As in the other models, severity is an important predictor of acute hospital readmission during PAC episodes. The odds of readmission increase with increasing severity. In the binomial logit model using both the MS-DRG severity measures and the HCCs (Set 3, Table 3-27), the odds of readmission for beneficiaries with MS-DRG w/MCC are 1.4 times the odds of readmission for beneficiaries without CCs. The odds of readmission are also higher for beneficiaries with particular HCCCs. The odds of readmission for beneficiaries with HCC80 Congestive Heart Failure are 1.3 times those for beneficiaries without this HCC.
The odds of readmission during a PAC episode also varied by demographic characteristics. The odds of readmission were higher for older beneficiaries, for beneficiaries enrolled in Medicaid, and for non-white beneficiaries. The post-acute care supply variables were not significant predictors of readmission. In looking at geography, there were very few significant differences compared to the New England region. In looking at the binomial model using the MS-DRG severity measure and the HCCs, the odds of readmission were significantly higher in the Middle Atlantic (OR=1.112) and the West South Central (OR=1.145) compared to New England. The beneficiary's index admission length of stay was also a significant predictor of readmission in this model (OR=1.033) (Set 3, Table 3-27).
Another interesting finding to note in this regression is related to the organizational relationship variables. The odds of readmission for beneficiaries discharged to colocated providers are not significantly different from those for other beneficiaries. Using the MS-DRG severity measure and the HCCs (Sets 2 and 3, Table 3-27), we found that the odds of being readmitted during a PAC episode was slightly lower than for other beneficiaries.