Technical documentation to the article titled:
"The Costs of Covid-19 Vaccinations – Should we Rethink the Policy?"
This post provides documention for the analysis process that was performed with the abovementioned article (this link), as well as some additional visualisations produced from the data downloaded from EMA and ECDC.
Vaccination Volume Data from ECDC
The data for COVID-19 vaccination volumes can be found here:
https://www.ecdc.europa.eu/en/publications-data/data-covid-19-vaccination-eu-eea
Individual Case Safety Reports from EMA
1. To download all ICSR reports, follow the steps below.
Open http://www.adrreports.eu/en/search_subst.html, navigate to 'C' and search for COVID-19.
To download ICSR reports for ALL forms of medication, one would need to visis the 4,000+ entries listed. As a workaround, I have a shortcut described below.
Since EMA does not allow the public to download all ICSR reports for all forms of medication in one go, I came up with a work around, by manipulating the url for entering their Oracle back end. By simply not specifying the identifier for a parrticular substance, all ICSR reports on all forms of medication get shown.
Shortcut to all Substances (incl. COVID19-vaccines)
Generate and download the ICSR Line listings for COVID, Measles and Influenza vaccines as well as the shortcut for all forms of medication, and repeat steps 2, 3 and 4 explained below, for each of the vaccines:
2. Export the ICSR reports to TAB-delimited text files as follows:
Navigate to the tab labelled 'Line Listing'.
For Geographic Region, select 'European Economic Area'.
Next, click on 'Run Line Listing Report'. This will open a new window and present you with the ICSR reports in tabular format. This may take a while.
Navigate all the way to the bottom and start the export as shown in the image below:
The export will be saved to the default Downloads folder of your web-browser.
3. Convert and rename all exported files to UTF-8 encoding (e.g. using Notepad++ for Windows).
In the instructions below, I used a naming convention as follows: MMMDD_export_[substance].csv
4. Load the exported file into PostgreSQL
Create a schema and table in PostgreSQL (I use an old version 9.1 on an old Windows Server 2008)
CREATE TABLE ema.sta_icsr
(
source varchar,
elocnr character varying,
reporttype character varying,
evgatewayrecdate character varying,
primarysourcequalification character varying,
primarysourcecountryforregulatorypurposes character varying,
literaturereference character varying,
patientagegroup character varying,
patientagegroupperreporter character varying,
parentchildreport character varying,
patientsex character varying,
reactionlist character varying,
suspectdruglist character varying,
concomitant character varying,
icsrform character varying
)
WITH (
OIDS=FALSE);
Description of the DOS command: For each export file (%f), skip header, add a column identifying the filename, replace some unreadable characters, prepend backslashes with a backslash and tell psql to execute the copy command.
5. De-duplication of records
To de-duplicate, we need a table that holds only unique entries. Also, we need to give priority / precedence to specific substances, before we load the rest (all substances thru that manipulated url).
id serial,
source varchar,
elocnr varchar
)
without oids;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%pfizer%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%astrazeneca%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%moderna%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%janssen%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%MMR%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%Influenza-split-virion-inactivated%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%influenza-live-nasal%'
;
insert into ema.dwh_geneloc (source, elocnr)
select
source, elocnr
from
ema.sta_icsr
where
not (elocnr) in (select elocnr from ema.dwh_geneloc group by elocnr)
and source like '%all_all%'
;
create index ix_staicsr_elocnr on ema.sta_icsr using btree (elocnr);
create index ix_geneloc_source on ema.dwh_geneloc using btree (source);
create index ix_geneloc_elocnr on ema.dwh_geneloc using btree (elocnr);
