Practical Implications Laura Lovato, Biostatistician Statistical and Practical Considerations When Terminating a Trial Early: Contrasting early stopping due to harm (ACCORD) versus for benefit (SPRINT) Practical Implications Laura Lovato, Biostatistician Liverpool, UK, May 8th, 2017 Invited Session 2.1 - Statistical Design and Trial Conduct

Coordinating center knowledge of the potential for early termination Strict Confidentiality Maintained Small group working on analyses, planning for early termination scenarios and developing materials for rapid cessation Biostatisticians Project Management Programmers

Planning for potential early termination 2-3 most likely scenarios Rapid plan for notifying Steering Committee, Clinical Sites, IRBs, and Participants Development of Operational Materials Data freeze and publication of results (data the DSMB used for decision, or updated data through stopping point, beyond?) Earlier transfer of participant care Training for closeout visits

Planning for potential early termination Website and data management systems modified in the background Documents (for sites, public, media) Data entry systems to capture transfer of treatment/care Closeout visit data entry and management Made ‘live’ once public announcement occurs ACCORD: continuation of Lipid and Blood Pressure trials SPRINT: follow up for cognition and kidney outcomes

Study Leadership Involvement ACCORD MASKED: Site staff and principal investigators UNMASKED: Some coordinating center staff, Some project office (NIH) personnel, and the Steering Committee Chair All actively participated in the discussions. SPRINT None of the project office personnel were unmasked except for the NHLBI representative; he was not otherwise involved in the study. The steering committee chair was masked. Site principal investigators were masked. This meant that the coordinating center had much less advice.

Materials Developed for Early Termination: Publication Documents: obvious! Press Release Journal Article Media talking points Letter to major medical societies

Materials Developed for Early Termination: Memo to Site PIs and Staff: explanation of result, reasons for stopping standard arm, overview of transition process, timeline for remainder of trial. Memo to CCN/Site IRBs: explanation of result, reasons for stopping standard arm, overview of transition process, timeline for remainder of trial. Q&A for Study Staff: anticipated questions and answers to address clinic operations during transition. Treatment plan for transition of standard arm participants to intensive goal (SPRINT) or backing off therapy from intensive arm participants (ACCORD) including written algorithm for implementing intensive arm BP target among standard arm participants.

Materials Developed for Early Termination: Modified Study Forms to be used during transition and remaining follow-up. Meet me slides: slides for presentation to clinic staff that explanation of result, reasons for stopping standard arm, overview of transition process, timeline for remainder of trial. Letter to drug companies currently donating medications: includes explanation of results and plans for transition and timeline for remainder of trial Participant information document for website: explanation of result, reasons for stopping standard arm, overview of transition process, timeline for remainder of trial. Model Consent Addendum

Materials Developed for Early Termination: Materials for dissemination of results to participants, including resources for group sessions. Letter to All participants explaining the result, clearly explaining what they should do next, and encouraging their continued follow-up (NEXT SLIDE) Letter to non-study care provider regarding trial results and recommendations. Talking Points for Participants: anticipated questions and answers for participants designed to be used in 1-1 discussion during transition visits. FAQ for General Distribution: anticipated questions and answers for site PIs and others to aid in explanation of results to general (lay) audience.

Elements of a Participant Notification Letter Introduction What is the news? What do these results mean for me? What should I do now? Thank you! Signatures of Project Office and local Principal Investigator (letter to hand to personal health care provider)

Participant Notification Letters ACCORD Template letters developed centrally, different for each study arm Overnight mail from clinical sites SPRINT Generic letter, same for both arms Printed, Labeled and Mailed from coordinating center to all participants, one day prior to clinical site notification

SPRINT Coordinating Center Staff – Labor Day weekend

Which events to include? Continued collection of events following early termination: implications for analysis Which events to include? Event Detected: prior to DSMB decision after DSMB decision Event Occurred: A BN BF F

Continued collection of events following early termination: WHAT DID SPRINT DO? DSMB Data Freeze July 5, 2015 NHLBI Director accepts DSMB Recommendation to Terminate Early August 20, 2015 NEJM Paper Data Freeze October 2015 NEJM Paper Published November 9, 2015 SPRINT Follow-up Competed June 30, 2016

Clinical Site Staff Concerns - stress ACCORD Devastated by news of possible harm Counseling services offered for staff at many clinical sites Emphasis on quickly and safely transferring participants to appropriate treatment (backing off intensive glycemia arm) SPRINT Emphasis on transferring participants to appropriate treatment, slower timeline Continue follow-up for other outcomes

Clinical Site Staff Concerns – job security ACCORD Glycemia trial stopped 17 months early, but follow-up continued to full duration for the Blood Pressure and Lipid trials SPRINT Intervention stopped 11 months early, and follow-up stopped 4 months early

Practical Considerations When Terminating a Trial Early: Contrasting ACCORD And SPRINT trials Early Termination due to harm (ACCORD) Early termination due to benefit (SPRINT) More sensitivity for study staff (counseling recommended) Follow other outcomes to complete the picture Rapid transition away from experimental treatment Slower transition; focus on explanation of study results and letter to primary care provider Publication based on data frozen for DSMB important to understand decision Publication contained events known at the time of publication that occurred prior to termination announcement

Special thank you to… Bob Byington (ACCORD) Sharon Wilmoth (ACCORD) Letitia Perdue (SPRINT) David Reboussin (SPRINT)