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

From |
Svend Juul <SJ@SOCI.AU.DK> |

To |
<statalist@hsphsun2.harvard.edu> |

Subject |
Re: st: Converting an odds ratio in a risk ratio - Logit |

Date |
Sun, 15 Jun 2008 09:41:15 +0200 |

Leda Inga wrote: I ran a logit with diferent dummies as you can see below: Y = f(X, D1_3, D4, D5, .., D10, D_more10) X: set of regressors D_: dummies that classify each indiviual by the number of visits made to the doctor. Since I ran a fixed effect logit, I only can get the odds ratio of each dummy. Nevertheless, I would like to convert them in risk ratios. I hope that someone has any idea or has been in a similar situation before. According to the recomendation of Jun Zhang and Kai F. Yu (http://jama.ama-assn.org/cgi/content/full/280/19/1690#ACK) it can be used the following: RR= OR/ (1 - Po + Po*OR) where Po is the incidence of Y in the "unexposed group". Since I used diferent dummies, I'm not sure what should be my "unexposed group" for each dummy. I know that the coefficient of each dummy must be interpreted versus the reference group (in my case, zero visits). But maybe to get the risk ratios of "X" visits versus the reference (zero visits) the unexposed group should be "X-1" visits or all but "X" (since each dummy takes values of zero for all individuals who didn't have "X" visits) ============================================================ findit relative risk led me to an excellent FAQ about it at http://www.ats.ucla.edu/stat/stata/faq/relative_risk.htm It shows the use of -glm-. Alternatively, you can use -binreg- with the -rr- option to obtain relative risks. Joseph Hilbe made the -oddsrisk- command; it uses the method described by Zhang and Yu. Find it by: findit oddsrisk The second question is about the choice of reference group. In my world (epidemiology), I don't think there is much doubt. The non- visitors is a natural reference group, and it should be the reference for each of the other exposure levels. Shifting reference groups makes the results quite difficult to interpret. Hope this helps Svend ________________________________________________________ Svend Juul Institut for Folkesundhed, Afdeling for Epidemiologi (Institute of Public Health, Department of Epidemiology) Vennelyst Boulevard 6 DK-8000 Aarhus C, Denmark Phone, work: +45 8942 6090 Phone, home: +45 8693 7796 Fax: +45 8613 1580 E-mail: sj@soci.au.dk _________________________________________________________ * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**R: st: one-way 95 CI for Ha: diff < 0** - Next by Date:
**RE: st: treatreg2** - Previous by thread:
**st: Converting an odds ratio in a risk ratio - Logit** - Index(es):

© Copyright 1996–2016 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |