Interpretation binary logistic regression in SPSS

#1
Hello!

I am doing research on colorectal cancer for my master's degree in medicine. I have a total of 177 patients, out of which 5 have residual malignant disease after resection (primary endpoint). Due to the small number of events I want to perform a binary logistic regression analysis. I have never done this before, but have given it my best try. I know that I can find whether or not the variable is a predictive/a risk factor in the table called ''Variables in the equation'' under ''Sig.''. I consider a p value <0.05 significant. I have a problem interpreting my results. One of the factors I am looking at has a p value <0.05 so is significant, but there are also two factors with p values of 0.998 and 0.997. Those two factors look at morphology of the colorectal polyp and the ASA-score, which are both categorical. All patients with an event fall into the same categories for morphology and ASA-score, which makes me wonder whether I should interpret those two p values as significant too. Can anybody please help me explain whether those p values are significant or not?

Thank you!
 
#3
Thank you for your response! I know I do not meet the rule of thumb of one predictor for every 10 events. However, my supervisor still asked me to perform binary logistic regression, and to just mention in the limitations that very few events happened. Do you know if I these high p values are considered to be significant too?
 

noetsi

Fortran must die
#4
P values that high would never be significant. But of course the p values might have no meaning at all because you are violating a central requirement for the method by having too few cases (although not a formal assumption). If you have theory saying they are significant, than this is the time to mention that :p