Analysis Data Reviewer’s Guide

XXX Company, Inc

YYYY Integrated Summary of Efficacy

Version 1

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 2 of 27

Analysis Data Reviewer’s Guide

Contents 1.  Introduction ........................................................................................................................................... 4 

1.1  Purpose .......................................................................................................................................... 4 

1.2  Acronyms ...................................................................................................................................... 4 

1.3  Study Data Standards and Dictionary Inventory ........................................................................... 4 

1.4  Source Data Used for Analysis Dataset Creation ......................................................................... 5 

2.  Protocol Description ............................................................................................................................. 6 

2.1  Protocol Number and Title ............................................................................................................ 6 

3.  Analysis Considerations Related to Multiple Analysis Datasets .......................................................... 7 

3.1  Comparison of SDTM and ADaM Content .................................................................................. 7 

3.2  Core Variables .............................................................................................................................. 7 

3.3  Treatment Variables ...................................................................................................................... 9 

3.4  Subject Issues that Require Special Analysis Rules ..................................................................... 9 

3.5  Use of Visit Windowing, Unscheduled Visits, and Record Selection .......................................... 9 

3.6  Imputation/Derivation Methods .................................................................................................. 10 

4.  Analysis Data Creation and Processing Issues ................................................................................... 11 

4.1  Split Datasets .............................................................................................................................. 11 

4.2  Data Dependencies ...................................................................................................................... 11 

4.3  Intermediate Datasets .................................................................................................................. 11 

4.4  Variable Conventions .................................................................................................................. 11 

4.5  Discrepancies with analysis performed on single studies ........................................................... 12 

5.  Analysis Dataset Descriptions ............................................................................................................ 13 

5.1  Overview ..................................................................................................................................... 13 

5.2  Analysis Datasets ........................................................................................................................ 13 

5.2.1 ADSL – Subject Level Analysis Dataset ................................................................................... 14 

5.2.2 ADEFBASE – Baseline Efficacy Endpoint AD ........................................................................ 14 

5.2.3 ADOA – Osteoarthritis History AD .......................................................................................... 15 

5.2.4 ADPAI – Questionnaire Pain Intensity AD ............................................................................... 15 

5.2.5 ADPAITTE – Time to Onset of Pain Relief AD ....................................................................... 15 

5.2.8 ADGIC – Questionnaire Global Intensity Change AD .............................................................. 16 

5.2.10 ADRESMED – Rescue Medications Usage AD ...................................................................... 17 

6.  Data Conformance Summary .............................................................................................................. 19 

6.2  Conformance Inputs .................................................................................................................... 19 

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 3 of 27

6.3  Issues Summary .......................................................................................................................... 20 

7.  Submission of Programs ..................................................................................................................... 21 

8.  Appendix ............................................................................................................................................. 23 

8.1  Appendix I: Process for deriving average weekly Rescue Medication Consumption ................ 24 

8.2  Appendix II: Proc MIXED option setting ................................................................................... 25 

8.3  Appendix III: Original eCRF Visit Nomenclature Standardization across studies performed in the SDTM datasets .................................................................................................................................. 26 

8.4  Appendix IV: Lookup dataset for Investigator Size Harmonization across studies .................... 27 

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 4 of 27

1. Introduction

1.1 Purpose

This document provides context for the analysis datasets and terminology that benefit from additional explanation beyond the Data Definition document (define.xml). In addition, this document provides a summary of ADaM conformance findings.

1.2 Acronyms

Acronym Translation

AUE Area Under the Effect

CDISC Clinical Data Interchange Standards Consortium

CGIC Clinical Global Impression of Change

ISS Integrated Summary of Safety

ISE Integrated Summary of Efficacy

NRS Numeric Rating Scale

OARSI Osteoarthritis Research Society International

SAP Statistical Analysis Plan

SDTM Study Data Tabulation Model

WOMAC Western Ontario and McMaster University Osteoarthritis Index

1.3 Study Data Standards and Dictionary Inventory

Standard or Dictionary Versions Used

SDTM SDTM v1.4/SDTM IG v3.2

ADaM ADaM Model Document 2.1

ADaM Implementation Guide v1.0

ADaM Basic Data Structure for Time-to-Event Analysis v1.0

ADaM Data Structures for Integration: General Considerations and Model for Integrated ADSL(IADSL) Draft June 2015

Controlled Terminology 2015-12-18

Data Definitions define.xml v2.0

Other standards (optional) None

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 5 of 27

1.4 Source Data Used for Analysis Dataset Creation

The ADaM datasets were derived from SDTM version 1.4 datasets. The datasets were derived from the final locked databases.

In addition to the clinical database, the source data include a lookup file that was used to harmonize the investigator name / country across studies in order to be able to classify subjects based on the size of the investigator site where subjects have been treated (see description of SITEGR4 and SITEGR5 variables in section 3.2 “Core Variables” and Appendix IV for more information).

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 6 of 27

2. Protocol Description

2.1 Protocol Number and Title

This ISE involves 3 studies. There were three double blind randomized trials: one phase III (study YYYY-001), one phase IIb (YYYY-002) and one phase II (YYYY-003). More details about the trial designs characteristics can be found in the ISE SAP and in the respective clinical trial protocols. Protocol Design in Relation to ADaM Concepts

Original treatment group assignment is stored in the SDTM DM domain. It was used in ADaM datasets to derive the treatment group (TRT01P, TRT01PN). These contain a single value per subject and were included on all datasets. The variables TRTP/TRTPN whenever used in other non-ADSL datasets contain the same values as TRT01P/TRT01PN.

The original study treatment groups (randomized arms) have been grouped into 5 treatment groups using the variable TRTIP (TRTIPN) available in the draft CDISC ADAM ADSL Integration Implementation Guidance (see reference in section 1.3):

- YYYY 10 and 20mg: from studies 001 and 002

- YYYY-001 40mg: from all studies

- YYYY-001 60mg: from study 001

- Placebo: from studies 002 and 003

- YYYY-001 IR 40mg: from studies 001 and 003

Similarly for actual treatment TRTIA (TRTIAN) was used.

Population flag variables were used to designate all randomized patients who received any amount of study medication (FASFL, Full Analysis Set) and all patients who received any amount of study

medication (SAFSFL, Safety Analysis Set).

Actual treatment is stored in TRT01A/TRT01AN derived from SDTM DM.ACTARM. All patients except seven received the planned treatment; seven patients did not receive any amount of study medication and therefore excluded from both FASFL and SAFFL

All outputs are based on the Full Analysis Set (FAS) population.

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 7 of 27

3. Analysis Considerations Related to Multiple Analysis Datasets

3.1 Comparison of SDTM and ADaM Content

Are data for screen failures, including data for run-in screening (for example, SDTM values of ARMCD=’SCRNFAIL’, or ‘NOTASSGN’) included in ADaM datasets?

No data for screen failures are used for analysis. Therefore, there are no records for screen failures in any analysis dataset

Are data taken from an ongoing study?

No. Data are not taken from an ongoing study

3.2 Core Variables

Core variables are those that are represented across all/most analysis datasets.

Variable Type Variables List

Subject Identifier USUBJID, STUDYID, SUBJID, and SITEID

Demographics AGE Age as collected

AAGE Re-derived age as per ISE/ISS SAP as years between date of birth and date of informed consent and rounded to the nearest integer

AGEGR1(N) Age in category

1=40-50; 2=51-64; 3=65-75; 4=>75 Years

SEX Original Sex

RACE(N) Original Race

ETHNIC Original Ethnicity

RACEGR1(N) Pooled Race

1=White vs 2=Asian vs 3=Non-White /Asian

SITEGR1 Investigator site country

Treatment and Periods

TRT01P(N) Original assigned treatment arm

ARM

TRT01A(N) Original actual treatment arm

TRTIP(N) Planned Dosing Group

1= YYYY-001 10 and 20mg

2= YYYY-001 40mg

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 8 of 27

Variable Type Variables List

3= YYYY-001 60mg

4=Placebo

5= YYYY-001 IR 40mg

TRTIA(N) Actual Dosing Group

TR01SDT Study drug medication start date

(TR01EDT) It is equivalent to TR01SDT since all studies drug administration are performed on day 1 only

Study Populations FASFL Y for subjects randomized and received any amount of study medication; N otherwise

SAFFL Y for subjects who received any amount of study medication; N otherwise

POOL1FL Y for studies contributing to the safety pool (ISS); Null otherwise

POOL2FL Y for studies contributing to the efficacy pool (ISE); Null otherwise

Other Baseline Characteristics used in the subgroup analyses

BMIGR1BL Baseline BMI (kg/m2) Group 1 - (BMIG1NBL)

1=Underweight; 2=Normal; 3=Overweight; 4=Obesity 1;

5=Obesity 2; 6=Morbid Obesity

OAGRADE(N) Kellgren-Lawrence Grade

From 0 to 4

…….

Other Baseline Characteristics that could be potentially used in the subgroup analysis

SITEGR2(N) Country/Region grouping

1=US

2=Canada

3=Australia/New Zealand

4=Hong Kong

5=Europe

SITEGR3(N) USA/Ex-USA Study Center

1=USA Study Centers vs 2=Ex-USA Study Centers

SITEGR4(N) It contains investigator name and country

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 9 of 27

Variable Type Variables List harmonized across all studies. This is needed in order to be able to group patients by size of the investigator site

SITEGR5(N) This is derived from SITEGR4 and it classifies subjects according to the size of the investigator site 1= Large Study Sites (>=19 randomized subjects) vs 2= Small Study Sites (<19 randomized subjects)

PAGR1BL Baseline Pain Categories on NRS (PAGR1NBL)

1=5-5.9; 2=6-6.9; 3=>=7

…………..

3.3 Treatment Variables

ARM versus TRTxxP

Are the values of ARM equivalent in meaning to values of TRTxxP?

Yes

ACTARM versus TRTxxA

If TRTxxA is used, then are the values of ACTARM equivalent in meaning to values of TRTxxA?

Yes

Use of ADaM Treatment Variables in Analysis

Are both planned and actual treatment variables used in analyses?

Yes. Planned treatment variable is used in all outputs based on the FAS population subset

3.4 Subject Issues that Require Special Analysis Rules

Seven Randomized patients did not receive the study drug injection and therefore excluded from all tables. Statistical analyses are completed in this ISE on the FAS Population (all randomized patients who received any amount of study medication).

3.5 Use of Visit Windowing, Unscheduled Visits, and Record Selection

Was windowing used in one or more analysis datasets?

Yes in ADPAI for calculating weekly means of the average daily pain intensity score (data were originally collected every day in the IVRS). All other endpoints in other ADaM datasets have

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 10 of 27

been analyzed and summarized according to the nominal visit as recorded in the CRF. Of note all visit nomenclature have been already standardized in the SDTM datasets (see the table in Appendix III for more details)

Were unscheduled visits used in any analyses?

Yes in ADGIC and ADWOMAC if the final visit occurred at a non-protocol specified time-point (i.e. for subjects withdrawn the study before the last protocol scheduled visit), a questionnaire may be reported under an unscheduled visit (i.e. VISIT is null in SDTM), therefore the unscheduled visit was “windowed” to the next planned visit and used in the analysis. For example if a subject withdrew before last planned visit i.e. week-24 in study 002 and 003, and last planned assessed (occurred) visit was week-16, the unscheduled, visit if occurred and after week-16, have been windowed to week-20 (the next planned visit) and therefore used in the analysis

3.6 Imputation/Derivation Methods

If date imputation was performed, were there rules that were used in multiple analysis datasets?

Date imputation was only needed in ADOA when the diagnosis date of Osteoarthritis index knee diagnosis was partial (i.e. either day or months was missing) in order to calculate the variable OADIAGY (Years since primary diagnosis of index knee OA). The following imputation method have been applied:

When only year was available, OADIAGY was computed as year of study drug administration – year of diagnosis

When only month and year was available, the day was imputed to 1 and OADIAGY was computed as (study drug administration - date of Osteoarthritis index knee diagnosis+1)/365.25

Otherwise if the diagnosis date was not partial the formula (study drug administration - date of Osteoarthritis index knee diagnosis+1)/365.25 was used

Additional Content of Interest

In ADWOMAC, whenever a missing item (unanswered question) could be imputed to the average of the other non-missing items (see section 5.2.7 for more details); DTYPE for the corresponding imputed item was set to AVERAGE.

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 11 of 27

4. Analysis Data Creation and Processing Issues

4.1 Split Datasets

There were no datasets that were split at the time of submission due to size constraints.

4.2 Data Dependencies

ADSL was used in the creation of all other analysis datasets. Prior to ADSL, ADEFBASE was created to derive the baseline “Weekly daily average pain” all baseline WOMAC sub-scale; all parameters were then transposed each of them represented by variable in ADSL. These baseline parameters are also re-calculated in ADPAI and ADWOMAC, therefore we verified that baseline calculation performed in ADSL matches baseline calculation performed in ADPAI and ADWOMAC.

ADPAI was used in the creation of ADTTEPAI and ADPAIAUE.

ADPAI, ADWOMAC and ADGIC were used in the creation of ADOMER.

No other data dependencies exist.

The following chart summarizes data dependencies.

4.3 Intermediate Datasets

ADEFBASE was created prior to ADSL. See prior section and section 5.2.2 for more details.

4.4 Variable Conventions

ANL01FL was used to flag records copied from SDTM but not used in any analysis i.e. if an observation is not flagged with ‘Y’ then it is not used in the analysis. The flag has been used in the following ADaM datasets:

- ADPAI and ADRESMED: any observation with a relative day not falling into any windowed visits (see table 6 in the ISE SAP), that is before day -7 and after day 168. Also for study 001 observations windowed after week 12 were not used given the fact in this study efficacy information were collected through week 12 only

ADSL

ADOA ADPAI ADGIC ADWOMAC

ADTTEPAI ADPAIAUE

ADRESMED

ADEFBASE

ADOMER

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 12 of 27

- ADPAI: observations occurred prior to end of wash-out period are also not flagged and not used in the analysis (i.e. in deriving weekly daily pain average)

- ADWOMAC and ADGIC: for study 001 questionnaires collected during visits week 1 and week 2 were not flagged and not used in the analysis since study 001 is the only study in the ISE having such a time-point

Moreover all observations with a missing value (AVAL=.) are also not flagged.

4.5 Discrepancies with analysis performed on single studies

The following tables list discrepancies with regards with outputs available in the single study clinical study reports and reasons.

ISE ADAM

Study

CSR Table Nr

Discrepancy and Reason

ADPAI

001

14.2.1.x and 14.2.2.x (all pain endpoint summaries)

The visit / week windowing in 001 final analysis was different i.e. week 1 was from day 1 to day 8 whereas in ISE (see SAP section10.1.1) we applied the windowing from day 1 to day 7 as we did for the analysis of study 002 and 003.

ADRESMED

001

14.2.1.10

Some of the special cases handled in ISE, and also during the analysis of the pivotal phase III study (003), were not handled same way during 001 final analysis.

See appendix I and ADRESMED section for more detailed conventions used for the derivation of rescue medications and detailed applied algorithm.

ADRESMED

002

14.2.7

The table in the CSR differs of few cases because in 002 final analysis we were only including records from the IVRS and records from the concomitant medications page if concomitant medication start date was on or after study drug administration. In the ISE, and also during the analysis of the pivotal phase III study (003), we updated the algorithm to use also concomitant medications reported in the week prior to study drug administration (i.e. CMSTDY>=-8 or (CMSTDY eq . and CMENDY >= -7)).

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 13 of 27

5. Analysis Dataset Descriptions

5.1 Overview

Do the analysis datasets support all protocol- and statistical analysis plan-specified objectives? Yes, only exposure table was created directly from the SDTM EX dataset and getting ADSL variables in the output program.

5.2 Analysis Datasets

Dataset – Dataset Label C

lass

Eff

icac

y

Saf

ety

Bas

elin

e or

ot

her

su

bje

ct

char

acte

rist

ics

PK

/PD

Pri

mar

y O

bje

ctiv

e

Structure

ADSL: Subject-Level AD

ADSL X X One record per subject

ADEFBASE: Baseline Efficacy Endpoint AD

BDS X One record per subject per parameter (baseline efficacy endpoint)

ADOA: Osteoarthritis History AD

Other X One record per subject

ADPAI: Questionnaire Pain Intensity AD

BDS X X

One record per subject per parameter (pain endpoint) per time-point (primary endpoint)

ADPAITTE: Time to Onset of Pain Relief AD

BDS (TTE)

X One record per subject per parameter (pain time to event endpoints)

ADPAIAUE: Questionnaire Pain Intensity AUC AD

BDS X

One record per subject per parameter (Pain Area Under the Effect) per aggregated time-point

ADWOMAC: Questionnaire WOMAC AD

BDS X

One record per subject per parameter (WOMAC questionnaire section) per time-point

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 14 of 27

Dataset – Dataset Label C

lass

Eff

icac

y

Saf

ety

Bas

elin

e or

ot

her

su

bje

ct

char

acte

rist

ics

PK

/PD

Pri

mar

y O

bje

ctiv

e

Structure

ADGIC: Questionnaire Global Intensity Change AD

BDS X

One record per subject per parameter (investigator and subject assessment) per time-point

ADOMER: OMERACT-OARSI Responder AD

BDS X

One record per subject per parameter (GIC, WOMAC and Pain average and derived OMERACT responder) per time-point

ADRESMED: Rescue Medications Usage AD

BDS X

One record per subject per parameter (daily rescue medication use and daily average) per time-point

5.2.1 ADSL – Subject Level Analysis Dataset

In addition to supporting all analyses, ADSL contains variables to support baseline characteristics and disposition analyses. Furthermore all subgroup baseline variables and baseline efficacy values for weekly mean average daily pain and all WOMAC sub-scales are also available (see ADEFBASE).

The population indicator variables are also defined in ADSL and copied into other analysis datasets together with key baseline variables. All subjects in DM were included in ADSL.

ADSL contains also new variables proposed in the new draft IG ADaM Data Structures for Integration, in particular variables for pooling group of study treatment doses (see section 3.2 for more details).

We develop one ADSL program to support both ISS and ISE, thus the structure of the two ADSL is same and therefore some variables might be not relevant for ISE (i.e. the safety population flag). The ADSL for the ISE contains only subjects randomized in study 001, 002 and 003, the three studies where efficacy endpoints were evaluated.

5.2.2 ADEFBASE – Baseline Efficacy Endpoint AD

The dataset is created before ADSL to be able to feed into ADSL all baseline efficacy endpoints related to Pain and WOMAC Questionnaires. The baseline weekly mean average daily pain and all WOMAC sub-

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 15 of 27

scale are represented by a parameter. Each of these parameters is then transposed into a horizontal structure where each parameter becomes a variable in ADSL.

5.2.3 ADOA – Osteoarthritis History AD

ADOA contains one record per subject with variables containing details related to the Osteoarthritis diagnosis information such as date of diagnosis, index knee and other Osteoarthritis related information such as surgeries.

5.2.4 ADPAI – Questionnaire Pain Intensity AD

The ADPAI contains the primary efficacy endpoint, that is the Change from Baseline in Weekly Mean of the Average Pain Intensity Score (CHG variable where PARAMCD=KNPAWK).

ADPAI contains one record per subject per daily pain assessment per windowed week. Weeks are derived (windowed) into analysis visits (represented by AVISIT and AVISITN) based on the date of the pain assessment and the date of randomization.

CRIT1/CRIT1FL, CRIT2/CRIT2FL, CRIT3/CRIT3FL were also created to flag assessment time-points where a 20%, 30%, 50% pain decrease from baseline was observed.

Records where ANL01FL=’Y’ are the ones that were used for analysis (see more details in section 4.4).

Change from baseline (CHG) is used as independent variable in all models.

The baseline value (BASE) is used as a covariate in the analysis together with treatment, site (SITEID) and week (AVISITN); the interaction site*week is also tested.

Source data can be traced back to the SDTM.QS domain from each study using USUBJID and QSSEQ.

Parameter Code / Name Description Usage / Additional Details

KNPA0101

Average Daily Pain Inten. Score

Daily assessment of the pain intensity

This is copied from SDTM.QS where QSCAT=Knee Pain Intensity

5.2.5 ADPAITTE – Time to Onset of Pain Relief AD

ADPAITTE is an analysis dataset following the ADaM TTE Structure (that is based on the BDS structure) supporting Pain time to event endpoint.

It is derived from ADPAI by selecting all evaluable (ANL01FL=”Y”) post-baseline (AVISITN> 2.1) individual daily intensity pain assessment (PARAMCD= "KNPA0101" and DTYPE=" ").

Source data can be traced back to the source analysis dataset from each study using USUBJID and SRCDOM (=ADPAI) and SRCSEQ variables

Parameter Code / Name Description Usage

TTOPR This is defined as the time from randomization to the first time

It is used in the survival model (PROC LIFETEST) with

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 16 of 27

Parameter Code / Name Description Usage

Time to Onset of Pain Relief (Days)

where a >30% pain decrease (CNSR=0) was observed (ADPAI.CRIT2FL=’Y’). Subjects without a >30% decrease were censored (CNSR=1) to the last daily pain intensity assessment

AVAL representing the ‘time-to’ information and CNSR the censoring flag (1=censor, 0= event).

DURPR

Duration of Pain Relief (Days)

This is defined as the time from pain relief (ADT where PARAMCD=TTOPR)) to the first time where a >30% pain decrease (CNSR=0) was NOT observed (ADPAI.CRIT2FL=’N’). Subjects without any decrease <=30% were censored to the last daily pain intensity assessment.

It is used in the survival model (PROC LIFETEST) with AVAL representing the ‘time-to’ information and CNSR the censoring flag (1=censor, 0= event).

The parameter is only applicable to subject where Pain Relief was observed (PARAMCD=TTOPR and CNSR=0)

WOM31TAV

WOMAC Total – Average

Derived parameter that reflects the average of all WOMAC sub-score i.e. sum of all subscale / 3. If any of the averaged subscale is missing, then the ‘Total-Average’ score is set to missing.

WOM31TSU (WOMAC Total - Sum) is also provided but it is not used in any calculation.

5.2.8 ADGIC – Questionnaire Global Intensity Change AD

The ADGIC contains the clinical and patient assessment of the Global Impression of Change (GIC). Given the fact the questionnaire assesses change (improvement), the questionnaire was assessed only after study drug administration therefore, although the dataset is using a BDS structure, it does not contain the CHG variable (and PCHG) and baseline records are not identified (it is not applicable).

The AVALCAT1 variable has been also added to further classify the Patient and Clinical GIC into the following three categories:

- A – Improved

- B – Minimal – No Change

- C – Worse

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 17 of 27

The proportion of “Improved” was analyzed using a LOGISTIC regression.

As mentioned in section 3.5, if the final visit occurred at a non-protocol specified time-point (i.e. for subjects withdrawn the study before the last protocol scheduled visit), a questionnaire could be reported under an unscheduled visit (i.e. VISIT is null), therefore the unscheduled visit was “windowed” to the next planned visit (AVISIT not null).

Records where ANL01FL=’Y’ are the ones that were used for the analysis (see more details in section 4.4).

Source data can be traced back to the SDTM.QS domain from each study using USUBJID and QSSEQ.

Parameter Code / Name Description Usage

CGIC0101

Clinical Global Impression Change

Original item score of Clinical judgment of Global Impression Change

This is copied from SDTM.QS where QSCAT= Clinical Global Impression Change

PGIC0101

Patient Global Impression Change

Original item score of Patient judgment of Global Impression Change

This is copied from SDTM.QS where QSCAT= Patient Global Impression Change

5.2.10 ADRESMED – Rescue Medications Usage AD

ADRESMED contains summary information about the usage of rescue medications.

Rescue medications were collected either in the IVRS as daily consumption (number of tables used by the patient; YR SDTM dataset) or in the concomitant medications page (CM SDTM dataset where CMDECOD=’Paracetamol’) when the patient was not able to report the information in the IVRS system; in such a case the information was reported to the investigator at the next visit and the rescue medications consumption was entered in the concomitant medications page with concomitant medication start-end date, dosage, unit and frequency.

In order to be able to pull the two information and derive the overall weekly rescue medications consumption, the rescue medications collected in the concomitant medication, as described above, was ‘converted’ into a number of tablets consumed using the assumption that 500 mg of Paracetamol are equivalent to 500mg. More details on other assumption and methods used to make this derivation are available in Appendix I.

ADRESMED is therefore one record per subject per day. Each single day is then assigned / windowed to a week (AVISIT/AVISITN). See the table below for more details about parameters available in ADRESMED.

Source data can be traced back to the source dataset domain from each study using USUBJID and QSSEQ variables (only for rescue medications collected in the IVRS).

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 18 of 27

Parameter Code / Name Description Usage

RMEDIVRS

Rescue Medications from IVRS (tablets)

Daily number of rescue medication tablets consumption collected in the IVRS (SDTM dataset YR)

Combined with RMEDCM to calculate the weekly average rescued medications consumption

RMEDCM

Rescue Medications from CM (tablets)

Derived daily number of rescue medication tablets consumption collected in the Concomitant Page (SDTM dataset CM where CMDECOD=”Paracetamol”). The parameter is derived using the assumptions described in Appendix I “transformed” into several records each one representing a day where rescue medications were used

Combined with RMEDIVRS to calculate the weekly average rescued medications consumption

RMEDAVG

Average Rescue Medications (tablets)

Derived weekly average rescue medications consumption combining daily consumption from the two different sources as described in the parameters RMEDIVRS and RMEDCM

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 19 of 27

6. Data Conformance Summary

6.2 Conformance Inputs

Were the analysis datasets evaluated for conformance with CDISC ADaM Validation Checks? Yes

If yes:

o Version of CDISC ADaM Validation Checks: 1.0

o Specify software used: Pinnacle 21 Community Version: 2.1.1

Were the ADaM datasets evaluated in relation to define.xml?

Yes see opencdisc-report-data

Was define.xml evaluated?

Yes opencdisc-report-definexml

6.3 Issues Summary

Dataset(s) Diagnostic Message and/or Check ID Severity Count and/or Issue Rate

Explanation

ADSL RACE value not found in 'Race' extensible codelist

Warning 7 MULTIPLE and OTHER are terminology allowed as per SDTM IG

……

7. Submission of Programs All programs used to derive the ADaM datasets are part of the submission, and available upon request. Output programs are also part of the submission, and available upon request.

All submitted programs will execute on a PC environment running Windows and SAS version 9.2 or later.

The intent is to give the reviewer the possibility to review how derivations has been done and models used in the analysis; however the programs have some macro and environment dependencies so they are not executable without making ad-hoc modifications.

Program Name Output Inputs+ Macro Used*

adefbase.sas adefbase

dm

qs (where qstestcd in ('KNPA0101') and (qsdy le -1 and qsdy ge -7)) or (index(qstestcd,'KOOS') and QSBLFL eq 'Y') or (qscat eq 'SF-12 Health Survey v2.0' and qsblfl eq 'Y') or (where QSCAT="Western Ontario and McMaster (WOMAC) Osteoarthritis Index v3.1')

ds (where dsdecod eq 'END OF WASH-OUT')

adsl.sas adsl

adefbase, dm, suppdm, ds, vs, qs, mg, zg, ex, suppex, zr, suppzr, sv, ae, cm, lb, zs, yg, suppyg

adpai.sas^ Adpai

adsl

qs (where qstestcd=”KNPA0101”)

adttepai.sas adpaitte

adsl

adpai (where paramcd = "KNPA0101" and dtype = " " and upcase(avisit) ne "BASELINE" and anl01fl = "Y")

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 22 of 27

Program Name Output Inputs+ Macro Used*

…..

* All programs make use of macro for setting up the environment (i.e. %SETUP), to generate XPT file (%_U_XPT) and to track program execution (%TRACK)

^ Primary Efficacy Endpoint

+ All output programs, tables and figures, make use of full analysis set population only (FASFL=’Y’)

See Appendix II for list of models and options used for each table generated containing inferential statistics.

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 23 of 27

8. Appendix

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 24 of 27

8.1 Appendix I: Process for deriving average weekly Rescue Medication Consumption

This document contains specific details used for the rescue medications derivation. It includes a step by step algorithm, including how to integrate rescue medications collected in the standard “concomitant medications” form with rescue medications collected in the IVRS system (daily consumption). The file is in the form of a memorandum agreed at the time of final analysis of study 003 and it has been applied the same way in this ISE when rescue medications consumption from study 001 and 002 were also integrated. The file is available and included in the package and referenced by the define.xml.

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 25 of 27

8.2 Appendix II: Proc MIXED option setting

The following table summarizes the use of TYPE= option used in the PROC MIXED for every table where a PROC MIXED was used. More details about PROC MIXED model used can be also found in the SAP.

Root Name  (Program Name) 

Output ID  Numbering  PROC MIXED TYPE = 

GIC  T‐GIC‐PATIENT‐POOL2‐FAS  Table 14.2.4.1.1.B  UN 

GIC  T‐GIC‐INV‐POOL2‐FAS  Table 14.2.4.2.1.B  CS 

PAIN  T‐PAIN‐MALE‐POOL2‐FAS  Table 14.2.1.1.1.B  TOEP 

PAIN  T‐PAIN‐FEMALES‐POOL2‐FAS  Table 14.2.1.1.2.B  TOEP 

PAIN  T‐PAIN‐AGE4050‐POOL2‐FAS  Table 14.2.1.2.1.B  TOEP 

…..  …..  …..  ….. 

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 26 of 27

8.3 Appendix III: Original eCRF Visit Nomenclature Standardization across studies performed in the SDTM datasets

VISITNUM

VISIT 001 002* 003* 004 005 006* VISIT (Original as in CRF if different)

1 Screening √ √ √ √ √ √ Day-21 to -1/SCR in 001

1.1 Day-1 √ Day-14 To -1/Screening in 002

2 Day 1 Baseline^ √ √ √ √ √ √

2.1 Day 2^ √ √ √

2.2 Day 3 √

2.3 Day 4 √

2.4 Day 5 √

2.5 Week 1^ √ Week 1 (Day 8) in 001

2.6 Day 14 √

2.7 Week 2 √

2.7 Week 2^ √ Week 2 (Day 15) in 001

2.8 Week 3 √

3 Week 4^ √ √ √ √ Week 4 (Day 29) in 001

3.1 Week 5 √

3.11 Day 42 √

3.2 Week 6 √ √ √

4 Week 8^ √ √ √ Week 8 (Day 57)

5 Week 12^ √ √ √ √ √ Week 12 (Day 85)

Final Visit in 005

6 Week 16^ √ √ √ √ Final Visit in 005

7 Week 20^ √ √ √ √ Final Visit in 005

8 Week 24^ √ √

^ Efficacy (Questionnaire) assessments; Week 1, Week2 assessment in 001 only were not included in the by time-point analysis

XXX Company YYYY NDA ISE Analysis Data Reviewer’s Guide

PDF created 2017-02-07 Page 27 of 27

8.4 Appendix IV: Lookup dataset for Investigator Size Harmonization across studies

The lookup dataset siteinfo has been used to harmonize site id across studies so that site can be classified according to their size as number of randomized patients contributing to the entire ISS. The dataset is available in the misc folder.