COVAX Facility

In response to the threat of vaccine nationalism, a coalition of international partners set up the COVAX Facility in April/May of last year to ensure that all countries–not just the rich ones–would have equal access to vaccines once they become available. Lead by the WHO, GAVI and CEPI among others, COVAX serves as a global purchasing pool that mixes more wealthy “self-financing participant” countries with lower-income publicly funded countries and promises all participants will have access to any the 18 different vaccine candidates being developed in by the pool of pharmaceutical companies in which COVAX has brokered Advanced Market Commitments (AMC). Given that viable vaccines are already on the market as of late-2020, and that more will emerge in the coming months, the big challenge for the COVAX facility is to ensure equitable access and fair allocation across countries.

In this Part I. blog post, I provide an overview of how the current allocation mechanism works and examine the data from the recently released COVAX distribution plan for the beginning of 2021 (Phase 1). In a follow-up blog post, I examine some commonly mentioned risk factors and country characteristics associated with COVID-19 and some of their implications when used as weighting adjustments for Phase 2 of the COVAX allocation mechanism.

Data

The data from this analysis comes primarily from two sources: the public documentation on GAVI’s COVAX Facility website and the latest consolidated country-level COVID data from Our World in Data (OWID). The original documents used for this analysis, along with the extracted data, are saved on my Github repo for ease of use. Additionally, I group countries for analysis using their most recent World Bank Country and Lending groups classifications rather than their WHO regional office groupings.

## 
##               AMC eligible                  Committed 
##                         92                         38 
##           Intent submitted Self-financing participant 
##                          6                         52

Countries with bilateral agreements for vaccines

According to the preliminary technical design for the COVAX allocation mechanism, the Facility will request that countries that have successfully concluded bilateral deals with manufacturers to cover a sizable share of their population (e.g. 20%) delay their receipt of COVAX distributed vaccines until other countries receive enough to cover their highest priority populations. While I do not have data on hand for all bilateral deals from all countries, the OWID data includes the total number of vaccinations already administered for each country in advance of COVAX distribution. These countries are indicated in the map above by their ISO abbreviation code. At present, 60 COVAX participant countries have already received vaccines from channels outside of the Facility. The number of total vaccinations already administered for these countries varies widely, as can be seen in the summary statistics below:

summary(dat_map$total_vaccinations)
##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max.     NA's 
##       30    49464   305371  1486803   746729 31200000      128

Phase 1 thresholds

In the plot above, I sum the populations of the current COVAX participant countries based on the UN’s 2019 Revision of World Population Prospects, and show the coverage needs for the different thresholds, up to 20%. Assuming vaccine needs are estimated as single doses to match population numbers one-to-one, this very simple calculation for 20% coverage for all participants comes to 1.45 billion doses, and 784 million doses for just the 92 AMC countries–both far under GAVI’s estimates of 2 billion and 1.3 billion, respectively. It should be emphasized that my calculations do not include the “8 economies (who are not United Nations Member States)”, nor the emergency buffer, though given the size of the difference in these calculations, it seems unlikely that this would be the only reason for the discrepancies.

Given that there will unlikely be enough supply to cover all participants, I also estimate the total dosage needs for different participation scenarios (e.g. only committed-countries receiving vaccines, only AMC-countries, countries with no other bilateral agreements, etc.). These alternate scenarios are meant to serve as a reference, as participants receiving allocations–especially early on–will likely be a mix of these different groups of countries and/or be included/not-included for reasons not mentioned here.

total_interim_doses <- 320064400

#From the total interim doses, I subtract the allocation for Non-UN Member states, and then divide by two to account for double dose regimen
((total_interim_doses - 1303200)/2) / sum(dat_dist$population2019) *100
## [1] 3.299806

dat_dist <- dat_dist %>% mutate(
  doses_pct = (doses_interim/2)/population2019 * 100
  )

summary(dat_dist$doses_pct)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   2.238   2.520   3.854   5.867   3.954  33.229

Interim allocation population coverage

We can see from the data that the coverage rate provided by the interim allocation forecast is not constant across all countries. In this section, I plot these coverage rates, grouped by region, to have a better understanding of the differences.

In the figure above, I plot the total deaths per million vs the share of the population covered by the interim allocation, broken up by region. The size of the points are scaled by the country’s population.

It’s clear from looking at the plot that the majority of countries have allocations that cover 2-5% of their populations, and there does not seem to be any relationship between the severity of the mortality rates in-country with the size of the allocation they received. In terms of the countries receiving significantly larger relative shares of the vaccine, by filtering the data for all countries with coverage above 5%, we learn that all of these cases pertain to small countries with populations of less than one million.

After we drop all countries with populations of over one million people, we can see a very clear negative linear relationship between population size and vaccine coverage from the COVAX allocations, as illustrated in the plot above. As we can also see, this relationship holds across all regions. Most likely, there are practicalities related to shipping or other logistics that makes it impractical to send small countries vaccines 2-3% proportional to their size.

In fact, when we look at just the countries in the interim allocation forecast with populations of over one million, we find that the mean coverage for all countries is 3.2%.

summary(dat_dist[dat_dist$population > 1000000, ]$doses_pct)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##   2.238   2.510   3.520   3.224   3.872   4.655       6

Countries with no interim allocations

From the 188 countries we have on our current COVAX participants list, 48 countries (26%) were not allocated any vaccines in this initial allocation plan. According to the forecast document, these countries “either exercised their rights to opt-out, have not submitted vaccine requests, or have not yet been allocated doses.” The list of participating countries with no initial allocations can be viewed in the table below.

The “Team Europe” EU27+ group is on this list, along with most other high-income and upper-middle income countries. There are only 4 low-income/AMC-eligible countries in this list–Burundi, Eritrea, Madagascar and Tanzania–and one would imagine their allocations are pending.

In addition, 36 of the 60 countries (60%) who have already begun COVID vaccinations in advance of COVAX distribution are included in this list of those who have opted out or are otherwise not receiving initial COVAX allocations.

Among the other countries who already have bilateral with vaccine manufacturers, it is interesting to see that New Zealand, Chile and Canada have chosen not to withdraw their names from this initial round of distribution despite having already secured enough doses outside of the COVAX platform to vaccinate their populations 2 to 5 times over (see EIU report figure above).

While COVAX was able to meet its 3% coverage target with this initial allocation plan, this was also done with 26% of COVAX participants not receiving their allocations in this round. Referencing the COVAX Phase 1 allocation needs calculations from above, current supplies put the Facility on the orange dotted line (“All (no bilats)”), far under the total dosages needed to cover all participating countries. It remains to be seen whether enough vaccines will be secured over the remainder of the year to move back to the blue line, where enough supply is on hand to distribute to all participants.