Creation of BDS Datasets- Day 05 of Onco-ADaM Project

Introduction

The Basic Data Structure (BDS) is a fundamental element within the Analysis Data Model (ADaM), which is a significant component of the Clinical Data Interchange Standards Consortium (CDISC).

BDS is the standard domain structure in ADaM, designed to facilitate the creation of analysis-ready datasets.

This structure allows for one or more records per subject per analysis parameter per analysis time point.

All the BDS datasets usually follow same variables, with a sprinkle of dataset specific suffix's. Examples are ADVS, ADLB, ADEG.

Some of the common variables in BDS datasets

Common variables in BDS datasets:

• STUDYID
• USUBJID
• SUBJID
• SITEID
• TRTP
• TRTPN
• TRTA
• TRTAN
• SCRNFL
• ENRLFL
• ITTFL
• SAFFL
• PKFL
• ADT
• ADTM
• ATM
• ADY
• VISIT
• VISITNUM
• AVISIT
• AVISITN
• PARAM
• PARAMCD
• AVAL
• DTYPE
• ABLFL
• BASE
• CHG
• EPOCH
• APHASE
• ANL01FL
• —ORRES
• —STRESC
• —STRESN
• —STRESU
• —SEQ
• —DTC
• —DY
• AVALCAT01

variables in ADLB dataset:

• LBSTAT
• LBNAM
• LBSPEC
• LBNRIND
• LBSTNRLO
• LBSTNRHI
• LBORNRLO
• LBORNRHI
• SHIFT1
• SHIFT1N

Variables in ADEG dataset:

• CHGCAT1
• CHGCA1N
• PCHG
• RHYOTH

Steps to create a BDS Dataset

1. Creating library.

2. Merging with ADSL dataset

   1. Sorting the LB dataset by USUBJID
   2. Merging with ADSL with full join
      • Creating all the renaming variables

3. Creation derived variables

   1. Analysis date time variable
   2. Defining Analysis visit variable
   3. Addressing parameter variables

4. Baseline concept

   1. Separating pre base variables (TRTSDT>ADT)

   2. Creating baseline flag and base variable. (Use LAST.PARAMCD statement sorted by USUBJID PARAMCD ADT ADTM)

   3. Sort the datasets both the base and master dataset by USUBJID PARAMCD —SEQ variables.

   4. Merging with master dataset

      1. Merge for Analysis baseline flag only (ABLFL)
         • keeping only USUBJID PARAMCD —SEQ from base dataset.
         • use only left join.
      2. Merge for Base variable only (BASE)
         • Keep only USUBJID PARAMCD.
         • Use left join.
      3. Creating the change (CHG) and percentage change (PCHG) variable

5. Creation of Derivation variable

   Usually depends by the requirement. for me in ADVS it was Last Observation Carry Forward (LOCF) and in ADLB it was Minimum observation, in ADEG it was Worst case (WOCF).

   Below is the example for ADLB DYPE derivation

   1. Creating a dataset by sorting the master dataset by PARAMCD and Date and time variable (USUBJID PARAMCD ADT ADTM).
   2. Defining the variables
   3. Append the above dataset with the master dataset.

6. Now create the remaining variables, here the variables were:

   1. Analysis Phase (APHASE)- in which phase the vitals are done (for e.g.,- Pre-tx, On-tx or Post-tx).
   2. Analysis Record Flag 01 (ANl01FL).

7. Sorting by Key variable

8. Creating final dataset

Note:

Here I have mentioned only ADVS dataset, but the concept for making BDS datasets is almost same just some dataset specific variable differs

Important

If this the first post you read, then i highly recommend to read my last blogs on this series, link is below down:

1. What is ADaM- Day:01 of Onco-ADaM Project
2. Creation of ADSL- Day: 02 of Onco-ADaM Project
3. Creation of OCCDS-ADCM Dataset - Day:03 of Onco-ADaM Project
4. Creation of OCCDS-ADAE Dataset- Day 04 of Onco-ADaM Project

Thank You for Visiting :)