Post-hospitalization referral to home health care and subsequent rehospitalization for Medicare patients with congestive heart failure

Abstract

A high percent of Medicare dollars are attributed to a small number of beneficiaries who have a recurring need for acute care intervention, often through multiple hospitalizations. The association between the use of home care services and hospital readmissions continues to have conflicting reports in the literature. The purpose of this study was to determine whether a referral to home care has an association with hospital readmission. The Behavioral Model for Health Services Use provided the conceptual framework for the study. Data for Medicare patients who had an index hospitalization in 1995 for congestive heart failure were obtained from the Connecticut Health Information and Exchange (CHIME). A total sample of 4094 subjects was randomly split into two subsamples to create a calibration sample and a validation sample. A multiple logistic regression analysis was conducted on the calibration sample of 2038 subjects to determine those factors predictive of a hospital readmission. An additional logistic regression analysis examined the 902 subjects who were readmitted to determine the factors associated with a prior referral to home care. When the predisposing variables age, gender and race were entered into the logistic regression with referral to home care, the referral was a statistically significant predictor of readmission (chi2(1) = 8.69, p =.003). In further analysis with additional factors, the number of prior admissions was the only statistically significant variable in predicting the likelihood of readmission (chi2(1) = 68.77, p = <.001). For those who were readmitted, older women with prior admissions were more likely to have received a referral to home care. Those who had lower ancillary hospital charges for the readmission also were more likely to have received a referral to home health care (chi2(1) = 12.17, p<.001). The regression equations were cross-validated on the validation sample of 2056 subjects. The logistic models were able to correctly classify approximately 61% of the subjects for both the calibration and the validation samples. The reduction in ancillary charges for readmitted patients who had received a referral to home health care is an interesting finding worthy of further investigation. A better understanding of the relationship between home care and hospitalization has important implications for public policy and nursing practice.