Summary: To address concerns that the Hospital Readmissions Reduction
Program (HRRP) unfairly penalized safety net hospitals treating patients with
high social and functional risks, Medicare recently modified HRRP to compare
hospitals with similar proportions of high-risk, dual-eligible patients
("peer group hospitals"). Whether the change fully accounts for
patients' social and functional risks is unknown.
Authors examine risk-standardized readmission rates (RSRRs) and
hospital penalties after adding patient-level social and functional and
community-level risk factors.
Using 2000–2014 Medicare hospital discharge, Health and
Retirement Study, and community-level data, authors identified latent factors
for patient social and functional factors and community factors. Authors
estimated RSRRs for peer groups and by safety net status using four
hierarchical logistic regression models: "base" (HRRP model);
"patient" (base plus patient factors); "community" (base
plus community factors); and "full" (all factors). The proportion of
hospitals penalized was calculated by safety net status. Authors looked at
20,255 fee-for-service Medicare beneficiaries (65+) with eligible index
hospitalizations.
Findings: Half of safety net
hospitals are in peer group 5. Compared with other hospitals, peer group 5
hospitals (most dual-eligibles) treated sicker, more functionally limited
patients from socially disadvantaged groups. RSRRs decreased by 0.7% for peer
groups 2 and 4 and 1.3% for peer group 5 under the patient and full (versus
base) models. Measured performance improved after adjusting for patient risk
factors for hospitals in peer group 4 and 5 hospitals, but worsened for those
in peer groups 1, 2, and 3. Under the patient (versus base) model, fewer safety
net hospitals (48.7% versus 51.3%) but more non-safety net hospitals (50.0%
versus 49.1%) were penalized.
Patient-level risk adjustment decreased RSRRs for hospitals
serving more at-risk patients and proportion of safety net hospitals penalized,
while modestly increasing RSRRs and proportion of non-safety net hospitals
penalized. Results suggest HRRP modifications may not fully account for
hospital variation in patient-level risk.