?????????????????????????????????? ?????? ScanTRIAD

1
????????????????????????????????????????
ScanTRIAD
??.??. ?????? ????????, PhD. (Statistics) ????????
?????????????????????????????????????????????? ???
???????????????????????????? ??????????????????
??????????????????
2
???????????????????????????????????
?????????????????? ???????? ?????????
?????????????????????????????????
??????????????????????????????
?????????????????
?????????????????????? Selection
Bias ??????????????????? Information
Bias ?????????????????? ?????????????
Confounding Bias
?????????????????????????????????????????
????????????????????????????
3
??????????????????????????????????????????????????
?
?????????/?????
??????
??????????????????????
Information Bias Validity reliability of
tools vs. Data management
Confounding Bias All related predictors vs.
what were collected
4
Systematic Approach Data Management and
Statistical Support
Availability of complete, clean data takes time,
effort, and attention to details.
??????????????????? Clean data
?????????????????????????????
Researcher Statistician ???????????????????????
???????????
?? Manual of Operation ????????? ??????? Plan
for data analysis
Statistician Data manager Programmer
??????????????????????????
????????????????? Documentation
Detect ???????????? ?????????????????????
5
???????????????????????
?????????????????????????????????? ICH-GCP
??????????????? http//www.ich.org
ICH International Conference on Harmonisation
of Technical Requirements for Registration of
Pharmaceuticals for Human Use GCP Good
Clinical Practice
?????????? European Union (EU), Japan, ??? USA
????? 1996
???????????????????? ???? Australia Canada the
Nordic countries WHO etc.
?????????????????????????????? ICH-GCP
?????????????????????????????????????
?????????????????? ???????????????????????????????
?????
6
????????????Quality assurance and/or control

• Prevent problems
• Detect problems
• Correct problems

7
Quality assurance elements

• Prevention
• Well-written protocol, manual of operations
• Collection limited to essential items,
  uncomplicated forms, criteria
• Pre-test study forms and procedures
• Investigators commitment to follow protocol
• Training and certification of all staff
• Data from central classification committees,
  specialized equipment (calibration), central
  laboratories or reading centers (internal
  replication external duplication or standards)
• Maintain study records audit trails, archiving

Adopted from Shrikant I. Bangdiwala, Ph.D.
8
Quality assurance elements

• Detection
• Central monitoring of data on individual
  subjects
• data entry system checks
• logical, consistency checks
• extreme values
• Site visits standard check-list, records audit
• Comprehensive performance-monitoring reports
  study overall, by site, by staff
• recruitment, follow-up, adherence, completion of
  procedures
• errors
• Statistical investigations of aggregate data by
  site, by staff
• identify unusual patterns
• lack of variability
• unusual relationships in the data

Adopted from Shrikant I. Bangdiwala, Ph.D.
9
Quality assurance elements

• Correction
• correct the errors and minimize the chance of
  future occurrences
• procedures must be implemented early in the study
• empower individuals, committees, centers to
  address problems
• effect of systematic errors, bias, violations of
  protocol
• address individual site or staff performance
• redress misconduct or fraud
• Document all actions

Adopted from Shrikant I. Bangdiwala, Ph.D.
10
RDM Processes
Data Entry Design Considerations
Design of data collection forms Paper-based
????????????????, ??????????, ??????????????????,
??????????? ??? Electronic-based CAPI, PDA,
Web-based, Applications, Optical Scan, etc Data
collection methods Self-administered ??????,
???????????????????????, ?????????????????????,
??? Interview Type of projects Single site
VS Multi-center Cross-sectional or
Longitudinal Routine data collection Others
Small size project VS Mega study VS Country
census Real-time monitoring, Urgent, Allow
sufficient time
11
RDM Processes
Data Entry Design
Portal of data entry Distributed data
entry Centralized data entry Design of
data entry interface Direct VS Via CRF Key
punching VS Mouse clicking VS Optical
scan Spread sheet style VS WYSIWYG Data entry,
validation, and verification methods Single
VS Double data entry Embedded validation at
entry VS Validation externally Verification
tools Paper VS Printout, Paper VS Screen,
Screen VS Screen, Two parts within a
screen, Data and images being
integrated (SD and CRF can be integrated)
12
RDM Processes

• Data cleaning
• All variables or key variables?
• How much computerized vs manual?
• Consistency checks across variables, across
  forms, across time, across similar individuals
• Frequency and timing given rate of accumulation
  and study needs
• Audit trail documentation ALL changes to
  original data specify what, when, why, by whom

13
RDM Processes

• Audit trail documentation ALL changes to
  original data specify what, when, why, by whom
• WHY?
• Monitor study integrity and quality assurance
• CC does this separately by personnel,
  collectively for trends
• Regulatory agencies wish to compare the
  information in original data collection forms
  with that in reports
• Usually, sample 10 of subjects in database, 100
  of data from sampled subjects, and often 100 of
  subjects for key variables
• Tolerance of errors lt 25/10000 fields 0.25

14
Main Tasks Data Management using

1. Scan
SCAN
Back Office
2. Verify
3. Purify
VERIFY
PURIFY
Front Office
State of the art for quality data entry
15
???????????????
1.Scan
Feed paper
2.Verify
3.Purify
16
(No Transcript)
17
Data Management System with Tools for Optical
Recognition, Verification, and Purification.
http//www.scantriad.com
18
?????????????????????????????????????? ???????
Lot ???????????????
19
????????? ??????? Lot ???????????????
20
?????????????
21
????????????? ????????????
22
Data Verification Center
23
Data Verification Center
24
Example that data verification is needed
25
Example that data verification is needed
26
Example that data verification is needed
27
Example that data verification is needed
28
Example that data verification is needed
29
Example that data verification is needed
30
Example that data verification is needed
31
Example that data verification is needed
?????? Verify