Spss cox proportional hazards model

Step-by-Step Guide to Assignment 9.3

Problem 3. Use SPSS to run a Cox Proportional Hazards test

If closed, open the Practice_Week09_PRACTICE_dataset.sav data set.

Step 1. Go to Analyze ( Survival ( Cox regression.

image1.png

a) Use stroke as the event and hypertension as the factor. (be sure to identify hypertension as a categorical variable). Be sure to include the output in your submission.

Step 2. Place the Time followed in Months variable in the Time box and the “Had stroke [stroke]” variable in the Status box. Click Define Event.

image2.png

Step 3. Select the Single value radio button and type 1in the Single value box. Click Continue.

image3.png

Step 4. In the Cox Regression window, move the Had hypertension [hypertension] variable to the Covariates box. Click Categorical.

Step 5. In the Cox Regression: Define Categorical Covariates, move Had hypertension [hypertension] from the Covariates to the Categorical Covariates box. In the Change Contrast area, select the radio button for First and click Change. In the Categorical Covariates box, “(first)” should follow hypertension (Indicator), as below. Click Continue.

image4.png

Step 6. In Cox Regression window, click Options. Check CI for exp(B) 95% in Model Statistics area. Click Continue.

image5.png

b) Create a Hazard Plot for Hypertension (select separate lines)

Step 7. In the Cox Regression window,click on the Plots button.Check the Hazard box for Plot Type. Move the hypertension (Cat) (Mean) variable to the Separate Lines for box as below. Click Continue.

image6.png

Step 8. In the Cox Regression window, click OK to run the analysis.

c) Interpret the results.

Step 9. Review the Output screen:

Variables in the Equation

B

SE

Wald

df

Sig.

Exp(B)

95.0% CI for Exp(B)

Lower

Upper

hypertension

1.396

.274

25.941

1

.000

4.041

2.361

6.916

image7.png

We are examining whether the hazard for stroke is different in those with and without hypertension. To do this, we only include those with events (stroke = 1). The hazard plot shows the difference in the hazard for stroke in those with hypertension (green line) and those without stroke (blue line) over the 32 month follow-up period.

Be sure to describe how the lines differ and whether this is statistically significant. Interpret the HR and the 95% CI.

d) Could the presence of ties in time followed (from the frequency table in 1c) affect the results? How might you accommodate ties in this analysis?