Time-to-event
Basic Medical Statistics Course: Module COctober 2010Wilma Heemsbergen
SPSS file: trial_rt.savInt J Radiation Oncology Biol Phys 2008; 72: 980-988.Update of Dutch Multicenter Dose-escalation Trial of Radiotherapy for Localized
Prostate Cancer.
Paste the syntax of the applied statistical procedures.
1. Construct a Kaplan Meier graph, similar to Figure 1-C.
- Generate the survival table.
- Look up: 5 y freedom from failure for both arms with standard errors.
2. Construct a similar Kaplan Meier graph for each prognostic group separately (risksf).
- Calculate the log rank statistic for each progn group separately, and an overall p value with the prognostic groups as strata.
3. - Perform a Cox regression with the variable arm.
- Perform a Cox regression with the variables arm and prognostic group.
Survival
Answers (total group)
Answers (total group)
EXP 78:5y = 60 months,estimate = 0.677, SE = 0.026
Answers (prog groups)
Answers (prog groups)
Answers (prog groups)
Answers (strata)
Answers (Cox Regression)COXREG efsh /STATUS=ffshi(1 THRU 3) /METHOD=ENTER arm /CRITERIA=PIN(.05) POUT(.10) ITERATE(20).
COXREG efsh /STATUS=ffshi(1 THRU 3) /METHOD=ENTER arm risksf /CRITERIA=PIN(.05) POUT(.10) ITERATE(20).
Syntax
Cox Regression (optional)
1. Construct the variables “death due to prostate cancer”, and “death due to other causes”.
2. Test in a univariate model whether the following factors have prognostic value (for each endpoint separately): rt dose, Gleason, T stage, baseline PSA, age, diabetes, smoking.
3. Construct a MV model with variables who have a significance level of p<0.1 in the MV model.