Architecture Design of Generic Outcome Adjudication in CTMS Wenle Zhao, PhD Medical University of South Carolina, Charleston, SC, 29425, USA Society for Clinical Trials 36th Annual Meeting Arlington, VA, USA - May 17-20, 2015 Contents 1. The Motivation 2. The Challenges 3. The Generic Database Model 4. The Automated Coordination 5. Summary 1.The Motivation 2. The Challenges 3. The Generic Database Model 4. The Automated Coordination 5. Summary Accurate Outcome ( Endpoint) Assessment is Important Patients Randomization Treatment A Treatment B Outcome A Outcome B Analysis The validity of the trial results relies on the accuracy of outcome assessment. Most Outcome Assessments are Subjective Outcome Safety Outcome Efficacy Outcome AE is Serious? Quality of Life AE is Expected? Modified Ranking Scale AE is Related to Study Treatment? NIH Stroke Scale Hamilton Rating Scale for Depression Glasgow Outcome Scale Extended Hamilton Rating Scale for Anxiety The Problem: The Solution: Uneven expertise and potential bias Independent central adjudication 1. The Motivation 2.The Challenges 3. The Generic Database Model 4. The Automated Coordination 5. Summary Example 1: A Simple SAE Adjudication CRF Completed? May be not. When it is expected? No schedule. Is serious? CRF Updated after submission? Could happen. SAE CRF Submitted Is expected? Done How to notify the adjudicator? Auto Email. Need more information? May be yes. Is related? Four challenges for coordinating outcome adjudication Adjudication procedures vary among projects No fixed model for every procedure. Timing of outcome event unpredictable No schedule. Narrow time window for completion of adjudication “Time is of the essence.” Data changing after adjudication procedure started The target is moving. 1. The Motivation 2. The Challenges 3. The Generic Database Model 4. The Automated Coordination 5. Summary The Generic DB Structure for Adjudication Procedures 1 Data (CRF) Form ∞ ID Form Name Table Name 1 Adjudication Procedure 1 ID Form ID Procedure Name 1 Form Table ID Subject ID Visit ID Q01 Q02 …… Updated By Updated On Adjudication Step The Model Allow multiple steps. Allow multiple updates. Limit ≤ 5 close-ended Qs. ID ∞ Procedure ID Adjudication Step Name Enable-logic condition Disable-logic condition Question 1 Answer 1 Code Group ID …… Question 5 Answer 5 Code Group ID Edit User Group ID View User Group ID 1 1 Form Freeze Table Freeze Record ID ∞ ID Subject ID Visit ID Q01 Q02 …… Updated By Updated On ∞ Adjudication Result ID ∞ Adjudication Step ID Record ID Freeze Record ID Answer 1 …… Answer 5 General Comments Locked 1 Two Tables are included in the Database Template One table specifies definitions of each adjudication step. The other table stores adjudication results. A freeze table is created for each data form table with adjudication Example 1: Definition of Three-step SAE Adjudication Adjudication Step Step 1 PM Review (S1) Enable logic Disable logic {S2 submitted after S1} {New SAE CRF submission} OR {S3 submitted after S1} {S1Q1=1} AND {S1Q2=1} Step 2 AND {S2 submitted after S1} MSM {no S2 submitted after S1} OR Adjudication AND {S3 submitted after S2} (S2) {no S3 submitted after S1} Step 3 PM Review (S3) Question 1 Question 2 (Q1) (Q2) CRF properly Requires MSM completed? review? 0 = No 0 = No 1 = Yes 1 = Yes Serious? 0 = No 1 = Yes MedWatch {S2 submitted} Form needed? AND {S3 submitted after S1} 0 = No {no S3 submitted after S2} 1 = Yes Question 3 (Q3) Related to study drug? Unexpected? 1 = Unrelated 0 = No 2 = Unlikely 1 = Yes 3 = Possible 4 = Probably 5 = Definitely Example 2: Definition of Six-step SAE Adjudication Adjudication Step Enable logic Disable logic Question 1 (Q1) Question 2 (Q2) Step 1 Project Manager {New SAE CRF submission} (PM) Review {S2 submitted after S1} CRF properly completed? 0 = No 1 = Yes Step 2 Clinical Event {S1Q1=1} AND {S1Q2=1} AND Coordinator (CEC) {no S2 submitted after S1} Review {S1 submitted after S2} OR {S3 submitted after S2} OR {S4 submitted after S2} OR Event type? Even packet complete? 1 = Neurological 0 = No 2 = Cardiac 1 = Yes 3 = Systemic Step 3 Adjudicator 1 Review (S2Q1=1) AND {no S5 submitted after S3} Step 4 Adjudicator 2 (ADJ2) Review (S2Q1=1) AND {no S5 submitted after S4} Step 5 Adjudicator 3 (ADJ3) Review {S3 submitted after S1} AND {S4 submitted after S1} AND {S1 submitted after S5} OR {S3Q1≠S4Q1 or S3Q2≠S4Q2} AND {S6 submitted after S5} {no S6 submitted after S5} AND Step 6 Adjudication Committee Chair (ACC) Review Requires CEC review? 0 = No 1 = Yes {S1 submitted after S3} OR {S5 submitted after S3} {S1 submitted after S4} OR {S5 submitted after S4} {S5 submitted after S1} AND {S5Q1≠S3Q1 or S5Q2≠S3Q2} AND {S6 submitted after S1} {S5Q1≠S4Q1 or S5Q2≠S4Q2} AND {no S1 submitted after S6} AND Outcome category=? 1 = ischemic stroke 2 = symptomatic hemorrhagic transformation of an ischemic stroke 13 = transient ischemic attack Death type =? 1 = ischemic 2 = hemorrhagic vascular 3 = nonvascular The Generic Table for Adjudication Results 1. The Motivation 2. The Challenges 3. The Generic Database Model 4.The Automated Coordination 5. Summary Automated Coordination of Adjudication Activities Site submits a Case Report Form N Form has adjudication procedure defined? Y System evaluates enable logics for each adjudication steps N Enable logic = true & disable logic = false? Y System sends email to [Edit] user group members User has adjudication step [Edit] permission & Enable logic = true & disable logic = false? Y System displays [Adjudication Edit] button for the user User adds/edits adjudication result and submits adjudication record The last freeze record is the same as the current CRF record? N User views a Case Report Form N System creates new freeze record, links it to the adjudication record. Y Example 1: Three-step SAE Adjudication Procedure Site adds/edits SAE CRF N S2Q1 = Yes & S2Q2 = Yes & S2Q3 = Possible/ Probably / Definitely? Site submits SAE CRF Y Auto email #5 to PM Auto email #1 to PM Step S1 = PM Reviews S1Q1: CRF properly completed? S1Q2: Requires MSM adjudication? S1Q1 = Yes? Y N Step S3 = PM Review S3Q1: MedWatch Form required? N S3Q1 = Yes? Y System composes MedWatch Form draft Auto email #2 to Site N Auto email #6 to Site S1Q2 = Yes? Y Auto email #3 to MSM Site edits MedWatch Auto email #7 to PM Step S2 = MSM Adjudication S2Q1: Serious? S2Q2: Unexpected? S2Q3: Related to study treatment? Auto email #4 to PM PM finalizes MedWatch and submits to FDA Auto email #8 to All Sites PM Closes the Adjudication Procedure Example 2: Six-step Outcome Adjudication Procedure System determines adjudicators 1, 2, and 3 based on S2Q2 and the Random number R. Site adds/edits SAE CRF and upload the event packet Site submits SAE CRF Auto email #5 to ADJ1 Auto email #6 to ADJ2 Auto email #1 to PM Step S3 = ADJ1 Review S3Q1: Outcome category? S3Q2: Death type? Step S4 = ADJ2 Review S4Q1: Outcome category? S4Q2: Death type? Step S1 = PM Reviews S1Q1: CRF properly completed? S1Q2: Requires CEC review? Y S3 = S4? N Auto email #7 to ADJ3 S1Q1 = Yes? N Auto email #2 to Site S1Q2 = Yes? Y Auto email #3 to CEC S2Q1 = Yes? N Auto email #4 to Site Step S5 = ADJ3 Review S5Q1: Outcome category? S5Q2: Death type? N Step S2 = CEC Adjudication S2Q1: Event packet complete? S2Q2: Event type? Y Y PM Closes this Adjudication Procedure S3= S5 or S4 = S5? N Auto email #8 to ACC Step S6 = Chair Review S6Q1: Outcome category? S6Q2: Death type? Auto email #9 to PM Y In this case, the Clinical Event Coordinator (CEC) completed the adjudication within few hours after the Study Manager (SM) ‘s completeness review passed. In this case, the procedure has not been completed yet after two months, due to CRF data change. 1. The Motivation 2. The Challenges 3. The Generic Database Model 4. The Automated Coordination 5.Summary Our Experiences Adjudication Procedure Three-step SAE adjudication Six-step outcome event adjudication From step To step Event count 1st quartile (25%) 2nd quartile 3rd quartile (50%) (75 %) hrs hrs hrs CRF submit PM review 1253 21 51 126 PM review MSM review 1226 5 13 35 MSM review PM close 1226 27 56 112 CRF submit PM review 609 26 120 477 PM review CEC review 599 96 197 364 CEC review ADJ1/2 review 468 47 204 892 ADJ1/2 review ADJ3 review 76 56 209 383 ADJ3 review ACC review 10 23 23 25 PM: project manager MSM: medical safety monitor CEC: clinical event coordinator ADJ1/2: adjudicators 1 and 2 ADJ3: adjudicator 3 ACC: adjudication committee chair The Strategy to Success Allows multiple adjudication procedures per CRF. Allows multiple steps for each adjudication procedure . Allows different questions be asked for each adjudication step. Allows multiple CRF submissions after adjudication started. Only close-ended questions are allowed. No more than 5 questions are allowed. Using automated emails to help adjudication coordination. Acknowledgement This research is partly supported by following NIH/NINDS grants: • U01NS059041 (NETT Palesch, Y. & Durkalski, V.) • U01NS062778 (ProTECT, Palesch Y.) • U01NS062835 (POINT, Palesch, Y.) • U01NS087748 (StrokeNet, Palesch, Y. & Zhao, W.) The followings are contributed to the design, development, validation and user feedback collection to this work: • Keith Pauls (DCU senior programmer) • Jaemyung Kim (DCU senior programmer) • Catherine Dillon (DCU senior data manager) • The Clinical Coordination Center of the NINDS NETT at University of Michigan Thousands investigators in hundreds institutions and thousands patients participating our clinical trials. Thank You! Contact me at: zhaow@musc.edu