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

From |
Roger Newson <[email protected]> |

To |
[email protected] |

Subject |
Re: st: Transformed values in logistic regression |

Date |
Fri, 27 Aug 2004 12:41:21 +0100 |

At 04:27 27/08/2004, you wrote:

Dear all, Please pardon this non-Stata question, but I have not found a satisfactory resolution. We recently submitted a manuscript for publication and received a favorable reply, except that one of the reviewers criticized a transformation that we performed on a continuous variable prior to fitting a logistic regression model. Specifically we were interested in modeling case-control status as a function of several patient covariates including serum creatinine which in our data ranges from 0.11 to 1.98. Because of skewness and to make the odds ratio independent of the units measurement, we decided to log-transform the creatinine values before entering them into our logistic model. However the reviewer wrote "Using a log-transform for creatine is absurd because a 1-unit increase in ln(x) is equivalent to increasing x by a factor of 2.718 which is in the realm of impossibility"

It is possible for a creatinine ratio to be exactly e (and therefore approximately 2.718). However, it is probably impossible to explain to a non-mathematician. I personally transform the X-variate to centred

binary logs. If -creat- is creatinine, and I think a good central creatinine value for healthy people is 1.2, then my do-file might contain lines like

gene l2creat=(log(creat)-log(1.2))/log(2)

lab var l2creat "Log_2(Creatinine/1.2)"

I would then enter -l2creat- into my logistic regression model as an X-variate, and compute a confidence interval for a per-doubling odds ratio, which is a scaling of the odds due to a doubling of creatinine.

The centering is done because usually, in a logistic regression, I want to see the baseline odds, which is in this case the odds of an event if creatinine is 1.2 units. To do this, I would type something like

gene byte baseline=1

lab var baseline "Baseline odds"

logit y baseline l2creat, noconst or robust

The parameter for -baseline- is the baseline odds, and the parameter for -l2creat- is the per-doubling odds ratio.

I hope this helps.

Roger

--

Roger Newson

Lecturer in Medical Statistics

Department of Public Health Sciences

King's College London

5th Floor, Capital House

42 Weston Street

London SE1 3QD

United Kingdom

Tel: 020 7848 6648 International +44 20 7848 6648

Fax: 020 7848 6620 International +44 20 7848 6620

or 020 7848 6605 International +44 20 7848 6605

Email: [email protected]

Website: http://www.kcl-phs.org.uk/rogernewson

Opinions expressed are those of the author, not the institution.

*

* 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/

**Follow-Ups**:**st: comparing coefficients***From:*Ricardo Ovaldia <[email protected]>

**References**:**st: time series data***From:*Christopher F Baum <[email protected]>

**st: Transformed values in logistic regression***From:*Ricardo Ovaldia <[email protected]>

- Prev by Date:
**Re: st: Transformed values in logistic regression** - Next by Date:
**st: conformability error in ivtobit** - Previous by thread:
**Re: st: Transformed values in logistic regression** - Next by thread:
**st: comparing coefficients** - Index(es):

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