Institutional Subscriptions   Individual Subscriptions 
Editors/Authors   Librarians   Newsletter   Site Tour   A-Z Index   About   Contact Us   Help 
Log In | Log In via Athens
Disable Autosuggest
Users' Guides to the Medical Literature
Guyatt G, Rennie D, Meade MO, Cook DJ
Part B Therapy
Chapter 9.2. Surprising Results of Randomized Trials
Christina Lacchetti, John Ioannidis, Gordon Guyatt
Table 9.2-3 Refuted Evidence From Observational Studiesa

Topics Discussed: aspirin, colorectal cancer, coronary artery bypass surgery, coronary heart disease, coronary revascularization, deep vein thrombosis, dementia, dexamethasone, diabetes mellitus, diuretics, duodenal ulcer, interpretation of results, ischemic stroke, leg ulcer, menopause, misleading results, myocardial infarction, observational studies, patient education, patient-important outcome, pulmonary embolism, randomized controlled trials, sodium fluoride, spinal fractures, statins, surprising results

Excerpt: "Table 9.2-3 demonstrates that the results of observational studies are often an inadequate guide for therapeutic decisions, even if they pertain to patient-important outcomes. Some investigators have suggested that usually randomized and observational evidence agree with similar evidence.57-59 An empirical evaluation, however, examined 45 topics for which both RCTs and observational studies were available on the same clinical question and used the same outcome. Observational studies showed, on average, larger benefits, and in 7 of these questions, the 2 designs gave results that were different beyond chance.60 Overall, observational studies may be subject to more noise in their estimates compared with randomized trials after accounting for differences in sample size.61 Some observational studies may use very large sample sizes (much larger than what randomized trials can achieve), and therefore they produce spuriously tight confidence intervals, whereas the true uncertainty associated with their findings is much larger...."
Log in to read the full chapter:
Subscriber Log In
Forgot your username/password?
Get full access to JAMAevidence two ways:
Subscribe to JAMAevidence
JAMAevidence is a subscription-
based website dedicated to the learning, teaching, and practicing of evidence-based medicine.
Pay Per View
Timed access to all of JAMAevidence
24 hours for $34.95
48 hours for $54.95
Copyright © American Medical Association. All rights reserved.  |  JAMA  |  McGraw-Hill Global Education Holdings, LLC.
Privacy Notice. Any use is subject to the Terms of Use and Notice. Additional Credits and Copyright Information.
Your IP address is