Frailty and comorbidity in predicting community COVID ‐19 mortality in the U.K. Biobank: The effect of sampling

  1. Jonathan K. L. Mak
  2. Ralf Kuja‐Halkola
  3. Yunzhang Wang
  4. Sara Hägg
  5. Juulia Jylhävä

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Background/Objectives

Frailty has been linked to increased risk of COVID‐19 mortality, but evidence is mainly limited to hospitalized older individuals. This study aimed to assess and compare predictive abilities of different frailty and comorbidity measures for COVID‐19 mortality in a community sample and COVID‐19 inpatients.

Design

Population‐based cohort study.

Setting

Community.

Participants

We analyzed (i) the full sample of 410,199 U.K. Biobank participants in England, aged 49–86 years, and (ii) a subsample of 2812 COVID‐19 inpatients with COVID‐19 data from March 1 to November 30, 2020.

Measurements

Frailty was defined using the physical frailty phenotype (PFP), frailty index (FI), and Hospital Frailty Risk Score (HFRS), and comorbidity using the Charlson Comorbidity Index (CCI). PFP and FI were available at baseline, whereas HFRS and CCI were assessed both at baseline and concurrently with the start of the pandemic. Inpatient COVID‐19 cases were confirmed by PCR and/or hospital records. COVID‐19 mortality was ascertained from death registers.

Results

Overall, 514 individuals died of COVID‐19. In the full sample, all frailty and comorbidity measures were associated with higher COVID‐19 mortality risk after adjusting for age and sex. However, the associations were stronger for the concurrent versus baseline HFRS and CCI, with odds ratios of 20.40 (95% confidence interval = 16.24–25.63) comparing high (>15) to low (<5) concurrent HFRS risk category and 1.53 (1.48–1.59) per point increase in concurrent CCI. Moreover, only the concurrent HFRS or CCI significantly improved predictive ability of a model including age and sex, yielding areas under the receiver operating characteristic curve (AUC) >0.8. When restricting analyses to COVID‐19 inpatients, similar improvement in AUC was not observed.

Conclusion

HFRS and CCI constructed from medical records concurrent with the start of the pandemic can be used in COVID‐19 mortality risk stratification at the population level, but they show limited added value in COVID‐19 inpatients.

Article activity feed

  1. Version published to 10.1111/jgs.17089
  2. ScreenIT

    SciScore for 10.1101/2020.10.22.20217489: (What is this?)

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

    Table 1: Rigor

    Institutional Review Board StatementIRB: The UK Biobank study was approved by the North West Multi-Centre Research Ethics Committee.
    Consent: All participants provided written informed consent for data collection, analysis, and record linkage.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.
    Sex as a biological variablenot detected.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    All analyses were performed using Stata v16.0
    Stata
    suggested: (Stata, RRID:SCR_012763)

    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:
    Nevertheless, there are several limitations to this study. Firstly, the HFRS was originally designed for hospitalized individuals rather than for the general population, although we attempted to account for individuals with missing hospital data and showed that it may be applied to community samples as well. Alternatively, future research may utilize frailty measures based on routine primary care data, such as the electronic FI [32] for assessing its predictive ability for COVID-19 mortality. Secondly, during the earlier periods of the epidemic, COVID-19 testing in the UK was largely restricted to hospitalized individuals, who have more severe course of the disease. As such, mild or asymptomatic COVID-19 cases may conceivably be missed, leading to an underestimation of COVID-19 positive cases. Thirdly, we modelled the outcome, COVID-19 mortality, as a binary trait rather than as time-to-event outcome because we could not ascertain the exact date of confirmed COVID-19 infection for several individuals in the COVID-19 positive subsample. However, as the follow-up time was limited, it could be considered essentially complete for most participants (i.e., minimal censoring due to migration and other deaths). Finally, UK Biobank is not a nationally representative sample, with generally healthier and less socioeconomically deprived participants than the UK average [33], thereby reducing the generalizability of our findings. In conclusion, HFRS and CCI, measures of frailty and comorb...

    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.

  3. Version published to 10.1101/2020.10.22.20217489v2 on medRxiv
  4. Version published to 10.1101/2020.10.22.20217489v1 on medRxiv