Biostatistics For Dummies, 2nd Edition (Monika Wahi, John Pezzullo)

CHAPTER 23 Survival Regression 337

Interpreting the Output of

a Survival Regression

Suppose that you have conducted a long-term survival study of 200 Stage 4 cancer

patients who were enrolled from four clinical centers (A, B, C, and D) and were

randomized to receive either chemotherapy or radiation therapy. Participants

were followed for up to ten years, after which the survival data were summarized

by treatment (see Figures 23-3a and 23-3b for the two treatments) and by clinical

center.

It would appear, from Figure 23-3, that radiation (compared to chemotherapy)

and clinical centers A and B (compared to C and D) are associated with better sur-

vival. But are these apparent effects statistically significant? PH regression can

answer these and other questions.

To run a PH regression on the data from this example, you must indicate the fol-

lowing to the software in your code:»

» ^{The time-to-event variable.}^{We named this variable}^Time,^{and it was coded}

in years. For participants who died during the observation period, it was

coded as the number of years from observation beginning until death. For

participants who did not die during the observation period, it contains

number of years they were observed.»

» ^{The event status variable.}^{We named this variable}^Status,^{and coded it as}

1 if the participant was known to have died during the observation period,

and 0 if they did not die.

FIGURE 23-3:

Kaplan-Meier

survival curves by

treatment and

clinical center.