|
Data collection, centralization and merging
To improve homogeneity in data collection and to
minimize subjectivity, EmoCard (see history for
details) was build up so that all the fields relevant to the national
registry were codified as yes/no or multiple choice variables; numerical
data were preferred wherever possible.
For each patient entered in the local database, EmoCard generates a unique personal ID code (EMO-ID)
to replace the patient's identity data when data are sent to the national
database. Every six month each HCT
automatically centralize anonymized records of
its patients. Once the information is gathered, the national database is
updated in accordance to the following rules:
1.
Each patient record in the national database is
updated in accordance to the last sent patient data.
2.
Any records refering
to patients not present in the national database are added as “new
patients”.
3.
Any patient in the national database not anymore
present in the last sent dataset is considered lost or withdrawn and is
checked as “deleted”.
This routine is separately run for each dataset
sent by each centre. Then a set of congruency checks are carried out on
the newly merged national database. All the records not passing this
stage are excluded from the analysis and queries are issued to the
relevant HCT in order to fix them. This usually occurs before the
following data collection.
Duplicate management
A common bias in the merged databases and data
analysis occurs when a patient is entered into the registry twice or more
because the same subject is managed in different centres. A duplicate
management routine was built up taking into account the need to preserve
patient privacy. Each HTC sends two
separate set of data, i.e. a dataset with records identified through EmoCard EMO-ID and an uncorrelated single column
table with a patient ID code (PTS-ID) built up on patient sensible data
(i.e. name initial, surname initial, birthdate,
code for birth place, sex). This last table, marked for each sending HCT,
is used to identify duplicate records sent by different HCT. Then, an
automatic routine is sent to all centres, that transparently marks the record with relevant PTS-ID in the local
database so that they will be merged in the subsequent data collection
based on their EMO-ID. Since the logic used to modify the local database
is not handled locally by EmoCard, and since
the set of PTS-ID are centrally destroyed after building up the updating
routine, patients privacy is fully preserved.
Data analysis validation
A predefined set of database queries is then run
on any newly updated database (working database) to automatically produce
national and local reports. These reports are published in password
protected pages on the AICE website (www.aiceonline.it), in order that
each HTCD can check and validate the local report. The incongruences found by each centre are considered,
verified and
eventually included in the working database that become
definitive. The same set of queries is then run again on the final
database version to produce the final reports.
Report publication and Website
The final reports (local and national) are then
published on the AICE website. Each HTCD can access to its own local
report and the national one. The website stores all biannual reports
since july-december 2004, and allows
comparisons between them.
|
