Clinical Data Management-I Unit 6 Unit 6 Discrepancy Management STRUCTURE 6.1 Introduction Objectives 6.2 OVERVIEW OF DISCREPANCY MANAGEMENT 6.3 DATA VALIDATION PLAN 6.3.1 BATCH VALIDATION 6.3.2 STUDY VALIDATION PLAN/EDIT CHECK 6.3.3 UNIVERSAL RULING 6.4 ............................................... DISCREPANCIES 6.4.1 HOW TO ACCESS THE DISCREPANCIES 6.4.2 IDENTIFYING THE PATIENTS FOR REVIEWING 6.4.3 Types of Discrepancies 6.5 DISCREPANCY REVIEW AND RESOLUTION STATUS CODES 6.6 AUDIT TRAIL 6.7 DATA CLARIFICATION FORMS SELF ASSESSMENT QUESTIONS 6.8 Summary 6.9 Terminal Questions 6.10Answers to SAQ and TQ 6.11ABBREVIATIONS Harstern Pubgya University Page No. 103
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Clinical Data Management-I Unit 6
Unit 6 Discrepancy Management
STRUCTURE
6.1 Introduction
Objectives
6.2 OVERVIEW OF DISCREPANCY MANAGEMENT6.3 DATA VALIDATION PLAN
6.3.1 BATCH VALIDATION
6.3.2 STUDY VALIDATION PLAN/EDIT CHECK
6.3.3 UNIVERSAL RULING
6.4 DISCREPANCIES6.4.1 HOW TO ACCESS THE DISCREPANCIES
6.4.2 IDENTIFYING THE PATIENTS FOR REVIEWING
6.4.3 Types of Discrepancies
6.5 DISCREPANCY REVIEW AND RESOLUTION STATUS CODES6.6 AUDIT TRAIL6.7 DATA CLARIFICATION FORMS
SELF ASSESSMENT QUESTIONS6.8 Summary
6.9 Terminal Questions
6.10 Answers to SAQ and TQ
6.11 ABBREVIATIONS
6.1 IntroductionThis unit essentially deals with various aspects of Discrepancy
Management. Various topics of discrepancy management like Data
Validation plan, discrepancies, discrepancy review and resolution status
codes are discussed.
Harstern Pubgya University Page No. 103
Clinical Data Management-I Unit 6
Objectives
After studying this unit, you should be able to:
Define Discrepancy Management
Describe the process involved in Data Validation Plan
Describe the type of Discrepancies
Define Data Clarification Form
6.2 Overview of Discrepancy Management
Discrepancy Management is a critical part of Data Management. It plays an
important role in cleaning and submitting the data for final study analysis in
a clinical trial. A discrepancy can be defined as a message flagged/stamped
either manually or programmed check following the batch validation process
if the response to a question is invalid or the data recorded is unacceptable.
Verbs can be TRANSITIVE or INTRANSITIVE. i) In ‘transitive’ the action passes from the subject to an object.
E.g.: My cat killed a rat.
The action of “killing” is passed from the
‘cat’ to the ‘rat.’ Here ‘killed’ is a transitive verb.
Harstern Pubgya University Page No. 104
Clinical Data Management-I Unit 6
6.3 Data Validation PlanData Validation Plan plays a key role in discrepancy management. It
involves:
Batch Validation
Study Validation Plan/Edit Check
Universal Ruling
6.3.1 Batch Validation:Batch Validation are programmed checks by Clinical Data Management
System (CDMS) to assure the validity and accuracy of the data.
Subject: The complete subject is the simple subject (a noun or a pronoun)
plus any words or group of words modifying the simple subject that tell who
or what the sentence is about. Thus, a subject is the person, place, or thing that acts, is acted on, or is described in the sentence.
Christopher Columbus discovered America
Subject Verb
The action of the sentence is expressed by the verb - ‘discovered.’
The noun ‘Christopher Columbus’ is doing the action of discovering.
Hence, ‘Christopher Columbus’ is the Subject in the sentence.
Sometimes the verb will express ‘being’ or ‘existence’ instead of action
It validates data against predetermined specifications and is primarily used
to check the efficacy data unique to the current study.
6.3.2 Study Validation Plan/Edit Check:Study Validation Plan is a document which records the edit checks for a
study in order to maintain consistency and validity of the data.
Harstern Pubgya University Page No. 105
Clinical Data Management-I Unit 6
1. Preposition of time : on, in, at, for, before, after, until, till, between, by,
upto.
E.g.: She was healthy till yesterday.
2. Preposition of place: to, at, from, away, on, onto, of, in, into, out, upon,
inside, within, by, over, above, on top of, behind, in front of, below,
beneath, across, through, all over, throughout, between, among.
E.g.: Where do you come from?
3. Preposition of method and manner: by, with
E.g.: The boys skipped going to school with audacity.
4. Preposition of reason and purpose: with, of, for,
E.g.: I rented a house for my holidays
5. Preposition of possession: of, with, by
E.g.: The tomb of Akbar is in Sikandarabad.
The document is normally reviewed by Database Programmer, Project Data
Manager, Biostatistician, Clinical Team Lead.
The edit check document may be modified depending upon the sponsor’s
requirement during the study. The final version of the document after
incorporation of the changes, if any, during the study, is approved before the
database lock. Comments are incorporated into the Draft version and
programming of the checks commence after sponsor approval. The checks
are put into production after validation.
Examples:
1. Inclusion Criteria- The subject should be of the age range 18-25 years.
An edit check will be written in the document and programmed for
subjects, whose age is not within the range 18-25 years.
A discrepancy will fire in the database for which a Data Clarification
Form (DCF) shall be sent.
Harstern Pubgya University Page No. 106
Clinical Data Management-I Unit 6
Some of the edit checks are repetitive in nature and hence can be
combined as per the event while programming. Generally, there are
around 100-250 checks, which can be written and programmed for a
study based on the number of pages and data points.
2. Please find below an example of an Edit Check document. The document
consists of edit checks for Demography Page and Concomitant
Medication Page. It briefly explains as to how the document should
appear.
Check No. CRF Item CRF
Page #Specific
visitCheck
Description DCF/Query text Discrepancy logic
Procedure Name Action
1 Birth date 1 Screening Birth date is missing
Birth date is missing. Please
provide the correct date.
Birth date should be provided.
DM_BLANK_01 DCF
2 Sex 1 Screening Sex is missingSex is missing. Please provide the sex.(M/F)
Response for sex should
be provided.DM_BLANK_02 DCF
3 Race 1 Screening Race is missingRace is missing. Please provide
the race.
Response for Race should be provided.
DM_BLANK_03 DCF
4 Start date 5 and 6 Screening
Name of the medication is
provided, however the start date is
missing.
Name of the Medication ( )
is provided, however the start date is
missing. Please provide the
date.
If Name of the
Medication is provided, Start date should be provided.
CR_BLANK_01 DCF
5 End date/Ongoing 5 and 6 Screening
Name of the medication is
provided, however the End date and
Ongoing is missing.
Name of the Medication ( )
is provided, however the
End date/Ongoing is missing. Please
provide the data.
If Name of the
medication is provided, the
End date /Ongoing should be provided.
CR_BLANK_02 DCF
Figure 6.1
Harstern Pubgya University Page No. 107
Clinical Data Management-I Unit 6
6.3.3 Universal Ruling:A Universal Ruling can be defined as a document, which allows the
authorized Data Management staff to make self-evident corrections to the
CRF and/or database without issuing a Data Clarification Form (DCF).
Example:For an indicator question, the response is missing, however, data is
provided on the CRF for the question.
Action: Check the [ ] Yes [ ] No
Universal ruling is referred to by different names as per preference such as
Global Ruling, self evident correction document etc. However, they all refer
to the same document, Universal Ruling.
6.4 Discrepancies6.4.1 How to access the Discrepancies:
CDMS Process Flow
Identify Patient (s) for Review
Identify Discrepancies
Review Discrepancies against Database and Case Report Form
Change Discrepancy Status
Create Data Clarification Forms
Harstern Pubgya University Page No. 108
Clinical Data Management-I Unit 6
Print Data Clarification Forms
Update the Database
6.4.2 Identifying the Patients for Reviewing: Patients for whom discrepancies need to be reviewed can be identified
in a number of ways based on present study needs.
All patients with unreviewed discrepancies.
Patients for a specific site with unreviewed discrepancies.
Patients with unreviewed discrepancies created during a certain time
period.
6.4.3 Types of DiscrepanciesA discrepancy is a variance between actual and expected responses as
defined in the Data Validation Document. When data fails a validation
check, the system generates and records a discrepancy in the CDMS.
There are four types of discrepancies, which are generated in CDMS, which
is system specific:
Univariate
Multivariate
Manual
Indicator
Univariate Discrepancy:Univariate discrepancy is generated when data is different from that defined
for the Data Collection Module (DCM) question (e.g., length or character in a
numeric field). Created during data entry, update or batch data load.
Univariate failure is errors that occur when data entered fail the predefined
Harstern Pubgya University Page No. 109
Clinical Data Management-I Unit 6
field parameters. A univariate error is system generated, and applies to only
one field.
Relative Pronouns: They are used for the nouns (antecedents) used
before them. They are used in the following:
1.1.1.1.1.1.1.1.1 Subject Object Possessive
For persons who, that whom/who, that whose
For things which, that which/that whose/of which
A relative pronoun must always be placed as near its antecedent as
possible. It must also agree with its antecedent in number, gender and
person
E.g.: This is the woman who stole the ring.
(ant.) (re. pro)
Generally, the relative pronoun in the objective case is omitted.
E.g.: The student (whom) you wanted to punish is absent today.
When the cursor is leaving a response field, the system validates the data
entered based, on the following criteria:
Mandatory value is entered
Length
Decimal precision
Data type (character vs number)
Dates are complete
Value is valid within a discrete value group, if applicable
Value is within predetermined range
Multivariate Discrepancy:Multivariate discrepancy is generated during Batch Validation when data
points are compared within/across visits does not meet criteria for a
validation procedure.
Harstern Pubgya University Page No. 110
Clinical Data Management-I Unit 6
Example;
Did the subject take the pills? [ ] Yes [ ] No
If yes, mention the number of pills taken_ __
Discrepancy message: No pills were taken, however the number of pills
taken is reported as ‘2’. Please check.
Manual Discrepancy:Any word that adds more meaning to the Noun is called an Adjective. It
qualifies a noun. Eg.: Ankur is a good player. The baby drank a little milk.
Correct Use of some adjectives:
a) Little (practically no chance) Deepak has little chance of being
elected.
A little (some chance) There is a little hope of his
success.
The little (whatever available) I shall give him the little money I
have.
b) Few (practically none) Few people are good.
A few (a small number) I have a few friends in my office.
The few (whatever available) I will pack the few things I have.
c) First (first in order) Yuri Gagarin was the first man to
go into space.
Foremost (leading, eminent) Einstein was the foremost scientist
of his day.
Indicator Discrepancy:Indicator Discrepancy is a check for a response to a question, which
determines a set of the remaining questions that require responses.
Harstern Pubgya University Page No. 111
Clinical Data Management-I Unit 6
If a follow-up question is either not collected when it should be, or collected
when it should not be, Clinical Data Management System (CDMS) creates
an indicator discrepancy during batch validation.
Example:
A) Did the subject take the pills? [ ] Yes [ ] No
If Yes, mention the number of pills taken _______
If No, check whether the subject is out of the study.
6.5 Discrepancy Review and Resolution Status CodesPlease find below list of Discrepancy Review Status, Discrepancy
Resolution Status, and Definition which can be viewed in Clinical Data
Management System (system specific).
Discrepancy Review status
Discrepancy Resolution status
Definition
Inv Review A discrepancy that is issued to a site per DCF.
Resolved Global/Universal Ruling To be used when a data change is made per Global/Universal Ruling and the discrepancy is no longer valid.
Resolved Data Entry Error To be used when a data change is made to fix a data entry error and the discrepancy is no longer valid.
Resolved No Action Required For any discrepancy that is valid, yet information is present on the CRF indicating the data is correct and a DCF should not be sent to the site.
Resolved Data Modified For a discrepancy attached to a DCF where the data was updated and the discrepancy is no longer valid.
Resolved Inv Confirmed For a discrepancy attached to a DCF where the data was confirmed and no updated were made.
Harstern Pubgya University Page No. 112
Clinical Data Management-I Unit 6
Resolved Inv-No Information To be used when a response on a DCF is not sufficient to close the discrepancy and the discrepancy is reissued on a new DCF.
Passive Review
A discrepancy that depends on the resolution of a different discrepancy for which a DCF is sent. This discrepancy is linked to the same DCF, but is not sent to the site.
On Hold A discrepancy that we are waiting for more information to process, e.g. a question is out to the client, waiting for another CRF page to come in.
Irresolvable* Inv-No Information To be used when a DCF with this discrepancy is never returned from the site and no changes are made to the database.
Unreviewed A discrepancy that has not been addressed by a CDC. This is the initial status of all discrepancies.
6.6 Audit TrailThe documentation that tracks the changes that have been made to
recorded data (e.g., case report forms) and/or databases. It is particularly
useful for reviewing the history of changes to data in a study.
6.7 Data Clarification FormsData Clarification Forms are queries, which are written to investigators for
clarifications, missing data etc. Edit check form the basis for writing a query.
The query text is written using discrepancies in the discrepancy module. An
Harstern Pubgya University Page No. 113
INV CORR Update made per DCF
UNIVERSAL/GLOBAL RULING Update made per Universal/Global Ruling
DATA ENTRY ERR Update made because of a DE error
CRA CORR Update made per site-DCF
SPONSOR CORR Updates made at the request of the sponsor without issuing a DCF
Clinical Data Management-I Unit 6
automated process of CDMS generates a form called Data Clarification
Form, containing the queries.
The DCFs are sent to the investigators on which they provide clarifications
on the form, sign with the date and send it back. These responses are
updated to the database appropriately.
Data Clarification Forms consist of:
Study Name
Investigator Name
Site/Patient ID
Patient Initials
Reviewer
Date: (Reviewer Date)
From (Clinical Research Organization (CRO) from where the Data
Clarification Form is generated).
Form Name/Visit Name
Page Number/Date
Question/Comments
Resolution
Investigator’s Signature
Date (Investigator signature date)
DCF ID
Few Examples of Data Clarification Forms (DCFs) are as follows:
Example 1On Medical History Page (), the response for Date of Birth is missing.
Please provide the missing response.
Example 2
Harstern Pubgya University Page No. 114
Clinical Data Management-I Unit 6
On Concommitant Medication Page (), the response for "Is the subject
currently taking medications" is yes. However, the data is missing. Please
clarify.
Example 3On Vital Signs Page (), the response for value of Diastolic Blood Pressure is
not within the normal range of 040 – 100 mmHg. Please provide the value
within 040-100 mmHg.
Example 4On Laboratory page (), the response for "Are there any positive results?" is
No. However ‘specify’ is provided. Please Clarify.
Example 5On Chest X-ray page (), "Are there abnormal findings on the Chest X-ray?"
is Yes. However, ‘specify’ is missing. Please provide the missing response.
Requery:Requery is sent when a resolution provided by the investigator requires
further clarification or if the resolution provided is not sufficient. For
example: DCF ID 548564
On Concommitant Medication Page (), the response for "Is the subject
currently taking medications?" is missing. Please provide the missing
response. If Yes, please provide the details.
Resolution: Please check as “Yes”.
Here in this case, the resolution has come back incomplete. The resolution
is Yes, however, details are not provided for the question as asked in the
query.
Hence, a requery will be sent asking for the details in reference to the first
query.
Harstern Pubgya University Page No. 115
Clinical Data Management-I Unit 6
As per DCF ID 548564, the response for "Is the subject currently taking
medications?" is Yes; however, the details are not provided. Please provide
the missing details.
Self Assessment Questions1. DCF stands for
a. Data Construction Form.
b. Data Creation Form.
c. Data Clarification Form.
d. Data Clearance Form.
2. Data validation is synonymous with
a. Data Cleaning.
b. Discrepancy management.
c. Both a and b.
d. None of these.
3. Discrepancies are primarily for any
a. Incorrect clinical data.
b. Illegible clinical data.
c. Inconsistent clinical data.
d. All of the above.
4. Discrepancy management primarily
a. Consists of electronic and manual checks on the data to assure the
validity and accuracy of the data.
b. Validate data against predetermined specifications.
c. Ensure consistency and data quality.
d. All of the above.
5. Which of the statement is false
a. Electronic and manual discrepancies will be stored within the
confines of a discrepancy management system.
Harstern Pubgya University Page No. 116
Clinical Data Management-I Unit 6
b. Data processing includes receiving, entering, cleaning and
transferring.
c. Univariate discrepancy appears post data entry.
d. Multivariate discrepancy appears post data entry.
6. DCF is raised to
a. Clinical site.
b. Sponsor.
c. Data Management Centre.
d. Pharmacovigilence.
7. Discrepancy management is primarily performed by
a. Clinical data Coordinator.
b. Data entry associate.
c. Database programmer.
d. Project Manager.
8. Universal rulings
a. Decrease total number of DCFs to the site.
b. Are self evident corrections.
c. Reduce time, labour and cost.
d. All of the above.
9. Discrepancy is a
a. Error.
b. Query.
c. Validation.
d. Resolution.
10. DCF consists of
a. Investigator name.
b. Investigator signature.
c. Resolution.
Harstern Pubgya University Page No. 117
Clinical Data Management-I Unit 6
d. All of the above.
6.8 SummaryDiscrepancy management is an integral part of Data Management. It
includes cleaning and reporting of data recorded at Clinical trials. It
undergoes intensive quality checks, performed as per the Data Validation
Plan to ensure very high and reliable data. Discrepancy Management forms
a foundation for Data Management in every project and are based on SOPs
followed for the data management operations.
6.9 Terminal Questions1. Define Discrepancy and briefly explain the types of discrepancies?
2. Briefly explain Data Validation Plan.
3. Briefly explain Data Clarification Form.
4. Briefly explain why a requery is sent.
5. Explain how an Edit Check Document is prepared and why?
6.10 Answers to SAQ and TQSAQ1-C 2-C 3-D 4-D 5-C 6-A 7-A 8-D 9-A 10-D
TQ1. Refer to Section 6.4
2. Refer to Section 6.3
3. Refer to Section 6.7
4. Refer to Section 6.7
Harstern Pubgya University Page No. 118
Clinical Data Management-I Unit 6
5. Refer to Section 6.7
The boy climbed when the bus stopped.
In the above example, there are two clauses. Only one of them is a