Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: epiconf: counfounding

 From Nick Cox To statalist@hsphsun2.harvard.edu Subject Re: st: epiconf: counfounding Date Tue, 8 Feb 2011 00:55:31 +0000

```-epiconf- is a user-written program (Z. Wang, STB-49).

Nick

On Mon, Feb 7, 2011 at 10:43 PM, SR Millis <aa3379@wayne.edu> wrote:
> I've been working with your Stata program, epiconf.  Using theaccompanying dataset and fitting the following logistic regression model:
>
> . epiconf dead ab_uria, con(age weight) cat(hichol hypert smoke sex) backward
>
> I obtained the following results.  How do you interpret the model when the changes in the absolute values of the change in the odds ratios are not monotonically increasing (in the case of backward selection) or, in the case of forward selection, monotonically decreasing?  That is, in this model, if we take 10% as the cut-point of importance, we need only adjust for age, smoking, and weight?  If so, how do you obtain the adjusted rate ratio and its confidence interval?
>
>
> Assessment of Confounding Effects Using Change-in-Estimate Method
> Exposure:   "ab_uria"
> N =          743
>
> Backward approach
> Potential confounders were removed one at a time sequentially
>
> Adj Var              95% CI     Change in OR      p
>
> Crude         2.35   1.24, 4.44          .         .
> -i.sex        2.26   1.21, 4.24       -3.6    0.47664
> -i.hichol     2.08   1.12, 3.86       -8.0    0.15144
> -weight       1.88   1.04, 3.40      -10.0    0.21915
> -i.hypert     2.00   1.13, 3.55        6.6    0.42812
> -i.smoke      2.24   1.28, 3.92       11.9    0.06456
> -age*         3.90   2.34, 6.52       74.4    0.00000
>
> *Crude estimate
>
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```