Version No: 1.0 / Clinical Research Centre / Page 1 of 5
Date: <dd/mm/yyyy> / Guideline on Clinical Data Management / WI-G-3-06-01

Property of CRC

May not be used, divulged, published or otherwise disclosed without the consent of

The Director, Clinical Research Centre

WORK INSTRUCTION FOR CLINICAL DATA MANAGEMENT

Instruction:

  • This work instruction describes the detailed processes for clinical data management.
  • Refer to SOP-R-G-4-02 for detailed procedures on study archiving.
  1. Study Database
  2. Plan and design the database for compilation of study data during the protocol development stage.
  3. The format of the database should take into consideration the following:
  4. Type of data
  5. Type of statistical analysis
  6. Software used for statistical analysis
  7. Number of records
  8. Reports to be generated
  9. More than 1 database may be required depending on the study design.
  10. Store in an appropriate server with appropriate access (e.g. networks access; internet access; etc).
  11. Have appropriate secured access (e.g. unique username and password; retina scan; thumb print scan; verification code through short messaging system; etc).
  12. If data entry is done by multiple individuals, it is recommended that same time multiple data entry be available.
  13. Perform regular back up of the data in appropriate secured media.
  14. Have disaster recovery plans (including preventative, detective and corrective measures if applicable).
  15. Ensure there is an audit trail of data corrections.
  16. Provide training for relevant staff on the use of the database.
  17. Allow direct access to the database by monitors, auditors and inspectors on a read only basis.
  18. If required, assure blind is maintain during data entry and processing
  19. Make sure that if data are transformed during processing, it is always possible to compare the original data in source document with the processed data.
  20. Ensure the database and server/computer operating system (if applicable) are checked, maintained and upgraded on a regular basis.
  21. Define database lock-down procedures to ensure access to the final dataset is restricted for final analysis and report preparation.
  22. Ensure database is archived properly after completion of study; refer SOP-R-G-4-02.
  1. Collects And Manages Data
  2. Ensure clear and complete audit trail of all transactions to the database (insert, update, delete) when processing study data.
  3. Apply ALCOA to assure data quality:
  4. Attribute: Ensure to record the author of every entry or revision.
  5. Legible: Ensure every entry or revision can be read.
  6. Contemporaneous: Ensure to collect current study data according to planned time frame.
  7. Original: Ensure to keep original study documents. If carbon copy was kept, ensure it can be traced back to original copy.
  8. Accurate: Ensure data are correctly recorded and no conflicting data recorded elsewhere especially for multicentre trial.
  9. The following standard practices should be observed when entering data or information:
  10. Data should be entered in a sequential manner, without leaving any empty spaces.
  11. Irrelevant field in a data collection form (DCF) (this includes case report form (CRF) for industry sponsored research) should be labelled with NA or N/A which indicates ‘not applicable’.
  12. Standardise the format of recording date (e.g. day/month/year) and time (e.g. 24 hours format or otherwise).
  13. All entries should be dated and signed by an authorised person.
  14. The date when the data were collected as well as the date of data entry, should appear when data cannot be entered in real time.
  15. Missing data or overlooked data cannot be inserted between existing lines or written in the margin, it should be inscribed following other entries with the notation of late entry.
  16. Data written by hand should be legible and written with permanent ink.
  17. If data entered by several team members, each entry should be signed and dated by the authorised person who made the entry.
  18. The following standard practices should be observed when correcting data or information:
  19. A single line is drawn through the data to be corrected (it should be possible to read the original data). Deleting a larger space or whole page may be completed with a diagonal line across the relevant section, and NA or N/A written above the line.
  20. The initials of the person who corrected the data and the date of correction should appear.
  21. As far as possible, corrections should be made beside the obsolete data, preferably, by the person who made the entry or by others authorised to do so.
  22. State reason for the change (if necessary).
  23. Use of liquid corrector or correcting material is prohibited (refer Section 2.4.1 for correction of data).
  24. Management of Attachment and Printout:
  25. Attachment to forms:
  26. Staple the attachment to the record; paperclips are not acceptable.
  27. Cross-reference the record and the form with each other, e.g. the record references the report number and the report references the record number.
  28. Attachment to workbooks/logbooks:
  29. Secure the attachment to the appropriate page of the workbook/logbook; use tape or staples, glue or paperclips are not acceptable.
  30. Do not obscure any data on either the workbook page or attachment.
  31. Ensure sufficient identification on the attachment to ensure traceability in the event it becomes separated (cross-referencing).
  32. Indicate on the original document that there is an attachment.
  33. Sign and date both the workbook and attachment.
  34. Thermal printouts:
  35. All printouts made on thermal paper must be copied before attaching to a report or filing. Indicate ‘copy of original ’or ‘true copy,’ on the copy and initial and date.
  36. Do not tape over information on thermal paper as the tape will cause the data to rapidly fade. After making a copy, secure the original and the copy with the report.
  37. Check DCF for missing data or incomplete responses or data outside of normal ranges. If any inconsistencies are found, these inconsistencies should be queried and resolved. All queries should be monitored.
  38. Transfer data from completed paper DCF into database. Data entry should be done by trained data entry staff. Perform single data entry with control checks or double data entry to reduce the incidence of errors.
  39. Single Data Entry with Control Checks: This method may be more suitable for smaller single centre studies with less staff available for data entry and/or less sophisticated database software. Once the data has been entered, a visual check can be done between what is recorded on the paper DCF and what was entered on screen.
  40. Double Data Entry: This method involves two people entering the same DCF data onto the database independently of each other. Depending on the software used, the data may be entered twice onto the database on two separate files, which are then compared by the system for accuracy. If the two entries do not match this would be flagged up by the database. Alternatively when the second data entry person enters the data, if it differs from that entered by the first person, a message immediately appears on screen and the original data can be checked. This method depends on the availability of a technically capable database.
  41. Perform data validation to ensure an accurate ‘clean’ set of data is provided for statistical analysis; one of the following can be used:
  42. Automatic data entry checks (if the study database has software enabled for this system of validation). It is advisable to set up warnings to alert data entry staff when values are entered outside of the expected range or if the type of value entered is incorrect (e.g. a numeric value entered rather than text). It is also useful to set up alerts for missing values where possible.
  43. Systematic post-entry computer tests when data have been entered and are available for the data manager (if applicable). Lists should be created (either through automatic database software system or manually) of the following data queries.
  44. All missing values
  45. All values outside of pre-defined range
  46. Logical checks to ensure consistent reporting between relevant fields and that there are no implausible difference between fields (e.g. male and pregnant).
  47. Define all checks before the study starts. Data validation should continue until all missing values and inconsistencies are corrected or clarified.
  1. Storage of DCFs
  2. Keep all paper DCFs in a secure environment such as a locked filing cabinet in a locked room. Protection against environmental damage such as damp, fire or pests. DCFs should not be kept below water sprinklers.
  3. If required, send copies of DCFs from site to the lead site via secure post or encrypted/password protected digital copy through email. The original DCFs must be retained by the principal investigator (PI) at site and the lead site must keep a log of all DCFs received.
  4. Store data in a secured central server under sole control of the lead site when electronic DCFs are used. Printout of the DCF should be signed and dated by the PI (certified copy) and retained in the investigator site file (ISF) at site prior to sending the DCF data to the PI or sponsor (if applicable).
  5. Ensure that investigator does not have access to the whole database for multi-site studies, if data entry is performed at the investigator site, to protect against bias occurring due to investigators making decisions based on interim data.
  6. Ensure that a backup system is in place to guard against loss of data due to software or environmental disasters.
  1. Database Lock
  2. Ensure that all study activities and related data entry are completed prior to database lock.
  3. All members of the study team must be notified of and agree to the proposed date of lock.
  4. Database must be “locked” to ensure access to the final dataset is permanently restricted so that final analysis and report preparation is not disrupted.
  5. Provide the sponsor with a copy of the locked study database following completion of the study and before the statistical analysis is performed (applicable for ISR).
  1. Data Monitoring Committee (DMC)
  2. It is highly recommended that a DMC be established to carry out reviews of study data at regular pre-defined intervals during the conduct of studies with the following characteristics:
  3. large multi-centre
  4. long duration
  5. Involve large number of participants
  6. Involve more than minimal risk
  7. Members should include experienced investigators, subject matter experts, statisticians and clinicians.
  8. If all committee members are independent to the research team, then the DMC is termed as independent data monitoring committee (IDMC).
  9. The role of the DMC is to review interim results and determine whether or not there are any safety issues or any reason why the study should not continue and protects against potential (unconscious) bias from investigators. In addition, DMC may also oversee the quality of data and compliance with regulatory requirements.
  10. The data reviewed by the DMC should be as up-to-date as possible and should be validated up to the point of the interim analysis to ensure it is of reliable quality.
  11. Where necessary, the DMC should recommend corrective and preventive actions. It is incumbent on the investigators to act on the recommendations.