Data_management - Dipen Khanna

Introduction to Data Management DIPEN KHANNA Manager – Data Review Pfizer Pharmaceutical India Pvt. Ltd.

21 April 2005

What is Data Management? 

Historically, Data Management has been thought of as “running edit checks” and “writing queries”.



While these two aspects of Data Management still exist as very important functions, Data Managers are responsible for many other aspects of managing the data, which may include…

21 April 2005

What is Data Management?       

21 April 2005

Setting up complex checks in either the Clinical Trial database or Data Browser programs Managing a vendor who is performing the data management for a study Interacting with other specialty outsource vendors (e.g., Central Labs) Acting as a key contributor to the larger Clinical Project team Oversight of Case Report Form and database development Knowledge of several complex systems for overseeing the data and data quality (e.g., OC, OC RDC, etc.) Others???

What is Data Management? Defined simply, Data Management is…

The entire process involved with taking original raw data from the clinical sites and compiling and validating it, so that it is suitable for reporting.

21 April 2005

What is Data Management? Ways in which data are managed:            21 April 2005

Monitoring and Source Document Verification Validation checks within a database EDC auto-hits at the clinical site Manual data checks Double data entry of all CRF data Blinded data review (BDR) Issue and resolve site queries Code Adverse Event & Medication terms Review of data listings Database QC audit Others???

Data Management Responsibilities Study Start-up:  Coordinate Protocol and CRF development  Develop data management documents – CRF Completion Guidelines – Data Management Plan – Self-evident Corrections – Data Entry Guidelines 21 April 2005

Data Management Responsibilities Study Start-up:  Oversee database development – I*NET – Oracle Clinical  Identify, oversee development, and test validation procedures  Oversee Imaging system set-up  Perform/oversee lab set-up

21 April 2005

Data Management Responsibilities Study Conduct:  Maintain data management study documents and ensure they are in the TMF  Oversee data management activities performed by a CRO/FSP and provide necessary study specific documents  Generate, distribute, and resolve queries

21 April 2005

Data Management Responsibilities Study Conduct:  Participate in dictionary coding – Adverse Events – Medications – Medical History  Request and track batch validations  Oversee electronic data loads 21 April 2005

Data Management Responsibilities Study Conduct:  Perform CRF page tracking  Oversee data flow  Participate in blinded data reviews (BDRs)

21 April 2005

Data Management Responsibilities Study Close-out:  Test break blind program  Oversee QC Audits  Perform database release and re-release  Document post-database release changes

21 April 2005

How Does DM Fit into Clinical Research? •

The data management function supports/oversees all data collection and data validation for a clinical trial program.

•

Data management is essential to the overall clinical research function, as its key deliverable is the data to support the submission.

•

Assuring the overall accuracy and integrity of the clinical trial data is the core business of the data management function.

21 April 2005

How Does DM Fit into Clinical Research? •

Data management starts with the creation of the study protocol

•

At the study level, data management ends when the database is locked and the Clinical Study Report is final

•

At the compound level, data management ends when the submission package is assembled and complete

21 April 2005

How Does DM Fit into Clinical Research? ICH Guidelines for Good Clinical Practice list requirements for how clinical trial data shall be validated and updated.

 ICH GCP 5.5  ICH GCP 8.3.14  ICH GCP 8.3.15 Example: 5.5.1 “The sponsor should utilize appropriately qualified individuals to supervise the overall conduct of the trial, to handle the data, to verify the data, to conduct the statistical analyses, and to prepare the trial reports.” 21 April 2005

Additional Guidelines -

21 April 2005

Importance of Effective Data Management

The quality of a clinical study is only as good as the weakest data point…

21 April 2005

Importance of Effective Data Management Statistical analysis – an accurate database is the basis for drug approval by the FDA – Adherence to federal regulation and guidelines mandate the safety and welfare of patients participating in trial and ultimately, the safety of patients prescribed the approved drug Marketing the drug – It is important for patients and physicians to clearly understand the indication for the treatment, potential side effects, and contraindications for the use of the product Post marketing surveillance – Drug companies are required to send safety information to the FDA after the drug has been approved and marketed 21 April 2005

Ensuring Quality Data

21 April 2005

The CRF or Other Data Collection Tool (DCT)… We all know the saying…

21 April 2005

The CRF or Other Data Collection Tool (DCT) … Protocol to CRF: • The study protocol dictates what data need to be collected. •

The CRFs or DCTs should be designed to collect only the data required to answer the study protocol’s research question(s).

•

Collecting the “nice to have” data that is not specified in the protocol should be avoided.

•

Standards should be adhered to.

21 April 2005

The CRF or Other Data Collection Tool (DCT)… CRF to database: • The CRF defines the overall design and structure of the clinical trial database. •

Data should only be collected in one place. Multiple sources of the same data introduces additional possibilities for error.

•

If a data point must be summarized it must be captured as a numeric or coded value. Free text cannot be summarized. 21 April 2005

The Clinical Trial Database Oracle Clinical (OC): Storing and validating the clinical trial data. An industry standard for managing clinical trial data.

OC is a fully validated system: Conforms to the software development life cycle (SDLC)

21 April 2005

The Clinical Trial Database Excel spreadsheets or Access databases should never be used to capture the clinical trial data, as they are not validated for that purpose and do not conform to GCP Guidelines. •

Excel and Access are not set up to require 1st and 2nd pass data entry

•

Excel & Access have no audit trail to track data changes

21 April 2005

Oracle Clinical Structure OC is set up to allow for easy data entry and data retrieval. The data entry screens are actually set up in sequential order of the CRFs, so that a Data Entry operator can enter the data quickly.

21 April 2005

Oracle Clinical

Data Verification  Verification:

check that what is in the source doc is on the CRF and what is on the CRF is in the database

 Verification

ensures that data are reported accurately on the CRFs and are consistent with the source data.

21 April 2005

Data Verification  Data

Verification is performed: – At the site via source document verification – In-house via: Double data entry Data reviews – BDRs, listings, etc.

21 April 2005

Data Validation  Validation:

check that what is in the database is logical, consistent, and analyzable

 Validation

ensures that data are:  Complete  Correct  Allowable  Valid  Consistent

21 April 2005

Data Validation  Data

Validation is performed: – At the site via CRF review for consistency and validity – In-house via: Programmed data checks within Oracle Clinical Manual data review via listing or edit check

21 April 2005

Verifying and Validating the Data  Potential

Sources for Error

– People – Data entry – Coding process

21 April 2005

Verifying and Validating the Data  Possible

Types of Error – Erroneous Data – Protocol Deviations – GCP Violations

21 April 2005

Data Management Documentation  Data

Management Plan may include: – Lists checks performed – Identifies which discrepancies can be solved inhouse (self-evident changes or no action required) – Identifies which must be queried to the investigator – Lists any assumptions that can be made during review/coding process

21 April 2005

Data Management Documentation Self-Evident Corrections (SEC): –Lists all changes that can be made to the data by sponsor without a query to the Investigator –Site is sent document prior to start of discrepancy management and at least one more time once the study has ended 21 April 2005

When to Query the Site •

Only a very limited number of corrections can be made to the data, without querying the clinical site.

•

For any data discrepancies that cannot be corrected as self-evident and are clearly data errors, a query, or Data Clarification Form (DCF) must be sent to the site.

21 April 2005

Data Queries  Definition

– Individual questions sent to investigative site

concerning a data discrepancy – Should be generated on ongoing basis – Should be resolved as early as possible – Query cycle time consuming and expensive  Commonly quoted each query cost $50-$75 to resolve  Query generation/resolution typically takes up 50% of the total data processing time 21 April 2005

Data Queries  Data

Discrepancy Flow Data Discrepancy

Generate Query Send to Site Site Responds with Answer Data Manager makes Change to Database 21 April 2005

Self-Evident Correction Data Manager makes Change to Database

When are the Data Considered Clean? •

All data has been received and entered

•

All DCFs and OC discrepancies have been addressed and resolved

•

A final QC has been performed across the entire study database

•

At this point the database may be locked and unblinding information is added.

•

Checks are run to validate the unblinding information.

•

The database is then frozen and released for analysis.

•

All SAEs are reconciled with the safety database. 21 April 2005

Database Release Making any post-release changes is highly discouraged, unless significant data issues are identified. There are very clear & detailed procedures on how to make a post-release change, should it be necessary.

21 April 2005

Oracle Clinical

Reporting the Data---Data Out A1 2 3 4 5 6 L is tin g o f B e s t S u b je c t R e s p o n s e w ith D e m o g ra p h ic D a ta S u b je ct In itia lsAg e ABC 56 DE F 48 G HI 62 K LM 46 NO P 53 RS T 38 UV W 59

• •

Sex M F F F M M F

R a ce T re a tm e n t C auc as ian O ur Drug A fric an-A m eric anO ur Drug A s ian O ur Drug C auc as ian Com petitor Drug H is panic Com petitor Drug A s ian O ur Drug A fric an-A m eric anCom petitor Drug

Be st Re sp o n se Du ra tio n o f Re sp o n se S table D is eas e 36 week s P artial R es pons e 22 week s P artial R es pons e 19 week s P artial R es pons e 13 week s S table D is eas e 17 week s P artial R es pons e 35 week s P rogres s ive Dis eas eN/A

This is a sample listing for the fictitious Oncology study A123456. This may be one of many tables used to perform the overall analysis of this trial’s data. 21 April 2005

Summary

21 April 2005

Flow Chart of Activities Protocol Protocol Design Design

Data Data entry entry

Data Data verification verification

Update Update database database with resolutions with resolutions and and close close out out queries queries

Database Database audit audit

CSR

21 April 2005

CRF CRF Design Design

Indexing Indexing

Data Data Entry Entry audit audit

Send Send to to sites sites thro’ thro’ CRAs/ CRAs/ DMs DMs

Database Database closure closure for for reporting reporting

Issue Issue of of final final biometrics biometrics tables tables and and report report text text

Database Database setup setup and and validation validation procedures procedures

Scanning Scanning

Merge Merge data data into into database database Generate Generate queries queries

Detailed Detailed statistical statistical analysis analysis plan plan written written

Monitored Monitored CRFs CRFs received received

Online Online dictionaries dictionaries automatically automatically applied applied Batch Batch validation validation

Programming Programming for for all all safety safety & & efficacy efficacy tables tables completed completed

Tables Tables generated generated

Review Review of of reports, reports, tables tables by by clinical clinical and and biometrics biometrics teams teams

Statistical Statistical report report on on methodology methodology and and findings findings

100% 100% QC QC of of tables tables

Data Management Workflow  Define

project specific data management requirements  With Study Start Up: – Develop CRFs – Develop database – Develop validation checks  CRF computerized tracking, entry and verification  Tracking electronic data loads 21 April 2005

Data Management Workflow  Validation

checks and data listings generated  Data review, coding, and query generation  Query resolution and database changes  Final database updates and verification  Final database quality check  Database lock and delivery to reporting  Post-release change management 21 April 2005

Whew!!!

21 April 2005

Q&A

21 April 2005

Thank you for your participation today!!!

21 April 2005