Different operational definitions of polypharmacy and their association with the risk of all-cause hospitalization: A conceptual framework using administrative databases

Abstract

Background: As in all pharmacoepidemiology studies, also in the cardiovascular field it is essential to take into account the clinical complexity of patients, which is very frequently estimated with polypharmacy. We aimed at describing the current heterogeneity of polypharmacy definition, and assessing the association of polypharmacy with clinical outcomes.
Methods: Using administrative databases of the local health unit of Bergamo (Lombardy), all subjects aged ≥40 years with at least one reimbursed drug prescription during the year 2017 were identified. We selected from literature relevant operational definitions of polypharmacy. First, we applied World Health Organization (WHO) definition (at least ≥5 different medications, ATC 4th level code). Second, we excluded drug prescriptions associated with short-term treatment. Third, we considered only the prescriptions of drugs with a total annual defined daily doses (DDDs) ≥60. All the approaches were evaluated within one year, one quarter, and one month. A multivariate logistic regression model was performed to estimate odds ratios (OR) and 95% confidence intervals [95% CI] for the association between polypharmacy and the risk of hospitalization for all-causes.
Results: Overall, 431,620 subjects were included in our cohort. The DDD-based definition led to estimates with little variability depending on the time windows (range 20.47%-21.16%), while the WHO definition determined the greatest variability (range 39.98%-31.24%). The DDD-based definition identified an older (mean age [SD], 72.6 [10.9]) and more complex cohort of patients (average number [SD] of previous hospitalizations 1.2 [1.7], average number of dispensed drugs 9.7 [3.5]). A dose-dependent increase in risk was observed as the number of the dispensed drugs increases regardless of definitions.
Conclusions: Different definitions of polypharmacy led to different prevalence estimates. All definitions showed a dose-dependent association with hospitalization risk, with the definition based on DDDs being the least heterogeneous. However, only a patient-by-patient approach can determine whether or not polypharmacy is appropriate.