Predicting COVID19 Critical Care Beds - The London North-West University Healthcare Trust Experience

  1. J.A Naude
  2. M. Taniparthi
  3. K. Adams
  4. J. Biggin-Lamming
  5. R. Elias
  6. P. Milner
  7. I. Mohammed
  8. M. Towers
  9. M. Kuper

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Our trust has an urgent need to make short-term (3-4 days in advance) informed operational decisions which take into account best-practice treatment regimens and known clinical features of COVID19 inpatients.

We believe that any model which is relied upon for operational decision making should have clinically identifiable parameters. Our model’s parameters take into account the conversion rates from acute wards into wards equipped with Non-Invasive Ventilation (NIV) and Mechanical Ventilation (MV), the typical time that these conversions take place and, the historical non-COVID usage of NIV and MV beds.

We have observed that this clinical performance is mathematically identical to a series of linear delays on the time varying inpatient level. High frequency inpatient data, sampled ∼4 hourly, has allowed our hospital trust to predict total critical care usage up to 4 days in advance without making any assumptions on upcoming inpatients. It is based entirely upon current bed occupancy levels and measured clinical pathways.

Through back-testing over the recent 4 months, the bounds of this model include 93.8% of all 4 day inpatient sequences. The average next-day error is 0.8 (95% CI: 0.44, 1.15) and so the system tends to over-predict the next day critical care inpatients by approximately 1 bed.

Potential extensions to the basic model include adjustments for seasonality, case mix, probabilistic marginalisation and known discharges.

Article activity feed

  1. ScreenIT

    SciScore for 10.1101/2020.11.20.20235226: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    Institutional Review Board Statementnot detected.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.
    Sex as a biological variablenot detected.

    Table 2: Resources

    No key resources detected.

    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: We detected the following sentences addressing limitations in the study:
    Limitations: This methodology can be reasoned to be ‘deductive’ in the sense that the prediction time horizon is intimately dependent on observed delays in the clinical evolution of inpatient numbers. If the time shift was not at all present, then future needs would depend on future inpatient numbers (which would be probabilistic). Currently, the system is configured to use up to the present patient numbers and hence the prediction time horizon is dependent on observed delays.

    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.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  2. Version published to 10.1101/2020.11.20.20235226v1 on medRxiv