A nomogram for predicting risk of death during hospitalization in elderly patients with Alzheimer's disease at the time of admission.

ABSTRACT

Background and objectives: Elderly patients with Alzheimer's disease (AD) often have multiple underlying disorders that lead to frequent hospital admissions and are associated with adverse outcomes such as in-hospital mortality. The aim of our study was to develop a nomogram to be used at hospital admission for predicting the risk of death in patients with AD during hospitalization. Methods: We established a prediction model based on a dataset of 328 patients hospitalized with AD -who were admitted and discharged from January 2015 to December 2020. A multivariate logistic regression analysis method combined with a minimum absolute contraction and selection operator regression model was used to establish the prediction model. The identification, calibration, and clinical usefulness of the predictive model were evaluated using the C-index, calibration diagram, and decision curve analysis. Internal validation was evaluated using bootstrapping. Results: The independent risk factors included in our nomogram were diabetes, coronary heart disease (CHD), heart failure, hypotension, chronic obstructive pulmonary disease (COPD), cerebral infarction, chronic kidney disease (CKD), anemia, activities of daily living (ADL) and systolic blood pressure (SBP). The C-index and AUC of the model were both 0.954 (95% CI 0.929-0.978), suggesting that the model had accurate discrimination ability and calibration. Internal validation achieved a good C-index of 0.940. Conclusion: The nomogram including the comorbidities (i.e., diabetes, CHD, heart failure, hypotension, COPD, cerebral infarction, anemia and CKD), ADL and SBP can be conveniently used to facilitate individualized identification of risk of death during hospitalization in patients with AD.