Model E5

Model E5

Description of the tool an Early Warning Score for COVID-19
This is a diagnostic tool calculating the probability that a patient has COVID-19 based on clinical variables

To which patients does this tool apply?
This tool was developed to aid in the COVID-19 diagnosis of patients who first come to the doctor.
no      yes
no      yes
no      yes
years
F      M
<37,8°C      >37,7°C
(number of symptoms)
<5,8      >5,7
What information will you need?
Signs of pneumonia on CT (yes or no)
History of close contact with COVID-19 patient (yes or no)
Fever (yes or no)
Maximum body temperature since coming to the hospital (>37.8°C, yes or no)
Neutrophil-to-lymphocyte ratio (>5.8, yes or no)
Sex (male or female)

How was this tool developed?
A total of 1311 patients who were admitted to the First Affiliated Hospital, School of Medicine, Zhejiang University and performed at least once SARS-CoV-2 nucleicacid detection for analysis were included in the study. The study included patients from Zhijiang district (designated district) and from the Qingchun District (outpatients)

Patients who tested positive were classified into the COVID-19 group (93), the rest into the non-COVID-19 group (1218‬‬). The following non-COVID-19 patients were excluded:

40% of patients outpatients between January 20th 2020 and February 5th 2020
60% of patients outpatients between February 6th 2020 and February 19th 2020
121 asymptomatic patients without history of exposure but had strong willingness for detection
111 patients with important information deficits

The data was split into a training set using time of enrollment, with the training data including patients between January 20th 2020 and February 5th 2020, and the validation data including patients between February 6th 2020 and February 19th 2020.

The data was collected from the electronic medical record system. Exposure to COVID-19 was defined as having contact with a COVID-19 confirmed patient, or exposure related to Wuhan, as in: traveled to, lived in, or had close contact with someone who has been to Wuhan.

Laboratory confirmation was achieved by real time reverse transcription-polymerase chain reaction (RT-PCR) assay for SARS-CoV-2 according to the protocol established by the WHO.

Mann-Whitney U-test were used to evaluate continuous data, chi-square test were used for categorical variables, and variables with P < 0.05 were selected to perform multivariate logistic regression analysis by using the forward stepwise method.

The model performed with a sensitivity of 93.2% and a specificity of 87.4%.

Further details about the modeling approach can be found in the published article.
https://www.medrxiv.org/content/10.1101/2020.03.05.20031906v1.full.pdf