Symptoms that predict positive COVID-19 testing and hospitalization: an analysis of 9,000 patients
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (ScreenIT)
Abstract
Purpose
To develop a reliable tool that predicts which patients are most likely to be COVID-19 positive and which ones have an increased risk of hospitalization.
Methods
From February 2020 to April 2021, trained nurses recorded age, gender, and symptoms in an outpatient COVID-19 testing center. All positive patients were followed up by phone for 14 days or until symptom-free. We calculated the symptoms odds ratio for positive results and hospitalization and proposed a “random forest” machine-learning model to predict positive testing.
Results
A total of 8,998 patients over 16 years old underwent COVID-19 RT-PCR, with 1,914 (21.3%) positives. Fifty patients needed hospitalization (2.6% of positives), and three died (0.15%). Most common symptoms were: cough, headache, sore throat, coryza, fever, myalgia (57%, 51%, 44%, 36%, 35%, 27%, respectively). Cough, fever, and myalgia predicted positive COVID-19 test, while others behaved as protective factors. The best predictors of positivity were fever plus anosmia/ageusia (OR=6.31), and cough plus anosmia/ageusia (OR=5.82), both p<0.0001. Our random forest model had an ROC-AUC of 0.72 (specificity=0.70, sensitivity=0.61, PPV=0.38, NPV=0.86). Having steady fever during the first days of infection and persistent dyspnea increased the risk of hospitalization (OR=6.66, p<0.0001 and OR=3.13, p=0.003, respectively), while anosmia-ageusia (OR=0.36, p=0.009) and coryza (OR=0.31, p=0.014) were protective.
Conclusion
Present study and algorithm may help identify patients at higher risk of having SARS-COV-2 (online calculator http://wdchealth.covid-map.com/shiny/calculator/ ), and also disease severity and hospitalization based on symptoms presence, pattern, and duration, which can help physicians and health care providers.
Article activity feed
-
SciScore for 10.1101/2021.08.09.21261729: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
Ethics IRB: After local ethics committee approval number 4.173.069, all information was gathered from the University of Campinas outpatient health center, Campinas, São Paulo, one of the biggest universities in Brazil. Sex as a biological variable not detected. Randomization not detected. Blinding not detected. Power Analysis not detected. Table 2: Resources
Software and Algorithms Sentences Resources All statistical analyses were performed using R version 4.0.2 (2020-06-22) on RStudio platform version 1.3.1073 and using the following packages: tidyverse, lubridate, forecast, quantreg, splines, ggmap, pracma, and janitor. RStudiosuggested: (RStudio, RRID:SCR_000432)Results from OddPub: We did not detect open …
SciScore for 10.1101/2021.08.09.21261729: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
Ethics IRB: After local ethics committee approval number 4.173.069, all information was gathered from the University of Campinas outpatient health center, Campinas, São Paulo, one of the biggest universities in Brazil. Sex as a biological variable not detected. Randomization not detected. Blinding not detected. Power Analysis not detected. Table 2: Resources
Software and Algorithms Sentences Resources All statistical analyses were performed using R version 4.0.2 (2020-06-22) on RStudio platform version 1.3.1073 and using the following packages: tidyverse, lubridate, forecast, quantreg, splines, ggmap, pracma, and janitor. RStudiosuggested: (RStudio, RRID:SCR_000432)Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).
Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.Results from TrialIdentifier: No clinical trial numbers were referenced.
Results from Barzooka: We did not find any issues relating to the usage of bar graphs.
Results from JetFighter: We did not find any issues relating to colormaps.
Results from rtransparent:- Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
- Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
- No protocol registration statement was detected.
Results from scite Reference Check: We found no unreliable references.
-
