LAUSR

As prescription drug use is on the rise, new medications are entering the market each year, and individuals with multiple comorbid conditions are living longer, the potential for drug–drug interactions (DDIs) is increasing. It has been estimated that DDIs are responsible for 1–5% of all hospital admissions [2, 3] and 13% of all adverse drug events in community-dwelling older adults [4]. With tens of thousands of potentially interacting drug pairs on the market [5], identifying and managing potential DDIs has become a daunting task for healthcare professionals and patients. Studies have found that clinicians’ knowledge about DDIs is limited [6, 7] and the reliability of DDI information sources is questionable [8, 9]. Once heralded as a solution to reducing exposure to DDIs [7, 10], information technology (IT)-based clinical decision-support systems generate an excessive number of DDI alerts, the overwhelming majority of which (C90%) are routinely overridden [11–13] and not always appropriately so [14]. There is a growing concern that DDI decision-support systems are failing to achieve their goal of reducing DDIassociated patient harm [15, 16]. Low alert specificity and ambiguity about the clinical relevance of DDIs are often reported as the major reasons for high override rates and the suboptimal efficiency of decision-support systems [17–19]. However, ensuring clinically relevant content is challenging because evidence of outcomes for most DDIs rarely extends beyond extrapolations from in vitro studies and case reports [18, 20]. Unfortunately, we know very little about the real-world consequences of DDIs. While strong evidence supports clinical outcomes associated with concomitant exposure to some drug pairs that have been found to interact pharmacologically (e.g., warfarin and certain antibiotics) [21, 22], many potential interactions likely do not manifest in clinical outcomes. This is good for patients who require concomitant use of these drugs or who inadvertently become exposed to them because of the failings of clinical decision-support systems. However, an absence of evidence about whether an interaction affects clinical outcomes not only contributes to DDI alert overload but can also itself result in adverse outcomes. Concerns about interactions for which no clinical outcome evidence exists might lead to underutilization of safe and effective medications. This, in turn, might result in suboptimal patient outcomes, even if the interactions themselves do not directly manifest in harm. Just as harm associated with DDIs is usually avoidable, we argue that adverse outcomes of suboptimal drug use that derives from a lack of evidence of clinical outcomes is also usually avoidable. We urgently need more and better pharmacoepidemiologic studies to understand the clinical impact, or lack thereof, of pharmacologically demonstrated DDIs. In this issue of Drug Safety, Meid et al. [1] report a study that aims to provide empirical evidence on the clinical significance of a pharmacodynamic interaction between QTc-prolonging medications. The clinical outcomes of QTc prolongation are potentially catastrophic as it can lead to Torsades de pointes (TdP) and sudden cardiac This commentary comments on ‘‘Investigating the Additive Interaction of QT-Prolonging Drugs in Older People Using Claims Data’’ (doi:10.1007/s40264-016-0477-y) by Meid et al. [1].