prognostic multivariable model on admission for hospitalized patients with COVID-19
This prediction model was developed by Xie J et al. (2020) to identify patients with lethal COVID19. Its practical reliance on commonly available parameters should improve usage of limited healthcare resources and patient survival rate.
Research authors: Xie J, Hungerford D, Chen H, Abrams ST, Shusheng L, Wang G, Wang Y, Kang H, Bonnett L, Zheng R, Li X, Tong Z, Du B, Qiu H, Toh C
Details Formula Study characteristics Files & References
Model author
Model ID
Revision date
MeSH terms
  • Coronavirus, SARS
  • Clinical Prediction Rule
  • Model type
    Logistic regression (Calculation)
    No Formula defined yet
    Condition Formula

    Additional information

    We developed and internally validated a multivariable logistic regression model to predict inpatient mortality in COVID-19 positive patients aged >18 years using data collected retrospectively from Tongji Hospital, Wuhan (299 patients). All patients were diagnosed by positive tests of novel coronavirus nucleic acids (SARS-CoV-2), according to WHO interim guidance. External validation was conducted using a retrospective cohort from Jinyintan Hospital, Wuhan (145 patients). Nine variables commonly measured in these acute settings were considered for model development, including age, biomarkers and comorbidities. Backwards stepwise selection and bootstrap resampling were used for model development and internal validation. We assessed discrimination via the C-statistic, and calibration using calibration-in-the-large, calibration slopes and plots.

    The final model included age, lymphocyte count, lactate dehydrogenase and SpO2 as independent predictors of mortality. Discrimination of the model was excellent in both internal (c=0·89) and external (c=0·98) validation. Internal calibration was excellent (calibration slope=1). External validation showed some over-prediction of risk in low-risk individuals and under-prediction of risk in high-risk individuals prior to recalibration. Recalibration of the intercept and slope led to excellent performance of the model in independent data.

    Preprint article by Xie et al, 2020 avaialble at MedRxiv. 

    Study Population

    Total population size: 299
    Males: {{ model.numberOfMales }}
    Females: {{ model.numberOfFemales }}

    Continuous characteristics

    Name LL Q1 Median Q3 UL Unit
    Age 54 65 73 years
    Days since onset 7 10 14 years
    SpO2 90 95 98 %
    SOFA score 2 2 4 points
    Systolic blood pressure 118 132.5 145 mmHg
    White blood cell count 4.7 6.8 10.6 10^9/L
    Lymphocyte count 0.50 0.75 1.11 10^9/L
    Platelet count 137.3 179.5 247.8 10^9/L
    ALT 16 26 40.8 U/L
    AST 24 33.5 56 U/L
    Bilirubin 7.6 10.3 14.6 umol/L
    BUN 3.9 5.9 9.2 mmol/L
    Creatinine 58 74 95 umol/l
    Creatinine kinase 70 131 380.8 U/L
    LDH 264 384 541 U/L
    CRP 27.4 65.3 113.9 mg/dl
    INR 1.04 1.13 1.25 ratio
    Procalcitonine 0.05 0.13 0.37 ng/ml

    Categorical characteristics

    Name Subset / Group Nr. of patients
    Hypertension yes 127
    no 172
    Diabetes yes 55
    no 244
    COPD yes 15
    no 284
    Heart failure yes 13
    no 286
    Smoking yes 10
    no 289
    Other co-morbidities yes 19
    no 280
    SpO2 ≥90% 229
    <90% 67
    D-dimer <0.5 mg/L 49
    0.5-1.0 mg/L 55
    ≥1.0 mg/L 154
    cTnI <28.0 pg/ml 173
    ≥28.0 pg/ml 69
    prognostic multivariable model on admission for hospitalized patients with COVID-19
    Refer to Intended Use for instructions before use
    Evidencio B.V., Irenesingel 19, 7481 GJ, Haaksbergen, the Netherlands

    Related files

    Supporting Publications

    Notes are only visible in the result download and will not be saved by Evidencio

    Predicted probability of mortality:

    {{ resultSubheader }}
    {{ chart.title }}

    Outcome stratification

    Result interval {{ additionalResult.min }} to {{ additionalResult.max }}

    Conditional information

    Result interpretation

    Intended use: 
    There is an urgent need for effective prediction models to identify patients who are most likely to develop respiratory failure and poor outcomes. This model was developed to predict mortality of COVID-19 hospitalized patients to guide decision-making between clinicians, patients and families.

    Model performance: 
    The model has been validated internally and externally with a cohort of 145 patients from a different hospital. Discrimination of the model was excellent in both internal (c-statistic=0.89) and external (c=0.98) validation. When applied to new patients, the model yielded probabilities of mortality that were too high for low risk patients and too low for high risk patients (calibration slope >1). After recalibration of the model to account for underlying differences in the risk profile of the datasets, calibration plots showed excellent agreement between predicted and observed probabilities of mortality. 

    Model limitations: 
    Recruited patients were mainly self-referrals, who are more likely to have severe symptoms that cause them to seek emergency medical help. Other hospitals are more likely to test patients who have more pronounced symptoms, thus creating an inherent bias in the patient sample. Therefore, the cohort may not accurately represent patients with mild or asymptomatic COVID-19. These results need to be further validated with patients who are hospitalized at different locations and for different severities of illness to confirm generalizability and robustness of the model. Finally, the accompanying publication describing the development and validation of the model has not yet been peer reviewed. 

    Preprint article by Xie et al, 2020 avaialble at MedRxiv. 

    {{ file.classification }}

    Calculations alone should never dictate patient care, and are no substitute for professional judgement. See our full disclaimer.

    Please enter a comment of rating
    Comments are visible to anyone

    Model feedback

    No feedback yet 1 Comment {{ model.comments.length }} Comments
    Not rated | On {{ comment.created_at }} {{ comment.user.username }} a no longer registered author wrote:

    Please sign in to enable Evidencio print features

    In order to use the Evidencio print features, you need to be logged in.
    If you don't have an Evidencio Community Account you can create your free personal account at:

    Printed results - Examples {{ new Date().toLocaleString() }}

    Evidencio Community Account Benefits

    With an Evidencio Community account you can:

    • Create and publish your own prediction models.
    • Share your prediction models with your colleagues, research group, organization or the world.
    • Review and provide feedback on models that have been shared with you.
    • Validate your models and validate models from other users.
    • Find models based on Title, Keyword, Author, Institute, or MeSH classification.
    • Use and save prediction models and their data.
    • Use patient specific protocols and guidelines based on sequential models and decision trees.
    • Stay up-to-date with new models in your field as they are published.
    • Create your own lists of favorite models and topics.

    A personal Evidencio account is free, with no strings attached!
    Join us and help create clarity, transparency, and efficiency in the creation, validation, and use of medical prediction models.

    Disclaimer: Calculations alone should never dictate patient care, and are no substitute for professional judgement.