The effect of severity on first site of PAC demonstrates that the odds of discharge to IRFs, LTCHs, and SNFs are much higher compared to the odds of discharge to hospital outpatient therapy or home health for beneficiaries of higher severity. For example, in the multinomial model run using the APR-DRG severity measures (Set 1, Table 3-24), the odds of discharge to LTCH for beneficiaries in APR-DRG severity level 4 (extreme) were 5.2 times the odds of discharge to hospital outpatient therapy. Using the APR-DRG severity levels variables, it is clear that the odds of discharge to inpatient post-acute settings increases with increasing severity. While similar patterns are observed in using MS-DRG severity measures (Set 2, Table 3-25), and MS-DRG and HCCs as severity measures (Set 3, Table 3-26), the magnitude of the odds ratios in these models were not as large. The model using the MS-DRG and HCC severit measures (Table 3-26), does highlight the role of particular HCCs in predicting first site of PAC. For example, the odds of discharge to IRFs for beneficiaries with HCC 158 Hip Fracture/Dislocation are 9.9 times the odds of beneficiaries without this HCC due to the rehabilitative services that these patients often require. Beneficiaries with HCC 79 Cardio-Respiratory Failure and Shock had 7.5 times the odds of being discharged to LTCHs compared with beneficiaries without this HCC and this may reflect the medical complexity of these patients.
Patterns of the effect of age varied across models and by first site of PAC (Tables 3-24, 3-25, and 3-26). Across all models, the odds of being discharged to SNFs increased with increasing age. Also across all models, the odds of being discharged to HHAs were lower for beneficiaries aged 85 or older compared with the odds for beneficiaries aged 75-84, indicating that these oldest beneficiaries may be more likely to be discharged directly to inpatient PAC following an inpatient stay.
The findings from the organizational variables indicate that organizational relationships do have some effect on the first site of care for beneficiaries discharged to PAC. In the models using the APR-DRG severity measures (Set 1, Table 3-24), we found that for beneficiaries discharged from an acute hospital with an IRF subprovider, the odds of being discharged to an IRF were 1.9 times the odds of being discharged to outpatient therapy. Similarly, for beneficiaries discharged from an acute with a colocated IRF, the odds of discharge to IRF were 2.3 times the odds of discharge to outpatient therapy. Similar results were seen across the models (Table 3-25 and Table 3-26). The organizational relationship variables also indicated increases in the odds of discharge to SNF for beneficiaries discharged from providers with SNF subproviders or colocated SNFs and an increase in odds of discharge to LTCH for beneficiaries discharged from acute hospitals with colocated LTCHs. The presence of organizationally related home health agencies did not increase the likelihood of discharge to HHA.