Idea: Eva Lopez
Research, data analysis and visualization, writing: Gianna-Carina Grün
Editing: Fabian Schmidt, Zulfikar Abbany
You can read the story in English, Deutsch, Español, Português do Brasil, Indonesia
This analysis is based on two different data sources that are structured differently and thus used to answer different questions as part of this analysis.
In the dataset provided by the WHO, each row corresponds to one developer and their vaccine candidate. In this analysis, this dataset is used to
- show how many vaccine candidates are in which phase
- classify the vaccine candidates according to their type, and subsequently also phase
In the "Clinical trials database" compiled by the London School of Hygiene and Tropical Medicine (LSHTM), each row corresponds to a registered clinical trial -- accordingly, a developer team testing their vaccine in multiple phases in parallel will see multiple entries in this dataset. In this analysis, this dataset is used to
- show trial sizes by phase by company
- provide estimates (min, max, median) of the clinical phases
Other media companies too have implemented vaccine development trackers (NYTimes, Guardian, Milken Institute, Zeit Online).
You will see that numbers differ, based on
- the sources used and
- whether trials or vaccine candidates are counted and
- based on the distinction between "in approval" and "approved"; we chose to not show the process, but only signify a vaccine as "approved" once the major authorities have done so
There are five phases of clinical trials: Phase I, Phase I/II, Phase II, Phase II/III, Phase III.
If a vaccine is in a dual phase like Phase I/II or in Phase II/III it is tested there simultaneously, but for analysis assigned to the higher phase: a vaccine candidate in Phase I/IIis assigned to Phase II, whereas a vaccine candidate tested in Phase II/III it is assigned to Phase III.
Beyond these dual phases, vaccine candidates can be in different clinical trial phases at the same time (e.g. Phase I and Phase III) with different trial parameters (age, pre-existing conditions). If that is the case for a candidate, it is shown in both phases.
Each country has their own national regulatory authorities (NRA) responsible for approving for new drugs in their country. There are several NRAs that are particularly relevant, among them the US-American FDA and the European EMA. We signify a COVID-19 vaccine as approved if one of these bodies approves a vaccine.
We will also classify a vaccine as approved if one of these two bodies greenlights a new vaccine on their "emergency use listing", or if the WHO does so.
| Vaccine Candidate | Approved by | on | notes |
|---|---|---|---|
| BioNTech-Pfizer | FDA | 2020-12-12 | for emergency use |
| BioNTech-Pfizer | EMA | subsequently | for emergency use |
| Moderna | FDA | 2020-12-19 | for emergency use |
| Moderna | EMA | subsequently | for emergency use |
| Oxford/AstraZeneca | EMA | 2021-01-29 | for emergency use |
| Janssen | EMA | 2021-03-11 | for emergency use |
| Sinopharm | WHO | 2021-06-03 | for emergency use |
| Sinovac | WHO | 2021-06-03 | for emergency use |
For the duration of clinical trial phases, we rely on the LSHTM data publishing the estimates provided by research groups. They proved a start date for their clinical trials and a primary completion date.
The primary completion date is defined as "The date on which the last participant in a clinical study was examined or received an intervention to collect final data for the primary outcome measure. Whether the clinical study ended according to the protocol or was terminated does not affect this date. For clinical studies with more than one primary outcome measure with different completion dates, this term refers to the date on which data collection is completed for all the primary outcome measures. The "estimated" primary completion date is the date that the researchers think will be the primary completion date for the study."
Datasets were downloaded from the above mentioned sources. The data provided by LSHTM did not undergo any form of pre-processesing as it already came in a machine-readable format.
As the data by WHO is published in form of a pdf, the dowloaded file undergoes a couple of pre-processing steps before analysis can start.
First, the pdf was converted into a machine-readable format using the software Abbyy Fine Reader and subsequently cleaned manually.
The datasets consists of two parts: the first table is labelled xx candidate vaccines in clinical evaluation , whereas xx corresponds to the number of rows in that table. It's followed by the second table yy candidate vaccines in preclinical evaluation that has a slightly different columns compared to the first one.
In combining both datasets, we established the one for analysis that holds the following columns
| column | description |
|---|---|
| COVID-19 Vaccine developer/manufacturer | Name of developer/manufacturer or team of developers/manufacturers, separated by / |
| Vaccine platform | Type of vaccine* |
| Preclinical | preclinical if True, else no value |
| Phase 1 | if there's a trial in this phase, cell holds one or multiple trial numbers |
| Phase 1/2 | if there's a trial in this phase, cell holds one or multiple trial numbers |
| Phase 2 | if there's a trial in this phase, cell holds one or multiple trial numbers |
| Phase 3 | if there's a trial in this phase, cell holds one or multiple trial numbers |
* For entries in the column Vaccine platforms we noticed different spellings of the same type, for example Protein subunit, Protein Subunit or Protein Sub-unit or Non-replicating viral vector or non replicating viral vector, which were unified to be spelled the same way before exporting the dataset for analysis.