What can trends in hospital deaths from COVID-19 tell us about the progress and peak of the pandemic? pandemic? An analysis of death counts from England announced up to 25 April 2020

  1. David A Leon
  2. Christopher I Jarvis
  3. Anne Johnson
  4. Liam Smeeth
  5. Vladimir M Shkolnikov

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Background

Reporting of daily hospital COVID-19 deaths in the UK are promoted by the government and scientific advisers alike as a key metric for assessing the progress in the control of the epidemic. These data, however, have certain limitations, among which one of the most significant concerns the fact that the daily totals span deaths that have occurred between 1 and 10 days or more in the past.

Data and methods

We obtained daily data published published by NHS England up to and including April 25 in the form of Excel spreadsheets in which deaths counts are presented by date of death according to age and region. Simple descriptive analyses were conducted and presented in graphical and tabular form which were aimed at illustrating the biases inherent in focussing on daily counts regardless of when the deaths occurred. We then looked at how a less biased picture could be obtained by looking at trends in death counts stratifying by individual period of delay in days between occurrence of death and when the death was included in the daily announcement.

Findings

The number of hospital COVID-19 deaths announced daily overestimates the maximum number of deaths actually occurring so far in the epidemic in the UK, and also obscures the pattern of decline in deaths. Taking account of reporting delays suggests that for England as a whole a peak in hospital COVID-19 deaths may have been reached on April 8 with a subsequent gradual decline suggested. The same peak is also seen among those aged 60-79 and 80+, although there is slightly shallower decline in the oldest age group (80+ years). Among those aged 40-59 years a later peak on April 11 is evident. London shows a peak on April 8 and a clearer and steeper pattern of subsequent decline compared to England as a whole.

Interpretation

Analyses of mortality trends must take account of delay, and in communication with the public more emphasis should be placed on looking at trends based on deaths that occurred 5 or more days prior to the announcement day. The slightly weaker decline seen at age 80+ may reflect increased hospitalisation of people from care homes, whereas the later peak under the age of 60 years may reflect the higher proportions at these younger ages being admitted to critical care resulting in an extension of life of several days.

Competing interests

All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years other than LS who reported grants from Wellcome, MRC, NIHR, GSK, BHF, Diabetes UK all outside the submitted work; no other relationships or activities that could appear to have influenced the submitted work other than LS who is a Trustee of the British Heart Foundation and AJM who is a member of the Royal Society Delve Committee.

Article activity feed

  1. ScreenIT

    SciScore for 10.1101/2020.04.21.20073049: (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:
    As we discuss below there are a number of caveats concerning how far hospital COVID-19 deaths can be regarded as reflecting the trajectory of the epidemic above and beyond the delays in reporting to the NHS that we have already discussed. However, as a first approximation, one may work back from a posible peak on April 8 to when the rate of community transmission began to fall. If we assume a median time of around 23 days following infection (5 days median incubation period (9) and 18 days from symptom onset to death(10)) one can track back to a date of infection of around March 16. This is a week prior to the hard lockdown announced on 23 March. However there is evidence of a decline in social contact and travel in the previous week (March 16-23) (11, 12). From another perspective, this illustrates just how long the delay is between a fall in rate of infection and it becoming manifest in a fall in deaths. To the 23 days one would need to add an additional 5 or more to be confident that the mortality signal was not due to delays in reporting, resulting in a 28 day gap between a decline in infections becoming apparent in deaths by date of occurrence. The dynamics of infection and subsequent mortality are importantly driven by changes in patterns of transmission in the community. However, it is important to note that community transmission is not the only source of infections that lead to death in hospital. There are also infections within care-homes, which will behave differen...

    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.
    • Thank you for including a protocol registration statement.

    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 10.1101/2020.04.21.20073049v2 on medRxiv
  3. Version published to 10.1101/2020.04.21.20073049v1 on medRxiv