Our findings demonstrate that PIMs continue to impose a substantial cost on public healthcare, with expenses remaining relatively stable over the past decade. Average per enrollee costs illustrate the evolution of PIM expenditures while also acknowledging for the increasing number of older adults due to aging population. Women, who outnumbered men, incurred higher total and per enrollee costs yet exhibited a more pronounced decreasing trend than men (in absolute and relative values). PPIs and benzodiazepines were the PIM classes that most significantly impacted the economic burden.
Cost analyses to evaluate financial burden of specific medications in Canada are scarce, particularly regarding recent estimates attributable to PIMs. PIM-related costs across Canada were estimated at $419 million in 2013; $75/enrollee [12] (inflated to 2024: $548 million; $98/enrollee), and at $1 billion in 2021; $140/enrollee [16] (inflated to 2024: $1.1 billion; $150/enrollee). In comparison, our data for Quebec shows $202 million; $169/enrollee, and $186 million; $119/enrollee for the same years, respectively. Our analysis included different PIM classes than those in these studies. The first study relied on the 2012 version of the Beers criteria, which did not include PPIs [26], a major contributor to PIM costs. The most recent study defined PIMs by combining hand-picked PIMs from various lists and included opioids, gabapentinoids, and cholinesterase inhibitors not included in the Beers criteria [16]. By removing these three medications from their study, Canadian PIM costs fall to $747 million; $104/enrollee [16] (inflated to 2024: $801 million; $112/enrollee). Moreover, there was no mention of exclusion of parenteral forms or consideration of concomitances (e.g. PPIs-NSAIDs), and both deductible and co-insurance costs paid by enrollees were included, likely increasing cost estimates compared to ours. Nonetheless, this could suggest that Quebec’s PIM expenditures were considerably higher than in the rest of Canada. Compared to national projections indicating a 33% decline in inflation-adjusted PIM costs between 2013 and 2021 [16], our data also point to a decreasing trend (−9.7%), though the absolute costs and slope differ due to the factors outlined above.
Other studies conducted in various countries have analyzed PIM reimbursement data among adults ≥ 65 years [27, 28]. Similar to our findings, prevalences were mainly attributed to PPIs and benzodiazepines in Lithuania (2015) [29], France (2017) [28], and United States (Medicare Part D) (2014–2018) [30]. A decrease in benzodiazepine use was also found in France when compared to anterior studies [28]. However, while our results showed higher PIM costs and prevalent use among women, these findings where not observed for prevalence in Lithuania [29] or for costs in France [28]. Reimbursement policies vary greatly across countries, including differences in exemptions or cost-sharing mechanism [31]. Differences in healthcare systems, disease epidemiology, prescribing habits and PIM identification tools (e.g. 2003, 2015, and 2019 Beers criteria, EU(7)-PIM list, PRISCUS list, REMEDI[e]S, and STOPP criteria) may contribute to these disparities. For instance, in the US, benzodiazepines were initially excluded from Medicare Part D coverage, but were added in 2013 [32]. Prescriptions are now valid for six months and can be refilled up to five times within that period. After six months, a new prescription is required, and this cycle can be repeated. Following this change, rates of benzodiazepines-related overdoses and fall-related injuries increased in older adults [33]. In France, benzodiazepine prescriptions are subject to strict time limits and regulatory conditions, including a maximum duration of 12 weeks for anxiolytics and four weeks for hypnotics [34]. In Lithuania, regulatory changes introduced in 2021 limited benzodiazepine prescriptions to a maximum of 30 days with a maximum of a 10-day period for dispensing, and a psychotropic medication prescription record [35]. These regulations aimed to limit long-term use and reduce the possibility of prescription accumulation across multiple prescribers. These systemic differences likely contribute to variations in PIM use and associated costs and may inform the development of more effective policy levers.
For both sexes, the most expensive PIM classes were PPIs and benzodiazepines during the whole study period. Several regulatory measures likely contributed to the decrease in PPI costs, despite an increase in the proportions of users. Over the decade, no new generic PPI was reimbursed by the public drug plan [36, 37]. However, a maximum payable price (maximal amount per pill that the public drug insurance agrees to reimburse) was introduced in 2013 [38], and was reduced in 2015 [39]. It is also possible that shorter treatment durations and fewer claims per individual by year contributed to the increase in PPI use prevalence, despite the decline in annual costs, suggesting potential changes in patterns of use. On the other side, the implementation of exception codes for PPIs in 2017, restricting their reimbursement to cases where the prescriber provided a valid indication associated with a specific code to limit their long-term use [14], likely led to an initial decrease in both costs and prevalence. However, this effect did not seem to persist, consistent with a recent Quebec study that showed exception codes were ineffective to prevent inappropriate PPI prescribing [40]. Since 2020, Quebec pharmacists can initiate PPIs and renew exception codes, which could also have contributed to their initiation and maintenance of these treatments [41, 42]. This pattern suggests that these regulatory measures may have had only a temporary impact and could potentially be circumvented by prescribers. One possible explanation is that PPI use has become normalized in clinical practice, and prescribers may feel ill-equipped or lack confidence in managing patients with alternative therapeutic or non-pharmacological options, particularly in complex older adults.
For benzodiazepines, both costs per enrollee and prevalence of use have decreased considerably, consistent with the trend observed in the last years [43, 44]. This decline mirrors awareness efforts by various organizations, including the Canadian Medication Appropriateness and Deprescribing Network and Beers criteria updates [9, 10, 13]. However, part of this decrease could also result from a shift towards other PIMs, such as antipsychotics and antidepressants. Interestingly, a shift towards Z-drugs was not observed in a study conducted across all age groups in Canada from 2016 to 2022 [45]. Despite these efforts, benzodiazepines and PPIs remain major contributors to the overall burden of PIMs in the older population. These findings underscore the need for sustained and multifaceted interventions, including patient and prescriber education, better enforcement of reimbursement restrictions, and regular medication reviews, to optimize medication use and reduce inappropriate prescribing in older adults.
From the public payor perspective, the economic burden of PIMs is likely much higher, as indirect costs such as hospitalizations or outpatient visits for adverse effects substantially add to medication-related expenses [46, 47]. Based on the RAMQ medication costs for adults aged ≥ 65 in 2021 [25] (inflated to 2024: $3.44 billion), the weight of PIM direct costs represented around 5.4%. This constitutes a significant savings opportunity, considering both direct and indirect costs. While some studies have attempted to quantify indirect costs, such as those associated with increased healthcare utilization, a comprehensive assessment of the full societal impact remains limited. In Canada, the estimated incremental health care utilization cost associated with PIMs was $1.4 billion in 2013 ($1.8 billion inflated to 2024). A retrospective study in Ontario using the STOPP/START criteria estimated a total cost of $1.5 billion (inflated to 2024) for newly prescribed medications (not necessarily PIMs), including hospitalizations and emergency department visits attributable to PIMs [48].
Strengths and limitations
To our knowledge, this is the first cost analysis conducted over a decade and encompassing nearly the entire Quebec population aged ≥ 65 with highly reliable data. Nonetheless, our study has limitations. The lack of clinical indications and diagnoses in administrative data prevented confirmation of the inappropriateness of the included medication, as clinical contexts could justify the use of some listed PIMs. However, we implemented several conditions to limit possible overestimates of PIM claims and costs. Besides, due to the absence of individual data notably about socioeconomic status and comorbidities, multivariate analysis could not be performed. However, we conducted stratified analysis for key variables such as sex and age. We also did not test for the assumption of independent observations, which could have led us to false statistically significant results and unreliable estimations. Additionally, our study focused exclusively on direct PIM costs and did not assess the additional indirect costs associated with PIM use in older adults, underestimating costs borne by public funds. These indirect costs were beyond the scope of the study, which relied on administrative medication claims data without linked clinical outcomes. Future studies incorporating comprehensive hospitalizations, adverse drug reactions, increased use of healthcare services, and productivity losses are needed to better capture the overall societal impact. A more complete understanding of both direct and indirect costs will help policymakers and clinicians appreciate the true magnitude of PIM-related economic burden and guide the design of targeted interventions to optimize medication use in older adults.