Figure 3.

Comparison of Savings Estimates From Primary Approach vs. Holding Baseline Assignments Constant vs. Using Prospective Assignments From Two Years Before Each Study Year^a

^aEstimates from analyses holding baseline assignments fixed over study period were inflated to correct for attenuation bias as described in the methods section. Similarly, estimates from analyses using prospective assignments based on claims two years before each study year were inflated to correct for attenuation bias introduced by prospective assignment. Of beneficiaries prospectively assigned to an ACO in a cohort in year t‐2, an average of 81% remained assigned to an ACO in year t; thus, inflation factors averaged 1.23 across cohorts.