Figure 3.

Multivariable forest plot of parametric predictors driving measles seronegativity. Derived from fully adjusted multivariable logistic regression modeling; independent markers are plotted on a logarithmic scale with their corresponding 95% confidence intervals.