Deprescribing in older adults with polypharmacy
BMJ 2024; 385 doi: https://doi.org/10.1136/bmj-2023-074892 (Published 07 May 2024) Cite this as: BMJ 2024;385:e074892
- Anna Hung, assistant professor1 2 3*,
- Yoon Hie Kim, assistant professor4*,
- Juliessa M Pavon, associate professor4 5
- 1Department of Population Health Sciences, Duke University School of Medicine, Durham, NC, USA
- 2Duke-Margolis Center for Health Policy, Durham, NC, USA
- 3Center of Innovation to Accelerate Discovery and Practice Transformation, Durham VA Health Care System, Durham, NC, USA
- 4Division of Geriatrics, Department of Medicine, Duke University School of Medicine, Durham, NC, USA
- 5Geriatrics Research, Education, and Clinical Center (GRECC) Durham VA Health Care System, Durham, NC, USA
- *Co-first authors
- Correspondence:
A Hung, anna.hung{at}duke.edu
Abstract
Polypharmacy is common in older adults and is associated with adverse drug events, cognitive and functional impairment, increased healthcare costs, and increased risk of frailty, falls, hospitalizations, and mortality. Many barriers exist to deprescribing, but increased efforts have been made to develop and implement deprescribing interventions that overcome them. This narrative review describes intervention components and summarizes findings from published randomized controlled trials that have tested deprescribing interventions in older adults with polypharmacy, as well as reports on ongoing trials, guidelines, and resources that can be used to facilitate deprescribing. Most interventions were medication reviews in primary care settings, and many contained components such as shared decision making and/or a focus on patient care priorities, training for healthcare professionals, patient facing education materials, and involvement of family members, representing great heterogeneity in interventions addressing polypharmacy in older adults. Just over half of study interventions were found to perform better than usual care in at least one of their primary outcomes, and most study interventions were assessed over 12 months or less.
Introduction
Polypharmacy, commonly defined as taking at least five drug treatments, is common in older populations.12 The prevalence of this condition, when combined across multiple countries, is about 45% in older populations (those aged 65 years or older) compared with 27% in younger populations (those aged under 65 years).1 Although older adults are more likely to have multiple chronic conditions, and guidelines often recommend multiple drug treatments, studies have found that polypharmacy can lead to problems such as adverse drug events, drug–drug interactions, drug treatment non-adherence,34 cognitive and functional impairment,5 and increased risk of frailty, disability, falls, hospitalizations, and mortality.678
Polypharmacy is also associated with the use of potentially inappropriate medications,910 which is linked to increased risk of hospitalization and visits to the emergency room.11 Furthermore, polypharmacy can result in higher healthcare costs; the total healthcare costs of patients with polypharmacy are twice as high as those of patients without polypharmacy.12 Affordability of drug treatments is already problematic, given that, for example, nearly one third of older adults in the United States with multiple chronic conditions spend more than 20% of their income on medical care.13
One way to tackle polypharmacy is via deprescribing, which can be defined as the “systematic process of identifying and discontinuing drugs when existing or potential harms outweigh existing or potential benefits within the context of an individual patient’s care goals, functional status, life expectancy, values, and preferences.”14 This process is often supervised by a healthcare professional, but shared decision making is critical, because deprescribing decisions need to consider patient and care partner values and preferences.1516 Contrary to a simplistic goal of reducing the number of drug treatments, the essence of deprescribing lies in achieving a nuanced balance: acknowledging the concept of “appropriate polypharmacy.”1718 This approach recognizes that certain medical conditions can warrant the use of multiple drug treatments to effectively manage symptoms, prevent complications, and improve overall patient wellbeing. Deprescribing, in this context, seeks not to diminish the drug treatment count arbitrarily, but rather to ensure judicious and contextually appropriate drug treatment use.
Coined in 2003, the term “deprescribing” has witnessed significant growth in usage, research, and the development of supporting networks, and has gained particular momentum since 2016.19 This evolution highlights the growing acknowledgment of deprescribing as a crucial strategy in optimizing drug treatment regimens to meet the specific needs and goals of individual patients. Concurrently, the substantial growth in randomized controlled trials evaluating deprescribing interventions attests to the increasing emphasis on evidence based approaches. Several excellent reviews of deprescribing interventions in older adults exist, but they differ slightly in focus; for example, some are limited to comprehensive medication review interventions, or target a specific population such as community dwelling, frail, or multimorbid.20212223 This narrative review summarizes randomized controlled trials that assess deprescribing interventions addressing polypharmacy in older adults.
Prevalence of polypharmacy
Over the past few decades, polypharmacy has become a common and widespread problem among older adults. In the US, for example, the prevalence of polypharmacy among adults aged 65 and older increased from 13% in 1998 to 43% in 2014,2425 with the most recent estimates from 2017-2018 at 45%.26 This increase was driven in particular by the growing use of cardioprotective and antidepressant drug treatments,24 and the highest prevalence of polypharmacy is seen among populations with heart disease.26 Additional reasons for increases in the use of drug treatments include the start of Medicare Part D (ie, prescription drug coverage for older adults) in 2006, as well as quadrupled spending on direct-to-consumer pharmaceutical advertising (from $985 million in 1996 to $4.24 billion in 2005).27
Similarly, a survey conducted across 17 European countries and Israel also revealed a high prevalence of polypharmacy in adults aged 65 and older, varying from 26% to 40%, with the lowest estimates in Switzerland, Croatia, and Slovenia and the highest estimates in Portugal, Israel, and the Czech Republic.28 In other regions of the world, the prevalence of polypharmacy was 46% in Hong Kong, 39% in Taiwan, 32% in South Korea, and 20% in Australia.29
Whereas polypharmacy is a pervasive phenomenon in contemporary healthcare, particularly as individuals contend with complex and coexisting health conditions, a distinction must be made between inappropriate polypharmacy and appropriate polypharmacy, which ensures optimal patient care. Inappropriate polypharmacy refers to the excessive, unnecessary, or unmonitored use of drug treatments, potentially leading to adverse effects, drug–drug interactions, and diminished overall wellbeing. On the contrary, appropriate polypharmacy entails the deliberate and evidence based use of multiple drug treatments to serve the specific needs of a patient, often stemming from the complexity of their medical conditions. In the context of deprescribing, the focus shifts to identifying instances of inappropriate polypharmacy and streamlining drug treatment regimens, to align with the principles of appropriate polypharmacy. Thus, the objective of deprescribing is to tailor treatment regimens to individual patient needs, minimize potential adverse effects, and ensure that the remaining prescribed drug treatments contribute meaningfully to a patient’s overall health and quality of life, while accounting for the principles of evidence based practice.
Barriers to deprescribing
Patient barriers
Despite the need to tackle inappropriate polypharmacy, deprescribing is not commonplace. Published literature has identified a myriad of barriers at the patient, provider, and system levels. One example of a patient barrier is uncertainty of outcomes from deprescribing (and a fear of negative consequences). Other patient barriers could be poor access to deprescribing education, or difficulty in understanding medical jargon.3031 Deprescribing of a drug treatment can also represent a radical change for patients with a chronic condition, who might have been told for a long time to focus on achieving adequate disease control (eg, low A1c in the case of diabetes).32
Provider barriers
Meanwhile, providers can also lack training and evidence on how to deprescribe (eg, how to taper and monitor), and can also lack knowledge and experience in engaging in complex conversations about deprescribing.31333435 Providers themselves might fear the potential consequences of discontinuing drug treatments (eg, potential adverse drug withdrawal events, inadequate disease control) especially if they do not have guaranteed follow-up with the patient. Another provider barrier is inertia around making changes to drug treatments when patients are not experiencing any drug treatment related problems. Additional provider barriers include consultation constraints (eg, insufficient time or resources to support deprescribing), uncertainty around discontinuing a drug treatment that, for example, was initiated by another provider, uncertainty around roles and responsibilities among providers, and poor communication and collaboration among providers.31333435 Many of these barriers are due to the healthcare environment in which providers work.
System barriers
Many system barriers exist. One example is a culture of diagnosing and prescribing in medicine.313334 Evidence based guidelines often recommend prescribing medicine but do not make recommendations on when to later stop the medicine. Evidence based guidelines also focus on a single disease, and evidence based guidelines for older adults with multimorbidity are lacking. Care itself for these older adults is fragmented, with multiple specialists and potential delays and/or mishaps in communication, which can further contribute to inappropriate polypharmacy. Additionally, the payment incentive structure in healthcare does not encourage complicated conversations around deprescribing, in-depth consideration of patient care goals, and often there is a lack of shared decision making tools and resources.313334 Also, performance metrics for providers and insurance plans can inadvertently encourage inappropriate overuse of drug treatments (to ensure adequate disease control) and make it difficult for deprescribing to attain individual patient goals.36
Barriers across settings
These barriers are not uniform across healthcare settings; primary care, hospitals, long term care, and private settings face unique challenges.303537 For instance, primary care confronts challenges using a diagnostic and prescribing culture, while hospitals contend with competing inpatient tasks, patient resistance during hospitalization, and post-discharge follow-up concerns. Long term care settings grapple with resource shortages and coordination gaps, and private settings face obstacles related to awareness, commitment, and incentives for deprescribing. Healthcare providers, including physicians, pharmacists, and nurses, exhibit varied attitudes and knowledge gaps, influencing deprescribing implementation. Specific drug treatment classes, such as benzodiazepines, introduce unique challenges, including patient beliefs, lack of knowledge, and emotional attachments to the drug treatment.3839
Implementation strategies
Overcoming these barriers is paramount for successful deprescribing, and implementation science can offer valuable tools and strategies. Key implementation strategies for successful deprescribing include structured education and training for providers and patients to enhance knowledge and confidence, patient centered approaches like shared decision making and motivational interviewing, offering non-pharmacological alternatives to reduce drug treatment reliance, and providing resources such as clinical decision support tools and deprescribing algorithms.40 Additionally, fostering improved communication and collaboration among healthcare settings and providers is crucial for maintaining continuity and consistency of care. Integrating implementation science principles requires understanding healthcare system complexities, securing provider buy-in, and tailoring interventions to specific contexts, as recent research underscores the value of this approach in overcoming barriers and optimizing facilitators in diverse healthcare landscapes.
Facilitators to deprescribing
Studies have identified patient, provider, and system facilitators that cover a wide spectrum.30313335 Most older adults (84-88%) are willing to deprescribe their drug treatment, based on provider recommendations.4142 In fact, 67% of older adults report wanting to reduce the number of drug treatments they are taking, and this desire increases when they are taking more drug treatments.43 Patient willingness and attitudes toward deprescribing vary, but it should be noted that this desire on average does exist.
Shared decision making processes with clear communication and a gradual introduction of the topic, as well as active patient and caregiver involvement, can enable deprescribing. Examples of provider facilitators include training to achieve in-depth drug treatment knowledge related to deprescribing outcomes (eg, potential adverse drug withdrawal events) along with multidisciplinary provider teams that have clear roles and responsibilities related to deprescribing decisions.
Examples of system facilitators include resources and tailored support tools to enable deprescribing conversations, deprescribing guidelines, and evidence on optimal deprescribing practices.30313335 The greater availability and acceptability of non-pharmacological alternatives, which allows for substitution of the deprescribed drug treatment, can also allow for patients and providers to feel more comfortable with deprescribing. At the broader system level, insurance plans have medication therapy management programs and have been incentivized through quality measures to monitor and encourage discontinuation of potentially inappropriate medications.44
Finally, increasing awareness of drug treatment overload; for example, media campaigns and reducing industry influence can also help facilitate a culture that normalizes deprescribing.45 Facilitators and barriers to deprescribing are important to consider in deprescribing interventions, and in our review of studies, we extracted various intervention components that would deal with some of these facilitators and barriers.
Source and selection criteria
A literature review was conducted to identify published randomized controlled trials of deprescribing interventions in older adults with polypharmacy. Searches were conducted in PubMed and Embase on 2 April 2023 using the free text search terms, “deprescribing” and “polypharmacy”, and filtering on randomized controlled trials, controlled studies, systematic reviews, and meta-analyses. We included studies that were published from 1 January 2012 to the day of the search and were in English. We limited our search to randomized controlled trials that enrolled those at least 65 years of age and with polypharmacy (defined as taking at least five drug treatments). We excluded pilot/exploratory randomized controlled trials and ancillary publications of the randomized controlled trial (eg, trial protocol only, process evaluation). We also manually searched through references of identified studies and systematic reviews of broader scope interventions for older adults with polypharmacy, and included studies if deprescribing was a part of the intervention (either explicitly by the term “deprescribing” or implicitly by reporting reduction in drug treatments as a trial outcome). Two reviewers reviewed each article and any discrepancies were resolved through discussion. We described intervention components and summarized key findings based on primary outcomes and any key secondary outcomes that were reported in the publication abstract. This distinction was made because many of the trials included a very long list of secondary outcomes. We then categorized whether the study intervention was found to have a statistically significant effect (p<0.05) on: (a) at least one primary outcome; (b) if not, at least one of the key secondary outcomes; or (c) neither primary nor key secondary outcomes. We acknowledge that if a randomized controlled trial tested many outcomes, then there could be multiplicity concerns. For this narrative review, we report on these outcomes without adjustment and are limited by what was reported on in the randomized controlled trial.
Secondarily, we searched for ongoing randomized controlled trials of deprescribing interventions in older adults with polypharmacy and applied the same inclusion/exclusion criteria as aforementioned. We defined the randomized controlled trial as “ongoing” if the primary outcome results had not yet been published as a manuscript. To identify these ongoing studies, we documented any trials that had published protocols from the aforementioned PubMed and Embase searches that did not yet have primary outcomes results reported in a manuscript, and also searched through the ClinicalTrials.gov and the International Standard Randomised Controlled Trial Number Registry databases.
When conducting this narrative review, we identified several completed studies4647484950 and ongoing studies5152 that were excluded because they included slightly younger populations (minimum of 50, 55, or 60 years of age). Similar to prior reviews, we chose an age cutoff of ≥65 because interventions and outcomes can vary for different ages.202122 For example, the Beers criteria that are intended for use in older adults aged ≥65 define a commonly used list of potentially inappropriate medications that might be used if interventions are targeting older adults.53 Similarly, additional studies were excluded if they did not specify that the recruited patients had to be using a minimum of five drug treatments (ie, the most common definition of polypharmacy), because the scope of this narrative review was older adults with polypharmacy.
Results
Out of 433 articles identified, 21 articles met all eligibility criteria (fig 1). Of these articles, 15 took place in primary care (11 in primary care practices and four in community pharmacies), four focused on hospital transitions, and two on nursing home settings (fig 2). Most studies took place in Europe or North America, and over half (12 of 21) were cluster randomized controlled trials (see supplementary table 1). While the majority (12 of 21) targeted patients aged ≥65, four studies targeted patients aged ≥70, and five studies targeted patients aged ≥75 (figs 3-7). The majority (12 of 21) examined populations with at least five drug treatments, while eight studies used higher thresholds (eg, at least 7, 8, 10, or 15 drug treatments). Three studies additionally targeted deprescribing of specific drug classes (eg, benzodiazepines, psychotropics, anticholinergics). In addition to targeting older populations with polypharmacy, five studies further focused on frail populations or those with multiple chronic conditions, alongside specific disease states such as dementia or cardiovascular disease.
Summary of findings from published deprescribing randomized controlled trials in older adults with polypharmacy (part 1). *Unless noted, the control arm was usual care. †Results on secondary outcomes were included if they were notable enough to be included in the abstract. Focus on adjusted results and intention-to-treat analyses when available.
DRP=drug related problem; EQ=EuroQol; HRQoL=health related quality of life; MAI=Medication Appropriateness Index; PCP=primary care physician; PIM=potentially inappropriate medication; SD=standard deviation; SPPiRE=Supporting Prescribing in Older Adults with Multimorbidity in Irish Primary Care.
Summary of findings from published deprescribing randomized controlled trials in older adults with polypharmacy (part 2). *Unless noted, the control arm was usual care. †Results on secondary outcomes were included if they were notable enough to be included in the abstract. Focus on adjusted results and intention-to-treat analyses when available.
CI=confidence interval; DBI=drug burden index; DRP=drug related problem; GP=general practitioner; GPGP= Good Palliative-Geriatric Practice; HRQoL=health related quality of life; OR=odds ratio; PIM=potentially inappropriate medication; STOPP/START=Screening Tool of Older Person's Prescriptions/Screening Tools to Alert Doctors to Right Treatment.
Summary of findings from published deprescribing randomized controlled trials in older adults with polypharmacy (part 3). *Unless noted, the control arm was usual care. †Results on secondary outcomes were included if they were notable enough to be included in the abstract. Focus on adjusted results and intention-to-treat analyses when available.
ADE=adverse drug event; ADWE=adverse drug withdrawal event; CI=confidence interval; DRP=drug related problem; GP=general practitioner; HRQoL=health related quality of life; MAI=Medication Appropriateness Index; PACIC=Patient Assessment of Care for Chronic Conditions; SD=standard deviation; STOPPFrail=Screening Tool of Older Persons Prescriptions in Frail adults with limited life expectancy; STRIPA=Systematic Tool to Reduce Inappropriate Prescribing Assistant; TRIM=Tool to Reduce Inappropriate Medications.
Summary of findings from published deprescribing randomized controlled trials in older adults with polypharmacy (part 4). HR=hazard ratio; PIM=potentially inappropriate medication; S$=Singapore dollar; VAS=Visual Analogue Scale.
All studies compared the intervention with usual care. We found a wide variety of interventions, with most including more than one component to overcome different barriers to deprescribing. Common intervention components explicitly described in studies included medication reviews, shared decision making processes and/or a focus on patient care priorities, active involvement of a multidisciplinary team, training for healthcare professionals, electronic clinical decision support tools or websites, patient facing deprescribing educational materials, and involvement of family (fig 7). Below, we summarize the intervention and key findings from each randomized controlled trial according to setting. We highlight whether study interventions were found to have a significant impact on at least one primary outcome assessed; or if not, at least one key secondary outcome; or none, and report on that impacted outcome (figs 3-7). Within setting, we also organize the studies based on the components included in the intervention. This categorization is helpful in framing our findings, but not absolute, as there could be overlap and variation in how much focus was given to each of the components.
Primary care
Historically, medication reviews with the patient have been a common approach to optimizing drug treatment management among older adults with polypharmacy in the primary care setting. These medication reviews typically consist of a healthcare professional, such as a pharmacist, reviewing a patient’s list of drug treatments, identifying potential drug related problems, meeting with the patient, and providing recommendations to the patient, physician, or both. Medication reviews can be broad in nature and include patient education and/or drug treatment changes such as reductions in dose, discontinuing a drug treatment, or adding drug treatments when appropriate. In our review, two thirds (10 of 15) of primary care interventions centered around medication reviews, and all affected at least one primary or secondary outcome (with a majority (7 of 10) affecting the primary outcome; figs 3-7).
Medication reviews can vary depending on how the drug treatment data are collected, who is performing the review, and the extent to which follow-up is performed, along with whether additional components (such as training for the healthcare professional) are provided. In one study, a medication review conducted by a pharmacist based on a questionnaire completed by patients led to a larger reduction in the number of drug treatments in the intervention group versus the control group at 12 months (p<0.046; primary outcome).60 In another study, pharmacist conducted medication reviews were preceded by a three day training course for pharmacists and succeeded by six month follow-up with the patient.61 Recommendations were made to patients, their physicians, or both, and this intervention led to a greater reduction (0.21 ± 0.06 drugs, 95% confidence interval 0.092 to 0.335) in the number of drug treatments in the intervention group compared with the control group, improved quality of life in the intervention group (p<0.001), and a dominant strategy in a cost–utility analysis (all primary outcomes).
In addition to healthcare professional training, another component commonly added to medication review interventions was the active involvement of a multidisciplinary team, which can improve communication and minimize provider barriers around uncertainty in role regarding changes to the patient’s drug treatment regimen. In one such medication review intervention, a primary care provider would send records to a homecare specialist, who would collect information to help inform a pharmacist’s medication review. This medication review was then followed up with recommendations that were sent back to the homecare specialist and primary care provider.62 This intervention was found to decrease the mean drug treatment appropriateness index score (indicating improvement in appropriateness) from the control phase to the intervention phase at 15 months (p≤0.001; primary outcome). In another medication review intervention, the multidisciplinary team included a geriatrician and a family practitioner who met after a medication review and clinical geriatric assessment. This intervention led to a higher mean health related quality of life (HRQoL) score in the intervention group versus the control group (p=0.03; primary outcome).63
Yet another component added to medication reviews was an electronic decision support website and a focus on patient care priorities or shared decision making, which can align deprescribing decisions with patient goals, facilitate communication between patients and providers, and reduce uncertainty in the deprescribing process. In one study, general practices were given access to a website that provided an educational module and template for general practitioners to conduct medication reviews that identified potentially inappropriate medications, opportunities for deprescribing, and patient priorities for care.64 This study found a larger reduction in the number of drug treatments in the intervention group versus control group at six months (p=0.045; primary outcome). In a second study, general practices were given a deprescribing guideline and training sessions on family conferences, during which a medication review took place with a frail patient and their family caregivers.65 While no effect was found on the primary outcome (number of hospitalizations in 12 months), the number of drug treatments (p=0.001) and number of potentially inappropriate medications (p=0.04; secondary outcomes) were lower in the intervention group than in the control group at six months.
Four more medication review interventions involved both multidisciplinary teams and focused on patient goals or shared decision making. In one of these studies, community pharmacists conducted medication reviews with patients with a drug burden index of ≥1 and subsequently met to discuss with the patient’s general practitioner, with patient expectations and desires as key elements in final decisions.66 This study did not find an impact in the proportion of patients who reduced their drug burden index by ≥0.5 at three months (primary outcome), but did find an impact on cognitive function (p=0.021) and fewer sedative side effects (p=0.024; secondary outcomes). Another study included a medication review by three experts (internal medicine specialist, clinical pharmacologist, and evidence based medicine expert) who provided specific recommendations for drug discontinuation to a general practitioner, who was invited to act on the recommendations based on a shared decision making process with the patient.67 This intervention was not found to affect hospital admissions or death (primary outcome) but did lead to fewer falls (p=0.04; secondary outcome). In a third study, a pharmacist performed a medication review with the final decisions agreed upon by both the physician and the patient in a face-to-face visit. This intervention led to fewer drug treatments (p=0.001) immediately after the review, and a greater likelihood of drug treatment discontinuation, dose adjustment, and substitution at three, six, and 12 months (primary outcomes). In a fourth study, the medication review conducted by the pharmacist incorporated goal setting by the patient and was followed up with face-to-face meetings with the patient’s general practitioner and discussion and follow-up with the patient. Compared with the control group, this intervention led to an improvement in HRQoL (3.4 points, 95% confidence interval 0.94 to 5.8, p=0.006) and a reduction in the number of health problems with moderate to severe impact on daily life in the intervention group (−0.34 problems, 95% confidence interval −0.62 to −0.044, p=0.024) at six months (primary outcomes).
Beyond medication reviews with patients, other interventions focused on patient facing deprescribing educational materials,2 deprescribing clinical decision support tools,3 or both.1 Patient facing deprescribing educational materials can empower patients to take a more active role in deprescribing decisions, thereby serving as facilitators to deprescribing. Of the two interventions focused on patient facing deprescribing educational materials, one was focused on benzodiazepines and provided specific information around the risks of benzodiazepine use and a stepwise tapering protocol.68 This study was conducted by community pharmacies and reduced benzodiazepine use (27% of the intervention group discontinued benzodiazepine use v 5% of the control group, risk difference 23%, 95% confidence interval 14% to 32%) at six months (primary outcome). The other study mailed a general deprescribing educational brochure and questionnaire to patients with dementia or mild cognitive impairment and their family members prior to a primary care visit.69 This was coupled with clinician notification and deprescribing tip sheets, but the intervention was not found to impact the primary outcomes (ie, number of long term drug treatments or proportion of individuals prescribed ≥1 potentially inappropriate medication) at six months.
Deprescribing clinical decision support tools coupled with shared decision making processes can help overcome patient, provider, and system barriers to deprescribing. Two such studies applied clinical decision support tools that reviewed drug treatments, and identified potential problems for providers to then make subsequent treatment changes based on shared decision making with the patient.7071 Neither study was found to impact their primary outcome (ie, hospital admission or death within 24 months, drug treatment appropriateness at 12 months); however, one study did find a greater reduction in number of drug treatments (p<0.001), which was a secondary outcome.70 A fifth study combined use of a clinical decision support tool in primary care clinics with individualized patient facing materials that included brief coaching on how to discuss concerns with the provider.72 The intervention was found to be associated with significantly more active patient communication, facilitative clinician communication, and drug treatment related communication among patients and clinicians (primary outcomes).
Hospitals
At hospital discharge, four study interventions focused on sending deprescribing reports to clinicians2 and medication reviews.2 Interventions generating individualized deprescribing reports can help providers with specific, actionable steps in deprescribing, such as which drug treatment to deprescribe, level of priority (among multiple drug treatments that should be deprescribed), and how to taper and monitor. In one study, a deprescribing plan was generated by a study physician,73 and in the second study, a report providing deprescribing opportunities was generated by clinical decision support software.74 In the first study, the report was sent to the patient’s physician, and in the second study, the report was sent to physicians, pharmacists, patients, and family members. Although the second study was not shown to reduce adverse drug events (primary outcome), both interventions led to deprescribing, as evidenced by a greater reduction in the mean number of drug treatments after three months (−2.6 intervention group v −0.36 control group, mean difference 2.25, 95% confidence interval 1.18 to 3.32; primary outcome for the first study) or increased deprescribing within 30 days (29.8% control group v 55.4% intervention group, adjusted risk difference 22.2%, 95% confidence interval 16.9% to 27.4%; secondary outcome for the second study).
Two additional studies involved either a pharmacist75 or a physician76 conducting a medication review, and then sending recommendations to the patient’s general practitioner. In the second study, two intervention arms were tested to allow for further patient engagement in one of those arms (eg, a drug treatment record sent to the patient).76 Across both studies, the interventions were not found to affect prescribing appropriateness, prescribing quality (eg, number of drug treatments, number of potentially inappropriate medications), or HRQoL (all primary outcomes). One study did report on implementation outcomes; for example, that 750 causes of drug related problems were identified, and that general practitioners implemented 42.4% of recommendations.75
Nursing homes
The two interventions targeting nursing home populations were both medication reviews involving multidisciplinary teams including pharmacists, physicians, and nurses. The first intervention also optionally involved family members for patients with cognitive impairment, and was not found to reduce falls at six months (primary outcome), but was found to reduce mortality (p<0.001) and hospitalization (p<0.001; secondary outcomes).77 The second study focused on older populations (aged ≥75 years) taking ≥10 drug treatments (specifically, ≥3 psychotropics), and was found to reduce the number of patients with ≥1 potentially inappropriate medication and the number of patients using ≥10 drugs at two months (p=0.007 and p=0.001, respectively; primary outcomes).78
In summary, various deprescribing interventions in older adults with polypharmacy have been tested in randomized controlled trials across different settings, with most (15 of 21 studies) conducted in primary care settings. The majority of interventions centered around medication reviews, with a mix of including additional components, such as a focus on shared decision making and/or patient care priorities, multidisciplinary teams, training for healthcare professionals, electronic clinical decision support tools or websites, patient facing deprescribing educational materials, and involvement of family. Such diverse intervention components represent how complicated deprescribing is and what is needed to overcome the myriad of existing patient, provider, and system barriers to deprescribing. Medication reviews with patients were generally successful in primary care settings (7 of 10 studies showing an impact on at least one primary outcome, and all showing an effect on at least one primary or key secondary outcome), but not always successful in other settings such as the hospital (two of two studies not showing an effect on primary or key secondary outcomes), although the number of studies was very small. Great variation in the mix of intervention components, as well as diversity in study population, led to small sample sizes within each stratum, making it difficult to determine if one particular component was considered to be more or less successful. Ongoing randomized controlled trials of deprescribing interventions in older adults with polypharmacy will provide more evidence, and further patterns should emerge as the number of studies grows.
Primary outcomes assessed across studies included drug treatment related outcomes (13 studies), HRQoL or health problems with impact on daily life, clinical outcomes such as falls, emergency room visits, hospitalizations, and deaths, cost, and outcomes related to shared decision making or patient/clinician communication (fig 8), and even greater diversity was observed in the secondary outcomes collected (see supplementary table 1). Of the 21 studies, just over half11 reported a significant effect on at least one primary outcome. Another six showed an effect on at least one of the key secondary outcomes, and four did not show an effect on primary or key secondary outcomes. The variability in results can be attributed to not only heterogeneity in intervention, but also the choice of primary outcomes. Notably, several studies emphasized the inadequacy of using clinical outcomes like hospitalization and death as a sole measure for deprescribing success, owing to challenges in translating the impact of discontinuing drugs into such downstream outcomes. Several studies using HRQoL as their primary outcome, and one study focusing on shared decision making and patient-provider communication as its primary outcomes, further underscored the need for a more comprehensive understanding of appropriate key clinical endpoints in deprescribing interventions.
Frequency of primary outcome assessed and if an effect was shown.
Furthermore, variability in drug treatment related outcomes across deprescribing studies adds to the complexity. Among primary outcomes, drug treatment related outcomes included number of drug treatments; number or proportion of potentially inappropriate medications; medication appropriateness index, drug burden index or other prescribing appropriateness criteria; drug related problems or adverse drug events; drug treatment changes (such as discontinuation, dose adjustment, or substitution; and proportion with ≥10 drug treatments (fig 9). This diversity hinders the comparison and generalization of findings, emphasizing the need for standardized measures. To tackle these problems, the use of core outcome sets can provide a standardized approach to outcome measurement and enhance comparability across diverse populations and settings in deprescribing research.798081 Commonly recommended outcomes within core outcome sets include drug treatment appropriateness, drug related problems, and hospitalizations associated with drug treatments, ensuring a more consistent and comprehensive evaluation of deprescribing interventions.
Frequency of specific drug treatment related primary outcomes.
In addition to outcome variability, difficulties such as overestimating intervention effects, limited provider acceptance, intervention spillover effects, and challenges in tracking pharmacy adherence in real world settings contribute to study variability. Recognizing the necessity of longer observation periods aligns with the complex nature of deprescribing interventions over time. Best practices for successful deprescribing interventions require targeting provider behavior change, employing methods to limit contamination effects in trials (eg, cluster randomized trials), powering studies for clinically meaningful effect sizes, and emphasizing longer follow-up periods. These practices contribute to a more robust understanding of deprescribing interventions and guide the development of effective strategies. While study follow-up periods varied (fig 10), with six months being the most commonly used timeframe, and 22 of 25 primary outcome measures within 12 months, the question of whether deprescribing interventions have long term effects remains unanswered. Two studies with 24 months of follow-up conducted in general practices indicated effectiveness in reducing falls and the number of drug treatments taken, suggesting potential long term benefits. However, further evidence is needed to establish the sustained efficacy of deprescribing interventions over extended periods.
Frequency of follow-up period for primary outcome.
Emerging interventions
Four ongoing randomized controlled trials are testing deprescribing interventions in older adults with polypharmacy in Canada, Germany, and Ireland, and are all focused on older adults who are being seen in primary care practices (table 1). The first two randomized controlled trials are investigating training for healthcare professionals. In the SPIDER cluster randomized controlled trial, targeting those taking ≥10 drug treatments in Canada, primary care practices will develop interprofessional learning collaboratives and work with quality improvement coaches.5455 The intervention will be compared with usual care, and the primary outcome will be reduction of potentially inappropriate medications at 12 months. Secondary outcomes include cost effectiveness, as well as patient and care provider perceptions of the intervention.
Characteristics of ongoing deprescribing randomized controlled trials in older adults with polypharmacy
Next, the PARTNER cluster randomized controlled trial in Germany targeting patients using ≥1 psycholeptic, sedative, or anticholinergic drug treatments comprises a multilevel, multicomponent intervention.56 Firstly, educational material will be provided for general practitioners and pharmacists on deprescribing psycholeptics, sedatives, and anticholinergics, and a moderated interprofessional workshop will take place. Secondly, leaflets will be used to encourage patient empowerment during a patient-pharmacist consultation. Thirdly, a patient-general practitioner consultation involving shared decision making will take place. This intervention will be compared with usual care, plus a patient-pharmacist consultation updating plans for drug treatment (but without a focus on psycholeptics, sedatives, or anticholinergics). The primary trial outcome will be reduction in exposure to psycholeptics, sedatives, or anticholinergics as measured by a point reduction of ≥0.15 in drug burden index at six months. Secondary outcomes include drug treatment related outcomes, such as frequency of deprescribing and frequency of other potentially inappropriate medications as well as quality of life and clinical outcomes such as falls, fall related hospitalizations, cognition, and insomnia.
Two additional trials are testing interventions focused on medication reviews. In the TAPER randomized controlled trial targeting those ≥70 years in Canada, a multidisciplinary comprehensive medication review focused on reducing drug treatment burden and conducted by pharmacists and physicians with patients will be compared with usual care.57 The primary outcome will be the number of drug treatments at six months and there are numerous secondary outcomes related to drug treatment use (eg, successful discontinuation or dose reduction, changes in drug treatment side effects and symptoms), quality of life, clinical outcomes like cognition, physical functional ability, and falls, healthcare resource use and cost effectiveness, and patient-reported deprescribing measures, such as confidence in drug treatment discontinuation and experience with the deprescribing process.
Lastly, a cluster randomized controlled trial in Ireland will use targeted medication reviews, in which pharmacists will meet with both patients and prescribers.5859 The intervention will be compared with usual care, and the primary outcome will be the number of potentially inappropriate medications at four months. Secondary outcomes include drug treatment related outcomes (eg, polypharmacy, drug treatment changes, adverse drug reactions, adverse drug withdrawal events), quality of life, and patient reported deprescribing measures (eg, patient attitudes toward deprescribing and the necessity of drug treatments).
In summary, four ongoing randomized controlled trials are being conducted in primary care practices, with a mix of intervention components such as medication reviews, patient empowerment leaflets, and training for healthcare professionals. The primary outcomes are all drug treatment related, such as reduction in potentially inappropriate medication or based on a score measuring drug treatment appropriateness (eg, medication appropriateness index or drug burden index), and are measured over 4-12 months of follow-up. Secondary outcomes are numerous, and include newer patient reported deprescribing measures (eg, attitude toward deprescribing and confidence in drug treatment discontinuation), as well as implementation outcomes (eg, costs, and patient/provider perception of, experience with, and satisfaction with the deprescribing intervention).
Guidelines
Deprescribing guidelines, algorithms, and additional resources have been developed in recent years for specific drug classes such as proton pump inhibitors,82 diabetes drug treatments,83 antipsychotics,84 benzodiazepines and Z drugs,85 and dementia drug treatments86 by the deprescribing.org research team.87 Another research team has also developed deprescribing guides for antidepressants, anticholinergics for Parkinsonism, antimuscarinics, sedating antihistamines, and opioids.88 Medstopper is a freely available web based tool that provides recommendations related to tapering and more.89 Additionally, countries such as Wales have produced guidance documents around management of polypharmacy in older adults, including recommendations on how to deprescribe specific drug classes.90 Not all drug classes have deprescribing guidelines, and evidence is still needed to support recommendations around optimal tapering and monitoring strategies. While a previous review identified several drug classes at increased risk of adverse drug withdrawal events,91 further evidence (such as through deprescribing trials) is needed to better understand the risk and severity of adverse drug withdrawal events after deprescribing, how this varies, and how to minimize these potential risks.92 To help fill these evidence gaps, multiple deprescribing networks are growing in Canada,93 Australia,94 the US,95 and Europe.96
Additionally, disease specific guidelines are also beginning to include consideration of discontinuation or de-intensification of treatment, especially among older adults with multiple comorbidities and polypharmacy. For example, the American Diabetes Association guidelines and the American Geriatrics Society guidelines both recommend de-intensification of diabetes drug treatments and individualized glycemic targets among older adults with multiple comorbidities.9798
Polypharmacy in older adults is associated with an increased risk of using potentially inappropriate medications,910 and several guidelines relate to potentially inappropriate medications. These guidelines include the Beers criteria in the US,99 the STOPP/START (screening tool of older persons’ prescriptions/screening tools to alert doctors to right treatment) criteria in Europe100 and Japan,101 the PRISCUS list in Germany,102 the European Union (7) potentially inappropriate medications (EU(7)-PIM) list which was developed by experts from seven European countries,103 STOPPFrail (screening tool of older persons’ prescriptions in frail adults with limited life expectancy) criteria in frail populations with limited life expectancy,104 RASP,105 and NORGEP.106 Other approaches and tools developed to help clinicians determine drug treatment appropriateness include the drug burden index,107 the anticholinergic risk scale,108 the medication appropriateness index,109 and the medication regimen complexity index.110 Across study interventions identified in this review, the most commonly used guidelines to identify potentially inappropriate medications were the STOPP, STOPP/START or STOPPFrail criteria (7 of 21) and Beers criteria (6 of 21; fig 11). Other ways to identify potentially inappropriate medications were based on country specific guidance, drug interaction databases (eg, LexiComp, UpToDate), EU(7)-PIM, NORGEP, or targeting specific drug classes such as benzodiazepines or opioids.
Frequency of guideline used to identify potentially inappropriate medications. STOPP=screening tool of older persons’ prescriptions; START=screening tool to alert doctors to right treatment study. *Included country specific guidelines, the EU(7)-PIM list, NORGEP, drug-drug interaction databases (eg, LexiComp, UpToDate), and targeting specific drug classes like benzodiazepines and opioids. †Study interventions could apply more than one guideline.
Furthermore, quality measures have also been used over the years to discourage potentially inappropriate medication use. For example, in the US, the Centers for Medicare and Medicaid Services has evaluated potentially inappropriate medication use in older populations through various quality measures for Part D prescription drug insurance plans, either generally (eg, high risk drug treatments in older people largely based on Beers criteria), or focused on specific drug classes and disease states (eg, antipsychotic use among older adults with dementia, multiple anticholinergic drug treatments, or multiple central nervous system active drug treatments in older adults with polypharmacy).44 Prescription drug plans are incentivized to reach out to beneficiaries and reduce potentially inappropriate medication use to perform well on these measures, because these measures can potentially influence beneficiary enrollment and bonus payments. In 2022, a new quality measure focused on deprescribing benzodiazepines in older adults was developed,111 indicating that future deprescribing quality measures could be used alongside guidelines to incentivize deprescribing at a large scale. Similarly, the Centers for Medicare and Medicaid Services also applies quality measures for nursing homes to monitor antipsychotic and sedative use.112
In summary, recently developed deprescribing guidelines and algorithms have focused on collating evidence and providing recommendations (eg, whether and how to taper) for specific drug classes. A variety of potentially inappropriate medication guidelines to help identify inappropriate drug treatment use exists, and quality measures can be used to incentivize reduction of inappropriate drug treatment use. As disease specific guidelines begin to include deintensification of treatment and the number of studies testing various deprescribing interventions grows, future guidelines should report on the intervention components and implementation strategies that have led to successful deprescribing interventions, and how these vary across setting.
Conclusions
This review summarizes published and ongoing deprescribing randomized controlled trials in older adults with polypharmacy, and describes important barriers and facilitators to deprescribing, as well as guidelines and resources that can support deprescribing. Medication reviews in primary care settings continue to represent a majority of deprescribing interventions, with many interventions incorporating a mix of components. Other intervention components included a focus on shared decision making and/or patient care priorities, active involvement of multidisciplinary teams, training for healthcare professionals, patient facing educational materials, and involvement of family members, which shows how complicated deprescribing can be. Just over half of interventions were found to be efficacious based on assessment of primary outcome alone, suggesting that applying such intervention components can help to overcome barriers in deprescribing. Most outcomes were assessed within 12 months of follow-up, and varied widely: drug treatment related (eg, number of drug treatments or number of potentially inappropriate medications), clinical outcomes (eg, hospital, death), HRQoL, and more. Ongoing trials are focused on primary care settings and are adding newer patient reported deprescribing outcome measures (eg, attitude toward deprescribing and confidence in drug treatment discontinuation), as well as implementation outcomes (eg, costs, and patient/provider perception of, experience with, and satisfaction with the deprescribing intervention).
Questions for future research
How effective are deprescribing interventions in the long term (>2 years)? What patient, provider, and system factors are associated with reinitiating treatments following deprescribing, and how should these be dealt with? How sustainable are deprescribing interventions in real world clinical settings, and what are the associated costs (eg, to support staff)?
What is the ideal primary outcome and core outcome set? How might these outcomes be based on setting, deprescribing intervention component, and study population?
To what degree have patients been involved as stakeholders in the design of deprescribing interventions, and to what extent do they affect effectiveness and implementation? How might patient input affect measures around adverse drug events?
How does deprescribing differ across populations; for example, in under-resourced populations? How do social determinants of health and the influence of cultural traditions and provider bias limit the effectiveness and implementation of current deprescribing interventions?
How do system level barriers like uncoordinated health services delivery affect deprescribing? How do disparate healthcare access, mistrust in health systems, poor communication, and provider negligence in considering patients’/caregivers’ values, healthcare beliefs, and literacy levels affect patients’/caregivers’ deprescribing decisions?
What types of culturally tailored interventions, shared decision making, and educational tools specific to deprescribing are most effective?
How patients were involved in the creation of this article
We thank the US Deprescribing Research Network Stakeholder Engagement Core and the two patient stakeholders who provided valuable feedback, including specific future directions for the ‘‘Questions for future research’’ section.
Acknowledgments
AH acknowledges the US Deprescribing Research Network Junior Investigator Intensive Program and the AGING Initiative Multiple Chronic Conditions Scholars Program.
Footnotes
State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors.
Funding: AH is supported by VA HSR&D (Career Development Award IK2 HX003359) and by the US Deprescribing Research Network (R24AG064025). JMP is supported by a Career Development Award K23AG058788 from the National Institute on Aging.
Sponsor’s role: The funders had no role in the design, analysis, or preparation of the paper. The contents do not represent the views of the US Department of Veterans Affairs or the US Government.
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: AH reports funding from the US Deprescribing Research Network (R24AG064025), honorariums and conference support from Academy of Managed Care Pharmacy, and honorariums from the Journal of Managed Care and Specialty Pharmacy. No other authors declare interests.
Contributing author statement: All authors made substantial contributions to the conception of the work, drafting/reviewing for important intellectual content, final approval of the version to be published, and agree to be accountable for all aspects of the work. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Provenance and peer review: commissioned; externally peer reviewed.