To provide protection against
model overfitting, the number of parameters in
the model for comparison of cause-specific mortality
due to CNS injury was reduced because of
the small number of events. To avoid an increased
type 1 error rate from multiple testing, we provide
only descriptive estimates for proportionate
mortality; we did not adjust for multiple comparisons
of cause-specific mortality. We compared
the time to death among the three periods with the
Kaplan–Meier method and used the Wilcoxon test,
because this test applies more weight to early
deaths than does the log-rank test. We considered
a two-sided P value of less than 0.05 to indicate
statistical significance.
Results