This work may not be copied, distributed, displayed, published, reproduced, transmitted, modified, posted, sold, licensed, or used for commercial purposes. By downloading this file, you are agreeing to the publisher’s Terms & Conditions.

Original Research

Risk Score for Predicting Mortality in People With Dementia: A Nationwide, Population-Based Cohort Study in Taiwan With 11 Years of Follow-Up

Chih-Ming Cheng, MDa,b,c,g; Wen-Han Chang, MSa,d; Yu-Chuan Chiu, MDe; Mu-Hong Chen, MDa,b,c; Wei-Ting Liao, MDa; Cheng-Hung Yang, MDa,b; Jen-Ping Hwang, MDa; Cheng-Ta Li, MD, PhDa,b,c,h; Ya-Mei Bai, MD, PhDa,b,c; Shih-Jen Tsai, MDa,b; Tzeng-Ji Chen, MDf; and Chia-Fen Tsai, MD, PhDa,b,c,*

Published: June 11, 2019

Article Abstract

Objective: Many researchers and physicians attempt to determine the prognosis and short- and long-term mortality risks of dementia for formulating suitable care plans for patients and their families. However, the published prediction models have been insufficient for this purpose and have worked only in certain specific populations. For medical autonomy and end-of-life decisions, an informative tool to predict 6-month, 1-year, 2-year, 3-year, and 5-year mortality rates for dementia patients merits further investigation.

Methods: Patients aged ≥ 65 years who received ICD-9-CM diagnoses of dementia between 2002 and 2009 were identified from Taiwan’s National Health Insurance Research Database and followed until the end of 2013. Patient characteristics and comorbidities that were considered potential risk factors for mortality were assessed. Mortality-predicting risk scores were developed using a regression coefficient-based scoring approach. In total, 6,556 patients were identified and then randomly divided into a derivation cohort (n = 4,371) and validation cohort (n = 2,185).

Results: By the end of the study, 1,693 of the 4,371 dementia patients (38.7%) in the derivation cohort were deceased. Mean duration of follow-up was 6.26 years. Eleven acute and chronic factors were identified for building the predictive score model, which produced scores from 0 to 24 points (higher scores indicated higher mortality). The score model exhibited good predictive power for various life expectancies (area under receiver operating characteristic curve: 6-month = 0.852, 1-year = 0.779, 2-year = 0.725, 3-year = 0.721, 5-year = 0.703) and good calibration in the validation cohort (Hosmer-Lemeshow test, χ2 = 4.709, P = .788).

Conclusions: The developed predictive score model may be the first tool that uses the same clinical factors to determine both short- and long-term mortality risks in patients with dementia.

Volume: 80

Quick Links:

Continue Reading…

Subscribe to read the entire article


Buy this Article as a PDF