Research Methods & Ethics · 1976
Large simple trials and trial overviews (Peto and colleagues)
ISIS, International Studies of Infarct Survival
Clinical trials in the early 1970s were typically sized to detect large treatment effects, because the assumption was that any treatment worth using would produce obvious benefits. Richard Peto, working in the Imperial Cancer Research Fund epidemiology unit at Oxford, began arguing that this assumption was wrong. Most useful treatments, he reasoned, produce modest effects, perhaps a 20-25% reduction in relative risk. A trial enrolling a few hundred patients has almost no power to detect that reliably. His papers in the British Journal of Cancer in 1976 and 1977, co-authored with colleagues including Michael Pike, formalized both the statistical argument and the logrank test as an appropriate tool for time-to-event data.
The design philosophy that emerged from this work was deliberately stripped down. A large simple trial would collect only the information necessary to answer the primary question: randomize patients, record the outcome of interest, and avoid the elaborate sub-protocols and extensive data collection that made conventional trials slow and expensive to run. The goal was to enroll tens of thousands of patients, a scale that could detect a 20% mortality reduction with adequate confidence. What was sacrificed in granularity was more than compensated by statistical power and generalizability.
The ISIS trials, launched from the Oxford Clinical Trial Service Unit in 1981, translated this framework into practice on acute myocardial infarction. ISIS-1 examined atenolol; ISIS-2, published in Lancet in 1988, is the landmark result. It enrolled 17,187 patients across 417 hospitals in 16 countries and showed that aspirin alone cut 35-day vascular mortality by about 23%, streptokinase alone by about 25%, and the two together by about 42%. The additive effect was unanticipated and immediately changed prescribing practice in coronary care units worldwide.
The same methodological papers introduced the concept of the systematic overview, which Peto distinguished from a narrative review by its insistence on finding all trials, including unpublished ones, and combining their results with formal statistical methods. What would later be called meta-analysis became, through this tradition, the expected form of evidence synthesis in cardiovascular and oncology research. The Cochrane Collaboration, established in 1993 largely on these principles, extended the approach across medicine.
Rory Collins joined Peto and became the operational leader of subsequent ISIS trials, extending the megatrial design to examine ACE inhibitors, fibrinolytics, and antiplatelet therapy in infarction. The ISIS programme and its intellectual successors, including the GISSI and GUSTO trials run on similar principles in subsequent years, produced much of the evidentiary base for modern acute coronary syndrome management.
Key People
- Richard Peto — Oxford statistician who developed the logrank test and the large simple trial framework.
- Rory Collins — Peto's collaborator; led the ISIS trial series from their Oxford Clinical Trial Service Unit.
- Michael Pike — Statistician who co-authored the foundational 1976-77 trial design papers with Peto.
- Peter Sleight — Oxford cardiologist who helped design and run the ISIS acute MI trials.
Br J Cancer 1976-1977; ISIS programme from 1981
Related landmarks
- 1972 · Cochrane, 'Effectiveness and Efficiency' (Research Methods & Ethics)
- 1972 · Tuskegee Syphilis Study Exposed (Research Methods & Ethics)
- 1964 · Declaration of Helsinki (Research Methods & Ethics)
- 1962 · Kefauver-Harris Drug Amendments (Research Methods & Ethics)