Critical aspects | Solution |
---|---|
Definition of the objectives of the registry | Discussion on the objectives in a working group involving different stakeholders, including patient representatives |
Definition of the population under study | |
Definition of inclusion criteria | Extensive literature research, retrieval of necessary information from existing registries, harmonisation of criteria made by a working group, adoption of an operational definition that could be used as inclusion criteria for the registry purposes |
Assessment of whether patients registered meet the inclusion criteria | Ideally, recording of all the information necessary to check diagnosis, but, operatively, assessment delegated to the data contributors who have to confirm that the inclusion criteria are met |
Definition of what to measure and how to do it | |
What to measure | Review of literature and discussion on variables definitions in a small working group of experts |
How to measure | Start data collection of few variables and test with a pilot study the applicability of their definition |
If the definition used is not the same across countries: | |
• try harmonisation by making the definition more generic | |
• involve stakeholders to discuss change of definitions and agree on a shared definition | |
• if definitions can be assimilated, report differences of definitions in the publications as caveats | |
Data management and data quality controls | |
Data management | Shared electronic platform for data collection with automatic computation of derived variables, allowing both direct data entry and remote data upload. |
Use of technology (such as XML) that ensures that required data format and coding is used. | |
Data quality controls | Automatic and immediate data quality controls on entering (plausible ranges, intra-record data coherence, and consistent information across years.) |
Use of drop-down menus with fixed input possibilities (e.g. yes/no/unknown) | |
Agreed controls with national registries in order to avoid duplication of identical data quality control processes. | |
Use of refined data controls based on age-and-sex-specific reference values | |
Set up of a data error procedure that uses a software that automatically warns and points the user to the data to correct | |
Handling of missing data | User-friendly software and useful feedback to contributors to encourage data entry |
Clear definitions, but attainable in daily clinical practice | |
Unequivocal exhaustive variable coding with no pre-set values | |
Avoid the use of tick boxes that code missing answers and negative answers the same way | |
Working with existing registries to accommodate definitions | |
Maintaining patient confidentiality | Separate storage of encrypted personal data and anonymous centre numbers |
Pseudo-anonymisation to allow contact with centre for error correction | |
Dissemination of data | Code of conduct document concerning publication rights, authorship and data access – preferably set up very early in the process |