Reference no: EM132322690

Assignment

Need fully worked out answers with graphs and the stata code used?

One hundred and fifteen patients were randomised to each arm of a chemotherapy trial.

The variables in the data set : 'patid'

- unique patient identifier; 'treat'

- treatment identifier (0 = standard, 1 = new); 'age'

- age at randomisation (years); 'ecog'

- ECOG performance status at randomisation (0, 1 or 2) 

- survival time from composite outcome of disease progression or death (in years); 'm_progdeath'

- composite outcome identifier (0 = no event observed (censoring), 1 = event was progression of disease or death); 's_death'

- survival time from death (in years); 'm_death'

- death identifier (0 = no event observed (censoring), 1 = patient died)

1) Produce a typical RCT "Table 1" to compare demographics (age and ECOG) between the treatment arms. A simple randomisation procedure was used to allocate patients to the two groups. Given the results, do you think another method of randomisation should have been used?

2) Run a Cox regression for the overall survival outcome with the treatment variable as the only covariate. Interpret the HR, the 95% confidence interval and the p-value. Are there any concerns with the proportional hazards assumption?

3) (In this question and for questions 4, 5 & 6, fit models with only the treatment covariate) What is the best fitting proportional hazards parametric model to the overall survival outcome?

4) What is the best fitting accelerated failure time parametric model to the overall survival outcome?

5) What is the best fitting proportional hazards parametric model to the progression free survival outcome?

6) What is the best fitting accelerated failure time parametric model to the progression free survival outcome?

7) Interpret the treatment HR for the best fitting proportional hazards model

8) Interpret the treatment coefficient for the best fitting accelerated failure time model

9) For the best fitting model overall, use a partitioned survival approach to compare a) life years and b) QALYs between the standard and new treatment arms. Use utilities of 0.85 and 0.6 for progression free and progression, respectively.

10) How plausible are the extrapolated survival curves for your best fitting model? In your answer you can refer to literature on the epidemiology of ovarian cancer and HTA literature on best practice for extrapolating survival curves.

Attachment:- Data Set File.rar