All age groups could face significant mortality increase in 2020: Georgian case

As the number of fatalities due to the Coronavirus pandemic rises worldwide (over 50 thousand as of April 2, and doubling every 5.5 days), more discussions focus on the different scenarios countries are going to face in the coming weeks and months.

Up until now Georgia’s place in these discussions was limited to local and international praise for its effective handling of the situation with zero fatalities (as of April 2) and moderate increase of cases compared to most European countries. At the very least, Georgia has been able to buy itself valuable time to prepare for what is still to come.

Swift action by the authorities and aggressive contact tracing (partly facilitated by the capabilities of the Tbilisi based Lugar Research Center) seem to have been responsible for the success so far. However, popular support for the imposed social distancing measures will likely become a crucial factor in slowing down the outbreak of the disease as Georgia enters the phase of uncontrolled internal spread.

On March 28, with 90 registered cases, Amiran Gamkrelidze, Head of the National Center of Diseases Control and Public Health, confirmed it is “highly likely that there might be a precedent of domestic transmission” after two newly confirmed cases had contracted the virus from apparently unidentifiable sources. On April 2, seven employees of the National Bank of Georgia tested positive for Covid-19, with speculation arising about a possible transmittion by touching cash. While the authorities are still trying to trace every case, the growing number of people in quarantine (over 5’000) makes this look like a sisyphean task.

Since uncontrolled internal transmission is now likely underway, I look at the impact an outbreak could have on mortality among the Georgian population. In a worst case scenario, what is the risk of dying within one year which Georgians of different age groups face in 2020, and how does this risk compare with the risk of death in a normal year (2018)?

Age is one (but not the only) main factor increasing fatality risk in patients suffering from Coronavirus.

The Centers for Disease Control and Prevention has listed age as a major risk factor for experiencing critical illness associated with COVID-19, particularly at age 65 and older. Older adults have a harder time resisting infection as the white blood cells needed to identify and eliminate infections decline in number and efficiency. Risk clearly exists among younger adults as well, although at significantly lower levels.

Gallup, 11 Million in U.S. at Serious Risk If Infected With COVID-19

However, are older people actually facing a steeper increase in mortality during the Corona pandemic than younger people? The answer to this question is surprisingly nuanced, as we will see.


Key takeaways:

  • Independently of the chosen scenario, adults of all age groups experience roughly the same increase in risk of death.
  • In a worst case scenario, adult Georgians may be about twice as likely to die in 2020 than in a normal year.

The model is based on Georgian demographic data (to determine the risk of death in a normal year) and Chinese case fatality data for Covid-19 (sample size 44’672). We grant, that the average fatality rate for an outbreak in Georgia may differ from the one observed in China due to factors such as differing overall health of the population, preparedness of the health care system, available treatment, etc. Moreover, the model takes into account the growing consensus that around half of the infections are never registered as such (for example asymptomatic infections) so long as there is a lack of comprehensive testing.

The worst case scenario assumes a herd immunity approach to handle the pandemic, for which to succeed, up to 60% of the world’s population would need to get the infection. This strategy aims to slow down (‘flatten the curve’) rather than outright stop the spread of the virus. In this way, the hope goes, a total collapse of healthcare systems as well as the economy can be averted, while the growing number of recovered patients will naturally reduce the power of the virus to spread:

If the virus keeps spreading, eventually so many people will have been infected and (if they survive) become immune that the outbreak will fizzle out on its own as the germ finds it harder and harder to find a susceptible host.

MIT Technology Review, What is herd immunity and can it stop the coronavirus?

In the early days of the european Covid-19 outbreak, herd immunity seemed to have been accepted by some governments (notably the british one) as a possible exit strategy from disaster. However, this approach immediately came under fire, and it is easy to see why: the death toll of such a scenario could potentially reach dimensions that no democratically elected government can afford to justify.

Let us assume 60% of the Georgian population (2.2 Mio.) will have had the virus in a year from now. According to current estimates, almost half of the infections take an asymptomatic course. Of the remaining cases, around 20% (more than 220’000 Georgians) are expected to require hospitalisation, and 5% (over 55’000) will need intensive care – all of it concentrated in a time span of just a few months. In such a scenario, the Georgian healthcare system would almost certainly be stretched far beyond its limits. This prompts us to assume an overall fatality rate twice as high as in the Chinese case (where 2.3% of recorded cases died), closer to what countries like Italy, Spain or France seem to be experiencing at the moment. Adverse health conditions, such as Georgia’s high level of antibiotics usage and the country having one of the highest numbers of smokers in the world, could contribute to that as well. It is worth keeping in mind, however, that at such an early stage of the pandemic fatality rates are notoriously hard to project and estimates may well be lowered considerably in the weeks and months to come.

Such a worst case scenario would lead to a marked mortality increase for adults of any age, as the following figures show (see methodology section below for how I obtained these):

Left: Average risk of dying in the next year for different age groups of the Georgian population — blue: in a normal year (data from 2018) — red: in 2020, assuming a worst case scenario for the Covid-19 pandemic.
Right: Increase (in %) of the risk of death in 2020 in a Covid-19 worst case scenario. The black error bars account for the uncertainity (95%-confidence interval) of the chinese fatality data.

As can be seen, the average risk of death for all adult age groups increases by about 100% compared to a normal year. (I excluded minors from these figures, since the associated uncertainties for that age group are too large.) If any distinction can be made at all, it would seem that adults in their twenties and sixties experience the highest increase in risk, while adults in their fourties and eighties face a more moderate increase. Given several sources of uncertainity, I would, however, not go as far as to claim that as a fact. (See discussion in the methodology notes.) The important lesson was aptly summarised in a blog post on the Brookings Institution website:

In China, more than 50 percent of COVID-19 deaths have been people who are over 70 years old even though most COVID-19 cases are being contracted by people below 70. Young people have interpreted this as an indication that they do not need to worry about the virus, while old people are being warned to take extreme caution. It is true that young people have a smaller chance than old people of dying from COVID-19, but they have a smaller chance of dying of almost any other health risk as well. [emphasis mine]

Katharina Fenz and Homi Kharas, A mortality perspective on COVID-19: Time, location, and age

This simple fact may have ethical implications for the handling of the Coronavirus pandemic in the medium run. Once drastic measures succeed at curbing the first wave of the disease, what is the way forward until a vaccine is available? Surely we do not want to uphold the current level of social distancing for over a year? One proposed solution with a growing number of advocates is to slowly normalise life for people of working age who would then take the brunt of the pandemic, while seniors keep social distancing. While this approach makes sense economically and would indeed reduce overall mortality by a lot (compared to the worst case scenario), it would also end up exposing the younger generations to a disproportionately increased risk of death when compared to older generations.

In reality, no country or region so far has come even close to the catastrophic scenario outlined above. Even in the densely populated city of Wuhan (11 Mio. inhabitants) where the novel Coronavirus originated from, only about 0.5% of inhabitants are estimated to have had Covid-19 by the end of the outbreak. If there is no second wave of infections, the Coronavirus could account for as little as 2% of all deaths in Wuhan this year. The worst case scenario simply reminds us why the social distancing measures are deadly serious. In Georgia alone they may help save 50’000 lives in 2020.

The eventual scale of the pandemic in Georgia remains highly unpredictable at this stage. However, no matter the eventual death toll, any increase in mortality will likely affect all age groups similarly, unless one age group is being shielded from the virus much more than others.


Methodology notes:

I use the following age-dependent quantities based on publicly available data:

CFR : Case fatality rate — the share of registered infections which result in death. Based on publicly available Chinese data. Differs for all age groups (average CFR = 2.3%).
ROD18 : Risk of death within one normal year for Georgians. Based on Georgian demographic data (Geostat) from 2018 (number of deaths divided by the average of the population sizes on 1/1/18 and 1/1/19). Differs for all age groups.

Our model (or ‘back-of-the-envelope calculation’, as mathematicians would call it) has the following parameters which are assumed to be independent of age:

I : Probability of contracting the virus within the next year. This is equal to the share of the population which will get infected within the next year.
NR : Share of infections among the chinese reference population which were not registered as such (e.g. asymptomatic cases).
FM : fatality multiplier — factor by which the overall case fatality rate in Georgia differs with respect to the the chinese reference population.
PD : Share of all preventable deaths attributed to Covid-19 (fatalities which would not occur were it not for the pandemic).

The risk of death in the next year due to Covid-19 is RODcov = I*(1-NR)*CFR*FM . The combined risk of dying from any cause in the next year is ROD18 + PD*RODcov . The relative increase in percentage points is thus PD*(RODcov/ROD18)*100% .

For the worst case “herd immunity scenario” presented above, I chose the following parameter values:

  • I = 0.6 (60%) : Complete herd immunity is reached when a fraction of about 1-1/R0 of the total population has reached immunity. Here, R0 is the average number of people an infected person transmits the virus to. For Covid-19, this is currently estimated to take a value of around 2.5, from which we derive I = 1-1/2.5 = 0.6 .
  • NR = 0.4 (40%) : This number is still a big unknown. Comprehensive testing is necessary to get a better understanding of how many people unknowingly carry the virus. For the passengers on the Diamond Princess cruise ship (a sample population which underwent comprehensive testing) the observed value was NR = 0.18, while some estimate the number for China to be around 0.5. A highly contested and not yet peer-reviewed article by researchers from the University of Oxford even speculated that it could be above 0.9. It is also plausible, that NR is in fact lower in older people than in the general population. I decided to stick with an age-independent number close to the current empirical estimates.
  • FM = 2 : In a worst case scenario with prolongued severe overload of the healthcare system, overall fatality is very likely to be equal or higher (FM > 1) than in China. There are currently many countries recording exorbitant fatality rates, including Italy (FM = 5.0), Spain (FM = 3.7), the Netherlands (FM = 3.2), France and the UK (FM = 2.8). Even though these numbers are highly unreliable at this stage and other countries report much lower fatality rates, I do find it justified to assume FM = 2 for a worst case scenario, which corresponds to an average CFR of 4.6% across all age groups.
  • PD = 0.9 (90%) : I assume that most deaths attributed to Covid-19 are preventable in the absence of the virus and would not occur anyway at a later point in the same year. This is certainly plausible for younger patients. In reality, PD should be somewhat lower for older patients, which we neglect.

An important observation should be made: no matter the eventual scale of the death toll (which is mostly governed by the choice of I), in our model the comparison between generations does not change. When in our worst case scenario people in their twenties experience twice as much risk increase than people in their eighties, this remains true in any other scenario as well, even if the overall death rates are much lower. The reason is, of course, that the model assumes the parameters I, ND and PD to be independent of age when in fact they are presumably not. If we were to adapt the model accordingly, lower ND for older people would result in somewhat elevated risk increase for seniors, while lower PD for elders has precisely the opposite effect. The dependency of I on age is more debatable and could go in both directions: on the one hand, seniors might self-isolate to a higher degree than the younger population, but on the other hand they might also be more likely to still attend mass on a Sunday morning. In the absence of drastic policy measures to differentiate between generations, the age-dependency is probably not significant enough to change our overall conclusion, namely that older people do not face a steeper increase in mortality risk.

For the calculations and source code visit: https://github.com/lo-hfk/covid-in-georgia


In the next post I will look at Georgia’s growth rate of confirmed cases and how it compares to other countries.

2 thoughts on “All age groups could face significant mortality increase in 2020: Georgian case

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.