(ASPOT28) The Economic Burden of Drug-Drug Interactions in an Older Population acutely admitted to Hospital in Ireland: A Propensity Score Pharmacoeconomic Analysis.

Background: Adverse drug reactions (ADRs) are associated with greater healthcare costs, particularly in the older population where medication-related harm presents a growing challenge for healthcare systems. Drug-drug interactions (DDIs) are an often predictable cause of ADRs. To date, the economic burden of DDIs is unclear.

Objectives: We estimate the average treatment effect (ATE) of DDI-exposure on an ADR-related hospital admission in an older population, and compare in-hospital stay costs between ADR and non-ADR groups per DDI exposure.

Methods: This was a cross-sectional study among 798 older individuals admitted acutely to hospital in Ireland between 2016 and 2017. Full medication history and objective clinical real-world data were available. DDIs were identified from a list of 28,225 “severe” DDIs (i.e. the result may be a life-threatening event/permanent detrimental effect). For hospital services, we used national unit costs, with overall patient cost expressed per total length of stay (LOS) in hospital. Directed acyclic graphs were developed to evaluate confounding between DDI-exposure and ADR-related hospital admission. The doubly robust augmented Inverse probability of treatment weighting estimator was used to balance treatment and control groups, and to estimate the ATE. Costs were compared using a generalised estimating equation, with Inverse probability of treatment weighting. A sensitivity analysis for costs was performed using 1:1 propensity score matching. DDI prevalence, ATE estimates (causal risk difference and causal risk ratio), estimated using SAS Proc Causaltrt, 95% bootstrapped (1000 samples) confidence intervals (CIs), and costs per hospital services are reported. Analysis was performed using SAS (v9.4) and R (v4.1.2).

Results: N=782 (98%) patients using ≥2 drugs were included for analysis. Mean age was 80.9 (SD ±7.5) years (range: 66–105); 52.2% were female; 36.2% polypharmacy (5-9drugs); 56% major polypharmacy (≥10 drugs); and 45.1% (n=353) had an ADR-related admission. The median LOS was 7 (IQR 4–15) days. At admission, n=316 (40.4%) patients had at least one DDI and n=466 had no DDI; n=113 (14.5%) had at least one DDI which increases the risk of bleeding, while n=669 were unexposed. The ATE of any DDI exposure on ADR-related hospital admission was estimated to be 0.04 [95% CI: -0.05, 0.13]; causal risk ratio of DDI exposure on ADR-related hospital admission was aRR= 1.17 [95% CI: 0.96, 1.43]. For patients exposed to a DDI which increases the risk of bleeding, the ATE on ADR-related hospital admission was estimated to be 0.27 [95% CI: 0.17, 0.37]; causal risk ratio aRR=2.83 [95% CI: 2.29, 3.49]. The cost of ADR-related hospital admission was on average €1,545.51 [95% CI:-5,629.00, 8,720.02] more expensive for those with any DDI compared to those without a DDI; and €3,370.28 [95% CI: -4,920.57, 11,661.00] more expensive for those with a DDI which increases the risk of bleeding compared to those without one of these DDIs. These findings were robust to sensitivity analysis.

Conclusions: Older individuals exposed to DDIs which increase the risk of bleeding carry a 2.8-fold increased risk of ADR-related hospital admission, and represent substantial cost to the healthcare system. These DDIs should be the focus of medication optimisation strategies in the older population.