1. Are Medicare Advantage Patients Less Likely to have a Readmission? Bernard Friedman, PhD Joanna Jiang, PhD John Bott MSSW, MBA Claudia Steiner MD, MPH AHRQ Conference, 2010

2. Background Hypothesis: the Advantage plans have both the motive (capitated revenue) and the means to reduce readmissions in comparison to FFS Hypothesis: the Advantage plans have both the motive (capitated revenue) and the means to reduce readmissions in comparison to FFS use 2006 data use 2006 data 14 HCUP states distinguish FFS and MA payer on discharge summary 14 HCUP states distinguish FFS and MA payer on discharge summary 5 states in HCUP also had reliable person identifiers. AZ, CA, FL, Mass, TN. 5 states in HCUP also had reliable person identifiers. AZ, CA, FL, Mass, TN. NY could now also be used NY could now also be used

3. Data Preparation Selecting Cases Selecting Cases – elderly Medicare beneficiary patient from urban CBSAs – initial admission Jan-Sept of 2006 – match to Linkage file (work-saving tool, uses no actual dates, releasable => replicable study, protecting identity) – use primary and secondary payer – discharged alive from initial admission Screening out re-hospitalizations Screening out re-hospitalizations – same day transfer (using Linkage File) – re-hospitalization says transferred in from another hospital – pregnancy or maternity-related – trauma-related principal condition (CCS 225-236)

4. Data Preparation Dependent Variables: at least one readmission within 30 days, 90 days Dependent Variables: at least one readmission within 30 days, 90 days Notes: Notes: – Each person enters only once, from initial admission. – Not addressing the “burden of readmissions”, i.e., multiple readmissions within any fixed period, overlapping episodes for a person during the year. – ED revisits not included

5. Initial comparisons 266K MA initial discharges, 868K initial FFS discharges 266K MA initial discharges, 868K initial FFS discharges 30 day readmission rate: 13.7% vs. 14.4% 30 day readmission rate: 13.7% vs. 14.4% 90 day readmission rate: 23.8% vs. 26.2% 90 day readmission rate: 23.8% vs. 26.2% However, Medicare Advantage patients tend to be However, Medicare Advantage patients tend to be – younger – less severely ill even when hospitalized – less likely to be Medicaid enrrolled as well – less likely to have a major operative procedure. for more detailed comparison over 13 states, see Stat Brief #66. for more detailed comparison over 13 states, see Stat Brief #66. Other work: AHIP study of CA and NV found a lower rate of preventable admissions for Advantage enrollees. Other work: AHIP study of CA and NV found a lower rate of preventable admissions for Advantage enrollees.

6. Problem: Self-Selection Instrumental Variable approach: Instrumental Variable approach: – What can we measure that could affect enrollment in an Advantage plan rather than FFS? – Instrument should not itself affect the likelihood of readmission R = f (M, X) R = f (M, X) M= g(Z) M= g(Z) – Z affects M but not R (standard tests for the performance of Z set) Z: CBSA characteristics. HMO penetration among younger insured, number of competing hospitals, available medical specialists, Z: CBSA characteristics. HMO penetration among younger insured, number of competing hospitals, available medical specialists,

7. Results of Tests Preliminary Preliminary Controlling for self-selection, Medicare Advantage patients had a 9% higher likelihood of readmission in 30 days. Controlling for self-selection, Medicare Advantage patients had a 9% higher likelihood of readmission in 30 days. Tests of the Instruments are passed Tests of the Instruments are passed Caveats: high users, ED visits Caveats: high users, ED visits Looking for a Reconciliation Looking for a Reconciliation – Enrollees do not have data to compare readmission rates when choosing a plan – Health Plans do not have adequate measures of risk-adjusted readmission rates for quality improvement efforts – Saving money on readmissions is not the only way to save.