Use propensity scores in combination with Cox models for time (fu) to death, to re-evaluate the effect of donor type (txtype) on survival following pediatric kidney transplant from Problem 9.5. Using...

Use propensity scores in combination with Cox models for time (fu) to death, to re-evaluate the effect of donor type (txtype) on survival following pediatric kidney transplant from Problem 9.5. Using your propensity scores, check balance, overlap of the living and cadaveric donor groups, and evidence for positivity violations. Implement models using quintile, decile, and a 5-knot restricted cubic spline in the propensity scores. Are the results consistent with standard adjustment? What would you do to address evidence for lack of overlap?

Problem 9.5

Using the UNOS data on the book website, estimate the marginal effect of donor type (cadaveric vs living) on 5-year mortality risk, adjusting for recipient age and sex, donor age (age don), HLA match (hlamat), graft status (graf stat), and previous treatment (prev ki).