 |
|
Using Observational Data to Estimate Treatment Effects—Reply
|
|
| Since this article does not have an abstract, we have provided the first 150 words of the full text and any section headings. |
|
|
|
In Reply: The letter by Dr Stukel and colleagues helps to clarify their use of instrumental variable analysis. We do not, however, find their points convincing enough to resolve problems and concerns.
The claim that their instrumental variable estimates are unbiased seems to be a definition and not a validation. Their point about including regional catheterization rates in a propensity analysis, and that the variable "regional catheterization rates" was not related to patient health status, does not seem to take into account that regions with better health facilities and better trained physicians usually have better health outcomes, and these are the regions that are more likely to have higher rates of catheterization.
Furthermore, their discussion concerning survival bias and why this may be the cause of the overestimate of benefit from the traditional observational study methods (that patients who were sick and who did not survive to receive the treatment . . . [Full Text of this Article]
Ralph B. D’Agostino, Jr, PhD
Department of Biostatistical Sciences
Wake Forest University
School of Medicine
Winston Salem, NC
Ralph B. D’Agostino, Sr, PhD
ralph@bu.edu
Department of Mathematics and Statistics
Boston University
Boston, Mass
RELATED LETTER
Using Observational Data to Estimate Treatment Effects
Therese A. Stukel, Elliott S. Fisher, and David E. Wennberg
JAMA. 2007;297(19):2078-2079.
EXTRACT
| FULL TEXT
RELATED ARTICLE
Estimating Treatment Effects Using Observational Data
Ralph B. D’Agostino, Jr and Ralph B. D’Agostino, Sr
JAMA. 2007;297(3):314-316.
EXTRACT
| FULL TEXT
|
