dh02 - the secure data office concept

DH02 - THE SECURE DATA OFFICE CONCEPT Annemie Van Dijck, PhD

Team Manager SD Office Programmers

SGS Life Science – Secure Data Office

PhUSE 2017- 10/OCT/2017

2

OVERVIEW

n  Introduction

n  Secure Data Office tasks §  Randomisation Data §  The Pharmacokinetic (PK) Data flow §  Other unblinding data streams §  Cleaning of unblinding data §  Deliver blinded result transfers §  Deliver data to unblinded teams

n  Conclusion

3

OVERVIEW

n  Introduction

n  Secure Data Office tasks §  Randomisation Data §  The Pharmacokinetic (PK) Data flow §  Other unblinding data streams §  Cleaning of unblinding data §  Deliver blinded result transfers §  Deliver data to unblinded teams

n  Conclusion

4

INTRODUCTION

n  Double blind study: §  Subject and investigator are unaware of treatment status §  Parties without access to unblinding data: subjects,

investigators, data management, statistics, PK analysts, sponsor

§  Parties with access to unblinding data: pharmacists, bio-analysis lab, unblinded CRA

n  Unblinding data: §  Any data that can unblind the study

•  Treatment information •  Pharmacokinetic data •  Immunogenicity data •  Biomarkers •  ....

5

INTRODUCTION (2)

n  Downside: §  Unblinding data is not cleaned §  Delay in clinical database lock or unblinding data are not

included in lock §  Crunched timelines for some parts of the analysis

6

Who should handle randomisation data and other unblinding data (e.g. PK, biomarker,

immunogenicity data)?

7

SD OFFICE

n  Dedicated team, physically separated from other study team members

n  Secure server locations and separate databases

n  Clearly defined processes in SOPs and WIs

8

OVERVIEW

n  Introduction

n  Secure Data Office tasks §  Randomisation Data §  The Pharmacokinetic (PK) Data flow §  Other unblinding data streams §  Cleaning of unblinding data §  Deliver blinded result transfers §  Deliver data to unblinded teams

n  Conclusion

9

RANDOMISATION DATA

n  Generation of randomisation lists §  Documented randomisation list specifications §  By SAS programmers with statistical background §  Distribution to required recipients

n  Creation of SDTM datasets containing randomisation information based on following input §  Receive randomisation information: from external vendor or

internally created §  Subject information: from data management

10

PK DATA FLOW

n  PK data is always unblinding

n  PK data (Identifiers and results): bio-analysis lab SDO

n  Blinded transfers: SDO Data Management (DM)

n  Cleaning is performed by DM

n  DM delivers clean PC file to SDO §  Contains CRF and sample tracking data §  Without results

n  SDO delivers clean PC file with results to include at time of DB lock

11

Complete PK flow

Central lab

Investigational site

Data management SD Office

BAN labShip

sample and shipping list

Ship sample, req. Form and shipping list

Capture eCRF data into clinical database

Transfer ST file for

reconciliation

Send clean PC without concentrati

ons

Create merged PC

Transfer SAMPID file

1

2

3

6

5

Clinical Pharmacology,

Pharmacometrics

9

11

12

Transfer Results file

Send identifier file

Create PP dataset

4

7

13

8

NCA input file

NCA output file

Send locked Clinical

database

10

ZOOM 1

12

Central lab

Investigational site

Data management SD Office

BAN labShip

sample and shipping list

Ship sample, req. Form and shipping list

Capture eCRF data into clinical database

Transfer ST file for

reconciliation

Send clean PC without concentrati

ons

Create merged PC

Transfer SAMPID file

1

2

3

6

5

9

Transfer Results file

Send identifier file

4

7

8

Creation of PC dataset

13

Complete PK flow

Central lab

Investigational site

Data management SD Office

BAN labShip

sample and shipping list

Ship sample, req. Form and shipping list

Capture eCRF data into clinical database

Transfer ST file for

reconciliation

Send clean PC without concentrati

ons

Create merged PC

Transfer SAMPID file

1

2

3

6

5

Clinical Pharmacology,

Pharmacometrics

9

11

12

Transfer Results file

Send identifier file

Create PP dataset

4

7

13

8

NCA input file

NCA output file

Send locked Clinical

database

10

ZOOM 2

14

CREATION OF PK INPUT FILES AND PP DATASET

Data management SD Office

Create merged PC

Clinical Pharmacology,

Pharmacometrics

9

11

12

Create PP dataset

13

NCA input file

NCA output file

Send locked Clinical

database

10

15

CREATION OF PK/PD INPUT FILES BY SDO

Start programming

input file

Issues in DB?

Final PK/PD input (NCA/NONMEM) file ready

PK/PD modeling Results

DB re-opening

Dat

abas

e lo

ck

Without SD Office

With SD Office

Start programming

input file

Issues in DB?

Final PK/PD input (NCA/

NONMEM) file ready

PK/PD modeling Results

Solved before DB lock

Dat

abas

e lo

ck

Time Gain

16

OTHER UNBLINDING DATA STREAMS: LAB DATA

n  Similar to PK data stream

n  Only unblinding analytes via SDO

n  SDTM domains: LB, PD, IS

n  DM unblinded at release §  Yes: combine LB parts by DM §  No: combine LB parts and release by SDO

17

OTHER UNBLINDING DATA STREAMS: PROTOCOL DEVIATIONS

n  Identification of unblinding protocol deviations (usually medication kit errors)

Protocol deviation defined asAny medication kit

misallocationOnly treatment misallocations

Medication kit number unblinding?

Y SD Office involved SD Office involvedN Detection of

medication kit misallocations by data

management

SD Office involved

18

OTHER UNBLINDING DATA STREAMS: PROTOCOL DEVIATIONS

n  Assigned medkit = planned data

n  Dispensed medkit = actual data

n  Reconciliation of planned and actual data by SDO

n  SDO creates part of the DV domain: §  Deviations related to medkit or treatment misallocations

19

CLEANING OF UNBLINDING DATA

n  Only if cleaning of the unblinding content is required

n  Cleaning requirements: in secure data handling plan

n  Cleaning Checks are: §  Defined §  Documented §  Programmed §  Validated §  Ran

n  Review output of checks

n  Write and follow up queries: blinded/unblinded

20

DELIVER BLINDED RESULT TRANSFERS

n  PK data (Identifiers and results), other unblinding data: external vendor SDO

n  Blinded transfers: SDO Data Management (DM)

n  Cleaning is performed by DM

n  Results transfer with dummy identifiers/scrambled : SDO statistics, PK analysts: dry runs

21

DELIVER DATA TO UNBLINDED TEAMS

n  PK data (Identifiers and results), other unblinding data: external vendor SDO

n  Blinded transfers: SDO Data Management (DM)

n  Cleaning is performed by DM

n  Results transfer with real production data: SDO safety commitee, DRC, SSG

22

OVERVIEW

n  Introduction

n  Secure Data Office tasks §  Randomisation Data §  The Pharmacokinetic (PK) Data flow §  Other unblinding data streams §  Cleaning of unblinding data §  Deliver blinded result transfers §  Deliver data to unblinded teams

n  Conclusion

23

CONCLUSION

n  Role of SD Office §  Create randomisation lists and datasets §  Clean data before database lock

•  PK data •  Medication kit misallocations •  Other unblinding data

§  Prepare input files for PK/PD analysis upfront -> analysis can start earlier

n  Results §  Time gain after lock §  Higher quality §  Unblinded data available before lock for DRC, SSG, dry runs §  Off the critical path and key results can be met

24

THANK YOU FOR YOUR ATTENTION

Annemie Van Dijck Life Sciences Team Manager Secure Data Office Programmers SGS Belgium NV Generaal de Wittelaan 19A b5 2800 Mechelen - Belgium Phone: + 32(0)488 245 699 Email: annemie.vandijck@sgs.com

clinicalresearch@sgs.com EUROPE: +32 15 27 32 45 AMERICAS: + 877 677 2667 www.sgs.com/cro

Meet us at the PhUSE Booth 2