Tips and Techniques for Improving the Performance of Validation Procedures in Oracle Clinical

Tips and Techniques for Improving the Performance of Validation Procedures in Oracle Clinical

Steve Rifkin, Lead Technical Analyst, Clinical Data Management & EDCLife Sciences Business Unit

facebook.com/perficient twitter.com/perficient_LSlinkedin.com/company/perficient

Welcome & Introduction

Steve RifkinLead Technical AnalystClinical Data Management & EDCLife Sciences, Perficient

• Extensive Oracle Clinical/RDC experience– 18+ years of experience with Oracle Clinical/RDC

– Member of the team that implements, supports,

enhances and integrates Oracle Clinical/RDC solutions

– 50+ implementations and integrations

Perficient is a leading information technology consulting firm serving clients throughout

North America and Europe.

We help clients implement business-driven technology solutions that integrate business

processes, improve worker productivity, increase customer loyalty and create a more agile

enterprise to better respond to new business opportunities.

About Perficient

• Founded in 1997

• Public, NASDAQ: PRFT

• 2013 revenue ~$373 million

• Major market locations throughout North America• Atlanta, Boston, Charlotte, Chicago, Cincinnati, Cleveland,

Columbus, Dallas, Denver, Detroit, Fairfax, Houston, Indianapolis, Los Angeles, Minneapolis, New Orleans, New York City, Northern California, Philadelphia, Southern California, St. Louis, Toronto and Washington, D.C.

• Global delivery centers in China, Europe and India

• >2,100 colleagues

• Dedicated solution practices

• ~85% repeat business rate

• Alliance partnerships with major technology vendors

• Multiple vendor/industry technology and growth awards

Perficient Profile

BUSINESS SOLUTIONSBusiness IntelligenceBusiness Process ManagementCustomer Experience and CRMEnterprise Performance ManagementEnterprise Resource PlanningExperience Design (XD)Management Consulting

TECHNOLOGY SOLUTIONSBusiness Integration/SOACloud ServicesCommerceContent ManagementCustom Application DevelopmentEducationInformation ManagementMobile PlatformsPlatform IntegrationPortal & Social

Our Solutions Expertise

Life SciencesPr

actic

es /

Solu

tions

ImplementationMigration

Integration

ValidationConsultingUpgrades

Managed ServicesApplication Development

Private Cloud Hosting

Application SupportSub-licensingStudy Setup

Services

Deep Clinical and Pharmacovigilance Applications Expertise

Clinical TrialManagement

Clinical Trial Planning and BudgetingOracle ClearTrial

CTMSOracle Siebel CTMS / ASCEND

Mobile CRA

Clinical Data Management & Electronic Data Capture

CDMSOracle Clinical

Electronic Data CaptureOracle Remote Data Capture

Oracle InForm

Medical CodingOracle Thesaurus Management System

Safety &Pharmacovigilance

Adverse Event ReportingOracle Argus Safety Suite

Oracle AERS / Empirica TraceAxway Synchrony Gateway

Signal ManagementOracle Empirica Signal/Topics

Medical CodingOracle Thesaurus Management System

Clinical DataWarehousing & Analytics

Clinical Data WarehousingOracle Life Sciences Data Hub

Clinical Data AnalyticsOracle Clinical Development Analytics

JReview

Data Review and CleansingOracle Data Management Workbench

Clients

Agenda

• What is an Oracle Clinical Procedure• Review of PL/SQL Concepts

– Cursors and Package Structure• Oracle Clinical Cursors• Structure of Oracle Clinical MAIN Procedure• Retrieval Techniques

– Correlations– Qualifying Expressions– WHERE clause extension– Visit Limitations

• Performance Considerations

What is an Oracle Clinical Procedure?

Definition => Validation Procs => Procedures

Definition => Validation Procs => Procedures

It’s a program!


• Procedure templates are completed • Procedure is “generated” and


A PL/SQL Program is created!

To get the most out of OC procedures, you need to understand the structure of the program you create.

Review of PL/SQL Concepts

• Detailed knowledge of PL/SQL syntax is not required for understanding the overall program structure.– However, knowledge of PL/SQL cursors and of the

PL/SQL Package structure is required.

PL/SQL Cursors

In SQL*Plus, a “select statement” retrieves an entire set of records:

SQL> select patient, patient_position_id from patient_positions; PATIENT PATIENT_POSITION_ID---------- -------------------100 5001101 5101102 5201103 5301104 5401105 5501106 5601107 5701108 5801109 5901

Cursors also retrieve of a set of records; but unlike SQL*Plus, cursors allow access to one record at a time for procedural operations on data within the record.

• Cursors are objects in a PL/SQL program• To use, cursors must be

1. Declared (to define the set of results)2. Opened (to initialize)3. Fetched (to retrieve one row for operations)4. Closed (to release)

PL/SQL Cursors

• Cursor declaration defines the set of records to be retrieved from the database:

• Like a “file” a cursor must be opened before it can be used:

CURSOR cHeight IS SELECT height, height_unitFROM height_table WHERE patient=‘100’ ORDER BY visit_number;

OPEN cHeight;

PL/SQL Cursors

PL/SQL Cursors

• A “fetch” on the cursor retrieves a single row from the set of selected records and places the result into program variables:

FETCH cHeight INTO height_var, height_unit_var;

CLOSE cHeight;

• Cursors are closed to release memory and perform internal cleanup:

PL/SQL Cursors

• Cursors are often fetched in a “loop” which processes the record:

LOOP FETCH cHeight INTO height_var, height_unit_var ; IF cHeight%NOTFOUND THEN

EXIT LOOP; ELSE

<process current height_var, height_unit_var> END IF;

END LOOP;

PAT_LOOP FETCH cPatient_data into patient_var;

IF cPatient_data%FOUND THEN

ELSE EXIT PAT_LOOP;

END IF; END PAT_LOOP;

HEIGHT_LOOP FETCH cHeight INTO height_var, height_unit_var ;

IF cHeight%FOUND THEN<process patient_var, height_var, height_unit_var>

ELSE EXIT HEIGHT_LOOP;

END IF; END HEIGHT_LOOP

Nesting PL/SQL Cursors

Cursors used by Oracle Clinical

• Patient Information Cursor– Retrieves a record with all the information recorded for a

single patient• DCM Cursor(s)

– A DCM cursor for each “alias” listed on the procedure question group form

– Retrieves the DCM responses (for the questions on the procedure question screen) and DCM header information

Fields Retrieved by the Patient Cursor

• V_CLINICAL_STUDY_VERSION_ID• V_DATA_MODIFIED_FLAG• V_PROCEDURE_ID• V_PROCEDURE_VERSION_SN• V_PROCEDURE_TYPE_CODE• V_USERNAME• V_DEBUG• V_LAST_BATCH_TS• V_CURRENT_LOCATION• V_MODE• V_LAB_DEPENDENT_FLAG• REPORTED_SEX• REPORTED_BIRTH_DATE• PATIENT_POSITION_ID

• CLINICAL_STUDY_ID• REPORTED_DEATH_DATE• PATIENT• INVESTIGATOR_ID• SITE_ID• EARLY_TERMINATION_FLAG• PATIENT_ENROLLMENT_DATE• CLINICAL_SUBJECT_ID• INCLUSION_EXCLUSION_DATE• REPORTED_PATIENT_REFERENCE• REPORTED_INITIALS• REPORTED_DATE_LAST_PREGNACY• FIRST_SCREENING_DATE• TERMINATION_DATE

Fields Retrieved by a DCM Cursor

• ACTUAL_EVENT_ID• CLIN_PLAN_EVE_ID• CLIN_PLAN_EVE_NAME• DCM_DATE• DCM_ID• DCM_SUBSET_SN• DCM_TIME• INVESTIGATOR_ID• LAB• LAB_ID• LAB_RANGE_SUBSET_NUM

• QUALIFYING_VALUE• RECEIVED_DCM_ENTRY_TS• RECEIVED_DCM_ID• REPEAT_SN• SITE_ID• SUBEVENT_NUMBER• VISIT_NUMBER

Using Patient Cursor Fields

• To use a value for a “V_” field, prefix the field name with “rxcpdstd.”– e.g., rxcpdstd.v_data_modified_flag

• To use all other variables, prefix the name with “rxcpdstd.patients_rec.” – e.g., rxcpdstd.patients_rec.reported_birth_date

Fields retrieved in the patient information cursor can be used to perform tests or calculations in your procedures or can be

used as arguments to custom functions you may write.

Structure of PL/SQL Packages

• Generation of an Oracle Clinical procedure results in a PL/SQL package

• PL/SQL packages contain one or more functions and/or procedures and have two sections– Specification

• Lists all procedures and functions in the package• Declares all cursors and variables to be available to all the

functions and procedures (i.e. the “common” area)

– Body • Contains all the procedural code for the procedures and

functions listed in the specification

Package Specification in Oracle Clinical

CREATE PACKAGE rxcpd_xxx_xx as CURSOR 1 is . . . . CURSOR 2 is . . . .

CURSOR n is . . . .

PROCEDURE main (… … …); PROCEDURE insert_discrepancy(… … …); PROCEDURE exception_handling (… … …); END rxcpd_xxx_xx;

. . .

Package is named with internal id and version numbers

Definition and number of cursors declared depends on the number of “aliases” on the procedure question group form

Three PL/SQL procedures are part of each Oracle Clinical package

Body for a Package in Oracle Clinical

INSERT_DISCREPANCY and EXCEPTION_HANDLING procedures call Oracle Clinical built-in packages and cannot be readily changed

CREATE PACKAGE BODY rxcpd_xxx_xx as PROCEDURE main (… … …); <Program Code >END main; PROCEDURE insert_discrepancy(… … …); < Program Code>END insert_discrepancy; PROCEDURE exception_handling (… … …); <Program Code>END exception_handling; END rxcpd_xxx_xx;

Program code in the MAINprocedure depends on options chosen on the procedure definition forms

• MAIN procedure is a series of nested loops• For a simple procedure, outermost loop fetches the

patient information– Next loop fetches DCM information as defined by the

first procedure question group definition• After fetching the innermost loop, the expression

details are evaluated

Simplified Structure of a MAIN Procedure

LOOP<Patient> Fetch Patient cursor

LOOP <DCM1> Fetch DCM1 Cursor

Execute Details

End <DCM1> LoopEnd <Patient> Loop

Simplified Structure of a MAIN Procedure …

One DCM Procedure Question Group

LOOP<Patient> Fetch Patient cursor


Execute Details

End <DCM1> LoopEnd <Patient> Loop


End <DCM2> Loop

Two DCM Procedure Question Groups

Simplified Structure of a MAIN Procedure

Some Techniques to Limit Retrievals

• Correlate inner loop with a more outer loop– On event– On qualifying question value– On question values

• Qualifying expressions• WHERE clause extension• Limit visits retrieved by the cursor

Correlations

• Can correlate on events– Inner cursor retrieves only a visit retrieved by the outer correlated cursor

• Can correlate on qualifying questions– Inner cursor only retrieves DCMs which have the same value for the DCM

qualifying question

• Can correlate on question values– Inner cursor fetches only if a question’s value is equal to a question’s value

in the outer cursor

Correlation limits the fetches for inner cursors based on data obtained by a fetch in a more outer cursor.

Fetch data for Patient 100Fetch SYSTOLIC_BP from VITALS DCM for Visit 1, Pt 100

Correlation on Event

Retrieval with No Correlation

Fetch HYPERTENSIVE_YN from HYPERTENSION DCM for Visit 1, Pt 100 Test

Fetch SYSTOLIC_BP from VITALS DCM for Visit 2, Pt 100Fetch HYPERTENSIVE_YN from HYPERTENSION DCM for Visit 1, Pt 100

Test



Fetch data for Patient 100Fetch SYSTOLIC_BP from VITALS DCM for Visit 1, Pt 100


Retrieval with Correlation on Event


Fetch SYSTOLIC_BP from VITALS DCM for Visit 2, Pt 100


Fetch HYPERTENSIVE_YN from HYPERTENSION DCM for Visit 2, Pt 100 TestX

Fetch HYPERTENSIVE_YN from HYPERTENSION DCM for Visit 1, Pt 100 TestX


Definition => Validation Procs => Procedures, Press [Q-Groups]

Alias to correlate toInner loops use the same event (including subevent) as retrieved for the outer (correlated) loop

Correlation on Qualifying Question Value

• Assume two qualified DCMs recorded at a visit– Vitals DCM: qualified at Pre-Dose, 1 Hr Post-dose– EKG DCM: qualified at Pre-Dose, 1 Hr Post-dose

• Correlate on both visit and qualifying question

Fetch data for Patient 100Fetch SYSTOLIC_BP from VITALS DCM for Visit 1, Pt 100, Pre-Dose

X

Fetch EKG from EKG DCM for Visit 1, Pt 100, Pre-doseFetch EKG from EKG DCM for Visit 1, Pt 100, Post-doseFetch EKG from EKG DCM for Visit 2, Pt 100, Pre-doseFetch EKG from EKG DCM for Visit 2, Pt 100, Post-dose

XX

Fetch SYSTOLIC_BP from VITALS DCM for Visit 1, Pt 100, Post-Dose

Fetch SYSTOLIC_BP from VITALS DCM for Visit 2, Pt 100, Pre-Dose


XX

XFetch SYSTOLIC_BP from VITALS DCM for Visit 2, Pt 100, Post-Dose

X


X

X

X


XX

Correlation on Qualifying Question Value

Correlating Questions

• Use responses to DCM questions to correlate between outer and inner fetch loops– May want to fetch AE records where AE description

matches the indication given for ConMed


Definition => Validation Procs => Procedures, Press [Q-Groups], [Correlating Questions]

Inner loop will fetch records only if the values of the responses to the two pairs of questions are the same!

Outer LoopOuter LoopInner Loop

Correlating Questions

Correlations

• Can correlate on Events– Inner cursor retrieves only a visit retrieved by the outer correlated cursor

• Can correlate on Qualifying Questions– Inner cursor only retrieves DCMs which have the same value for the DCM

Qualifying Question

• Can correlate on Questions– Inner cursor fetches only if a question’s value is equal to a question’s value

in the outer cursor

Correlation limits the fetches for inner cursors based on data obtained by a fetch in a more outer cursor

Effect of Correlation on Event

ACTUAL_EVENT_ID is retrieved as part of the standard cursor variables in the outer cursor

…….

Outer Cursor Definition

Cursor variable i_actual_event_id specified when cursor is opened

used in the where clause

…….

Inner Cursor Definition


Call to open inner (CM) cursor uses the actual_event_id retrieved by the outer_cursor

Fetch on the outer (AE) cursor retrieves a value of actual_event_id

);


Effect of Qualifying Question Value and Question Value Correlation

• Other types of correlation (on qualifying question value or question values) work in the same way– When procedure compiled using these techniques, the

procedure question group cursors are restricted– Opening of cursors in the MAIN procedure use cursor

variables

Qualifying Expressions

• Allow selection of records based on any information retrieved by the current or any more outer cursor– Can join values between cursors in the qualifying

expression

Continue only if medication on the AE form is not null

Qualifying Expressions

• Allow selection of records based on any information retrieved by the current or any more outer cursor– Can join values between cursors in the qualifying expression

Continue only if Medication on the AE form is not null

Patient Record Loop

Details

Fetch Qualify Loop

Inner CM Cursor Loop

Outer AE Cursor Loop

Effect of Qualifying Expressions

Fetch an AE record

Exit AE Fetch Qualify loop if no more AEs

Exit AE Fetch Qualify loop if fetch qualifies

Process inner loops/details or exit the cursor loop

Effect of Qualifying Expressions

• Limit cursor fetch with SQL expression comprised of key DCM fields– An LOV is available for fields which can be used in the

Extension– With caution, can use other variables in the

RECEIVED_DCMS and RESPONSES tables, but only for the current cursor

• One way to limit retrievals to specific visits

WHERE Clause Extension

Can join only fields from the current cursor

WHERE Clause Extension

• Limit cursor fetch with SQL expression comprised of key DCM fields– An LOV is available for fields which can be used in the

Extension– With caution, can use other variables in the

RECEIVED_DCMS and RESPONSES tables, but only for the current cursor

• One way to limit retrievals to specific visits

WHERE Clause Extension …

Extension of “where visit_number=1”

Added to the default cursor definition

Effect of Where Clause Extensions

• Can limit to a range of visits fetched by providing the names of the first and last visits on the procedure question group form

By default, all patient visits will be retrieved by the DCM cursor

Controlling Range of Visits


Only DCMs recorded for visits starting at “Visit 1” and going through “Visit 3” will be retrieved

Controlling Range of Visits

• Can limit to a range of visits fetched by providing the names of the first and last visits on the Procedure Question Group form

By default, all patient visits will be retrieved by the DCM cursor

Controlling Range of Visits …

Effect of Range of Visits Control

These are passed as input when the DCM cursor is opened by the main procedure

Performance

• Generated PL/SQL code contains a multitude of loops– Cursors are opened, fetched and closed many times in

the internal fetch loops• Use as much correlation as possible to improve

performance– Some types of correlations affect performance more

than others– Correlation may be necessary to make sure the

procedure does what is expected

Effecting Performance

Event Range In DCM Cursors Large Correlation

Event In DCM Cursor LargeQualifying Value In DCM Cursor Large Correlating Questions In DCM Cursor Large

Where Clause Extension In DCM Cursor LargeQualifying Expression Test after DCM Fetch Smaller

Where Effect on Parameter Implemented Performance

Note: In many cases, parameters must be changed so procedure performs correctly

Summary

• Procedure generation produces a complex PL/SQL program

• Program makes extensive use of cursors and looping structure

• Program performance can be changed by correct use of many techniques available from the form templates

www.facebook.com/perficientwww.perficient.com

www.twitter.com/perficient_LS

Thank You!For more information, please contact:

[email protected]@perficient.com (Sales)

+44 (0) 1865 910200 (U.K. Sales)+1 877 654 0033 (U.S. Sales)

Technology

Tips and Techniques for Improving the Performance of Validation Procedures in Oracle Clinical