Survival Analysis:

Data are provided for 628 patients who were randomly assigned to one of two different residential

drug treatment programs that differed in length: (a) short program, and (b) long program. Time

until return to drug use was measured for all patients.

The SAS dataset that you will use for this exercise is called uis. It contains the following variables:

Variable Age in years at study enrollment

site: Two sites; 0=Site A; 1= Site B

treat: 0= short program; 1= long program

ndrugtx: Number of previous drug treatments

time: Time (in days) until return to drug use

censor: 0=censored; 1= target event observed

herco: 1= both heroin and cocaine use in the last three months

2 = either heroin or cocaine use in the last three months

3 = neither heroin nor cocaine use in the last three months

Question is

a. Fit an exponential survival regression model with treat, age, site, ndrugtx and herco as the

covariates. Would you reject that the scale parameter is equal to 1?

SAS code:

proc lifereg data = uis;

model time*censor(0)=treat age site ndrugtx herco /distribution=exponential;

run;

Analysis of Maximum Likelihood Parameter Estimates Standard Parameter DF Estimate Error 95% Confidence Limits Chi-Square Pr > ChiSq Intercept 5.0194 0.2639 4.5021 5.5367 361.69 <.0001 treat age site ndrugtx herco scale o weibull shape multiplier statistics parameter chi-square pr> ChiSq Scale 5.3598 0.0206Analysis of Maximum Likelihood Estimates Parameter Standard Hazard Parameter DF Estimate Error Chi-Square Pr > ChiSq Ratio Label treat -0.25541 0.09102 7.8737 0.0050 0.775 Treatment Randomization Assignment age -0.02228 0.00751 8.8151 0.0030 0.978 Age at Enrollment site 1 -0. 18815 0. 10156 3.4321 0.0639 0.828 Treatment Site ndrugtx 1 0.03573 0.00766 21.7704 <.0001 number of prior drug treatments herco>