Simply Out of Date
The ‘Scientists for EU’ Facebook page posted a graph of excess deaths. This graph showed a measure of excess deaths in 19 countries. It claimed: “And this is where we now are.”
The graph had no source or links. At the shared time, the graph was already out of date. Statistical agencies will continue to publish mortality statistics. As a result, such comparisons are provisional and changeable.
A cold trail
The graph compared excess deaths per million people “during Covid outbreaks”. Facebook users shared this graph over 3,600 times. The ‘Scientists for EU’ post commented:
And this is where we now are.
The graph contains no links or sources. The image itself appears to mimic a Financial Times data visualisation. Since the initial publication:
Spain made a revision to its mortality estimates, adding 12,000 to its toll of excess deaths from coronavirus in a one-off adjustment to 43,000.
The Instituto Nacional de Estadística — Spain’s statistical office — is less timely. Their latest provisional death figures are for the first half of 2019.
Some countries do not publish recent statistics, or make those figures hard to find. John Burn-Murdoch (Financial Times) shows the difficulty in getting Italian stats:
On 28th May, the Financial Times graph was:
The statistics are after the Spanish update, but before the latest ONS report.
What are excess deaths?
Excess deaths means deaths from all causes above a baseline. That baseline is often an average of previous years. In its reports, the Office for National Statistics uses the average of the prior five years.
Excess mortality is difficult to interpret. By definition, the value concerns deaths from all causes above a given baseline.
During this pandemic, excess deaths are not “deaths from coronavirus”.
Excess deaths should be a consistent measure. For example, deaths above the average of past years means the same thing across countries.
There are issues with these provisional comparisons:
- Consistency: some institutes may use a modelled value for the baseline. Other statistics offices (such as the ONS) use averages of prior years.
- Interpretation: some deaths may occur due to the pandemic’s strain on health services. These deaths are not direct from the virus. State lock-downs may influence what types of deaths occur. There are ‘natural’ excess deaths, from stalled mortality rates and increased populations.
- Availability: countries may not publish their recent all-cause mortality statistics. The Financial Times article highlights this problem. The graph is for those chosen countries with comparable statistics, not “the world”.
- Timeliness: statistical agencies do not publish mortality statistics at the same time. Some offices publish weekly. Others publish less often. Annotations should make clear when latest figures for each country covers.
- Harvesting: some people may be frail, with the pandemic ‘bringing forward’ their death. We need to look at death statistics once the current pandemic has waned to understand this issue.
The pandemic persists. Comparisons of excess deaths between countries are provisional and changeable.
On social media, graphs should include links to sources. Maintaining that trail demonstrates trustworthiness to readers.
Excess deaths are a difficult measure to interpret. This key measure needs accurate description.