Figure 1.

Observed and Predicted Variability in Thirty-Day Mortality Based on SEER-Medicare Data from 2000 to 2005 across the Thirty Census Tracts with at Least Twenty CRC Patients Based on the Multivariable Model Note. Predicted 30-day mortality, based on the multilevel logistic regression model adjusted for sociodemographic, census-tract, hospital, clinical, tumor, and treatment characteristics, are plotted as circles above the encrypted census tract identifiers along the horizontal axis. Error bars indicate 95 percent confidence intervals. Census-tract-level predicted values were computed by averaging the patient-level predicted probabilities for all patients who resided in that census tract. Census tracts 1 through 12 had zero observed deaths; thus, there is no bar associated with these census tracts.