[R] replicating the odds ratio from a published study- post # 2

Bob Green bgreen at dyson.brisnet.org.au
Fri Jan 26 22:46:28 CET 2007

Peter & Michael,

I just came across the following on another mailing list and realized that 
my use (and the authors of the article use of the term 'odds ratio' ) is 
probably incorrect. I believe my interest is in the 'odds' of schizophrenia 
among the population of homicide, rather than a comparison of odds .

"Yours seems a good idea, Kevin, if you are only interested in computing 
the ODDS of disease (not the odds RATIO). The odds of disease
equal the probability of disease divided by the probability of non-disease, 
i.e. p/(1-p), where p is the proportion of cases at or above the cutoff
point. An odds RATIO is the ratio of two odds, e.g. the odds for vaccinated"

If it is the odds advice regarding the appropriate R script would be useful.


Bob Green

., At 02:57 PM 26/01/2007 +0100, Peter Dalgaard wrote:
>Michael Dewey wrote:
> > At 09:04 26/01/2007, Bob Green wrote:
> >
> >> I wanted to compare odds ratio across studies and tried to replicate
> >> the results from a study but have not been able to determine how to
> >> do this in R.
> >>
> >> The study reported a sample of 961 men, of whom 41 had a disorder.
> >> The reported raw odds ratio was 6.52 (4.70-9.00)
> >>
> >
> > For an odds ratio you require two odds from which you form the odds ratio.
> > You only have one odds.
> > Do you have another one lying around somewhere?
> >
>Alternatively, the odds ratio presumably compares two groups. If you
>know the group sizes, the two odds ratios may be reconstructed. If I
>make a wild guess that the groups are roughly equal, I might get
> > M <- cbind(c(460,460),c(6,35))
> > M
>      [,1] [,2]
>[1,]  460    6
>[2,]  460   35
> > fisher.test(M)
>         Fisher's Exact Test for Count Data
>data:  M
>p-value = 7.406e-06
>alternative hypothesis: true odds ratio is not equal to 1
>95 percent confidence interval:
>   2.393209 17.104976
>sample estimates:
>odds ratio
>   5.824317
>Judging by the c.i., the groups are probably *not* of similar size. I
>suppose that the high-incidence group is a bit smaller so that the count
>of advverse events is more similar. M <- cbind(c(803,117),c(21,20)) is a
>bit more like it, but your (Bob's) confidence interval is narrower even
>than this.
>    O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
>   c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
>  (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
>~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907

More information about the R-help mailing list