Rationale and design of a navigator‐driven remote optimization of guideline‐directed medical therapy in patients with heart failure with reduced ejection fraction

Abstract Although optimal pharmacological therapy for heart failure with reduced ejection fraction (HFrEF) is carefully scripted by treatment guidelines, many eligible patients are not treated with guideline‐directed medical therapy (GDMT) in clinical practice. We designed a strategy for remote optimization of GDMT on a population scale in patients with HFrEF leveraging nonphysician providers. An electronic health record‐based algorithm was used to identify a cohort of patients with a diagnosis of heart failure (HF) and ejection fraction (EF) ≤ 40% receiving longitudinal follow‐up at our center. Those with end‐stage HF requiring inotropic support, mechanical circulatory support, or transplantation and those enrolled in hospice or palliative care were excluded. Treating providers were approached for consent to adjust medical therapy according to a sequential, stepped titration algorithm modeled on the current American College of Cardiology (ACC)/American Heart Association (AHA) HF Guidelines within a collaborative care agreement. The program was approved by the institutional review board at Brigham and Women's Hospital with a waiver of written informed consent. All patients provided verbal consent to participate. A navigator then facilitated medication adjustments by telephone and conducted longitudinal surveillance of laboratories, blood pressure, and symptoms. Each titration step was reviewed by a pharmacist with supervision as needed from a nurse practitioner and HF cardiologist. Patients were discharged from the program to their primary cardiologist after achievement of an optimal or maximally tolerated regimen. A navigator‐led remote management strategy for optimization of GDMT may represent a scalable population‐level strategy for closing the gap between guidelines and clinical practice in patients with HFrEF.

support, or transplantation and those enrolled in hospice or palliative care were excluded. Treating providers were approached for consent to adjust medical therapy according to a sequential, stepped titration algorithm modeled on the current American College of Cardiology (ACC)/American Heart Association (AHA) HF Guidelines within a collaborative care agreement. The program was approved by the institutional review board at Brigham and Women's Hospital with a waiver of written informed consent. All patients provided verbal consent to participate. A navigator then facilitated medication adjustments by telephone and conducted longitudinal surveillance of laboratories, blood pressure, and symptoms. Each titration step was reviewed by a pharmacist with supervision as needed from a nurse practitioner and HF cardiologist. Patients were discharged from the program to their primary cardiologist after achievement of an optimal or maximally tolerated regimen. A navigator-led remote management strategy for optimization of GDMT may represent a scalable population-level strategy for closing the gap between guidelines and clinical practice in patients with HFrEF.

| INTRODUCTION
Although optimal pharmacological therapy for heart failure with reduced ejection fraction (HFrEF) is carefully scripted by treatment guidelines, many eligible patients are not treated with guidelinedirected medical therapy (GDMT) in clinical practice. 1 In data recently published from the CHAMP-HF (Change the Management of Patients with Heart Failure) registry of ambulatory heart failure patients in the United States with HF and reduced EF, roughly one-third of eligible patients were not receiving beta-blockers (β-blockers), one-fourth were not receiving angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor blockers (ARB) or angiotensin receptor-neprilysin inhibitors (ARNI), and two-thirds were not prescribed mineralocorticoid receptor antagonists (MRA). Amongst those receiving these therapies, the vast majority are dosed below guideline-recommended targets, with only 1% of patients eligible for all classes of medication receiving target doses of all three medication classes. Since appropriate application of GDMT is associated with considerable reductions in heart failure-associated morbidity and mortality, these data suggest a considerable opportunity for quality improvement. 2 Although prescription or dose titration of GDMT may in some cases be limited by blood pressure, heart rate, renal function, or serum potassium, medical contraindications are not always apparent, suggesting that other factors may be responsible for the implementation gap. Possible alternative explanations include lack of familiarity with guideline recommendations, infrequent clinic-based follow-up, uncertainty regarding the value of dose titration, limited opportunity to make dose adjustments in the clinic setting, concerns about tolerability, a focus on arbitrary numerical values for discrete endpoints, opportunity costs to patients and physicians, and difficulty in implementing adequate laboratory surveillance. 3 To overcome some of these barriers, we designed a strategy for remote optimization of GDMT on a population scale in patients with HFrEF leveraging nonphysician providers in a collaborative practice model. In this manuscript, we summarize the details of the design and implementation of this program, as well as preliminary enrollment data supporting the feasibility of this approach.

| PROGRAM DESIGN
As part of a broader effort at quality improvement in population health, we launched the Virtual Heart Failure Clinic (VHFC) at Brigham and Women's Hospital in 2017. The overarching goal of the program is to systematically identify patients with heart failure and reduced ejection fraction who are longitudinally managed by Brigham and Women's Hospital providers and facilitate remote optimization of GDMT through a telephone-based, navigator-led approach. Eligible patients were identified through a search of electronic health records (EHRs), and included women and men ≥18 years of age with a diagnosis of chronic heart failure and left ventricular ejection fraction ≤40%. All patients had to have an established relationship with a cardiology provider at our center, defined by at least two previous visits including one within the 18 months prior to enrollment. Patients with end-stage HF requiring inotropic support, mechanical circulatory support, transplantation, and those enrolled in hospice or palliative care were excluded.
Detailed inclusion and exclusion criteria are summarized in Table 1.
We developed a search strategy to identify suitable patients with heart failure from the EHR. The initial approach used billing codes to derive a set of coded inclusion and exclusion criteria to identify patients with likely heart failure and creation of a data mart of all patients who met these criteria since 1990. A clinical subject matter expert then reviewed the medical charts for 250 patients randomly selected from the data mart. This review created a gold standard which was used to train a statistical model to predict the presence or absence of HF at a positive predictive value threshold of 90%. We further refined this data-mart using natural language processing to identify patients who were most likely to meet the eligibility criteria. 4 Baseline characteristics of the patients recruited into the study are included in Table 2.
Patients identified through the EHR-based search were contacted via phone by a navigator who completed a medication reconciliation and verification of eligibility for participation in the remote optimization program. Treating providers were then approached for consent to adjust medical therapy according to a sequential, stepped titration algorithm modeled on the current ACC/AHA HF Guidelines. The program was approved by the institutional review board at Brigham and Women's Hospital with a waiver of written informed consent. All patients provided verbal consent to participate. Patients and providers who declined to participate in the remote optimization program served as a reference group. This workflow is detailed in Figure 1.

| Drug titration
For patients enrolled in the remote optimization program, medication titration was overseen by pharmacists practicing under a Collaborative Drug Therapy Management (CDTM) agreement. Protocols for the initiation, discontinuation, and titration of β-blockers, ACEI, angiotensin II receptor blockers, ARNI, aldosterone antagonists, sinus node inhibitors, hydralazine, and isosorbide dinitrate were developed by a team of pharmacists, nurses, general cardiologists, and cardiology heart failure specialists to approval through multidisciplinary review at the BWH Pharmacy and Therapeutics Committee. These protocols were heavily based on published guidelines and formed the basis of the CDTM agreement. When the sequence of introduction of therapy was not explicitly defined in guidelines, our team made these decisions based on the ACC expert consensus statement and clinical practice. 5,6 The CDTM agreement allowed pharmacists to initiate, discontinue, and titrate all medication classes outlined in Figure 2. We developed a software application to generate a HF medication change based on patient-specific information and to longitudinally monitor each partici- Medication initiation and titration orders were dictated by the algorithm (Figure 2). Titrations proceeded until patients reached the guideline-directed target doses, reported intolerable symptoms, or met criteria for no further adjustment, which was generally dependent on blood pressure, serum potassium levels, and renal function (Table S1). Specific rules governing sequencing and titration of each drug class are provided in Appendix 1.

| Follow-up
Patients were considered to have graduated from the VHFC once they

| OUTCOMES
The primary goal of the intervention was to enhance the proportion of patients receiving >50% of guideline directed doses of GDMT at 3 months following initial contact in the remote medication optimization group compared with the reference group of patients who declined to participate in the medication titration intervention. Key safety outcomes of interest included the proportion of emergency department visits, hospitalizations, and deaths during study follow-up in both groups.

| STATISTICAL CONSIDERATIONS
As this study was organized as a quality improvement intervention rather than a clinical trial, no formal power calculation was performed.
Based on anticipated recruitment, our sample targeted 1000 patients, baseline utilization of GDMT at >50% of target doses in 20% of patients, and projected enrollment of 25% of subjects in the remote medication optimization arm, we anticipate the study will provide >80% power to detect an absolute improvement of 10% in utilization of GDMT using this approach.

| DISCUSSION
Optimization of GDMT has been associated with reductions in cardiovascular and heart failure morbidity and mortality in numerous clinical trials, registries, and meta-analyses. 7-21 However, clinicians frequently fail to implement guideline directives in practice. 22 These gaps in care have been attributed to numerable factors, such as inertia, reluctance to increase medication burden, cost, lab monitoring, requirements for insurance pre-authorization, and lack of knowledge about rapidly evolving evidence. 23,24 There is a substantial opportunity for meaningful improvement in clinical outcomes amongst HF patients, however, the 2013 AHA/ACC guidelines for the management of heart failure encourage strategies to close the gap between current practice and guideline recommendations. 1 Even when clinicians apply HF medications as directed by guidelines, medications are frequently not dosed to guideline-recommended targets, and infrequent clinic-based contact means that the medical regimen evolves over a protracted time interval, with many months lapsing between medication titration. Given that the benefits associated with deployment of GDMT are often seen early, this may reflect a missed opportunity to improve patient outcomes. [25][26][27] Moreover, deployment of invasive strategies for HF including ICD and/or CRT is intended to follow on medical optimization, since this therapy may in many cases result in reverse remodeling that can lead to improvements in EF over time and obviate the need for device therapy. 28,29 Unfortunately, data suggests that most patients who receive ICD or CRT do not optimize GDMT prior to device implantation, reflecting another missed opportunity for these patients. 30 These gaps in care are associated with significant mortality for patients with HFrEF. 31 A number of approaches to enhance GDMT utilization and address gaps in implementation have been explored. Research initiatives aimed at understanding and addressing gaps in care (summarized in Table 3) have failed to consistently and reproducibly change behavior and impact outcomes. Educational strategies focused on patients and providers to emphasize the value of guideline-driven care are clearly important, but the ability of these initiatives to rapidly drive changes in clinical practice is unclear. 31 Although traditional multidisciplinary HF disease management programs do achieve higher utilization and less discontinuation of GDMT, such programs are not accessible to the vast majority of HF patients, and rates of optimal GDMT utilization in these clinics still falls well below guidelinerecommended targets. 6,35,45,46 However, research suggests that improving upon current rates of GDMT is possible and innovative approaches to improving optimal rates of adoption and goal dosing have shown promise. 36,47,48 Early experience suggests that integration of pharmacists in collaborative practice agreements may facilitate optimizing medical therapy in HFrEF patients, but systematic exploration of these efforts at scale has not been completed yet and have not incorporated the use of nonclinician navigators nor expanded to include full complement of GDMT for HFrEF. 33,49,50 Since algorithms for initiation, titration, and even discontinuation of medical therapy for HF are detailed in major society guidelines, there may be an opportunity to improve appropriate application of GDMT on a population scale by leveraging nonphysician providers to supplement the work of dedicated HF clinicians. Such collaborative practice models may enable more rapid evolution of the medical regimen outside the clinic setting, while muting practice variation with regard to drug titration and laboratory surveillance. As well, they may Patients who were on an evidence based β-blocker began titrations from their current dose, unless there was evidence or concern that the patient would not tolerate further titration. All nonevidence based β-blocker doses were converted to equivalent doses of metoprolol succinate, in the absence of contraindications. Patients who were currently taking sotalol at any dose level were considered to have met their maximally tolerated dose.
ACEI and ARB titration: Patients who were naïve to ACEI or ARB therapy were initiated on lisinopril at a starting dose of 2.5 mg QD, all other patients began titrations from their current dose of their existing ACEI or ARB, unless there is evidence or concern that the patient would not tolerate further titration. Patients who experienced an intolerable cough on an ACEI were converted to an equivalent dose of losartan. Lab surveillance was conducted 10 ± 3 days after each medication titration. An increase in serum creatinine greater than 30% from baseline or an increase in potassium greater than 15% from baseline prompted repeat lab surveillance in 7-10 days. In patients who could not tolerate the lowest dose of an ACEI or a dose reduction was indicated due to systolic blood pressures below 90 mmHg or symptomatic hypotension, the dose of β-blocker was preferentially reduced and the patient was re-evaluated in 1 week. If the patient was still unable to tolerate these doses, the ACEI or ARB was removed and the patient was re-evaluated in 1 week. If symptoms resolved and/or systolic blood pressure returned to ≥95 mmHg the β-blocker was resumed at the previously tolerated dose. If symptoms did not resolve or systolic blood pressure did not return to ≥95 mmHg adjustment of the β-blocker continued until no further adjustment was indicated per Table S1.
ARNI titration: Patients who were eligible for ARNI therapy per the current ACC/AHA HF Guidelines and who proved tolerability to an ACEI or ARB at daily doses of >10 mg or 160 mg, respectively, were transitioned to a middose (49/51 mg) ARNI. Patients who met the guideline criteria for consideration of ARNI therapy but who could not tolerate an ACEI or ARB at daily doses of >10 mg or 160 mg, respectively, or who had an eGFR of <30 were initiated on low dose (24/26 mg) ARNI therapy. Lab surveillance was conducted 10 ± 3 days after each medication titration. An increase in serum creatinine greater than 30% from baseline or an increase in potassium greater than 15% from baseline prompted repeat lab surveillance in 7-10 days.