Explaining the Recent Utilization Trends in Ultra-High Rehabilitation Therapy Services in Skilled Nursing Facilities

In recent years, Medicare has observed increased billing for therapy services by skilled nursing facilities (SNFs). Under the SNF prospective payment system, Medicare reimburses SNFs on a per diem basis, with the amount paid based on a patient’s “resource utilization group” (RUG). The percentage of SNF admissions for rehabilitation that were assigned to “ultra-high” therapy RUGs, defined as receiving 720 or more minutes of therapy in a week across at least two therapy disciplines (speech, occupational, or physical therapy) increased from 32.3% in 2008 to 61.7% in 2014, among the beneficiaries aged 65 and older. Notably, SNFs receive considerably higher payment for providing high levels of therapy. Due to concerns of upcoding, the Centers for Medicare and Medicaid (CMS) will soon revamp the SNF payment system by eliminating categorization by therapy minutes and instead focusing on patients’ clinical acuity and needs. Our paper investigates whether the increase in SNFs’ use of ultra-high therapy in the past decade is explained by an increasing acuity of SNF admissions versus other factors.

We first identified a set of potential acuity measures that do not rely on SNFs’ patient assessments. We then investigated the performance of different sets of these acuity measures in predicting the use of ultra-high therapy in a base year (2008) and then compared actual versus predicted ultra-high status in subsequent years (2009-2014). We examined other factors that may have been related to growth in ultra-high therapy assignment, including facility characteristic and changes in access to other intensive rehabilitation. We also investigated whether the increase was associated with reductions in other health care use or improvements in episode outcomes and cost.

Our study sample included episodes of SNF care initiated between 2008 and 2014 for individuals aged 65 and older. In each case, patients were discharged to a SNF immediately after hospitalization, which did not occur within 30-days of a prior SNF discharge.

We found that the actual overall growth rate between 2008 and 2014 in ultra-high assignment (29.4 percentage points) was much higher than that predicted based on patient acuity (only 0.4 percentage points). We found that for-profit SNFs and SNFs with a higher volume of rehabilitation admissions were more likely to assign patients to ultra-high therapy. We found mixed evidence that increased assignment to ultra-high therapy was associated with reductions in hospital readmissions or better patient outcomes. Rather, we found that the growth in ultra-high therapy assignment was primarily associated with higher Medicare SNF and overall episode spending.

Our results suggest that the increase in the percentage of SNF rehabilitation admissions assigned to ultra-high status between 2008 and 2014 was not driven by changes in patient composition and severity. Our empirical evidence supports CMS’s new Medicare SNF payment system starting from fiscal year 2020, the Patient-Driven Payment Model (PDPM), which is not based on therapy minutes but instead focused on patients’ clinical acuity and needs. What will happen to SNF therapy provision under the PDPM remains to be seen.