Primary Care Provider-Led Approach to Deprescribing Potentially Inappropriate Medications in Older Adults

Introduction

Older adults on potentially inappropriate medications (PIMs) are susceptible to adverse drug events including falls, delirium, and hospitalization (Pravodelov, 2020). The American Geriatric Society BEERS Criteria (2019), developed in 1991 and most recently updated in 2019, offers guidance to prescribers by categorizing drugs by their potential for harm in older adults. Despite the availability of this resource, studies across healthcare settings confirm high rates of PIM prescribing (Blachman et al., 2017; Kim et al., 2017; Stuckey et al., 2018). Several studies have found deprescribing in older adults is safe and feasible (Ibrahim et al., 2021; Wu et al., 2021) and that successful deprescribing can reduce adverse outcomes, including mortality, in community-dwelling older adults (Bloomfield et al., 2020). Given the risks of continued PIM use and the potential benefits of deprescribing, several recent initiatives have focused on reducing harmful medications in primary care. The Institute for Healthcare Improvement and the John A. Hartford Foundation’s Age-Friendly Health Systems Initiative includes reduction of PIMs as one of the four pillars of evidence-based care in older adults (K. S. Mate et al., 2018). Centers for Medicare & Medicaid Services also placed a call to action by assigning high priority to the 2020 Merit-based Incentive Payment System (MIPS) Clinical Quality Measure #238: Use of High-Risk Medications in the Elderly (American Medical Association, 2019).

Studies aimed at decreasing the number of PIMs in primary care have heavily focused on pharmacist-led, inter-professional, and computerized support interventions, with mixed results (Campbell et al., 2021; Cossette et al., 2019; Deyo et al., 2020; Martin et al., 2018; Stuckey et al., 2018; Zillich et al., 2008). While quality improvement work often requires a multidisciplinary approach, there are compelling reasons to focus deprescribing efforts on provider-level interventions. Polypharmacy and PIM use result from prescribing practices that rely on the motivations of and the relationships between patients and providers (Steinman et al., 2021). Primary care providers (PCPs) have the ultimate responsibility for managing a patient’s long-term medication list (Kvarnström et al., 2018). Patient-centered studies have previously shown that a majority of older adults would be open to deprescribing if recommended by their primary physician (Gaurang et al., 2021; Reeve et al., 2018). Despite patient willingness, common barriers to provider deprescribing have been identified as perceived lack of knowledge, skill, time, and organizational support (Pravodelov, 2020). In this study, our primary aim was to implement a provider-driven approach to deprescribing that focused on PCP education, dedicated time for panel review of PIMs, and a decision tree model of intervention and our secondary aim was to improve medication list accuracy.

Methods

This quality improvement intervention was conducted from April to October 2021 in an outpatient internal medicine clinic affiliated with an academic hospital system. The clinic population is mainly non-Hispanic white (73.4%) and English-speaking (97.4%) with an average age of 56.5 years. A Doctor of Nursing Practice student and their advisor led a short, 20-min educational didactic on PIMs at a monthly staff meeting in April 2021. The selected medications were benzodiazepines, sedative-hypnotics, tricyclic antidepressants, paroxetine, systemic estrogens, and glyburide. These were based on both the MIPS Clinical Quality Measure #238 (American Medical Association, 2019) and the Age-Friendly Health Systems’ list of high-risk medications (K. Mate et al., 2021). The presentation shared the risks of the selected medications and the aims of this quality improvement study. Providers were given access to resources for deprescribing (Deprescribing.org, 2022; US Deprescribing Research Network, 2022) and a handout developed by the study team in conjunction with a pharmacist on safer alternative medications. A study team member ran a report from the electronic health record to find all patients aged 65 and older who had one or more of the selected medications on their active medication list. A provider-specific spreadsheet was privately shared with PCPs electronically to complete their review. Providers were instructed to select an intervention from the decision tree for each PIM identified on their patient list. Interventions included remove from medication list, taper, schedule a visit with PCP or specialist, and other. Reasons for other included but were not limited to patient on hospice, medication for rare use, no alternative medication, and previously discussed risks/benefits with patient and wishes to continue.

Three provider-specific reminder emails were sent to encourage completion of the panel review within 3 weeks. After 3 weeks, clinic care coordinators contacted patients who required a visit or referral per the decision tree. Three attempts were made to reach each patient. Reports were run at 3 and 6 months to analyze the number of identified PIMs that had been removed from medication lists. Chart reviews assessed if medications were tapered or a visit with their PCP or specialist occurred. All reviewers followed a standardized protocol for the chart review process. Percentage change from baseline was calculated to identify potential improvement from this intervention. RStudio (RStudio, PBC, Boston, MA) was used to perform a test of binomial proportions to determine any statistical significance at α of .05. This study was reviewed and deemed exempt by the Institutional Review Board (IRB) as a quality improvement study, which does not require IRB oversight.

Results

Thirty-eight providers received the training and 33 had at least one patient with a PIM. Most PCPs (97%) completed the panel review. Out of 4570 patients, 9.7% had at least one of the identified medications on file. Patients with a PIM were predominately non-Hispanic white (86.9%), English-speaking (98.0%), and female (61.7%) with an average age of 73 years.

Among 449 reviewed medications, 233 required provider action (Figure 1). Scheduling a PCP visit and discontinuing medication were the most common interventions selected resulting in a 22.4% and 65.0% reduction at 6 months, respectively (Figure 1). Benzodiazepines and sedative-hypnotics represented 87% of the identified medications at baseline from participating providers. After 6 months, 46.2% of the benzodiazepines and 40.8% of sedative-hypnotics were deprescribed (Table 1).

OPEN IN VIEWER

Figure 1.

Flowchart with the selected interventions by providers for all patients with identified potentially inappropriate medications.

OPEN IN VIEWER

Table 1.

Potentially Inappropriate Medications by Drug Class at Baseline, 3 Months, and 6 Months.

Drug class	Baseline				Follow-up
	Month 0				Month 3		Month 6
	Total medications	Medications from participating providers	Exempt medications	Remaining medications	Tapered	D/C	Tapered	D/C
Benzodiazepines	247	228	124	104	6	35	7	41
Sedative-hypnotics	174	159	61	98	2	31	1	39
Tricyclic Antidepressants	33	32	18	14	1	1	1	3
Paroxetine	16	15	8	7	0	2	0	2
Estrogen	14	13	4	9	1	1	0	3
Glyburide	2	2	1	1	0	0	0	0
Total	486	449	216	233	10	70	9	88

Overall, this provider-driven deprescribing approach discontinued 88 and tapered 9 PIMs—a 41.6% reduction. After 6 months, 8.2% of the original patients had at least one PIM on their medication list showing a significant decrease from a baseline of 9.7% (p = .02).

Discussion

Practical solutions to tackling polypharmacy are needed and require prescriber engagement to effect lasting change. In this study, we implemented a provider-driven quality improvement intervention that resulted in high provider participation and a significant decrease in active PIMs. Identified medications were consolidated on a single document per provider with a decision tree model of interventions, allowing providers to quickly choose an appropriate patient-specific intervention, reducing the time demand to review and increasing provider participation.

Accurate medication lists are necessary to prevent medication errors, especially during transitions of care. Medication reconciliation has been a Joint Commission National Patient Safety Goal for Ambulatory Health Care since 2005 (Nassaralla et al., 2009). Despite the push for accuracy, prior studies have shown large discrepancies between what medications a patient actually takes at home and the medications found on the medication list (Rose et al., 2018). PCPs are the keepers of medication lists and it is their responsibility to ensure lists are regularly updated. It is notable that in our intervention, 23.2% of the actionable identified medications were removed from medication lists by PCPs without the need for a visit by the end of the study period. This highlights the fact that PCPs know their patients’ medications and can quickly identify medications that were outdated and no longer accurate.

PCPs scheduled separate visits to discuss medication use, which resulted in significant reductions in medication use through both discontinuation and tapering. This included benzodiazepine and sedative-hypnotic medications, which are often regarded as some of the most difficult medications to deprescribe. This indicates that active efforts to talk to patients about the risks and benefits of medications can lead to meaningful change.

Study limitations include implementation at a single academic clinic; though there was a large enough sample to detect significant change. Clinic care coordinators were used to contact patients and schedule visits, which may not be feasible for some clinics. The intervention included a single 20-min education session on the intervention and no-cost, online resources were made available to assist clinicians with deprescribing, which should lower the barrier for implementation. While we did allow for PCPs to choose an exemption if the identified medication was prescribed by a specialist, this was done for only 30 out of the initial 449 medications. Future deprescribing interventions should include tools to help PCPs collaborate with specialists who are prescribing PIMs to their patients.

Conclusion

Overall, this deprescribing approach catered to PCPs by highlighting PIMs on one document to review, and enlisted the help of clinic care coordinators to schedule visits and referrals specifically to discuss medication use. Future work is needed to assess the impact of this deprescribing approach in other primary care settings.