TTEST vs Logistic Regression

I have seen some posts here with the same title but I believe my problem is different:

I am unsure whether to conduct a 2 sample ttest or a logistic regression because I am unsure of which variable should be my outcome and which should be my predictor variable.

The problem:
There are 5 cadavers chosen to be studied. The lumbar vertebrae are chosen from each cadaver to be studied. An operation is performed on the vertebrae and we are interested to find out if the operation fractured the bone or not.

The question:
Is age an important predictor of fracturing a vertebrae?

My approach:
At first I ran a logistic regression because I treated the fracture as an outcome and age as a predictor in that model.
But someone else told me they did a 2 sample ttest where they took all those that are fractured and all those that are not fractured and determined to see if the mean age was different for these two groups of bones.

The results are different (one significant and one not significant) and we can't decide which result to rely on.

I would greatly appreciate any help!




Less is more. Stay pure. Stay poor.
Logistic all the way. You said age predicts fracture. You can report the means for the two groups or boxplot, etc., but the outcome should be conducted with logistic. You will end up with: for every blank unit increase in patient age the odds of fracture increases (or decreases) by blank (95% CI) percent.