Continuous Covariate (RHS - Right Hand Side) models:

    Based on the figures below, proposed possibilities for the RHS formulations for the continuous covs. are:

            1.) Age at Death:  Linear 

            2.) Months from Diagnosis to Death: Piecewise Linear, knots at 12 & 24 months.

            3.) Census Tract Median Income:  Piecewise linear or cubic

            4.) Census Tract % High School Grads: Piecewise linear or cubic

Continuous covariate investigations using smoothers to detect nonlinearity:

Marginal Covariate Relationships: E(Yearly Reimbursements) = smooth(covariate)

Each Figure Contains:
                                    Upper Left Graph:    Actual Data & smooth
                                    Other Graphs:          Smooth Function & 95% Confidence Limits with
                                                                    different degrees of freedom (lower = more smooth).

Figure 1:  Age at Death Smooths

Figure 2:  Months from Diagnosis to Death Smooths

Figure 3:  Census Tract Median Income Smooths

Figure 4:  Census Tract % High School Graduates Smooths