Decreasing polypharmacy in older adults with cancer: A pilot cluster-randomized trial protocol

Abstract

Introduction:

Polypharmacy is prevalent in older adults with cancer and associated with multiple adverse outcomes. A single-site, cluster-randomized clinical trial will enroll older adults with cancer and polypharmacy starting chemotherapy and will assess the effectiveness and feasibility of deprescribing interventions by comparing two arms: a pharmacist-led deprescribing intervention and a patient educational brochure.

Materials and Methods:

The study will be conducted in two phases. In phase I, focus groups and semi-structured individual interviews will guide adaptation of deprescribing interventions for the oncology clinic (phase Ia), and eight patients will undergo the pharmacist-led deprescribing intervention with iterative adaptations (phase Ib). In phase II, a pilot cluster-randomized trial (n = 72) will compare a pharmacist-led deprescribing intervention with a patient education brochure, with treating oncologists as the cluster. Both efficacy (relative dose intensity of planned chemotherapy, potentially inappropriate medications successfully deprescribed, chemotherapy toxicity, functional status, hospitalizations, falls, and symptoms) and implementation outcomes (barriers and facilitators) will be assessed.

Discussion:

This study is anticipated to provide pilot data to inform a nationwide randomized clinical trial of deprescribing in older adults starting cancer treatment. The cluster randomization is intended to provide an initial estimate for the intervention effect as well as oncologists’ intra-class correlation coefficient. Deprescribing interventions may improve outcomes in older adults starting cancer treatment, but these interventions are understudied in this population, and it is unknown how best to implement them into oncology practice. The results of this trial will inform the design of large, randomized phase III trials of deprescribing.

1. Introduction

Polypharmacy (PP), the concomitant use of multiple medications, is prevalent in older adults with cancer. Increasing multimorbidity, “prescribing cascades” (medication initiated to treat the adverse effects of another medication) [3], fragmented care across multiple specialists [4], and hesitancy of providers [5] and patients [6] to discontinue medications all contribute to PP. Among older adults with cancer, 60–84% of patients are taking ≥5 medications [7], compared to 30–60% of older adults without cancer [8]. PP increases the risk that one or more medications are “potentially inappropriate” (PIMs), with potential risk higher than anticipated benefit [9]. In older adults with cancer, PP/PIMs have been associated with lower doses of cancer therapy [10–12], chemotherapy toxicity [13–16], impaired functional status [17–20], unplanned hospitalizations [21,22], falls [23–25], and mortality [26,27]. PP/PIMs also increase the number of clinically significant drug-drug and drug-chemotherapy interactions [28–32].

Deprescribing is “the process of withdrawal of an inappropriate medication… with the goal of managing PP and improving outcomes.” [33] Most studies to date in the general population of older adults have focused on reduction in number of medications or specific drug classes [34–38]. For example, the phase III EMPOWER trial used a patient brochure to direct deprescribing of benzodiazepines, and patients receiving the brochure had eight-fold higher odds of discontinuing benzodiazepines compared to patients in the control group [39]. Data for improved distal health outcomes (such as falls, hospitalization, and survival) are lacking, due to small sample sizes and limited follow-up [40]. However, deprescribing is acceptable to patients; in a survey of older Medicare beneficiaries (n = 1981), 92% reported they would be willing to stop medications if recommended by their physician, and 67% wanted to reduce the number of medications they were taking [41]. Multiple types of deprescribing interventions have been developed, which can be generally grouped by the initiator or promoter of the intervention: patient [39,42], physician [43], pharmacist [44,45], or specialized multidisciplinary team [46]. Interventions range from simple and low-resource (e.g., providing a one-time educational handout to patients [39]) to multi-level and high-resource (e.g., the review of medications over multiple clinic visits by a large multidisciplinary team [46]).

For patients with cancer, deprescribing has largely been studied in the context of palliative care and end-of-life when patients no longer have the life expectancy to benefit (“time-to-benefit”) from many chronic medications. In a randomized trial of 381 patients with a life expectancy of ≤1 year (49% of whom had a cancer diagnosis), discontinuing statins did not worsen survival or increase cardiovascular events, but it did improve quality of life [47]. Deprescribing approaches in the oncology clinic, however, are understudied, particularly for patients on active cancer treatment. In our pilot study of 26 older adults with cancer referred for geriatric oncology consultation, pharmacist-led deprescribing was shown to be feasible, requiring an average of 30 min per patient to complete (range 18–77 min). A mean of five PIMs were identified per patient, and 73% of PIMs were successfully deprescribed. Patients reported high satisfaction with the intervention, and 52% of patients reported no perceived barriers to deprescribing. For patients who perceived a barrier, fear of symptom recurrence or disease worsening was the most common [45].

Oncology clinics may be in a unique position to deprescribe: oncology teams see patients receiving cancer treatment frequently and may have easier access to pharmacist collaborators given chemotherapy workflows. It is common for oncologists to assume primary care duties during active cancer treatment [48], and primary care providers (PCPs) report they do not participate in assessment and management of adverse events from chemotherapy, which may be precipitated by PP/PIMs [49]. However, barriers and facilitators to deprescribing in the oncology clinic have not been fully elucidated. Multiple qualitative and mixed-methods studies with patients and PCPs have identified barriers and facilitators of deprescribing for the general population [5,6,50,51], but data incorporating specialists (like oncologists) are limited [52]. From the patient perspective, impressions of medication necessity, pressure to continue medications by physicians or family members, lack of time and support to discuss medication concerns in clinic, and worry about the effects of discontinuing medications were the primary barriers identified in one systematic review [6]. Trust in physicians and beliefs in medication “unnaturalness” or harm have been cited as facilitators [6,50]. From the perspective of PCPs, barriers to deprescribing include a fragmented, complex system of care, hesitancy to challenge a specialist’s recommendations, time pressures, and knowledge gaps [50]. Facilitators include access to pharmacist consultation and information technology systems, but these are often unavailable [53] or poorly implemented [54].

To better understand both the implementation and effectiveness of deprescribing among older adults receiving active cancer treatment, a clinical trial protocol was developed. The conceptual model (Fig. 1) shows the factors considered in the development of this multi-phase hybrid implementation-effectiveness trial [55]. Initial focus groups with key stakeholders will inform adaptation of trial procedures. A “lead-in” phase will allow for iterative adaptation of a pharmacist-led deprescribing intervention. Finally, a cluster-randomized clinical trial will compare the pharmacist-led deprescribing intervention with a patient educational brochure, collecting both effectiveness and implementation outcomes.

Fig. 1.

Conceptual model. The lightning bolt represents potential barriers to receipt of chemotherapy, which would result in lower relative dose intensity (RDI).

2. Methods

2.1. Aims

The primary or effectiveness aim is to investigate the effects of deprescribing interventions on relative dose intensity (RDI) in older adults undergoing chemotherapy. This outcome was chosen as it is associated with both the effectiveness and the toxicity of chemotherapy. PP has been associated with reduced RDI [56,57], which in turn has been associated with decreased survival [58]. Other effectiveness outcomes of interest include number of PIMs deprescribed, chemotherapy toxicity, functional status, hospitalizations, falls, and symptoms after three months of chemotherapy. The secondary or implementation aims are: (1) to adapt and refine potentially scalable deprescribing interventions and (2) to identify barriers and facilitators of implementation of deprescribing interventions.

2.2. Study Design and Setting

The study will be conducted in two phases (Fig. 2). In phase I, focus groups and semi-structured individual interviews will guide adaptation of deprescribing interventions for the oncology clinic (phase Ia), and eight patients will undergo the pharmacist-led deprescribing intervention to assess acceptability and feasibility and perform iterative adaptations (phase Ib). In phase II, a pilot cluster-randomized trial will compare a pharmacist-led deprescribing intervention with a patient education group. In this single-site pilot trial, a total of 72 older adults with cancer and PP starting chemotherapy at the University of Rochester Wilmot Cancer Institute and affiliated community oncology clinics will be enrolled across at least 12 oncologists (at least six oncologists per arm) with randomization at the level of the oncologist (i.e., cluster is oncologist). Both efficacy and implementation outcomes will be assessed.

Fig. 2.

Study schema.

2.3. Interventions

Two arms will be developed and compared, each adapted from a prior intervention reported in the literature: (1) pharmacist-led deprescribing and (2) patient education brochure.

Pharmacist-led deprescribing.

The proposed pharmacist-led intervention will be adapted from our pilot work [45]. A pharmacist will review the patient’s current medication list and develop a list of targeted recommendations for deprescribing. The pharmacist will review the recommendations with the treating oncologist, and then conduct a virtual visit with the patient via video-conferencing software either during a scheduled visit to the oncology clinic (i.e., before or after the clinic appointment) or when the patient is at home. For an in-clinic visit, the pharmacist will be located outside the clinic space, and the patient will use a tablet computer loaded with a HIPAA-compliant video-conferencing software.

During the virtual pharmacist visit, the pharmacist will verify the medication list and adherence for all medications. Once the medication list is verified, the pharmacist will utilize their expertise in addition to validated criteria for assessing medication appropriateness. Validated criteria utilized in both the preliminary pilot and current studies include the Beers criteria [59], an explicit list of PIMs developed, reviewed, and published by the American Geriatrics Society based on expert consensus; the Screening Tool of Older Person’s Prescriptions (STOPP) list [60], another explicit set of criteria defining PIMs in older adults; and the Medication Appropriateness Index (MAI) [61], a set of implicit criteria evaluating the appropriateness of medications based upon current indication, duration, interactions (drug-drug, drug-disease, and drug-condition), dosage, administration instructions, and effectiveness to generate recommendations.

Recommendations will be entered into a standardized communication template that is reviewed and co-signed by a geriatric oncologist. This template will be sent to the patient’s treating physicians (via electronic health record [EHR] inbox or fax) and will be available to the patient participant via the EHR patient portal. The pharmacist will also document written instructions in the patient participant’s after-visit summary where applicable. If discontinuation is contingent upon another clinician’s approval, this will be expressly stated in both verbal and written recommendations. The pharmacist and/or principal investigator (PI) will contact the clinician via phone, email, or EHR inbox to discuss the recommendation. If the clinician disagrees, the medication will not be discontinued. Final recommendations will be conveyed to the patient participant in a follow-up phone call. If, upon review of recommendations, another clinician on the patient’s care team disagrees with a recommendation, the PI will discuss it with the clinician until consensus is reached. The consensus recommendation will be conveyed to the patient participant by phone and in writing, including specific and tailored information about how and when to adjust medications. Any recommendations that are disagreed upon and ultimately decided against will be documented. Type of medication and reason for refusal of deprescribing will be recorded. The pharmacist will conduct a follow-up phone call weekly for four weeks post-intervention to answer patient participant questions and assess for any problems (such as worsening symptoms or withdrawal). Via the electronic health record (EHR), the pharmacist will send the initial evaluation and recommendations, as well as brief notes on follow-up telephone conversations, to the oncologist and other clinicians including the PCP.

Patient education brochure.

The proposed patient education intervention will be adapted from the phase III cluster-randomized EMPOWER trial [39], in which patients were given targeted educational material in the form of brochures. The brochures detail the risks of medications (such as benzodiazepines) in older adults, tapering schedule, and alternative therapies to consider [39]. They are available for use and download. In conjunction with pharmacist, oncologist, and patient participants in phase Ia, these brochures will be adapted to create a general educational brochure about PP and deprescribing. The brochure will also undergo separate expert review by a Health Literacy Specialist at our institution’s medical library to ensure reading comprehension at a fifth-grade level. These brochures will be provided to patient participants by the research staff.

2.4. Study Procedures

Phase Ia:

In phase Ia (Fig. 1), the PI will conduct focus groups (5–8 participants per group, all with the same role) with the following stakeholders: pharmacists, oncologists, oncology nurses, PCPs, and patient advocates. These focus groups will be audio recorded and transcribed into deidentified transcripts. After preliminary analysis of focus group transcripts, a report of major themes elicited will be sent to each group’s members for comment (member checking) [62]. Individual semi-structured interviews will be used to follow up on themes needing further elucidation if necessary.

Phase Ib:

Eight patient participants will be enrolled as a pre-pilot cohort testing the pharmacist-led deprescribing intervention to assess acceptability by patients [63] and feasibility. The intervention will be as described in “Study Design and Setting” above, with possible ongoing adaptations informed by phases Ia and Ib. Fig. 3 shows possible iterative adaptations that may occur based on the data collected in these phases. The PI or a research coordinator will directly observe each virtual pharmacist visit (in the same room as the patient participant or joining as a virtual participant in the Zoom call if the entirety of the patient participant’s clinic visit is conducted virtually). The observer will conduct 5–10 min exit interviews with the patient participants and caregivers (if present) to assess acceptability and barriers to a virtual pharmacist visit. Post-intervention semi-structured interviews with the patient participant will be conducted by a trained research coordinator at 1–2 weeks and 4–6 weeks post-intervention.

Fig. 3.

Example of possible iterative adaptation throughout phases Ia and Ib, focusing on a communication template for deprescribing recommendations. Iterative adaptation may be performed on any aspect of the intervention workflow.

EHR = electronic health record.

Phase II:

A two-arm, cluster-randomized trial will deliver and compare the adapted intervention strategies (pharmacist led deprescribing versus educational brochure). Oncologist clusters will be enrolled using a “matched pairs” design that matches two oncologists who see similar types of patients and allocates them randomly to different arms [64]. The interventions will be delivered as outlined above.

2.5. Eligibility Criteria

Oncologists are the “cluster” for the planned cluster-randomized trial. Oncologists will be recruited from Wilmot Cancer Institute (WCI) and associated regional sites. Oncologists must be a current faculty member within the WCI and/or the affiliated community oncology group and see ≥20 patients per year with any combination of patient eligibility diagnoses. Oncologists must not be planning to retire within the study period.

Patient study participants must be 65 years or older; have a diagnosis of malignancy including aggressive lymphoma or cancers of the breast, gastrointestinal system, genitourinary system, or lung; be initiating chemotherapy, alone or in combination with other systemic antineoplastic agents for a period of at least three months within four weeks of enrollment; screen positive for PP/PIMs (Table 1); be able to read and write English; and be able to give informed consent. Patient study participants must not be planning to receive a cancer treatment regimen that does not include standard cytotoxic chemotherapy (e.g., only targeted therapies, hormonal therapies, monoclonal antibody therapies, immunotherapy); have surgery or radiation without concurrent chemotherapy planned within three months of consent; have a planned referral to the geriatric oncology clinic within one month of treatment initiation; or lack decisional capacity (unless a legally authorized representative is available to sign consent and participate in study visits and follow-up phone calls).

Table 1:

Screening Instrument. High-risk medications were compiled from the Beers Criteria and STOPP, based on clinical practice guidelines published by the Canadian Deprescribing Network [1,2].

2.6. Study Measures

For the cluster randomized trial, the primary outcome will be RDI of chemotherapy. The secondary outcomes will include both efficacy and implementation outcomes. Efficacy outcomes include number of PIMs successfully deprescribed, chemotherapy toxicity, functional status, hospitalizations, falls, and symptoms. Implementation outcomes are barriers and facilitators of implementation, measured qualitatively through direct observation of the intervention arm, interviews of participating oncologists, pharmacists, and patients (first 1–2 enrolled per cluster; 8–10 per arm), and audio transcripts of the first oncologist visit post-intervention. Covariates include demographics and cancer variables, comorbidities, patient self-reported attitudes toward deprescribing, and patient trust in the oncologist and medical team. Table 2 shows the measures collected for each patient participant.

Table 2.

Study measures.

Measures or Outcomes	Instrument(s)	Base-line	3 Month
Demographics	Age, sex, race/ethnicity, marital status, education, SES	X
Cancer variables	Cancer type, stage, prior and current treatment	X
Comorbidity	Diagnoses and severity (CIRS-G)	X
Patient attitudes	rPATD, WFPTS	X
RDI	Formula based upon standard therapy dosing		X
PIMs deprescribed	Number of meds discontinued		X
Chemo Toxicity	grade 2–5 by CTCAE, v5		X
Functional Status	ADL, IADL, SPPB	X	X
Hospitalizations	0 or ≥ 1		X
Falls	0 or ≥ 1	X	X
Symptoms	PRO-CTCAE	X	X
Barriers and facilitators	Interviews of patients and oncologists, direct observation	X	X

SES = socioeconomic status; CIRS-G = Cumulative Illness Rating Scale – Geriatrics; rPATD = Revised Patient Attitudes Toward Deprescribing questionnaire; WFPTS = Wake Forest Physician Trust Scale; RDI = Relative Dose Intensity; CTCAE = Common Terminology Criteria for Adverse Events; ADL = Activities of Daily Living; IADL = Instrumental ADL; SPPB = Short Physical Performance Battery; PRO-CTCAE = Patient-Reported Outcomes version of the CTCAE.

RDI will be based upon standard dosing, with dosing prospectively collected from the patient’s chart by study staff. The National Comprehensive Cancer Network guidelines [65] will be utilized to capture the standard dosing of chemotherapy regimens. The planned chemotherapy regimen (individual drugs, doses, and schedule) will be captured at the beginning of the study from the primary oncology team. The cumulative dosages per unit time of the individual drugs in the regimen will be calculated for the first three months of treatment: (total mg of drug in all cycles/m² body surface area)/(total days of therapy/7). The denominator is based on total days on treatment (from day one of cycle one through one cycle length after the date of the last treatment), reflecting all dose delays. The RDI is calculated as the ratio of the amount delivered to the amount intended based on standard guidelines. The actual dose delivered (in the numerator of RDI) will account for chemotherapy dose reductions. The RDI is calculated for each cytotoxic drug in a multidrug regimen, which are averaged to derive the average RDI. Dose reductions, dose delays, or discontinuation of the chemotherapy course will be captured, as well as the cause (i.e., relationship to chemotherapy toxicity).

Grade 2–5 chemotherapy toxicity will be prospectively assessed for the first three months by the research coordinator, based on Common Terminology Criteria for Adverse Events (CTCAE) v.5 [66]. The research coordinator will contact the primary treatment team after each chemotherapy treatment to capture type and grading of chemotherapy toxicity. The medical record will also be reviewed to capture each clinical encounter (scheduled or emergency visits). This will include a review of the clinic notes, emergency room visits, and hospitalizations. If the patient seeks emergency care outside of the primary institution, the patient’s permission will be obtained to review these outside records. Details regarding the overall category of toxicity (hematologic or non-hematologic) and specific type of toxicity will be captured.

Functional status will be assessed by the Activities of Daily Living (ADL) [67] and Instrumental Activities of Daily Living (IADL) [68], which are patient-reported tools, and Short Physical Performance Battery (SPPB) [69], an objective test assessing balance, gait speed, and repeated chair stands. Symptoms will be assessed by selected elements of the National Cancer Institute Patient Reported Outcomes CTCAE (PRO-CTCAE) [70]. Assessments of functional status and symptoms will occur at baseline and three months post-intervention.

2.7. Data Analysis and Monitoring

2.7.1. Sample Size Determination

The effect size of deprescribing interventions on outcomes in older patients with cancer is unknown, and this pilot trials aims to estimate effect sizes for a larger cluster-randomized trial. In the phase III GAP (URCC 13059) trial, which enrolled patients with advanced cancers, the percentage of patients with RDI ≥85% at three months was 47%. This percentage is considerably higher in many randomized controlled trials enrolling fit older patients with curable cancers. However, in “real-world” cohorts of older adults with curable cancers, the proportion of patients achieving RDI ≥85% is similar to the GAP study or even less [71–73]. A sample size of n = 32 evaluable patients per group will provide 80% power to detect a medium effect size (h = 0.6, representing an increase in proportion of patients attaining ≥85% RDI from 47% to 77%, or difference of 0.3) at a two-sided significance level of 0.10. Assuming 10% attrition, we will enroll 72 patients.

2.7.2. Planned Statistical Analysis

Underlying assumptions to all statistical analyses will be checked by graphical and numerical methods [74]. In the case of serious violation of assumptions, proper transformations or nonparametric methods will be used [75,76]. If outlying observations are encountered, data will be checked, and if no errors are found, we will conduct sensitivity analyses with and without the outlying data. Treatment groups will be assessed for baseline imbalance in patients’ demographics, disease, treatment, and functional measures. Since this is a pilot study, we will use significance level = 0.10 to avoid false negative results [77,78].

To assess the association of treatment arm with RDI ≥ 85%, we will employ logistic regression models where treatment arm is included as a fixed effect. We will then investigate the effect of cluster randomization by a generalized linear model with logit link, including arm as a fixed effect and the oncologist as a random effect. Estimation will be performed using pseudo-likelihood methodology, and inferences will be performed using the Kenward-Roger procedure [79]. We will assess the effect of clinically relevant covariates (e.g., cancer type and biological variables of age and sex) by adding them to the model as fixed effects. Covariates for which baseline imbalance between treatment arms was found will be added to the model as well; non-significant covariates will be subsequently removed. In addition, following an analogous approach, we will analyze RDI as a continuous measure using linear modeling methods of instead of logistic regression. These analyses will allow us to obtain estimates of the intervention effect as well as oncologists’ intra-class correlation coefficient (ICC), to inform a future phase III R01 study.

The same strategy as for the primary outcome will be used to evaluate the association with treatment arm and dichotomous outcome variables (chemotherapy toxicity, functional status [ADL, IADL], hospitalizations, and falls). Linear mixed effects models will be used for continuous variables, such as medications deprescribed, functional status (SPPB), and symptoms.

2.7.3. Planned Qualitative Data Analysis

Focus groups and qualitative interviews will be audio-recorded, transcribed, and imported into MAXQDA software for sorting, coding, and analysis. Focus group transcripts will be supplemented by field notes taken by a research coordinator observing and not moderating the discussion [80]. Inductive content analysis will use a systematic classification process of coding to extract themes [81]. Two independent coders will code data [82] until thematic saturation (i.e., the point at which no new data emerge) [83]. Coding will be compared, and discrepancies resolved through an iterative process of interpretation and constant comparison to raw data [83]. The PI (Ramsdale) will train coders and guide the process. Adaptations will occur iteratively throughout phase I and are anticipated to primarily affect the pharmacist-led arm (Fig. 2).

Audio-recorded clinic visits will be transcribed and coded similar to focus groups, using open inductive coding. Post-intervention interviews with patients, oncologists, and pharmacists may be audio-recorded, transcribed, and coded. If the interviews are very brief and contain no information beyond “yes/no” answers, they will not be transcribed. Within each arm, we will select “high-performing” and “low-performing” oncologists based on average number of medications deprescribed, and we will use data from direct observation, clinic transcripts, and post-intervention interviews to further elucidate the barriers and facilitators of implementation.

2.7.4. Data and Safety Monitoring

The PI or the participant’s physician will determine if an adverse event requires expedited review. The James P. Wilmot Cancer Center Data Safety Monitoring Committee (DSMC) will serve as the DSMC of record for this study. The study PI will conduct continuous review of data and participant safety. The PI will submit semi-annual progress reports of these data to the DSMC for review. The review will include for each treatment arm: the number of participants enrolled, withdrawals, significant toxicities as described in the protocol, and serious adverse events both expected and unexpected. The PI will maintain a database of all adverse events with toxicity grade and information regarding treatment required, complications, or sequelae. The PI will submit a copy of the adverse events spreadsheet along with the Progress Report to the DSMC for review.

Serious adverse events (SAE) that are related and expected or unrelated and unexpected will be reported to the DSMC for review at a quarterly meeting. SAE reports are expected to include sufficient detail so that the DSMC can determine the severity, toxicity grade, expectedness, treatment required, and a follow-up report documenting resolution of if there are sequelae. Unless otherwise specified in the protocol, serious adverse events that require detailed reports (but not necessarily expedited) are expected, related, non-hematologic toxicities of grade 3, 4, or 5.

The Safety Coordinator administratively coordinates reports and data collection and prepares documents for the DSMC Chair and committee review. The Safety Coordinator will administratively monitor adverse event rates utilizing the report from the study database. If any study has had two or more of the same SAE’s reported in a month or more than six of the same SAE’s in six months, the DSMC will review the summary of SAEs, discuss events with Study Chair, and conduct a more detailed review with the Study Chair. The Data Safety Monitoring Chair will determine if further action is required.

3. Discussion

Polypharmacy and PIMs are prevalent and consequential issues in older adults with cancer, associated with numerous adverse outcomes in these patients. For patients starting cancer treatment, the initiation of antineoplastic therapies and supportive care medications further increases the risk of medication interactions and adverse medication effects [32]. The diagnosis of cancer itself may change the appropriateness of medications: pharmacodynamics may be altered by the metabolic effects of cancer (such as weight loss and sarcopenia), and patients may no longer have the life expectancy to benefit from medications for chronic conditions.

Deprescribing has been studied in the context of specific medications (such as statins [84] or benzodiazepines [39]) and in general populations of older adults in the hospital and outpatient settings [85]. It is highlighted by the American Board of Internal Medicine Choosing Wisely^® initiative and supported by professional organizations in multiple countries. However, data for deprescribing in patients with cancer are limited, as it has mostly been studied in the context of end-of-life care [86]. Deprescribing may have the potential to mitigate adverse outcomes in older adults with cancer and PP initiating cancer treatment.

Systematic review and analysis of deprescribing interventions are also limited by heterogeneity in implementation of deprescribing. However, one recent meta-analysis suggested that comprehensive medication review by a pharmacist can reduce adverse outcomes such as PIM use and mortality [85]. Another meta-analysis showed a reduction in adverse outcomes for personalized (patient-specific) deprescribing approaches, versus no reduction for general education interventions like educating clinicians on deprescribing [36]. Pharmacists are already well-integrated into oncology clinical practice, based on their role in preparing and monitoring chemotherapy, and a pilot study in patients with cancer indicated that a pharmacist-led deprescribing approach is feasible [45]. Therefore, pharmacist-led deprescribing was selected as the main intervention for this protocol.

This protocol has several strengths. It was designed up-front to have multiple phases and to allow for iterative adaptation of an understudied intervention. Feedback from multiple stakeholders, including patients and their caregivers, is incorporated throughout the trial. The randomized clinical trial is a hybrid implementation-effectiveness trial, allowing for collection and analysis of both clinical and implementation outcomes, including patient-reported outcomes that are prioritized by older adults with cancer. It is also designed as a cluster-randomized trial, with the treating oncologist as the cluster, to permit calculation of an initial intra-class correlation coefficient to inform sample size estimation for a larger, national trial. Finally, the decision was made to compare the pharmacist-led study to a patient education brochure control group, rather than usual care; if the pharmacist-led intervention is not supported by the implementation and clinical outcomes, the brochure can be tested in a future intervention. The results of this study will impact care for a growing and highly vulnerable population who are typically excluded from clinical trials.