After weeks of talk about reaching the „peak“ of the Covid-19 epidemic, Georgia has started to relax restrictions on public life. Has the country conquered the epidemic? What can be expected from the months to come?
Easter celebrations have passed without a spike in new cases, and last week‘s lifting of the car travel ban was the government‘s first step in their 3-months plan of returning to normalcy. However, a certain fatigue with lockdown measures is palpable, and recent farmer‘s protests show the economic costs of the emergency measures are increasingly hard to bear for many Georgians.
Time to take a closer look at the current epidemic situation. Here are the figures of confirmed cases from the past 2.5 months, as reported by the National Center for Disease Control (NCDC):
The cumulative number of cases from the first recorded internal transition (27 March) until Easter Sunday was closely following an exponential trendline with a doubling time of 10.5 days. Had the country stayed on this trajectory, the cumulative number of cases would have already surpassed 1400 to date, and we would now be seeing 50 to 100 new cases a day.
Instead, as the bar chart at the bottom shows, daily new confirmed cases have stagnated since easter (19 April), well before they could reach a critical number. In terms of this metric, Georgia might indeed have surpassed the peak two weeks ago.
From the point of view of the healthcare system an equally important metric is the number of currently infected patients (red area in the upper chart). That number has only reached a peak earlier this week since recoveries naturally lag some time behind new infections.
Despite these promising figures, a decline of case numbers in May is not a foregone conclusion, as a quick look at the international situation shows: several countries have experienced a second, stronger wave of infections after a brief lull. These include Singapore and Japan, while Georgia‘s neighbour Armenia is currently showing signs to follow such a scenario as well.
Georgia‘s trajectory of confirmed cases has visibly flattened, and it is now only slightly steeper than many of the countries who are starting to emerge from lockdown: Germany, Norway or my native country Switzerland, for example. At the same time, it is markedly flatter than the curve of Sweden, which has tried to manage the situation using less strict measures than most other European countries.
Let‘s also look at daily new confirmed cases, where Georgia‘s curve is now showing first signs of a decline:
Some countries (Italy, Germany, Switzerland, Turkey, Estonia) are seeing a long and slow decline in new cases, while others are still in the growth phase of the epidemic (Brazil, India, Russia). Some countries have brought the spread under control very quickly (South Korea, Taiwan, to some extent China), while others are experiencing a long and drawn out peak (the US and Sweden, for example).
It should be emphasised that all these charts rely on reported cases. It has become increasingly clear that in many countries these are only the tip of the iceberg. Hence, the number of reported cases should not be mistaken for the true number of infections. However, as long as there is no drastic change in reporting policy or a sudden saturation of testing capacities, the trajectory of a country adequately represents the general trend of its epidemic.
I will look at the issues of underreporting and testing in a separate post.
What now? The era of R-engineering
If the number of new infections in Georgia indeed starts to decline soon, how long will cititzens and businesses have to remain patient until they can start to transition back to their usual mode of life?
There are two possible scenarios:
1) The government decides to attempt a complete elimination of the disease within the country‘s borders. This requires the continuation of tough measures until the number of new cases is practically down to zero. If successful, life could go back to normal while international travel is minimised and extremely stringent border controls are put in place in order to avoid re-importing the virus.
2) The epidemic is suppressed until it reaches a level which is permanently manageable (to be determined by the health care specialists). Subsequently, a balance between normal life and social distancing has to be found which allows for the epidemic to continue at this level without growing.
In most of East Asia as well as New Zealand, the first option seems to be favoured by governments. In much of Europe and America, however, the sheer size of the epidemics make such a strategy very time-consuming and costly. Many health experts advocate for the first option, while proponents of business have been pushing heavily for the second approach.
For example, the German government opted on April 15, under mounting public pressure, for a partial relaxation of restrictions and thus for the second strategy. In doing so, it went against the express advice of the Helmholtz research association, the largest scientific organisation of the country.
Next door to Georgia, the Armenian government has recently announced it is shifting from an eradication strategy to what seems to be a containment strategy.
If the second strategy is to prevail in Georgia as well, there will soon be a lot more talk about the reproduction number R. This is the average number of people an infected person infects.
For the novel coronavirus, R was found to be around 3 in most countries before the introduction of social distancing measures. In other words, a person with Covid-19 would typically pass on the virus to 3 other people before they recovered.
The aim of the lockdown measures was to push R below 1. If R is smaller than 1, the virus does not find enough hosts to reproduce, and the epidemic eventually fizzles out. A decline in new cases would seem to indicate that R has fallen below 1.
If R is equal to 1, the epidemic neither grows nor subsides. Strategy 2 thus amounts to keeping R very close to 1. This requires a fine-tuning of measures which will be an enormous challenge for governments in the months to come.
According to research at the University of Hong Kong, the extremely strict lockdown conditions in China have achieved minimal R-values between almost 0 and 0.4 in most cities. In contrast, the lockdowns in most European countries have brought down R to values between 0.5 and 1. This suggests that there may be little wiggle room to relax restrictions without a backlash, and the situation will need to be constantly monitored.
Specifically with respect to Georgia, this raises several questions: First, what value of R has been achieved during the lockdown? Secondly, which of the government‘s many measures were most effective in bringing down the transmission rate, and which ones may have been marginal or even obsolete?
Comparison of different countries (for example South Korea versus China) shows that vastly different approaches can lead to similar overall outcomes. Clearly, it would be beneficial to determine a set of minimally invasive measures which are just strong enough to keep R around 1. Much of cititzens‘ personal freedom and economic prosperity over the course of this year (and perhaps beyond) may depend on how well authorities and the scientific community worldwide manage to tackle this problem.
It remains to be seen how smoothly Georgia can transition from the current heavy-handed but successful approach to a more subtle strategy with less side effects. The peak of infections may have passed already, but the toughest challenge facing the country may still lie ahead.
A call for open data
In some countries, updates on the development of R have already become a regular part of coronavirus news. In Germany, statisticians of the Robert Koch Institute for infectious diseases publish an estimate every day, and there are at least two other research institutes in the country which do the same.
In contrast, based on currently public data it is virtually impossible to get an accurate estimate of how R has developed in Georgia during the state of emergency.
A relatively small sample size, such as the 600 Georgian Covid-19 patients, is not necessarily an impediment to a reasonably precise calculation of how R developed over time. Statisticians at ETH Zürich have computed R for the Swiss regions, many of which have sample sizes comparable to the Georgian one. The same is true for the research on Chinese cities mentioned above.
The obstacle is data quality rather than quantity. The statistical procedures to compute R accurately require data which include, among others, detailed information about the time spans between onset of symptoms and reporting, and (if known) the likely date of infection. Data about epidemiological links between different patients have proven to be particularly useful in the Hong Kong study.
The low number of cases in Georgia likely means that the NCDC is safeguarding one of the most detailed and complete collections of Covid-19 patient-level data in the world, including many precise epidemiological links between cases.
While recognising the need to protect sensitive patient data, it seems worth for the NCDC to consider opening an anonymised and redacted version of their patient database to the public. In the age of open data, and especially when it comes to a global issue such as the coronavirus, there are hundreds of research institutes and thousands of capable data scientists and statisticians who are motivated to take on the challenge and help us all gain the most insight from the data that already exists.
In my next post, I am going to look at Georgia’s performance when it comes to PCR testing.