Objectives: 1) Identify demographic, clinical social support, functional, and psychological factors about which data are available within 24 hours of hospital admission associated with emergent unscheduled readmission for a group of older general medicine patients; 2) develop a model to predict emergent readmission.
Design: Interview- and chart-based study of emergent admissions that occurred within 60 days of discharge.
Setting: General medicine wards of the Memphis Veterans Affairs Medical Center, an 862-bed university-affiliated tertiary care facility.
Patients/participants: General medicine patients greater than or equal to 65 years old (n = 173). Inclusion criteria were willingness to participate, written consent (patient or family member), and patient interview within 36 hours of admission.
Measurements and main results: The dependent variable was emergent readmission within 60 days of discharge from the hospital. Independent variables included demographic (age, race, income, education), social support (marital status, living arrangements), psychological (cognition, depression), activities of daily living functioning, and clinical (diagnoses, type and source of admission, length of stay, numbers of hospitalizations and days of hospitalizations in the past year, illness severity) parameters. Readmitted patients were emergently admitted and more severely ill, had more diagnoses of chronic obstructive pulmonary disease (COPD) or congestive heart failure (CHF), less ischemic heart disease, and more hospitalizations and hospital days in the past year (all p less than 0.05). Logistic regression identified diagnostic group (COPD or CHF), emergent admission, and admission severity of illness as predictive of readmission. The likelihood of being readmitted was 5.4. Accuracy of the three-variable model was 76%, predicted value positive, 73%, and predictive value negative, 77%.
Conclusions: Chronically ill patients who are severely ill at index admission and who have had several hospitalizations in the past year tend to be readmitted. Using this model, high-risk patients may be prospectively targeted to reduce readmissions.