Non significant regression

#1
Hi all,

I have what is surely an embarrassingly basic question.

I am looking at the relationship between a specific medical diagnosis and unplanned admission to the hospital following outpatient surgery. My dataset has 100 patients. 77 without the diagnosis, 23 with. 5 patients in the dataset were admitted following surgery. All of them have the medical diagnosis.

Thus, it would seem that the diagnosis increases the odds of admission. However, when I perform the logistical regression analysis, it’s not significant with a p value of nearly 1. Why is this?

Thank you!!
 

hlsmith

Not a robit
#2
Not near a computer, so i cant run these numbers. Given the rarity of the outcome perhaps you cannot rule out chance given the small dataset.Though I am surprised the model converged. Can you post the model's output to confirm it ran correctly.

Typically when you have a small sample and complete separation you need to use the Firth penalization.
 
#3
Thanks for your response! Here is my SPSS output. Your idea that it's due to the small sample size and chance is what I was thinking, considering the Classification tables for both the block 0 and block 1 are the same, even with the inclusion of the medical condition (OSA).

Any idea how to run the Firth penalization in SPSS?

Many thanks


Block 1.PNG
Block 2.PNG
Block 3.PNG