Clustering of age standardised COVID-19 infection fatality ratios and death trajectories
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (ScreenIT)
Abstract
Background
An accurate measure of the impact of COVID-19 is the infection fatality ratio, or the proportion of deaths among those infected, which does not depend on variable testing rates between nations. The risk of mortality from COVID-19 depends strongly on age and current estimates of the infection fatality ratio do not account for differences in national age profiles. Comparisons of cumulative death trajectories allow the effect and timing of public health interventions to be assessed.
Our purpose is to (1) determine whether countries are clustered according to infection fatality ratios and (2) compare interventions to slow the spread of the disease by clustering death trajectories.
Methods
National age standardised infection fatality ratios were derived from age stratified estimates from China and population estimates from the World Health Organisation. The IFRs were clustered into groups using Gaussian mixture models. Trajectory analysis clustered cumulative death rates in two time windows, 50 and 100 days after the first reported death.
Findings
Infection fatality ratios from 201 nations were clustered into three groups: young, medium and older, with corresponding means (SD) of 0.20% (0.03%), 0.38% (0.11%) and 0.93% (0.21%).
At 50 and 100 days after the first reported death, there were two clusters of cumulative death trajectories from 113 nations with at least 25 deaths reported at 100 days. The first group had slowly increasing or stable cumulative death rates, while the second group had accelerating rates at the end of the time window. Fifty-two nations changed group membership between the time windows.
Conclusion
A cluster of younger nations have a lower estimated infection fatality ratio than older nations. The effect and timing of public health interventions in preventing the spread of the disease can be tracked by clustering death rate trajectories into stable or accelerating and comparing changes over time.
Article activity feed
-
-
SciScore for 10.1101/2020.08.11.20172478: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
No key resources detected.
Results from OddPub: Thank you for sharing your code and data.
Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:Our study has some limitations. Age stratified IFRs relative to the 80+ age group may differ from those in China or Italy, particularly in countries where health system support is limited, overwhelmed or inequitable. The assumption that infection rates are equal across age groups may be met only in nations with large outbreaks and high death rates or with high inter-generational mixing [19, 22-23]. However, cluster group membership in …
SciScore for 10.1101/2020.08.11.20172478: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
No key resources detected.
Results from OddPub: Thank you for sharing your code and data.
Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:Our study has some limitations. Age stratified IFRs relative to the 80+ age group may differ from those in China or Italy, particularly in countries where health system support is limited, overwhelmed or inequitable. The assumption that infection rates are equal across age groups may be met only in nations with large outbreaks and high death rates or with high inter-generational mixing [19, 22-23]. However, cluster group membership in the trajectory analysis did not depend on weighting for age adjustment. COVID-19 mortality data may be under-reported and the calculated IFRs may be under-estimates or lower bounds [18]. Conversely, mortality from COVID-19 may reduce throughout the pandemic as more effective treatments for the disease are discovered [24] and the calculated IFRs may become upper bounds. The IFR estimates were produced from data that was available in February 2020, before large scale seroprevalence studies had been conducted [3-6]. If more up to date age stratified IFR estimates become available, the analysis can be updated. The spread of disease through a population may also depend on international mobility, climate or regional susceptibility [25]. Finally, the association between the IFR clusters and the change in death rate trajectories between time points may be biased by socioeconomic factors or missing death rate trajectories in countries excluded due to low numbers of reported deaths.
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.
-