how age demographic data shows Covid-19 passes like a tidal wave through the population
and how the inherent ‘sociability’ within age bands acts as a multiplier of infection rates
Introduction and Summary
How we interact with others is key to understanding how a virus like Covid-19 spreads through communities and within this article I’m going to demonstrate how this has impacted the spread of Covid-19 within England.
I am going to use actual case data to demonstrate the relentless underlying march of the virus but how different age bands see different growth rates of transmission due what I call a multiplier effect. An effect that reflects the underlying social interactions of the age group.
What emerges is a common sense picture, that time spent together and the frequency of meeting different people is key to viral transmission and how this is dramatically different between children, the working population and retirees.
What also becomes clear is that there is no one particular age group within the population that drives viral spread. Indeed, I shall demonstrate how clearly visible outbreaks within the student population have no apparent impact on the trend of the virus in other age groups and how the same appears to be true of school-aged children.
The Story of Covid-19 told in age bands
Starting with the under 20's
I’m going to introduce the data using the under 20 age bands — which are in isolation both the ones easiest to misinterpret, and the ones that skew the overall data the most.
This chart shows the new cases by day and an average for the week to aid interpretation. The weekly daily average excludes Christmas Day and New Years day because these days were artificially low. I have also used an extended 9 day ‘week’ for the period from the 28th December to the 5th January followed by a shorter 5 day week to reflect the unusual nature of the extended festive period.
In terms of this chart:
- perhaps most obvious in this chart is the spike in 15 to 19 year olds. A spike that coincides with the start of the academic year on the 28th September;
- less obvious is that infections within school aged children grows more quickly during term times — 2nd September to 23rd October and 2nd November to 18th December; and
- there is an obvious and significant step up over Christmas. Something that is most marked in the more independent 15 to 19 year old group (an 80% increase compared with 40% in the younger groups).
Moving up the ages, and now using just the weekly average, we begin to put the above in context as within the following picture:
The university impact is again obvious as is the even more marked increase at Christmas in the older age groups, the first real evidence of the multiplier effect.
We can also begin to imagine that the 15 to 24 year old populations included two groups, university students and other young people — the former showing marked increases, the latter following a trend more akin to the younger groups, something that becomes even more apparent as we move up the ages once more.
Adults Over 25
Both the true underlying trend of the virus and the impact of the multiplier effect becomes exceedingly clear in the over 25’s:
What also becomes apparent is the significance of retirement, the 60 to 64 year age band, which is a key transitioning life stage, after which the multiplier effect reduces significantly.
Within all groups there is noticeable and marked impact of the festive period, more or less doubling the cases in each group. This is a clear reflection of how quality time spent together over the festive period gives the virus the opportunity to infect close friends and family.
The Science supporting the Theory
To summarise the charts:
- there is a clear underlying and consistent wave across all age bands as the virus spreads through the community;
- the impact on each age band changes by a ‘multiplier’ effect, meaning that the most social groups (essentially the working age population) see a bigger impact than the least social ones (children and those well into retirement);
- the 60 to 64 year old, retirement population, was a key transition age.
So, what explains this pattern?
There are essentially three factors to take into account:
Firstly , the number of interactions people have with others. The following chart, taken from a study found on Research Gate on the Mixing Patterns Between Age Groups demonstrates how this changes with the ages:
This chart clearly ties in with the new case data we’ve just seen on Covid-19 in England.
Secondly, and from the same study, the fact that people tend to mix most with people of their own age — the one exception being families.
Again explaining the multiplier effect within the age bands.
Thirdly, the time spent together, and its importance in driving viral transmission:
Taken together these all explain the increase in infection we see at Christmas and New Year.
It also explains why children are often seen as key spreader of disease but in fact are not. Instead they are simply more likely to be exposed to the virus because they cross two vectors — their own peer group and their parents.
So overall, what we clearly see in the data is a single underlying wave representing the spread of the virus from the return of the virus with the winter season.
Then because of various multiplier effects within each age group, the impact is magnified in certain groups (essentially the working/active element of the population). Hence the virus reaches more of the relevant population demographic as it spreads.
Additionally, there are key times, like Christmas which act as a significant multiplier/spreading event, this is because of the enhanced amount of quality time spent together, that does not occur at other times of the year.
Finally — it is also worth pointing out that what appears as two waves of cases in the overall England data is actually two different outbreaks — the first the return of the virus in the major conurbations of the Midlands and North of England, the second an outbreak in December in London and the Commuter belt — I shall add a separate article on this to demonstrate this fact as soon as I can.
You can find me on Twitter https://twitter.com/BottoMatt
Relevant/Related Articles/Further Information
(PDF) Mixing patterns between age groups in social networks
We present a method for estimating transmission matrices that describe the mixing and the probability of infection…
Age Differences in Adults' Daily Social Interactions: An Ecological Momentary Assessment Study
Prevailing research suggests that social relationships get better with age, but this evidence is largely based on…
This latter study is relevant because it adds further weight to the impact of the duration of contact/length of exposure being critical to viral transmission:
and the role of family and friends:
- All of the data is sourced from the UK’s Coronavirus Reporting Webside — coronavirus.data.gov.uk.
- To make it easier to see the underlying trends, a weekly daily average has been calculated and shown on the charts. The weekly daily average excludes Christmas Day and New Years day because these days were artificially low due to festivities.
- The averages also include an extended 9 day period from the 28th December to the 5th January followed by a shorter 5 day week. This is to reflect the differing nature of the festive period.
- New cases data reflects when individuals with symptoms become apparent, actual infection will have occurred some 3 to 6 days prior to the onset of symptoms.