I am evaluating the association between levels of a biomarker A and outcome(occurrence of a disease). The levels of a biomarker is a continuous variable. To assess the association between the levels of a biomarker A and the odds of the outcome, i've used restricted cubic spline analysis with logistic regression as below in R:
d<-datadist(df)
options(datadist="d")
fit <- lrm(outcome~rcs(levels,4), data=df)
summary(fit)
anova(fit)
And the results were shown as
> summary(fit)
Effects Response : outcome
Factor Low High Diff. Effect S.E. Lower 0.95 Upper 0.95
levels 72.5 73.1 0.6 -0.004536 0.13166 -0.26259 0.25352
Odds Ratio 72.5 73.1 0.6 0.995470 NA 0.76906 1.28850
> anova(fit)
Wald Statistics Response: outcome
Factor Chi-Square d.f. P
levels 6.28 3 0.0986
Nonlinear 2.38 2 0.3050
TOTAL 6.28 3 0.0986
How can I interpret the results?
Is it right that Restricted cubic spline analysis demonstrated a linear relationship between the levels of A and theoutcome (p-value for nonlineairty 0.3050) but the levels did not significantly associated with the outcome (OR 1.0, 95% CI 0.77-1.29).