I am planning to start a study to evaluate a measure of sarcopenia (x) to predict 6-months mortality in hospitalised patients with a disease (D) for which previous studies showed around seventy percent 6-month mortality rate. I intend to run survival analysis with cox proportional regression and asses x as a predictor for 6-months mortality in D patients. My questions are:
1- What would be the best study design (Cohort of D cases and asses outcome or should it be something like case-control given that healthy controls are unlikely to die in 6 months!)
2- Is survival analysis better than logistic regression in this context?
3- How to calculate sample size with the power of 80% and a significance level of 0.05 given that the 6-months mortality in D patients population was about seventy percent in previous studies?
Thank you so much for your valuable insights…