Role and Function of DSMBs
[email protected]
References • Ellenberg SS, Fleming TR, DeMets DL. Data Monitoring Committees in Clinical Trials. Chichester. John Wiley & Sons Inc. 2003. • Guidance for Clinical Trials Sponsors. Establishment and Operation of Clinical Trial Data Monitoring Committees. March 2006. www.fda.gov • American Heart Journal. 2001;141(1,2 and 3)
DSMB Nomenclature • • • •
DMC (Data Monitoring Committee) IDMC (Independent Data…..) DSMC (Data and Safety Monitoring ...) DSMB (…. Monitoring Board)
DMC History 1950’s – Modern RCTs 1967 – Greenberg Report to NIH a) need for expert advisory group, not directly involved in the trial, to review protocol and advise NIH on conduct b) need for mechanism for early termination, if evident trial cannot meet objectives or has become superfluous
DMC History Mid 60’s – CDP (lipid lowering) Policy Advisory Board (subcommittee of which considered efficacy and patient safety and recommended early termination of 3 of 5 arms to PAB and NIH) a) early data trends can be very unstable (small number of events, play of chance) b) complexity of decision-making process
DMC History late 60’s – several NIH cardiovascular trials, all with DMCs mid 70’s – NIH routinely required DMCs in all clinical areas, also VA Cooperative Studies Program early 80’s - NCI early 90’s – UK MRC early 90’s – Pharmaceutical Industry trials
DMC History FDA 2001 Guidelines 2006 Guidelines International Conference on Harmonisation 1998 CIHR “describe DSMB if applicable” Committee to advise on funding of clinical trials
DMC Responsibilities to: • Patients in the trial • Future patients • Trial investigators, sponsors, scientific community, society
DMC Decisions regarding: • Safety SAEs, unexpectedly negative trend for principal outcome or some component of outcome • Early efficacy Marked early benefit overall or in some component of outcome • Futility No evidence of benefit is likely to be found if trial goes to completion, or external events dictate results will be superfluous
DMC Recommendations • Continue or terminate trial or some component of it Must consider the totality of the evidence (all outcomes, adverse events, external data)
• Modify protocol • Study conduct (recruitment, adjudication, particular centres, intensity or focus of monitoring, interim results)
DMC Membership • Chair • Clinicians • Biostatistician(s) • (Ethicist) Appropriate disease expertise, clinical trial and DMC experience, free of COI, decisions by consensus (odd number of members) • (Lay member)
DMC Interactions with Study • Appointment of chair – by study chair and/or sponsor • Appointment of other members consultative • PI, Study statistician • Independent study statistician • Recommendations from DMC • Formal interim analyses
Statistical Monitoring and Stopping Guidelines
• Safety • Early Efficacy • Futility
Zvalue
Information fraction
Symmetric Group Sequential Monitoring Boundaries
Interim analyses
Zvalue
Information fraction
Symmetric Group Sequential Monitoring Boundaries
Reject H0
Interim analyses
Information fraction
Zvalue
Reject H0
Symmetric Group Sequential Monitoring Boundaries
DMC Charter • • • • •
Primary responsibilities Organizational diagram Membership (including COI) Timing and purpose of meetings Procedures to ensure confidentiality and proper communication • Statistical monitoring guidelines • Content of open and closed reports
Do All Clinical Trials Require a DMC? • All clinical trials (phase 1, 2 or 3) require safety monitoring (may be an independent “safety officer” or small committee) • A formal, external DMC is required in the following circumstances: Large trials (long duration, multicentre) whose outcomes include irreversible major morbid events Definitive studies that evaluate new therapies with potentially severe risks Trials that evaluate a known therapy intended for a new use and characterized by a significant risk profile Potentially fragile study population (children, pregnant women, elderly)
DeMets.Clin Trials 2006;3:522
DeMets.Clin Trials 2006;3:522
DeMets.Clin Trials 2006;3:522
Schulz and Grimes. Lancet2005;365:1591