Risk Stratification Using a Novel Polypharmacy Assessment Score

Polypharmacy is a global health priority, characterized by an escalating prevalence of people using multiple medicines. Identifying people with problematic polypharmacy is complex, as multiple medicines can be suitable for people with several chronic conditions requiring more treatment. We present the development and internal validation of a novel approach to identifying potentially problematic polypharmacy. Medication count was modelled using zero-inflated negative binomial regression with age, gender, 37 long-term conditions and their grouped interactions as dependent variables. The "Polypharmacy Assessment Score" was then constructed through calculating the difference between the observed and modelled count of prescribed medications, thereby highlighting groups with unexpected levels of prescribing. This exhibited superior predictive validity for several clinical outcomes, compared with current approaches of using simple counts. Next steps to advance clinical utility include assessment for bias in certain groups and optimizing risk stratification groups to benefit patient care.