Improved prediction of 5-year mortality by updating the chronic related score for risk profiling in the general population: lessons from the italian region of Lombardy

• Published in: Frontiers in public health Vol. 11; p. 1173957
• Main Authors: Corrao, Giovanni, Bonaugurio, Andrea Stella, Chen, Yu Xi, Franchi, Matteo, Lora, Antonio, Leoni, Olivia, Pavesi, Giovanni, Bertolaso, Guido
• Format: Journal Article
• Language: English
• Published: Frontiers Media S.A 30.08.2023
• Subjects: comorbidities; healthcare; Public Health; risk profiling; score; stratification
• ISSN: 2296-2565; 2296-2565
• DOI: 10.3389/fpubh.2023.1173957

Objective The aim of this study was to improve the performance of the Chronic Related Score (CReSc) in predicting mortality and healthcare needs in the general population. Methods A population-based study was conducted, including all beneficiaries of the Regional Health Service of Lombardy, Italy, aged 18 years or older in January 2015. Each individual was classified as exposed or unexposed to 69 candidate predictors measured before baseline, updated to include four mental health disorders. Conditions independently associated with 5-year mortality were selected using the Cox regression model on a random sample including 5.4 million citizens. The predictive performance of the obtained CReSc-2.0 was assessed on the remaining 2.7 million citizens through discrimination and calibration. Results A total of 35 conditions significantly contributed to the CReSc-2.0, among which Alzheimer's and Parkinson's diseases, dementia, heart failure, active neoplasm, and kidney dialysis contributed the most to the score. Approximately 36% of citizens suffered from at least one condition. CReSc-2.0 discrimination performance was remarkable, with an area under the receiver operating characteristic curve of 0.83. Trends toward increasing short-term (1-year) and long-term (5-year) rates of mortality, hospital admission, hospital stay, and healthcare costs were observed as CReSc-2.0 increased. Conclusion CReSC-2.0 represents an improved tool for stratifying populations according to healthcare needs.