(154) Identifying Hospitalizations in Patients with Medicare Supplemental Insurance

Publication Number: 485

Background: Length of hospitalization and rate of readmission are important hospital performance measures. It is unclear if hospitalized days that are fully covered by Medicare Part A, which requires no co-pay/co-insurance for days 2-60, are consistently reflected in databases that rely on Medicare Supplemental insurance data that only contain claims for services covered at least in part by the supplemental plan.

Objectives: We compared length of stay (LOS) for hospital admissions for acute myocardial infarction (AMI) and 90-day readmission rates between fee-for-service (FFS) Medicare claims from CMS and FFS Medicare Supplemental claims from IBM MarketScan (MKS).

Methods: We conducted parallel analyses in MKS and a 20% sample of CMS Medicare claims. We identified patients’ (66+ yrs) first hospital admission for AMI between 2007 and 2017, requiring evidence of at least 365-days of continuous enrollment in an employer-sponsored supplemental FFS insurance plan and prescription drug coverage. Patients were followed until the first of hospital readmission, death, disenrollment, or reaching 90-days post discharge. We compared LOS, defined as the number of nights of the index admission, and 90-day readmission rates between the cohorts. We explored 2 methods of identifying inpatient stays in the MKS data. The first relied solely on admission and discharge dates from the inpatient admission and services files. The second incorporated inpatient-specific evaluation & management CPT codes with an inpatient place of service found in the outpatient services file. We evaluated the impact of the second method on the LOS and estimated 90-day risk of readmission.

Results: We identified 12,753 and 67,633 patients in CMS and MKS, respectively. Mean age (CMS: 77.8,MKS:77.6) and percent female (CMS:40.1%,MKS:41.4%) were similar. The mean LOS in CMS was 3.7d; 21.2% of patients were readmitted within 90 days. When relying only on the inpatient files in MKS, the mean LOS was 4.2d, and; 7.3% of patients were readmitted. When incorporating inpatient services from the outpatient file, the discharge date was adjusted for 11.8% of admissions. The mean LOS increased to 4.5d, while the readmission rate increased to 25.0%.

Conclusions: Due to the complex interplay in reimbursement between Medicare and Medicare Supplemental plans, readmissions occurring after patient deductibles have been met may not appear in inpatient files in Medicare Supplemental claims. This analysis of MKS Medicare Supplemental claims suggests that limiting hospitalizations to those in the inpatient file may substantially underestimate the risk of readmission.