Abstract
Health differences among the elderly and the role of medical treatments are topical issues in aging societies. We demonstrate the use of modern statistical learning methods to develop a data-driven health measure based on 21 years of pharmacy purchase and mortality data of 12,047 aging individuals. The resulting score was validated with 33,616 individuals from two fully independent datasets and it is strongly associated with all-cause mortality (HR 1.18 per point increase in score; 95% CI 1.14–1.22; p = 2.25e−16). When combined with Charlson comorbidity index, individuals with elevated medication score and comorbidity index had over six times higher risk (HR 6.30; 95% CI 3.84–10.3; AUC = 0.802) compared to individuals with a protective score profile. Alone, the medication score performs similarly to the Charlson comorbidity index and is associated with polygenic risk for coronary heart disease and type 2 diabetes.
Highlights
Health differences among the elderly and the role of medical treatments are topical issues in aging societies
We trained 28 medication score candidates (Supplementary Table S1) in the National FINRISK study[11] where we included a subset of 20,078 participants who were alive and at least 46 years old at the beginning of 2006
Our study suggests that long-term and large-scale health data can be distilled into a composite measure to infer health differences in the general aging population
Summary
Health differences among the elderly and the role of medical treatments are topical issues in aging societies. Several instruments have been developed to summarize disease diagnoses and exposure to medications into numeric scores using information from hospital databases and medication administration records, e.g. Charlson Comorbidity Index (CI)[4], Elixhauser Index (EI)[5], Rx-Risk-V6, Medication-Based Disease Burden Index[7], and Chronic Disease Score (CDS)[8]. These instruments customarily consider a limited number of predefined severe health conditions and consider typically short time windows of a few years, depicting primarily acute changes in health. We demonstrate the applicability of the resulting score in two fully independent prospective cohorts and investigate its relationship to known genetic predictors of late-onset diseases and diagnosis-based comorbidity index
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
