Professor Clinical Pharmacology, Pharmacy, and Environmental Medicine, Department of Public Health, University of Southern Denmark. Odense C, Denmark
Background: Studies consistently associate polypharmacy, often defined as the use of 5 or more medications, to mortality. However, such studies have generally not adjusted adequately for confounding and as they are highly prone to confounding from indication from the underlying disease(s) the medications are used to treat, these findings must be interpreted with caution.
Objectives: We investigated the association between the use of number of medications and mortality through a series of analyses to illustrate the inherent challenges in studying this association, by describing aspects that point to a non-causal interpretation.
Methods: We used data on a 20% random sample of all Danish residents moving into a nursing home 2015-2021 (n=16,649). Using a Kaplan Meier plot, we illustrated the 1-year mortality, stratifying by number of drugs filled six months prior to nursing home admissions. Further, we plotted a spline of the 1-year mortality as a function of number of drugs filled, as well as the association between number of drugs and 1-year mortality, with increasing levels of adjustment for covariates. Finally, we performed logistic regression associating the 20 most common individual drugs at baseline to 1-year mortality.
Results: The 1-year mortality was 31% (n=5,082). Mortality was highly dependent on the number of drugs used at baseline, with 17% 1-year mortality among those using 0-1 drugs, increasing gradually to 46% among those using ≥16 drugs. The spline of 1-year mortality increased almost linearly with the number of drugs used at baseline. In sex- and age adjusted analyses, the odds ratio (OR) of 1-year mortality increased almost linearly to more than 5.0 with use of more than 20 drugs. Upon additional adjustment for use of furosemide, this decreased to 4.0 and when adjusting for the Nordic Comorbidity Index, dropped to 2.2. Strong protective effects were seen with use of e.g. thiazide diuretics (OR 0.64), simvastatin (OR 0.66) and donepezile (OR 0.50), while increased risk of death was observed with use of morphine (OR 1.59), zopiclone (OR 1.13) and pantoprazole (OR 1.15).
Conclusions: While strong associations are observed between the use of multiple drugs and 1-year mortality, associations were markedly attenuated by adjusting for comorbidity and there are numerous paradoxical relationships with use of individual drugs and mortality.
