Behind the scenes of

What will it take to build Africa’s vaccine industry?

This notebook describes the methodology and data used for our data dive on what it will take to build Africa's vaccine industry.

You can see our data processes in our Vaccine Manufacturing GitHub repository.

We hold transparency and reproducibility central in all our data work. However, for this project we are unable to share the raw vaccine supply and demand datasets, as per the arrangements through which we received the data from WHO and Linksbridge.

Despite this limitation, here we go through each data source and explain where we got the data, how we use it, and why we use it that way.

The analysis in this Data Dive can be divided into the following four subjects / data sources: - Vaccine Supply: WHO MI4A Dataset (Provided by WHO) - Vaccine Demand: Linksbridge's Vaccine Almanac - Manufacturing Facility Commitments: PATH/CHAI - Gavi-related Statistics: Gavi

Vaccine Supply: Complemented WHO MI4A dataset

For vaccine supply (or production) data, we use the complemented version of the WHO Market Information for Access to Vaccines (MI4A) dataset.

The MI4A describes the total volume of vaccines distributed by manufacturers worldwide. It contains information on where these vaccines are distributed, the headquarter location of the manufacturer, price of vaccines, etc.

The publicly available dataset contains data provided by participating countries that have agreed to share vaccine price and procurement data. This dataset is incomplete and yields different results to WHO's annual Global Vaccine Market Report.

We reached out to the WHO to understand the differences.

WHO complements the publicly available dataset, reported by participating governments, with additional sources, including WHO's dataset on the global vaccine markets (which includes self-reported data from manufacturers themselves).

WHO kindly offered to share a view of this data with us. Given privacy considerations (including relationships with self-reporting private companies), this dataset cannot be shared publicly.

Data notes

Information contained in the MI4A database is provided by participating countries that have agreed to share vaccine price and procurement data. The data used in this analysis are complemented with additional sources, including WHO's dataset on the global vaccine markets.
The classification by continent is performed using the OWID convention. The WHO uses a different classification based on six world regions. The choice to use the classification by continent is the sole responsibility of the author (The ONE Campaign) and does not imply the expression of any opinion whatsoever on the part of the WHO concerning the legal status of any country, territory, city or area of its authorities.
Values are for 2022 reported in 2023. Data includes COVID-19 vaccines.

The data we received was structured as follows: - Number of vaccine doses distributed to World and to Africa by WHO Region and Continent - Figures including and excluding COVID-19 vaccines - Data for 2022

This data led to some important decisions.

We include COVID-19 vaccines in all supply/production analysis: the DataDive focusses on the productive capacity of Africa in relation to the rest of the world. By excluding COVID-19, we would omit a large portion of continents' vaccine production capacity, especially in 2022.

Our analysis is completed by Continent: The Partnerships for African Vaccine Manufacturing (PAVM) target focuses on Africa as a whole. It makes sense for our analysis to follow Our World in Data's definition of Africa instead of the WHO region which excludes North Africa.

This data is used in two data visualisations:

Current versus target production chart

This chart shows the scale of increase required for Africa's current production capacity to be able to cover 60% of their population by 2040.

The Africa Union and Africa CDC estimate that 60% of African vaccines will be around 1.5 billion vaccines in 2040.

From the MI4A data, we know Africa distributed 20.7 million vaccines to the world in 2022. Africa needs to increase this production capacity 73-fold to reach the 1.5 billion target.

Africa produces a fraction of the world's vaccines (bar chart)

The bar chart shows how Africa's production capacity compares to other continents'.

Each bar represents the continent's total number of vaccine doses distributed to the world (including COVID-19 vaccines), as a share of total vaccine doses distributed to the world in 2022.

Vaccine Demand: Linksbridge's Vaccine Almanac

For vaccine demand data, we use Linksbridge's Vaccine Almanac (VA). This data is not publicly available, so we cannot share the raw vaccine demand data.

The VA showcases market information across multiple vaccine markets. You can see an outline of the data available in the latest Vaccine Almanac report.

Data for the VA comes from the Global Vaccine Market Model (GVMM), a data-sharing collaboration supported by the Gates Foundation’s Global Delivery Programs (GDP) team using information from the foundation, CDC, CHAI, PAHO, PATH, UNICEF and WHO that is curated by Linksbridge SPC. Not all of the data available in the GVMM appears in the Vaccine Almanac.

The VA includes three vaccine demand data tables: "Total Required Supply", "Total Required Supply by Country", and "Childhood Immunisation Schedule".

Vaccine demand is estimated using a series of factors, including target population, immunisation schedule, new vaccine introduction, number of doses per vaccine, etc. You can see more detail in the 'Use of GVMM Data' slide deck (slide 9).

For our analysis, we use the "Total Required Supply by Country" data table. This has the number of vaccine doses required for each country, each year, by vaccine name for 2000 to 2030. It excludes COVID-19 data.

Linksbridge separates COVID-19 data into another dataset. They informed us that they are currently refining their COVID-19 forecast, and are planning on publishing COVID-19 demand data in Q3 2024.

As demand for COVID-19 becomes more clear (Q3, 2024), we will update our analysis.

It is unclear when data represents recorded or forecasted data. The Vaccine Almanac Report shows "a forecast of global public demand for vaccines for 2022 through 2033", suggesting data is projected from 2022. Based on exchanges with Linksbridge about the start date of forecasts, we have set all data points beyond and including 2024 as projections. Some projections could start from 2023, however.

We used this data to produce one visualisation:

The area chart shows Africa's increasing share of global vaccine demand.

For a more detailed understanding of the data processes behind this graph, please refer to the vaccine_demand.py file. In short, we: - Assigned individual countries to their respective continents, as per OWID's classification. We counted "global stockpile" data as its own "global" continent. Note: India's data is dissagregated by region in the format 'India: Region'. There is also entries for 'India' as a whole. After inspecting the data, the India entries appear to be for the centralised distribution of certain vaccines, while the regional entries contain a seperate set of vaccines. As such, we count all 'India' and 'India: Region' rows as India. - Aggregated data by continent and year. - Calculated Africa's share of total vaccine doses, by year, including global stockpile.

Manufacturing Facility Commitments: PATH/CHAI

As commonly reported, 30 vaccine manufacturing facility projects have been announced in Africa. This figure was meant to provide information on the location of the facilities, the manufacturer, and an update on the status of the project. The primary aim was to show the inconsistent progress of projects.

However, getting the most up to date data on the location and status of these projects is notoriously difficult to find.

Depending on the source, you can get different locations (see Gavi, figure 4, versus Gavi, Slide 3).

And desk-based research into updates of each projects had various degrees of success.

We reached out to PATH and CHAI, who have partnered to produce detailed research on these projects. They provided us with their most up-to-date list of manufacturing commitments, containing data on the country location of the project, the manufacturer, and the current status of the project.

This data is extremely helpful. We have utilised the country location and the manufacturing name in our data visualisation. However, the status of each project can quickly become outdated. PATH/CHAI are currently updating their research and plan to publish in upcoming weeks. While we wait for this updated status data, we have opted to include additional information for specific 'case studies' for the manufacturing commitments where we are most confident our research is accurate.

We will update our data visualisation with the new PATH/CHAI data once it is released.

Gavi-related Statistics: Gavi

"Thirty-nine African countries are currently supported by Gavi, the Vaccine Alliance, which accelerates access to new and under-used vaccines and improves childhood immunisation coverage. Six of those countries — São Tomé and Príncipe, Nigeria, Kenya, Ghana, Djibouti, and Côte d'Ivoire — are projected to transition out of Gavi support by 2030. These countries represented 18% of vaccine doses fully supported by Gavi funding in 2023"

We calculated this statistic using the 'Gavi Shipments 2023 Vaccines - All Regions' dataset.

For a more detailed understanding of the data processes behind this stat, please refer to the gavi_supply.py file. In short, we: - Use Python to scrape the PDF into a pandas DataFrame. - Remove non-vaccine rows from the dataset. This includes "AD-Syringe, 0.5 ml", "AD-Syringe, 0.1 ml", "RUP-2.0 ml", "RUP-5.0 ml", "Safety Box, 5 Litre" - Remove Co-financing rows. We focus only on vaccines fully funded by Gavi. - Aggregate by country. - From this data subset, we divide each countries total number of vaccine doses fully financed by Gavi by the global total. - Sum the shares of São Tomé and Príncipe, Nigeria, Kenya, Ghana, Djibouti, and Côte d'Ivoire. These are the six African countries are projected to transition away from Gavi support by 2030, as per Gavi, figure 9, p.28.