[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: ODDSRISK module to convert Logistic Odds Ratios to Risk Ratios

From   [email protected]
To   [email protected]
Subject   st: RE: ODDSRISK module to convert Logistic Odds Ratios to Risk Ratios
Date   Thu, 20 Dec 2007 12:00:23 EST

'ODDSRISK': module to  convert Logistic Odds Ratios to Risk Ratios now on SSC

I thought I had  posted a notice about this module to the list yesterday, but 
it seems that it  did 
not make it. I take the Digest, so do not see what is posted until the  next 
day. I apologize if the earlier one is posted today, as well as this one.  

Zhang and Yu (1998, JAMA) proposed an algorithm to convert binary  logistic 
regression odds ratios to estimated risk ratios. I have seen the method  used 
in a number of medical research articles, and had written about the method  in 
a book I am writing. I noticed a couple of Statalist communications during  
the past couple of weeks related to the relationship of odds ratio to risk  
ratio. I wrote about this relationship in my "Negative Binomial Regression", but  
not about this method and its rationale. Anyhow, I thought it might be helpful 
 to those interested in using this method to release a Stata implementation 
now  rather than with the forthcoming book several months from now. I wrote the 
 program in such a manner that the logistic odds ratio, estimated risk ratio, 
and  95% confidence intervals of the estimated risk ratios are displayed for 
each  predictor in the model. The incidence rate of the unexposed risk factor, 
or  primary predictor of interest (binary (1/0)), which is required for 
calculation,  is also displayed. 

The estimated risk ratios are particularly important  when we wish to use 
terms such as "likely" in place of "odds of" when  interpreting the odds ratio of 
a logistic model. This is primarily the case with  the odds ratio is under 
0.5 or over 2.5 and the incidence rate is greater than  10%. To say something 
like, "patients having an anterior infarct are 50% more  likely to die within 48 
hours of hospital admission than are patients having  just sustained an 
infarct at another primary site", is using risk language. Such  language may be 
justified for many logistic models, but not all.  Using the  estimated risk 
values and CI's calculated by -oddsrisk- presumably enhance the  justification for 
using risk language with logistic models when the conditions  specified above 

Kit Baum has graciously posted the program,  called -oddsrisk- and the 
associated help file to the SSC site.  You can  download it by typing 
. ssc install oddsrisk
on the Stata  command line. 

I have pasted the SSC site module description under my  name below. 

Best,  Joseph  Hilbe


oddsrisk converts logistic regression odds  ratios to relative
risk ratios. When the  incidence of an outcome is common in the
study  population; i.e. greater than 10%, the logistic  regression
odds ratio no longer approximates  the risk ratio. As the
incidence rate becomes  more frequent, the more the odds ratio
overestimates the risk  ratio when its value is greater than  1,
and the more it underestimates the risk  ratio  when under one. J.
Zhang and K. Yu  proposed a method of adjusting the logistic
regression odds ratio in a cohort study or clinical trial so  that
it approximates  the risk ratio.  This is particularly important
when the odds  ratio is greater than  2.5 or under 0.5. The  method
has also been shown to be applicable  for retrospective  and
observational  studies as well.

Author: Joseph M.  Hilbe
Arizona State  University
[email protected]   or   [email protected]

**************************************See AOL's top rated recipes 
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index