Did Site-Neutral Payment for Long-Term Care Hospitals Reduce Financial Incentives to Time Discharge?

OBJECTIVES

Historically, fee-for-service Medicare reimbursed long-term care hospital (LTCH) stays as a lump-sum payment, which was substantially reduced for discharges before the diagnosis-specific short-stay outlier (SSO) threshold day, leading to large spikes in discharges on the threshold day. The objective of this study was to examine if LTCHs similarly time discharge at the SSO threshold for blended site-neutral payment cases compared with standard payment cases.

DESIGN

Cross sectional.

SETTING AND PARTICIPANTS

Cohort of Medicare beneficiaries.

METHODS

We used the national 100% LTCH Medicare Provider Analysis and Review Limited Data Set for fiscal year 2017 and exploited differences in prior ICU days using a multinomial model adjusting for patient demographics and case-mix. We only included stays where payment status was solely determined by prior ICU days. Our primary outcome was the discharge on the SSO threshold day. The exposure was an indicator variable for ≥3 intensive care unit days in the preceding acute care hospital, where yes equated to standard payment and no was a blended site-neutral case.

RESULTS

Among 10,910 LTCH discharges (48% of cases were aged 65-74 years, 52% were female, 25% were non-white, 24% were blended site-neutral cases), we found that despite an approximately 50% reduction in payment increase, the spike in the adjusted probability of discharge on the SSO threshold day vs the day before was similar for blended site-neutral (20% vs 2%) and standard payment cases (16% vs 1%), with an adjusted difference of -3% (95% CI, -5% to 2%).

CONCLUSIONS AND IMPLICATIONS

Unwarranted spikes in discharge on the SSO threshold persisted despite a 50% reduction in the payment increase, a crucial insight for policymakers and payers who seek to avoid incentives to strategically time discharges. LTCH stays should be reimbursed without meaningful payment increases based on length of stay thresholds.