A narrative review of evidence to guide deprescribing among older adults

Abstract Potentially inappropriate prescription and polypharmacy are well‐known risk factors for morbidity and mortality among older adults. However, recent systematic reviews have failed to demonstrate the overall survival benefits of deprescribing. Thus, it is necessary to synthesize the current evidence to provide a practical direction for future research and clinical practice. This review summarizes the existing body of evidence regarding deprescribing to identify useful intervention elements. There is evidence that even simple interventions, such as direct deprescribing targeted at risky medications and explicit criteria‐based approaches, effectively reduce inappropriate prescribing. On the other hand, if the goal is to improve clinical outcomes such as hospitalization and emergency department visits, patient‐centered multimodal interventions such as a combination of medication review, multidisciplinary collaboration, and patient education are likely to be more effective. We also consider the opportunities and challenges for deprescribing within the Japanese healthcare system.


| INTRODUC TI ON
The consequence of population aging and multimorbidity includes polypharmacy and adverse drug events. 1 Polypharmacy may trigger drug interactions, adverse drug events, increased healthcare costs, and decreased adherence to medication. 2 Adhering to current clinical practice guidelines in caring for an older person with multimorbidity may lead to polypharmacy, 3 regardless of whether the prescription is to prevent life-threatening events [4][5][6][7] or to alleviate symptoms. This is because most clinical guidelines are aimed at a single disease status. 3 Many researchers have studied effective interventions for polypharmacy. However, recent systematic reviews regarding interventions to reduce polypharmacy have failed to demonstrate a benefit in clinical outcomes such as survival. [8][9][10] Since potentially inappropriate prescribing and polypharmacy have the potential for adverse clinical outcomes, it is necessary to synthesize the current evidence to provide a practical direction for future research and clinical practice.
The prevalence of polypharmacy among older Japanese adults is relatively high, 11,12 and the number of drugs prescribed tends to increase until people are aged over 90 years, despite the evidence that the use of multiple drugs after the age of 80 poses a higher risk of adverse drug events. 13 The Japanese healthcare system includes free access to healthcare facilities, which could lead to fragmentation of care without coordination ("polydoctoring"). 14  of polypharmacy is exceptionally high among older adults who visit multiple healthcare providers. 15 The revised Japanese Geriatrics Society guidelines explicitly disclosed the risks associated with multidrug use and listed potentially inappropriate drugs. 16 The risks of concomitant use of multiple drugs have also been covered in mass media, and awareness of the problem of polypharmacy is now widespread. 17 Considering the impact of healthcare systems and cultures on medical practice, a narrative review outlining existing evidence with a focus on the Japanese context would help healthcare providers in Japan.
This review summarizes the existing body of evidence regarding polypharmacy interventions to elicit useful intervention elements and potential disadvantages. We also describe the prospects for which intervention methods could be practical or what approaches are needed within the Japanese healthcare system.

| ME THODS
We conducted a review of studies examining the effects of polypharmacy interventions among older adults. As overall reporting guidelines, we partially adopted PRISMA 18 and the recommendation by Ferrari. 19 SANRA 20 was used for methodological rigor. A literature search of databases including PubMed, Embase, Google Scholar, Ichushi-Web (Japanese medical literature), and J-STAGE was conducted from inception to March 14, 2021. The researchers used "polypharmacy," "intervention," "deprescriptions," "potentially inappropriate medications," "medication reconciliation," "clinical decision support systems," "patient care teams," "interdisciplinary communication," "education," and "feedback" as keywords. We cited individual randomized controlled trials (RCTs), nonrandomized interventional studies, observational studies, systematic reviews, narrative reviews, qualitative studies, and published gray literature as appropriate.
Reference lists of the included studies were also used for further literature searches. We excluded studies that did not address any interventional aspects. Literature published in English and Japanese was adopted.
The Effective Practice and Organisation of Care (EPOC) taxonomy was adopted to classify the interventions into the elements of "delivery arrangements," "implementation arrangements," and "financial/governmental arrangements." 21 As this study was a narrative review of published sources, an ethical review was deemed unnecessary.

| RE SULTS
Polypharmacy interventions tend to be multifactorial. In a scoping review of intervention elements to reduce inappropriate prescribing by Lee et al., the included studies had an average of 2.5 intervention elements. 22 More than 80% of the trials used more than one element. 22 This review focuses on the major components of interventions and discusses them according to the EPOC taxonomy, classifying interventions as shown in Table 1. We focus especially on delivery arrangements and implementation arrangements to explore new perspectives on intervention approaches in clinical practice.

| Delivery arrangements
Delivery arrangements are defined as "changes in how, when and where healthcare is organized and delivered, and who delivers healthcare," called organizational interventions in the previous version of the EPOC taxonomy. 21 We will summarize the interventions by dividing them into those that use specific criteria and those that do not. interventions.

Direct deprescribing
The use of prophylactic medications such as cardiovascular drugs is common among older adults. Thus, healthcare providers need to continually evaluate on an individual basis whether continuing medication prescribing is still justified, in light of age-related comorbidities and changes in frailty. Potential triggers for prophylactic drug deprescribing include current or anticipated adverse drug events, medication duplication and errors, and limited life expectancy. 23 It is relatively straightforward to deprescribe prophylactic medications once the patient and the prescriber understand the risk-benefit of discontinuation. However, lack of understanding of the patient's perception of the medical condition and drug treatments can be a barrier to direct deprescribing. 24,25 Concerns about withdrawal symptoms, which are common with symptomatic medications, can also be a barrier to reduced prescribing. 26 Although there is limited evidence to guide the reduction or cessation of individual drugs, several key articles can be utilized for patient-centered communication and shared decision making (Table 2).

Explicit criteria-based interventions
These are methods that use criteria (e.g., Beers Criteria, 27 31 In Japan, the application of the Japan Geriatrics Society's Guidelines for Medical Treatment and its Safety in the Elderly 2015 16 was reported to be a useful tool for deprescribing. 32 However, the effect of explicit criteria intervention on clinically significant endpoints, such as death and rehospitalization, is unclear.

Implicit criteria-based interventions
This approach determines the appropriateness of each drug, including indication, safety, efficacy, and manageability, for each patient. preferentially from drugs with a higher risk than benefit and a lower likelihood of withdrawal symptoms or symptom recurrence, and (5) monitor carefully after the deprescribing. 33 In an RCT comparing the deprescribing protocol with usual care in 95 patients aged ≥65 years, the number of drugs taken was reduced without significant adverse effects on survival or other clinical outcomes (The mean change in number of regular medicines at 12 months was −1.9 ± 4.1 in intervention group and +0.1 ± 3.5 in control group). 34

Medication review
Medication review is a comprehensive intervention, a structured evaluation of a patient's regimen to optimize medication use and improve health outcomes. 35 In a 2020 review of inappropriate prescribing interventions for multimorbid older outpatients, 70% of studies included medication review, the most frequent of the 14 intervention elements. 22 A 2016 Cochrane review of the effect of medication review on hospitalized patients showed a reduction in emergency department , with a number needed to treat to prevent an ED contact of 37 for a low-risk population and 12 for a high-risk population (e.g., elderly patients, patients with multiple co-medications) over one year. 36 However, systematic reviews that did not limit the patient population to hospitalized patients did not show clinical benefit. [8][9][10] This body of evidence suggests that the effects of medication reviews' effects may be subtle unless targeted at high-risk populations.
In an RCT by Ravn-Nielsen et al., a composite of readmissions or ED visits were reduced in the multimodal intervention group utilizing medication review, combined with motivational interviewing and multidisciplinary team follow-up, compared with the usual care among patients admitted to an acute care medical ward (HR, 0.77; 95% CI, 0.64-0.93). 37 However, there was no significant difference in outcomes between the usual care and medication review alone groups. Similarly, interventions that combine patient interviews and patient education with medication review have been shown to reduce hospital visits and drug-related hospitalizations 38 and ED visits. 38,39 Patient-centered multimodal interventions, such as a combination of medication review, multidisciplinary collaboration, and patient education, may be more effective than medication review alone.

Clinical decision support system
A clinical decision support system (CDSS) is designed to improve medical decisions with targeted clinical knowledge, patient information, and other health information. 40 Evidence suggests that the use of CDSS was useful for reducing potential drug therapy problems in nursing homes 41 and new PIMs in the elderly. 42

| Implementation arrangements
Implementation arrangements for inappropriate polypharmacy involve a comprehensive, practice-oriented approach to changing the culture of individual healthcare professionals and organizations. 21 They include education and feedback for healthcare professionals and patient education.

Educational interventions
Educational interventions for medical professionals are often targeted at physicians. They range from simple (e.g., the use of explicit criteria such as the STOPP/START) to a more comprehensive set of educational sessions (e.g., the pharmacokinetics of the elderly, comprehensive geriatric assessment, basic knowledge of polypharmacy).
Pre-post comparison studies examining the effects of explicit criteria-based physician education have reported a decrease in the number of medications and PIMs and improved MAI scores. 56,57 However, the effects of educational interventions using an individualized implicit approach have been inconsistent. For example, a cluster RCT that educated clinicians regarding comprehensive geriatric assessment, the pharmacokinetics of the elderly, and PIMs through e-learning did not reduce PIMs. 58 Another cluster RCT that implemented a 10-h educational program and telephone consultation service targeted at physicians revealed a significant PIM reduction and drug duplication in the intervention group. 59

| Patient education
In addition to health professional education, patient education plays a crucial role in deprescribing. A Canadian cluster RCT that examined the effect of pharmacist-led patient education on patients with chronic benzodiazepine use showed a significant reduction in the benzodiazepine prescription, with a discontinuation rate of 27% in the intervention group and 5% in the control group. 68 In a similar cluster RCT of older patients taking PIMs, explaining drug information to patients using educational pamphlets significantly reduced inappropriate prescribing after six months. 69 An Australian report prioritized elements of practical interventions. "Pharmacist-led medication reconciliation for new residents," "facility-level drug audits and feedback to healthcare providers and staff," and "prescription scripting to support physician-patient discussions" were deemed to be high priorities for deprescribing among facility residents. 70  year. 81 This result may indicate that there are barriers for deprescribing such as the shorter hospital stay and "polydoctoring," or that the reimbursement system for deprescribing has not been widely recognized. Further dissemination of the policy and verification of effectiveness, as well as removing these barriers, are required.

Antihypertensive medications
Deprescribing of antihypertensive medications is considered when adverse drug events, such as dizziness, fainting, falls, and fall-related injuries, are possible. In RCTs that examined the effects of reducing antihypertensive medications in older adults, blood pressure increased by about 7-15 mmHg immediately after discontinuation, then gradually returned to baseline within nine months. 82,83 Patients with optimal blood pressure control may benefit from knowing that deprescribing may raise their blood pressure temporarily, but in time it may settle down.

Proton pump inhibitors
Dose reduction or cessation of proton pump inhibitors (PPIs) have been associated with symptom recurrence. A 2017 Cochrane review compared gastrointestinal symptoms between dose reduction and continuation in patients with long-term PPI use. 84 The study found that gastrointestinal symptoms could occur within two weeks of discontinuation. Tapering or on-demand deprescribing was also associated with symptom recurrence. In a Swedish RCT that examined PPI tapering in long-term users, only 27% of patients in the intervention arm could discontinue PPIs. 85 When deprescribing symptomatic medications, clinicians should explain to patients that symptom recurrence may occur and that they can restart the medication at any time if necessary. A common understanding is essential before initiating deprescribing.

| Withdrawal symptoms with deprescribing
Certain drugs can cause withdrawal symptoms upon discontinuation. The most famous drug class includes benzodiazepines, but several other drugs can also cause withdrawal symptoms 86 (Table 3).
Attention should be paid when reducing these medications in patients who have been taking the drug for a certain amount of time.

| Communication errors with patients
Patients with polypharmacy may feel that they have too many medications and be willing to discontinue them if they know that their physician can resume them as necessary. 87,88 Some patients are reluctant to reduce their medications if they expect them to have preventive effects in the future. 89 Moreover, if patients are anxious about discontinuing a drug, they may be more likely to experience side effects when the drug is reduced or discontinued (nocebo effect).
Healthcare providers should understand such feelings and expectations about the target medication. One study that evaluated opioid deprescribing in patients with chronic pain found that nocebo effects could be minimized by thoroughly educating patients about the benefits of medication reduction and reducing opioids more slowly than the standard duration. 90 The keys to successful intervention are to provide clear guidance on discontinuation, reduce patient anxiety, and create an individualized drug reduction protocol.
Clinicians need to balance these keys with the amount of time and effort required.

| Communication errors among healthcare professionals
Clinicians' attitudes toward prescription-related problems vary. In addition to physician factors (e.g., beliefs, attitudes, knowledge, skills, and behaviors), many external factors (e.g., work environment, healthcare system, culture) influence their clinical decision. 91 Clinicians' inertia is a characteristic that makes them more likely to continue potentially inappropriate prescriptions. It was reported that general practitioners were more likely to continue prescribing, because of uncertainty and lack of information. [92][93][94] They are also less likely to discontinue a drug prescribed by another specialist, [95][96][97] probably due to concern about deterioration in the physician-physician relationship. 98

| Summary
Evidence on hard endpoints such as mortality, hospitalization, and falls is scarce and further research is needed. Improving process indicators, such as the number of prescribed medications and PIMs, can be achieved with any approach in the EPOC taxonomy. CDSS, MDTM, and educational interventions are useful, but direct deprescribing targeted at Fall Risk Increasing Drugs and explicit criteria-based approaches can also effectively reduce PIMs. But if the goal is to improve clinical outcomes such as death, hospitalization, falls, and QoL, patient-centered multimodal interventions such as a combination of medication review, multidisciplinary collaboration, and patient education are likely to be more effective. Table 4

| What interventions are recommended?
A 2016 meta-analysis found no survival benefit for deprescribing. 102 However, subgroup analysis found that patient-specific

TA B L E 4 (Continued)
improve clinical outcomes. 103 A patient-centered, shared decisionmaking model of deprescribing is a possible model that has been proposed recently. 104 Problem-solving-oriented MDTMs are effective in reducing potentially inappropriate prescribing and healthcare costs but do not improve clinical outcomes. However, indirect benefits of MDTMs, such as nurturing trust and organizational culture in the workplace, could be expected in facilities that address polypharmacy. 22 The American Geriatrics Society has proposed five critical elements for effective multidisciplinary collaboration: (1) shared goals and objectives, (2) clarification of roles and responsibilities, (3) appropriate contributions of team members, (4) cooperation and coordination in activities, and (5) fostering mutual trust through ongoing relationships. 105

| Problems to be solved in the Japanese context
Japan has had universal medical coverage since 1961, and copayment is inexpensive by global standards. 106 Access to health care is good because patients can visit medical institutions without paying much attention to costs. However, this accessibility tends to lead to excessive medical care. 107 The rate of inappropriate prescribing was higher among those fully exempted from public payment for medical services. 108 Recently, public education via mass media has become widespread. This leads to public opinion that polypharmacy is a significant public health concern. 17 A resulting problem is the psychological avoidance of problems related to polypharmacy. 109 This is because oversimplified criticism of polypharmacy, which ignores each case's context, has sometimes been disseminated via mass media and medical professionals. However, it has been accepted that there are "appropriate polypharmacy" and "problematic polypharmacy" and that polypharmacy and the use of PIMs are not uniformly harmful. 110 Another disadvantage of free access is polydoctoring, that is, patients seeking care from multiple providers.
Since polydoctoring is a known risk factor for polypharmacy, 15 it is necessary to establish an environment in the Japanese healthcare system that encourages people to utilize primary care physicians as the point of contact for care, as well as better multidisciplinary cooperation. As stated before, the recent reimbursement system for deprescribing may not been widely recognized. Adequate public awareness campaign of the system, further incentives for deprescribing, and the implementation of hospital dashboard including the rate of acquisition of the deprescribing reimbursement would be expected to promote dissemination of the policy. Recently, pharmacists in clinical practice have become actively involved in pharmacotherapy through multidisciplinary collaboration. 111,112 Nevertheless, a lack of trust and communication among pharmacists and prescribers can be a barrier and should be addressed. 113

| Future implications
As mentioned, reducing the number of prescribed medications and PIMs does not directly lead to improved clinical outcomes. Thus, it is necessary to recognize that the number of drugs is no more than an intermediate factor. Nevertheless, using the number of drugs as a surrogate endpoint, patient-centered multifaceted intervention There is no single intervention that can be expected to improve clinical outcomes, and patient-centered multifaceted interventions combined with medication review may be effective. It is necessary to examine what mechanisms work behind such relationships and which factors play a crucial role in clinical effectiveness.
Fostering a culture of preventing and reducing inappropriate polypharmacy in medical practice Trust and "culture" between professions is a prerequisite for a useful medication review. What is necessary to develop such a culture? What process should be used to do so at each medical facility?
What are the subgroups for which interventions are most effective?
Assuming that the effect size of uniform polypharmacy intervention is not substantial, future studies need to clarify what kind of deprescribing approach is most likely to offer benefit for particular patient populations.

Studies using patient-centered outcomes
It is necessary to examine the effect of interventions on patient-centered outcomes such as health-related quality of life. Likewise, we recommend evaluating the quality of care from multiple perspectives, including QoL and patient experience.
Research from the perspective of medical practice in Japan It is worth examining whether concepts proposed and interventions proven in Europe and the United States are also useful in Japan's cultural context and the healthcare system. Evidence in the Japanese context, especially from qualitative and mixed methods studies, is needed.

Examining the effects of policy interventions
The impact of the 2016 implementation of the fee for deprescribing on physicians' prescribing behaviors, patient outcomes, and healthcare costs should be examined.
local formularies, may be vital. We have summarized the challenges for future polypharmacy interventions from the Japanese primary care perspective in Table 5.
Cross-disciplinary approaches, such as behavioral economics, may become more critical in developing practical approaches to inappropriate polypharmacy. This is because patients' health careseeking behavior and doctors' clinical decisions are often not based merely on scientific evidence. It is further necessary to educate the public so that polypharmacy is viewed as an opportunity to seek better medical care.

ACK N OWLED G EM ENTS
We would like to thank Helena VonVille and Junji Haruta for their assistance with the review methodology. The authors received no specific funding for this work.

CO N FLI C T O F I NTE R E S T
Authors declare no conflict of interests for this article.