Interpreting results of observational researchBMJ 1994; 309 doi: http://dx.doi.org/10.1136/bmj.309.6966.1439 (Published 26 November 1994) Cite this as: BMJ 1994;309:1439
Problem with confounders can be tackled
- David Buck,
- Matthew Sutton
- Research fellow, Research fellow, Centre for Health Economics, University of York, York YO1 5DD.
EDITOR,—Paul Brennan and Peter Croft argue that the application of tests of significance in observational studies is inappropriate because of the lack of randomisation and consequent likelihood of confounding.1 The authors conclude that neither the P value nor the 95% confidence intervals should be used as evidence for the validity of findings of an observational study. We believe that this conclusion is unwarranted and unwise.
While confounding is a major problem in observational studies, statistical methods exist to test for its presence …
Log in using your username and password
Log in through your institution
Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
Sign up for a free trial