A data-driven medication score predicts 10-year mortality among aging adults.

Area of Science:

• Gerontology
• Biostatistics
• Pharmacoeconomics

Background:

• Health disparities in aging populations are a growing concern.
• Understanding the impact of medical treatments on elderly health is crucial.
• Existing health measures may not fully capture medication-related risks.

Purpose of the Study:

• To develop and validate a data-driven health measure using pharmacy purchase and mortality data.
• To assess the association of this novel medication score with all-cause mortality.
• To evaluate the combined predictive power of the medication score and Charlson comorbidity index.

Main Methods:

• Utilized modern statistical learning methods on 21 years of data from 12,047 individuals.
• Validated the developed medication score using two independent datasets comprising 33,616 individuals.
• Analyzed associations with all-cause mortality, Charlson comorbidity index, and polygenic risk scores.

Main Results:

• The medication score strongly predicted all-cause mortality (HR 1.18 per point increase).
• Combining the medication score with the Charlson comorbidity index resulted in a six-fold higher mortality risk (HR 6.30).
• The medication score demonstrated performance comparable to the Charlson comorbidity index and linked to genetic predispositions for common diseases.

Conclusions:

• A medication score derived from pharmacy data offers a valuable tool for assessing health and mortality risk in aging individuals.
• This score can complement existing clinical assessments like the Charlson comorbidity index for more accurate risk stratification.
• The findings highlight the importance of medication data in understanding health trajectories in older adults.