1
2	"General Issues" General Issues Examples* *Examples 2 and 3 are drawn from two of the case studies in the forthcoming book, “Data Monitoring in Clinical Trials: A Case Studies Approach,” edited by David L. DeMets, Curt D. Furberg, and Lawrence M. Friedman, and published by Springer. The two case studies cited are “Lessons from warfarin trials in atrial fibrillation – missing the window of opportunity,” by Charles H. Tegeler and Curt D. Furberg; and “Data monitoring in the Heart Outcomes Prevention Evaluation and the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events trials – avoiding important information loss” by Janice Pogue, David Sackett, George Wyse, and Salim Yusuf.
3	Randomized Controlled Trials Influence clinical practice Address important questions in scientifically rigorous and ethical manner Require monitoring so that can modify or stop if needed
4	DSMB Issues (a) Was the study properly designed? Measures of efficacy and safety—are they appropriate? Why was the study started (and who is the sponsor)? Reduce morbidity/mortality (clinical/public health impact) Regulatory agency approval “Marketing” (e.g., many active-control trials)
5	DSMB Issues (b) Has the question been answered? How persuasive are the data? Test new intervention Confirm existing view Overturn established ideas Is there relevant external information? Should more data on subgroups or secondary outcomes be obtained?
6	DSMB Issues (c) Can the question be answered? Unconditional power Conditional power Data integrity and quality
7	DSMB Issues (d) Should the question be answered? Is the question still important or have the practice of medicine and technology passed it by? Is the study still ethical? SAEs Consent form (“contract”) Updates to participants
8	Example 1: Women’s Health Initiative Postmenopausal women Hormone Replacement Therapy (estrogen with and without progestin) Designed to look at outcomes of heart disease, hip fracture, breast and colon cancer Refs: JAMA. 2002;288:321-333; JAMA. 2004;291:1701-1712
9	Example 1: WHI (b) Approved drug Used for many years in many women Observational study data and prevailing wisdom indicated benefit for heart disease Clear benefit for symptoms associated with menopause
10	Example 1: WHI (c) Key outcome contrary to expected How real were the findings? Need to overturn established practice Heart and Estrogen/progestin Replacement Study (HERS) Serious adverse events (stroke) Consent form and subsequent communications with participants
11	Example 1: WHI (d) In considering persuasiveness of data and when to stop, DSMB weighed Evidence from several outcomes (including “global index”) Existing practice and belief and other clinical trial data
12	Example 2: Warfarin and Atrial Fibrillation Will warfarin prevent stroke in people with non-rheumatic, non-valvular atrial fibrillation? 1985 to 1987, five trials started Copenhagen AFASAK (Ref: Lancet 1989; 1:175-9) Stroke Prevention in Atrial Fibrillation (SPAF) (Ref: N Engl J Med 1990; 322:863-8) Boston Area Anticoagulation Trial for Atrial Fibrillation (BAATAF) (Ref: N Engl J Med 1990; 323:1505-11) Canadian Atrial Fibrillation Anticoagulation (CAFA ) (Ref: J Amer Coll Cardiology 1991; 18:349-55) Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation (SPINAF) (Ref: N Engl J Med 1992; 327:1406-12)
13	Example 2: Warfarin and AF (b) When is the answer known? AFASAK, SPAF, BAATAF reported benefit CAFA stopped by investigators without knowing interim data after AFASAK and SPAF reports (26% reduction, p = 0.25) SPINAF stopped by DSMB because observed benefit consistent with prior studies (p = 0.001)
14	Example 2: Warfarin and AF (c) Preponderance of external evidence led to early stopping of one trial (CAFA) and affected deliberations (and perhaps early stopping) of another (SPINAF). BAATAF deliberations affected by knowledge that SPAF had stopped.
15	Example 3: Heart Outcomes Prevention Evaluation (HOPE) Trial of ACE-I in patients with CAD Composite primary outcome (MI, stroke, cardiovascular death) Monitoring boundary for benefit crossed at 4^th interim analysis Questions: Are the results similar in components of the outcome? Are the same trends seen in subgroups? Ref: N Engl J Med 2000;342:145-53
16	Example 3: HOPE (b) DSMB decided to look at data after 4 more months Wanted clinicians to have sufficient evidence of consistency in subgroups, secondary outcomes, and over time Stopped 8 months ahead of schedule
17	Example 3: HOPE (c) Even though boundary for benefit for overall primary outcome was crossed, the DSMB thought it important to obtain additional information, so that the results would be more persuasive and more likely to change clinical practice
18	Conclusions No simple algorithm for when to stop a trial; complex because of numerous medical, ethical, and statistical factors Need committee of experienced, thoughtful individuals Need to consider the results of a trial in the context of all other evidence and current clinical practice Clinical trials have considerable impact on practice; they must be designed, conducted, and analyzed rigorously and ethically