Development and Validation of a Clinical Symptom-based Scoring System for Diagnostic Evaluation of COVID-19 Patients Presenting to Outpatient Department in a Pandemic Situation

Abstract

Background Preventive strategies in the form of early identification and isolation of patients are the cornerstones in the control of COVID-19 pandemic. We have conducted this study to develop a clinical symptom-based scoring system (CSBSS) for the diagnostic evaluation of COVID-19. Methods In this study, 378 patients presenting to screening outpatient clinic with clinical suspicion of COVID-19 were evaluated for various clinical symptoms. Statistical associations between presenting symptoms and reverse transcription-polymerase chain reaction (RT-PCR) results were analysed to select statistically significant clinical symptoms to design a scoring formula. CSBSS was developed by evaluating clinical symptoms in 70% of the total patients. The cut-off score of the CSBSS was determined from ROC (receiver operating characteristics) curve analysis to obtain a cut-off for optimum sensitivity and specificity. Subsequently, developed CSBSS was validated in the external validation dataset comprising 30% of patients. Results Clinical symptoms like fever >100⁰F, myalgia, headache, cough and loss of smell had significant association with RT-PCR result. The adjusted odds ratios (95% confidence interval [CI]) for loss of smell, fever >100°F, headache, cough and myalgia were 5.00 (1.78-13.99), 2.05 (1.36-3.07), 1.31 (0.67-2.59), 1.26 (0.70-2.26) and 1.18 (0.50-2.78), respectively. The ROC curve and area under the curve of development and validation datasets were similar. Conclusion The presence of fever >100°F and loss of smell among suspected patients are important clinical predictors for the diagnosis of COVID-19. This newly developed CSBSS is a valid screening tool that can be useful in the diagnostic evaluation of patients with suspected COVID-19. This can be used for the risk stratification of the suspected patients before their RT-PCR results are generated.

Keywords: covid-19; diagnosis; score; screening; validation.

Figure 1. ROC Curve for development dataset

Derivation dataset (70%), n=265 ROC: receiver operating characteristics

Figure 2. ROC Curve for validation dataset

Validation dataset (30%), n=113 ROC: receiver operating characteristics