-19 infected patients.

Our results corroborated regarding the sore throat, nasal congestion, headache, myalgia, and cough symptoms, but not fever symptoms (no difference between the two groups) nor rhinorrhea (greater prevalence in flu-like syndrome cases). The main discrepancies may arise from differences between the populations in the different studies, and in the SARS-CoV-2 virus variants. At the start of the pandemic, anosmia was considered a predictor of COVID-19 [19]; however, its prevalence decreased with the different variants of the virus (Cardoso et al, 2022 [20]). Our study demonstrates the decrease in anosmia prevalence in COVID-19 cases in 2022.

In clinical practice, the identification of factors associated with the disease may be used as guidance to healthcare professionals to conduct targeted investigations. They may also be used as an additional tool where diagnostic resources are limited [5,18]. Moreover, early diagnosis allows for the immediate implementation of isolation measures, contributing to disease transmission control and a decrease of its incidence [6].

This study presented some limitations. Approximately 42% of the cases were excluded, mostly for missing data, which demonstrates a lack of structuring of electronic medical records aimed at research. More structured fields and training for doctors in record filling could decrease the occurrence of missing data, while data imputation strategies could be used for future studies. The control group was limited due to the high prevalence of COVID-19 during the study period; clinical manifestations of COVID-19 may vary according to the viral strain (different variants of SARS-CoV-2) and the vaccination status of the patients (partial or complete vaccination schedule). The use of retrospective data may be affected by information bias, yielding less accurate results.

The use of a database from a telemedicine platform was adequate and sufficient to identify significant factors to differentiate COVID-19 from other flu-like syndromes. It is necessary to improve the acquisition of data through telemedicine by developing more structured electronic medical records. Nevertheless, the use of this data has advantages in terms of decreasing cost for studies with a large sample size, ease of conducting studies in adverse situations such as a pandemic, little need for on-site infrastructure, and helping with faster prevention and surveillance actions in the face of worsening conditions.

5. Conclusions

Our study identified independent predictors associated with flu-like syndromes and COVID-19 that could help distinguish between these infections. Fatigue/tiredness is associated with COVID-19, whereas the presence of rhinorrhea, ocular symptoms, abdominal pain, rhinosinusopathy, and wheezing/asthma/bronchospasm are associated with flu-like syndromes.