Understanding intent to treat analyses: An important lesson from the international cooperative study on the timing of aneurysm surgery

Abstract

Background and purposeIntent-to-treat analyses (ITT) are the best way to analyze randomized clinical trials because they preserve the benefits of randomization: to provide an unbiased assessment of relative treatment effects. Yet they play a more fundamental role, which can be demonstrated with observational studies. MethodsWe use a hypothetical RCT to explain why ITT analyses are more appropriate to analyze RCT results. We review the International Cooperative Study on the Timing of Aneurysm Surgery (ICSTAS), a landmark observational study on the management of ruptured aneurysm patients. We discuss the impact of the ICSTAS lesson on the interpretation of future observational studies using Big Data. ResultsPer-protocol (or as-treated) analyses can be misleading: The ICSTAS study provided ‘as-treated’ results clearly in favour of delayed surgery, while overall management or ITT results showed no difference between early and delayed surgery. A contemporary RCT showed that early surgery was best. ICSTAS’ lesson is that observational studies can provide misleading results when intent-to-treat categories are not predefined in the first place. ConclusionIntent-to treat analyses are the most appropriate way to analyze data, whether from randomized trials or observational studies. This observation has momentous consequences. A science of medical practice is impossible without predefined questions regarding optimal care.