RESUMO
BACKGROUND: Readmission is related to high cost, high burden, and high risk for mortality in geriatric patients. A scoring system can be developed to predict the readmission of older inpatients to perform earlier interventions and prevent readmission. METHODS: We followed prospectively inpatients aged 60 years and older for 30 days, with initial comprehensive geriatric assessment (CGA) on admission in a tertiary referral centre. Patients were assessed with CGA tools consisting of FRAIL scale (fatigue, resistance, ambulation, illness, loss of weight), the 15-item Geriatric Depression Scale, Mini Nutritional Assessment short-form (MNA-SF), the Barthel index for activities of daily living (ADL), Charlson Comorbidity Index (CCI), caregiver burden based on 4-item Zarit Burden Index (ZBI), and cognitive problem with Abbreviated Mental Test (AMT). Demographic data, malignancy diagnosis, and number of drugs were also recorded. We excluded data of deceased patients and patients transferred to other hospitals. We conducted stepwise multivariate regression analysis to develop the scoring system. RESULTS: Thirty-day unplanned readmission rate was 37.6 %. Among 266 patients, 64.7 % of them were malnourished, and 46.5 % of them were readmitted. About 24 % were at risk for depression or having depressed mood, and 53.1 % of them were readmitted. In multivariate analysis, nutritional status (OR 2.152, 95 %CI 1.151-4.024), depression status (OR 1.884, 95 %CI 1.071-3.314), malignancy (OR 1.863 95 %CI 1.005-3.451), and functional status (OR 1.584, 95 %CI 0.885-2.835) were included in derivation of 7 score system. The scoring system had maximum score of 7 and incorporated malnutrition (2 points), depression (2 points), malignancy (2 points), and dependent functional status (1 point). A score of 3 or higher suggested 82 % probability of readmission within 30 days following discharge. Area under the curve (AUC) was 0.694 (p = 0.001). CONCLUSIONS: Malnutrition, depression, malignancy and functional problem are predictors for 30-day readmission. A practical CGA-based 7 scoring system had moderate accuracy and strong calibration in predicting 30-day unplanned readmission for older patients.