This measure estimates the risk-standardised rate of unplanned, all-cause readmissions for patients (Medicare fee-for-service [FFS] beneficiaries) discharged from an Inpatient Rehabilitation Facility (IRF) who were readmitted to a short-stay acute-care hospital or a Long-Term Care Hospital (LTCH), within 30 days of an IRF discharge.

The numerator is mathematically related to the number of patients in the target population who have the event of an unplanned readmission in the 30- day post-discharge window. The measure does not have a simple form for the numerator and denominator—that is, the risk adjustment method used does not make the observed number of readmissions the numerator and a predicted number the denominator. Instead, the numerator is the risk-adjusted estimate of the number of unplanned readmissions that occurred within 30 days from discharge. This estimate includes risk adjustment for patient characteristics and a statistical estimate of the facility effect beyond patient mix.

The denominator is computed with the same model used for the numerator. It is the model developed using all non-excluded IRF stays in the national data. For a particular facility the model is applied to the patient population, but the facility effect term is 0. In effect, it is the number of readmissions that would be expected for that patient population at the average IRF. The measure includes all the IRF stays in the measurement period that are observed in national Medicare FFS data and do not fall into an excluded category.

The measure excludes some IRF patient stays; some of these exclusions result from data limitations. The following are the measure's denominator exclusions, including the rationale for exclusion: 1. IRF patients who died during the IRF stay. Rationale: A post-discharge readmission measure is not relevant for patients who died during their IRF stay. 2. IRF patients less than 18 years old. Rationale: IRF patients under 18 years old are not included in the target population for this measure. Paediatric patients are relatively few and may have different patterns of care from adults. 3. IRF patients who were transferred at the end of a stay to another IRF or short-term acute care hospital. Rationale: Patients who were transferred to another IRF or short-term acute-care hospital are excluded from this measure because the transfer suggests that either their IRF treatment has not been completed or that their condition worsened, requiring a transfer back to the acute care setting. The intent of the measure is to follow patients deemed well enough to be discharged to a less intensive care setting (i.e., discharged to less intense levels of care or to the community). 4. Patients who were not continuously enrolled in Part A FFS Medicare for the 12 months prior to the IRF stay admission date, and at least 30 days after IRF stay discharge date. Rationale: The adjustment for certain comorbid conditions in the measure requires information on acute inpatient bills for 1 year prior to the IRF admission, and readmissions must be observable in the observation window following discharge. Patients without Part A coverage or who are enrolled in Medicare Advantage plans will not have complete inpatient claims in the system. 5. Patients who did not have a short-term acute-care stay within 30 days prior to an IRF stay admission date. Rationale: This measure requires information from the prior short-term acute-care stay in the elements used for risk adjustment. 6. IRF patients discharged against medical advice (AMA). Rationale: Patients discharged AMA are excluded because these patients have not completed their full course of treatment in the opinion of the facility. 7. IRF patients for whom the prior short-term acute-care stay was for nonsurgical treatment of cancer. Rationale: Consistent with the HWR Measure, patients for whom the prior short-term acute-care stay was for nonsurgical treatment of cancer are excluded because these patients were identified as following a very different trajectory after discharge, with a particularly high mortality rate. 8. IRF stays with data that are problematic (e.g., anomalous records for hospital stays that overlap wholly or in part or are otherwise erroneous or contradictory). Rationale: This measure requires accurate information from the IRF stay and prior short-term acute-care stays in the elements used for risk adjustment. No-pay IRF stays involving exhaustion of Part A benefits are also excluded.

Statistical risk model

Claims, Other

Measure Retired and Endorsement Removed

www.qualityforum.org/QPS/2502 www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/IRF-Quality-Reporting/Downloads/PV-Copy-1-of-DRAFT-Specifications-for-the-Proposed-All-Cause-Unplanned-30-day-Post-IRF-Discharge-Readmission-Measure.pdf