logistic risk prediction model

[Carla Hope]

Hi,

I have developed a risk prediction model for Outcome (binary) based on 4 binary variables. I developed this on 70% of my dataset and kept 30% as the validation set (included below). AUROC was >0.8 and GOF was non-significant with good sensitivity and specificity.

Is there a way to calculate the OR of having just any one risk factor (bil/alp/ggt/CBDD) and then 2 risk factors, 3 risk factors, 4 risk factors?

logistic Outcome bil alp ggt CBDD
lroc
lsens
estat gof, group (10)
estat class, cutoff(0.2)
nomolog


I performed nomolog to get a graphical representation of the model and to be able to manually calculate for any individual but what I would like to say is your OR of Outcome is x if you have any one risk factor, x if you have 2 risk factors etc.

I would also like to plot predicted vs observed Outcome for my validation cohort using the model above if possible.
I tried the following for this but the output made sense (presumably because I have binary outcome??):
logistic Outcome bil alp ggt CBDD
predict predvar
predict resid, residuals
scatter resid predvar

Many thanks in advance,
Carla

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte CBDD float(bil alp alt ggt) byte outcome
1 1 1 0 0 1
1 1 1 1 1 1
0 1 1 1 0 0
1 1 1 1 1 0
1 1 1 1 1 1
1 1 1 1 1 0
1 1 1 0 0 0
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 0 0 1
1 1 1 1 1 0
1 1 1 1 1 0
1 1 1 1 1 0
0 1 1 1 1 0
1 1 1 1 1 1
0 1 1 0 0 0
0 1 1 1 1 1
1 1 1 1 1 1
0 1 1 0 0 0
0 1 1 1 1 1
1 1 1 1 1 0
0 1 1 1 1 0
0 1 1 1 1 0
0 1 1 0 0 1
0 0 1 0 0 0
1 1 1 1 1 0
0 1 1 1 1 1
0 1 1 0 0 0
0 1 1 1 1 0
0 1 1 1 1 0
0 0 1 0 0 0
0 1 1 1 1 1
0 1 1 1 1 0
0 1 1 1 1 0
0 1 1 1 1 0
0 1 1 0 1 1
1 0 1 1 1 1
0 1 1 1 1 0
0 1 1 1 1 0
1 0 1 0 0 0
0 1 1 1 1 0
0 1 1 1 1 0
0 1 1 1 1 0
1 0 1 0 1 0
0 1 1 1 1 0
0 1 1 1 1 0
1 0 1 1 1 0
0 1 1 1 1 0
0 1 1 1 1 0
0 1 1 1 1 0
1 0 1 0 1 0
1 1 0 0 0 0
0 1 1 1 1 1
0 1 1 1 1 0
0 0 1 0 0 0
0 0 1 0 1 0
0 0 1 0 0 0
0 1 1 1 1 0
0 1 1 1 1 0
1 0 1 0 1 0
0 1 1 1 1 0
0 0 1 0 1 0
0 1 1 1 1 0
1 0 1 1 1 0
0 0 1 0 0 1
0 0 1 1 1 0
0 0 1 1 1 1
0 0 1 0 0 0
1 1 0 1 1 0
0 1 0 0 0 0
0 1 0 0 0 0
0 1 0 0 0 0
1 1 0 1 1 0
0 0 1 1 1 0
1 1 0 0 1 0
0 0 1 0 1 0
0 0 1 1 1 0
0 0 1 1 1 0
1 0 0 0 0 0
0 0 1 1 1 0
0 1 0 0 0 0
1 0 0 0 0 0
1 0 0 0 0 0
1 0 0 0 0 0
1 1 0 1 1 0
0 1 0 0 0 0
0 0 1 1 1 0
1 0 0 0 0 0
0 1 0 1 1 0
1 0 0 0 0 0
0 1 0 1 1 0
0 0 1 1 1 0
0 0 0 0 0 0
0 1 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
1 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
end

------------------ copy up to and including the previous line ------------------

Listed 100 out of 226 observations
Use the count() option to list more

[Weiwen Ng]

The model you fit is:

logistic Outcome bil alp ggt CBDD

To this question:

Is there a way to calculate the OR of having just any one risk factor (bil/alp/ggt/CBDD)...

There isn't a way to do this with the model you fit, because you are treating each of those four risk factors as distinct and not identical. They might have effects that differ considerably. The issue applies to any combination of two or three of the risk factors. You could create a count of the 4 risk factors, and you would be getting something like the average risk of each risk factor. I think a lot of statisticians and clinicians would probably prefer each one modeled with a dummy.

Now, you might want to familiarize yourself with the margins post-estimation command. You can easily display the risk difference with respect to the base category. It's possible to get margins to display odds ratios, but it requires some work. You can then use marginsplot to plot. I think Richard Williams' primer on margins may be a bit easier to read than the manual examples, but those are also good.