Vanessa Montanari
By Fateh Cheklat, ARC/TEC/Data Manager chez Pharmaspecific
Dear CRAs, for this article I want to talk about a profession, little known in the field of Clinical Research, and yet, working closely with you throughout a clinical trial. The profession I exercise for 4 years now « Clinical Data Manager » or « Data Manager ». The CRAs do not always know what constitutes the profession of Data Manager, except via the queries left on the electronic CRF … so I would like to explain in this article what this profession is.
To become a Data Manager, you must have a level of study Baccalaureate +3 to +5, scientific or computer, in order to integrate training of one year at a private school or university: 6 months of theoretical courses are taught followed by 4 to 6 months of practical training. A traineeship report is required to validate the diploma (RNCP I).
The objectives of the Data Manager (DM) are to structure clinical data of patients enrolled in a clinical trial in a database, to verify the consistency of these data and their reliability, and clean up the database before it is sent to the statistician for statistical analysis. (1)
In clinical research, data management step is crucial for statistical validation. The more the database is clean, consistent, complete, and of high-quality, the more interpretation and analysis of data will be accurate and the decision will be better. The opposite will necessarily give a poor and incomplete analysis, and therefore, the decision will be bad: For this and like any player in clinical research, the DM must comply with Good Clinical Practices (http://www.arcat-sante.org/a/testing/schedules/bonne.html). Moreover, he must respect the personal CNIL directives (Law No. 78-17 of 6 January 1978 and amended on 6 August 2004: https://www.cnil.fr/fr/loi-78-17-du- 6-January-1978-modifiee) (see figure 1).
- The tasks of the Data Manager: (See Figure 2 and 3)
From the clinical trial protocol, the Data Manager identifies data to be collected at each visit and throughout the patient’s participation period (before ‘screening’, during and after treatment as part of patient monitoring). He will meet with the Head of the study project to validate the list of data to be collected, as well as the ergonomics of the case report form (CRF).
Once the list of data validated, the DM designs at first, according to the Good Practice of Data Management (http://libraries.ucsd.edu/services/data-curation/data-management/best-practices.html). Yes, the creation of a CRF still requires a paper copy, even in case of an electronic CRF.
Once the CRF paper is designed, the DM annotates it by presenting data in tables: demographic (DM), vital signs (VS), medical history (MH), concomitant therapies (CM), adverse events (AE), etc. Then he gives a name and a format or code for each data. For this, either the DM follows the CDISC standards or he agrees with…
The CDISC standards are standards that have been developed to harmonize practices of Data Managers.
Example: the annotation for the date of birth: belongs to the table « Demography » (DM). The corresponding name is « BRTHDT » (for « Birthday date). On the CRF, the DM notes next to the birth date « DM. BRTHDT « and its format is » DDMMMYYYY « (for Day-Month-Year).
The DM designs a database using software dedicated to the clinical data management (such as Oracle Clinical, Inform, Clintrial, CLINSIGHT, etc…) identical to the paper version of CRF.
These software include several modules for
Under « draft » mode and from annotated CRF, the DM built the database that is a set of entry screens (equal to the number of pages of the CRF paper) and he programs coherence and presence tests. This mode allows us to make changes if necessary, unlike the « production » mode.
The DM conducts tests on the basis using fictitious patients, to control its reliability and the one of the programmed tests by trying all possible cases.
If the test is good, the CRF will be compiled in the « production » used for real patients.
The final CRF can be either in paper form or electronic form:
1- If the project needs a paper version for data collection, the CRF will be printed in multiple copies according to the number of patients to be included (a CRF for each patient) and sent to the participating centers to fill.
In this case, data entry clerks will report patient data in the database designed by the DM, hence the usefulness of a base identical to CRF to facilitate the task of data entry clerks.
The latter verifies and validates the entry using data management software.
Also using the mentioned software, he runs the presence and consistency tests. These tests allow to check if all required data have been completed and if these data are consistent. For example, if a required data is not provided, a query will appear as an error message asking the filling of this data. Similarly, for a data which is not consistent with one or more other data (the test is scheduled on the two or more variables), such as a date of birth entered in 1800, for example.
The DM then prints queries and sends them to the concerned centers for correcting or explanation.
On the return of these queries, if the answer is that is to say that the explanations or the answer provided is adequate, the DM corrects the data on the basis. Otherwise he reminds the center until a clear and convincing answer is provided.
2- If the project needs a CRF online, the DM puts CRF pages in HTML format and puts everything into the accessible internet website.
In this case, there will be no need for data entry clerk, or to compare and validate data entries, since the investigator himself fills (or his CRA/CSC) and validates online CRF pages of his patients.
Moreover, in case of existence of the calculated data, the DM sets up automatically calculable fields by combining formulas with this data, which avoids calculation errors and save time for the investigator.
Similarly, for queries. The queries will be sent directly to centers online. The investigator or the CSC will answer these queries, either correcting or confirming the data. If the answer is « convincing », the query will vanish from the circuit upon restart of tests; otherwise the query will be recreated by the system.
There are tests such as text fields, where the center registered the name of lesions to follow, for example… In this case, a list of data to verify visually will be generated, a computer cannot analyze the data. If inconsistencies or errors are found, manual query is generated via the eCRF, which will be sent to the concerned investigator. If the response is good, the DM corrects and closes the query; otherwise he reminds the investigator.
Once the study is completed at the centers, the queries resolved, and the base clean and of good quality, the DM « freezes in locking it to prevent any changes of CRF while the data has already been validated. Data is accessible only in read only. The DM makes data extraction and transfer it to the statistician for analysis before the final stage, « the publication. » (See Figure 2 and 3)
- What do GCPs say:
Research management, data processing and archiving […] If the data in biomedical research are being processed or managed by computerized systems, the promoter
- Ensures and documents the fact that computerized systems used in research meet the requirements laid down by him for integrity, accuracy, data reliability and compliance with expected performance (that is to say validation);
- Implements and monitors standard operating procedures for the use of these systems;
- Ensures that the design of these systems allows changing of data so that the changes are documented and that no entered data is deleted (that is to say, keeps an audit trail of data and changes);
- Implements and monitors a security system that prevents unauthorized access to data;
- Maintains a list of persons authorized to modify the data;
- Performs appropriate backup copies of data;
- Preserve the unbeknownst, if any (e.g. during data entry and data processing);
- Ensures that the personal data processing carried out in the framework of the research are made under the conditions defined by Law No. 78-17 of 6 January 1978 referred to above and regulations made thereunder and, where applicable, the decision of the national Commission for Data Protection and liberties of January 5, 2006 on approval of a personal data processing methodology undertaken in the course of biomedical research.
If data are transformed during processing, it should always be possible to compare data and original observations and data after transformation. « (2)
I hope this article will help you to see more clearly in the profession of Data Manager. If you are Data Manager, share your experience in the comments. Yes if you’re CRA and you have questions, do not hesitate to let us know!
References:
Figure 1: http://www.acadm.fr/pages/metier-formation/qui-est-le-data-manager.html
Figure 3: unknown
