GSAC Election Analysis
Analysis of COVID-19 in Georgia
Photo byCovid-19 | Radoslav Zilinsky

Analysis of COVID-19 in Georgia

By Tamaz Khunjua | Edited by Veronika Malinboym #COVID-19

Since mid-August, Georgia has been faced with a second wave of COVID-19. Recently, the mark of 500 cases was breaches, which, in comparison with the statistic of this spring, is, indeed, a very tangible figure. However, the Georgian government says that the situation is under control. With the Parliamentary Elections coming up, there is a possibility that the government artificially mitigates the situation in order to avoid loosing in its rating. We decided to examine in detail what is actually happening and make our own conclusions. Also, we are launching a new, updated section with an independent statistics which will be updated on daily basis.

For the purpose of this comprehensive analysis, it is necessary to examine all the statistics from the first day of the epidemic, as well as assess the measures that the government has taken and continues to implement right now. In this analysis we will use the basic epidemiological parameter R0, based on the SEIR model as well as the statistical data. The method of assessing explained at the end of the article.

Georgia 1 wave (7.03 - 25.04)

During the first wave, R0 did not exceed 1.5. The government imposed a strict quarantine, which allowed it to be reduced below 1. Also, today we are well aware that COVID-19, like many other viruses, has a seasonal character. This property also played a positive role in stopping the spread of the disease. In a more nuanced analysis of this period, the impact of tougher temporary quarantine measures, such as curfews and restrictions on private and public transport, can also be seen. In retrospect, it is difficult to determine which of the measures has had the greatest impact on the suspension of the epidemic. However, we can say that in general, by the beginning of summer, the spread was stopped, which is clearly shown in the graph.


Georgia 2 wave (15.09 - today)

Today we find ourselves in the midst of the second wave of the disease. At the same time, according to the parameters, it is significantly more serious than the first one. Moreover, this time have little hope for a seasonal decrease in the activity of the virus. It is also surprising that, unlike the first wave, the Government does not recognize the gravity of the situation and has not yet imposed any quarantine measures. At the same time, over the past few weeks the value of R0 has reached a total of more than 2, which, according to the international regulations is characterized as epidemical.



Georgia (General Chart)

On the graph with the overall number of diseases, the difference in the scale of the two waves of epidemic can be seen with the naked eye.


The burden on the health system

The following graph shows the curve of the number of patients that are grouped as infected. It is evident that during the first wave the simultaneous number of cases did not exceed 600 people, whereas at the moment there are more than 4 000. According to the Government, they had prepared 3 000 hospital beds. However, it is not entirely clear what exactly do they mean by that? Are all of them equipped with mechanical ventilation? How many severe patients can the system take?



Testing volumes are one of the main questions that arises in terms of the government’s response to the second wave. At the moment, the ratio of the tests to the positive is at the level of 6-8. Sometimes the number of tests performed is below 4 000, while the appropriate number of tests would have been 20 000 – 25 000. The failure to meet this number would ultimately result in the so-called super-spread of the disease, and postponed treatment of the infected. Similarly, there are some questions about the volumes of testing in summer, as only 10 000 tests per a positive case were made.

The Government claims that it adheres to the protocol of tracking all contacts of the sick and testing them. However, the figures suggest otherwise. Public testing points are being practiced in many countries with this epidemiological load. The other day, the government reported a stockpile of 200 000 tests, however, with the proper volumes of testing, this number is extremely low. On top of that, a number of sources claim that it takes 3-4 days to get the test results, which is also quite dangerous given the overall situation in the country. Below, we present you the official statistics issued by Ministry of Health of Georgia.


centered image

Mortality rate

One of the most significant parameters in assessing the pandemic in the country is the mortality rate. This is a big challenge for the government, and for society it is an indicator of how well the system was prepared during the summer lull. We can already make some preliminary assessments. We know from the world's statistics that death from COVID-19, on average, occurs 14 days after the infection. At the time when the second wave hit, the number of patients in critical condition was small and the health system was doing well. Mortality was also very low.

In order to assess the mortality rates, a shift in dates must be taken into account (Kaplan – Meier analysis). Indeed, as of right now, the patients who die were, on average, tested 13 – 15 days ago. To average out the results, we divided the entire period into groups of five days and made a relative shift of three groups. The following table reflects the results of the calculations:

It is clear that the mortality rate right now is significantly higher than that witnessed in summer. This suggests that the system has already started to work to the limit of its capabilities. It is also clear that the three coming groups are significantly larger than the previous ones.


  1. According to R0, Georgia is in the growing phase of the epidemic. During the last week its value was 2.1 with a measurement error of 0.2. During the first wave, when the R0 was around 1.5, the government carried out very strict quarantine measures to stop the spread of the disease. So far, only partial measures have been introduced in the Ajar region.
  2. The lack of quarantine in the Ajara region is a matter of serious concern. To date, 80% of confirmed cases are confirmed there. It is obvious that without the introduction of testing of those leaving the region, the virus will spread to a much more densely populated Tbilisi.
  3. With today's number of diseases reaching 500 people, the number of tests to exit the "red" zone should be around 20-25 000. Being in the red zone ultimately means dangerously uncontrollable outbreaks and the occurrence of super-spreaders as well as the increased mortality due to late diagnoses.
  4. The government also needs to speed up the test vetting process. If the information that in some cases the tests are tested for 3-4 days is correct, then this creates an extremely dangerous situation.
  5. There is not enough statistics to properly estimate the mortality rates. However, the picture that exists today shows that this mortality is higher than that in the period prior to the hit of the new wave.
R0 analysis method

Epidemics are successfully described by empirical models. The very first and simplest of them is called SIR-model and was proposed in 1927 by Kermack and McKendrick. In this model, the population is divided into 3 separate groups - susceptible, infectious and removed. In the original state, the entire population is susceptible (except for the first infected person) and then, over the time, individuals begin to move into the infectious group, and after recovery in the removed group. This dynamic is well described by the differential equations that depend on the time. On top of the main variables in such equations, there is a number of parameters, that are responsible for the speed of getting infected and the recovery from the disease. Such variables are well known in studying the diseases. Moreover, such parameters can be combined in a way that will produce a single R0 number (reproductive number). Put simply, R0 is the number of people, that, on average get infected by the single infected person throughout the entire period. If this number is below 1, then the disease in the goven population is degrading, and, if it higher than 1 – the disease is growing. The higher is the number – the faster is the spread of disease.

The SIR model has many generalizations for different groups. In the case of COVID-19, the most important effect to consider is the asymptomatic period, i.e. the period in which a person is already sick but has no visible symptoms. This model is called SEIR-model, where E-exposed refers to a group of people in a latent phase. The existence of this group, in which the infected person is located for 5-7 days is what makes COVID-19 so dangerous.

On the one hand, empirical models offer an excellent opportunity to describe the epidemic, but on the other hand they do not take into account the impact of quarantine measures and restriction of mobility. That is, within this model, the population is seen as a closed system with unlimited contact capability. However, in reality, even without quarantine measures, people mostly live in small groups (clusters) and their mobility is naturally limited. Additional restriction of population mobility is the main method of combating the epidemic.

Thus, for real calculation of the R0 parameter, one should use a mixed approach, namely based on both the seironic model and real statistics. The solution of the SEIR model at the time of the increase of the disease is the exhibitor. That is, based on statistics, it is necessary to evaluate the indicator of exhibitors and known parameters of the disease that can be evaluated, as well as the R0. This approach is very much detailed in the work [1]. It has the following kind of formula for R0 in the SEIR model:


centered image

where gamma is the recovery rate, sigma is the rate that latent individuals leaving the class, and lambda is an exponential factor. The first two parameters are the reverse values of the average number of days that individuals spend in the appropriate class. These parameters are well studied, although today they have a large margin of error. We'll use the parameters from the summary table of the article [2] (γ = 1/6 +/- 0.03, σ = 1/6-0.03). The λ parameter we get from the fitting of statistics using the following function:


centered image

Source code on GitHub

  1. Estimating epidemic exponential growth rate and basic reproduction number. Junling Ma. Infectious Disease Modelling. Volume 5, 2020, Pages 129-141.
  2. A Systematic Review of COVID-19 Epidemiology Based on Current Evidence. Minah Park, Alex R Cook et. al. J Clin Med. March 2020. Volume 9(4).
  3. Serial interval of novel coronavirus (COVID-19) infections. Hiroshi Nishiura, Natalie M. Linton, Andrei R. Akhmetzhanov. International Journal of Infectious Diseases. April 2020. Volume 93, Pages 284-286.

Subscribe to receive our latest analyses weekly.