Within a single month, Georgia has gone from international Covid model country to high risk zone. The government has adopted a more relaxed approach than in spring, relying on cititzen’s cooperation rather than draconian lockdowns. But mobility data indicates that the eastern half of the country has been slow to adapt to the return of the pandemic.
New confirmed cases have reached a record 1351 on Wednesday, while the death toll rose to 172. Already on Sunday, Department Head Marina Endeladze from the Tbilisi Infectious Disease Hospital called the situation „alarming“.
As of Wednesday, average incidence per 100‘000 inhabitants in two weeks stands at 308 cases for the whole country, 12 times the EU green zone limit. Top of the regional incidence list is still Adjara with 1002 cases, followed by Imereti with 455 and Tbilisi with 349 cases.
The surge in cases is happening against the backround of a dramatic rise of coronavirus cases in Europe. Many cities and regions, for instance in Spain, France, Belgium, Poland, the Czech Republic, the UK and the Netherlands, are again in partial or full lockdown. As the incidence map shows, most of Europe is currently far outside the self-defined green zone. The Czech Republic leads the pack with a national 14-day incidence of 974 cases per 100’000 inhabitants.
Central and Eastern European countries are not spared by this second wave. In March and April, most of them successfully averted the catastrophic outbreaks seen in some Western European countries by shutting the borders early and imposing draconian lockdowns. Yet this time around they are in the eye of the storm. Reports indicate that healthcare systems in many of these countries are now stretched to maximum capacity, running out of hospital beds, doctors, nurses and tests.
The Covid histories of many Central and Eastern European countries sound remarkably similar: a strong resolve to suppress the pandemic at all cost in spring, followed both by premature declarations of victory over the virus as well as growing public doubt that the economic cost of such an approach was worth the success.
Does Georgia fit into this pattern as well? According to Surveys conducted by the Caucasus Research Resource Centers (CRRC) in May and June a majority of participants continued to support the government‘s response, but also expressed their belief that the economic burden of the pandemic was worse than the virus itself.
Perhaps most crucially, however, only a minority of respondents stated that they expect a second outbreak of Covid-19 in Georgia, casting the government‘s expectation management in a bad light.
Against the backdrop of these expectations, and with elections ahead on 31 October, the government was reluctant to clamp down hard when the pandemic hit again in September. Lockdowns were immediately ruled out, although the Prime Minister‘s latest remarks indicate he ultimately sees it in the hands of citizens to avoid such a scenario.
The government has thus far intervened with selective measures to slow the spread of the virus. Certain types of large gatherings have been banned nation-wide since 10 September, while restaurants, bars and other entertainment facilities are facing restricted opening hours in virus hotspots. Public transport in Adjara was suspended for three weeks, and various restrictions apply to the educational sector. But just how effective are those measures?
One possible indicator is cititzen‘s mobility as measured for instance by facebook‘s granular movement range data. Here, facebook‘s data scientists divide the Earth‘s surface in tiles of roughly 500×500 meters, and they track the number of such squares that an individual smartphone appears in on any given day. The data is made public in an anonymised form and aggregated over all users by municipality.
Since the outbreak began at the end of August, average movement range in Batumi has dropped by about 25% to below typical levels in winter. To a varying degree, a decrease in mobility can also be observed in most other places in western Georgia.
Some of the effect could be due to the ban on large gatherings after 10 September, although the premature end of the tourism season and voluntary behavioural changes might have contributed as well.
In Batumi, the effect of public transport suspension and restrictions on restaurants and bars can be seen in the mobility data. Their combined impact on people‘s movement range seems to have been relatively minor. At the same time, there was a noticeable stagnation of new cases in Adjara around one week after the measures were implemented. That should, however, be taken with a grain of salt due to a change in testing policy after 2 October.
Contrary to the epicenters of the outbreak in the west, localities in eastern Georgia have only experienced a very small change in mobility in the wake of the gatherings ban. In early September, one can even spot a temporary increase, perhaps associated with people returning from holidays in western Georgia. Unsurprisingly, in October the virus has spread at a faster rate in Tbilisi than in Batumi. It is too early to assess the effect of the latest restrictions imposed in the capital and in Imereti on Friday.
In the meantime, worrying reports about bottlenecks in the health-care system have been mounting. Hospital beds are becoming a scarce resource not only in Batumi but also in Tbilisi, and testing is running at what seems to be the maximum sustainable capacity of around 10‘000 PCR tests a day. Presumably as a consequence, the National Center for Disease Control announced a change of testing policy on 2 October, according to which contacts of confirmed cases are only tested once they show symptoms. This likely resulted in a deceptive slump in case numbers during the following week.
PCR testing should be an area of major concern. The test positivity rate is now over 10%. To put this into context, the EU green zone requirement is 3%. International experience in spring showed that countries with a positivity rate much above 10% tended to have uncontrolled outbreaks where contact tracing was virtually impossible.
Test positivity in most Central and Eastern European countries is currently between 13% and 25%, and contact tracing in several of them has already collapsed or never been fully implemented. As another point of reference, incidence in Switzerland is now at a similar level as in Georgia, and the test positivity rate has just surpassed 10%. Reports from many Swiss cantons indicate that the contact tracing system is severely strained and failing in some regions, with shortages both of staff as well as tests.
It is hard to assess from published information how well the contact tracing system is still working in Georgia at this point in time. However, if cases do not stop growing very soon, the system will inevitably reach a breaking point.
Despite the critical situation, the Georgian parliamentary elections are set to go ahead as planned on 31 October. The government is performing a delicate balancing act. The governing Georgian Dream party is favoured to win the elections, and the pandemic is thought by many to play rather to their advantage. In this context, any additional coronavirus restrictions would risk alienating voters besides derailing the economy. Yet, a failure to get the spread under control in the next few days could make another lockdown inevitable despite all the promises. Georgia would not be the first country to backtrack on that.
1) Incidence data for Georgia: The incidence data for October is based on the NCDC’s regional case numbers presented in the daily press briefings by the Tbilisi Hospital of Infectious Diseases. For July and April, I rely to a large extent on the NCDC Covid reports from May, June, July and early October. However, since no regional time series are presented there, some guesswork is needed in order to determine with high confidence in which color catagory some of regions fell at the time. A fairly consistent picture emerges by combining the NCDC’s total incidence figures published at various points in time, their figures on region of residency of Covid infected (which tend to overestimate incidence in rural regions), retrospective reports on clusters in highly affected municipalities, as well as local media reports at the time.
2) Mobility data: A detailed description of how facebook computes daily movement range in a municipality can be found here. My categories „West“ and „East“ correspond to an unweighted average of all municipalities in Western and Eastern Georgia for which data was continuously available over the whole period since February. I did not assign Samtskhe-Javakheti to either side, and for Racha-Lechkhumi-Kvemo Svaneti there was no data. The municipalities included are the following:
East: Akhmeta, Bolnisi, Dedoplis Tskaro, Dusheti, Gardabani, Gori, Gurjaani, Kareli, Kaspi, Khashuri, Khobi, Kvareli, Lagodekhi, Marneuli, Mtskheta, Sagarejo, Signagi, Tbilisi, Telavi, Tetri Tskaro.
West: Bagdati, Batumi, Chiatura, Khoni, Kobuleti, Kutaisi, Lanchkhuti, Martvili, Ozurgeti, Sachkhere, Samtredia, Senaki, Terjola, Tkibuli, Tsalenjikha, Tskaltubo, Zestaponi, Zugdidi.
After averaging over municipalities, I smoothed the time series with a Savitzky-Golay filter of polynomial order 1 and window size 7 days. The individual time series of Batumi and Tbilisi shown in the figure have also been smoothed in this way.