Systemizing the Evaluation of Acute Heart Failure in the Emergency Department: A Quality Improvement Initiative

INTRODUCTION

Introduction of the Hospital Readmission Reduction Program, imposing financial penalties for early readmissions after hospital discharge, has created a strong incentive for hospitals to reduce heart failure (HF) admissions, often by risk-stratifying patients prior to discharge for more intensive follow-up or disease management programs. A less explored option for reducing readmissions is risk-stratifying patients presenting to the emergency department (ED) to identify those who can be managed safely at home. Exploring this opportunity is important, given that the ED plays a pivotal role in HF care, with nearly 80% of patients evaluated being admitted¹. The current decision to admit these patients is based on clinical judgement, with large regional and hospital-level variability in readmission rates². At our institution, the rate of HF discharge from the ED across the 46 ED physicians in the 2016 calendar year was 20% and varied from 0 to 34%. To better identify patients warranting admission, we developed and implemented a standardized care path, CodeHF, for evaluating acute HF patients presenting to the ED. This report describes the creation and implementation of the CodeHF pathway, along with early insights into its promise and pitfalls.

GOALS

The primary goal of this initiative was to improve care for patients with HF exacerbations presenting to the ED through early identification, evidence-based risk stratification, education and timely follow-up.

LOCAL CHALLENGES IN IMPLEMENTATION

Despite leadership by an ED physician (AS), it was well recognized that there would be resistance to practice change due to poor coordination between medical specialties, lack of infrastructure for an evidence-based, standardized method of triage, minimal education for acute HF in the ED, and the perception that HF management had not been identified as an important opportunity for quality improvement. To address this, a multi-disciplinary team was formed that involved all relevant stakeholders (ED physicians, cardiologists, HF disease management nurses, administration and researchers). Serial biweekly, meetings were held to design the intervention and oversee its implementation.

DESIGN AND IMPLEMENTATION OF THE INITIATIVE

Development of the pathway

The CodeHF program was founded with a strong clinical logic. We considered that the primary indications for admitting patients with a HF exacerbation were to either improve survival or to improve their symptoms. To identify patients who might derive a survival advantage from hospital admission, we first sought to identify those at high risk for short-term mortality. Prior to launching this initiative, the Emergency Heart failure Mortality Risk Grade (EHMRG), which estimates a patient’s risk of 7-day mortality, had been published and subsequently validated^{3, 4}. In discussions with the entire multi-disciplinary team, a health system-specific consensus was obtained to ‘automatically’ admit patients with a 1% or greater risk of dying within the next week and to consider ED discharge for those with lower risks.

For patients with lower risk, the decision to admit was then based upon whether or not the patient improved with early diuretic therapy and whether or not an obvious cause for the exacerbation could be identified and addressed. Patients meeting these 3 conditions, low risk of dying, improved symptoms and clinically-logical explanation for the patients’ worsening, were felt safe to send home with close follow-up if patients and their family members were comfortable doing so after a detailed discussion regarding the risk/benefits of discharge. For those eligible for discharge, the CodeHF program provided both an educational module for the patient and their family, as well as a pathway for early follow up in cardiology clinic. The latter was supported both by a special referral number for the ED staff during regular business hours, and a structured phone call from our HF nurses within 24 hours of discharge. Thus, the CodeHF program, initiated at triage, aims at early identification of eligible patients and provides a clear care path within the ED to help physicians decide the optimal patient disposition. (Figure 1)

Figure 1: Structure of the CodeHF pathway and its early impact outcomes.

HF – heart failure; ED – emergency department; CHF – congestive heart failure; IV - intravenous

Implementation of the pathway

After developing a prototype of the CodeHF program, serial iterations were made through discussions with the ED staff to align the program with their culture and preferences. Examples of the adopted iterations are provided in Supplementary Table 1. The collective effort of these changes both improved efficiency and clinical acceptability. Once finalized, our Information Technology team implemented the standardized order-set (Supplementary Figure 1) and the EHMRG risk calculator (Supplementary Figure 2) within the electronic medical record (EMR).

The CodeHF Program was launched in January 2018 at the 4 Saint Luke’s Health System EDs and all consecutive patients, over the next 13 months, formed the program’s evaluation cohort. Eligible patients were those with a documented history of HF and a primary complaint of shortness of breath, excluding those with new-onset of HF and on hemodialysis. Institutional Review Board approval and informed consent for this project were waived as this was a quality improvement initiative. Funding was obtained from the health system.

Program Evaluation

To assess the program, we separately evaluated the completeness of its implementation, its association with processes of care, and its impact on outcomes. The last two evaluations were made by comparing those managed with and without the CodeHF pathway. The completeness of implementation was measured by capturing all patients with a history of HF who were discharged from the ED (either home or admitted) with a primary diagnosis of HF and examining the proportion of patients for whom the CodeHF order set was used. Variability in use was also described by both ED physician and site. The process of care, aimed at estimating efficiency, compared the door-in to door-out time from the ED. Similarly, the success of the new approach in offering cardiology follow-up for those being discharged home from the ED, compared the proportion of patients seen as an outpatient within 7 days.

The primary outcome was the proportion of patients discharged home from the ED who were and were not treated by the CodeHF pathway. To assess the safety of the pathway on those discharged from the ED, we compared the proportion of patients who were readmitted within 30 days after following the pathway, as compared with those discharged using usual care. To assess efficiency, we compared the number of ‘unnecessary admissions’ to the hospital, defined as a hospital stay of <48 hours. This threshold is also used to separate an observation stay from an inpatient admission and for which the hospital is paid significantly less by Medicare.

Analyses were confined to those with a primary encounter diagnosis of HF in the ED, identified using ICD 9 & 10 codes (428.XX & I50.XX). Comparisons were made using Pearson Chi-square and one-way ANOVA tests for categorical and continuous variables, respectively. Additionally, to account for patient and physician-level differences, hierarchical logistic regression analysis, with ED physician entered as a random effect, was performed with additional adjustments for patient age, race, insurance status, and diastolic blood pressure. All analyses were done using SAS version 9.4 (SAS Institute Inc).

RESULTS

A total of 2344 patients with a history of HF presented to the ED with shortness of breath, of which 1203 (51.3%) were excluded for having a primary encounter diagnosis other than acute HF and 41 (1.7%) were excluded for missing data. Among a final cohort of 1100, eligible patients (mean [SD] age, 72.3 [15.2] years; 74% Caucasian; 48.7% Male), 149 (13.5%) patients were managed with the CodeHF pathway (Supplementary Table 2). CodeHF was used by 28 (61%) ED physicians with wide physician-level (2–100%) variability and never used by 16 (36%) [Median odds ratio 3.3 (95% CI 2.21, 4.45)] physicians (Supplementary Figure 3). There was also substantial site-level variability (3.9–33.2%) across the four EDs. Among the patients treated by the CodeHF pathway, half (n= 74) had a high (≥1%) 7-day risk of mortality and 75 (50.3%) a low risk. When the CodeHF pathway was used, time in the ED was 54 minutes shorter (294 vs 240 minutes, p 0.016) than when it was not used. Among those who received cardiology follow up after ED discharge, CodeHF resulted in a 93% greater proportion of timely (within 7 days) follow up compared to usual care (50% vs 25.9%, p 0.008).

Improved outcomes were also noted in patients on the CodeHF pathway, with a 49% greater proportion of ED discharges (37.9% vs 25.2%; p 0.001). Supporting the safety of ED discharge using the CodeHF pathway, among those discharged, there was a 24% lower trend for re-admissions within 30 days (16.1%. vs 21.3%; p 0.33). Fewer patients who were admitted on the CodeHF pathway had a hospital stay < 48 hours (9.0% vs. 11.4%; p 0.79; Figure 1). Further, the CodeHF program resulted in a 19% relative reduction in hospital admissions (59.7% vs 73.7%; p 0.001) and among those admitted there was no difference in inpatient length of stay (4.02 vs 4.04 days), in-hospital (0.7% vs 1.6%; p 0.71) or 30-day mortality (4.7% vs 4.7%; p 0.98). The significant improvement in ED discharge rates persisted within the adjusted analysis (OR 1.63, 95% CI 1.04, 2.57; p = 0.034; supplementary figure 4).

TRANSLATION TO OTHER SETTINGS

Although this study cannot define the ultimate success or generalizability of CodeHF, we believe that it provides a novel model that other systems can consider in developing better ED management of decompensated HF. The key components required to create a similar pathway at other institutions include a method for evidence-based risk stratification, infrastructure for ED-based HF education, and early outpatient follow up. The EHMRG is a well-validated risk model that uses readily available patient and laboratory variables and was developed using a cohort of acute HF patients presenting to the ED.^3,4 For optimal implementation, a consensus regarding the threshold to define high-risk patients warranting hospital admission must be established and the pathway should be customized to local culture. Embedding the risk calculator, HF-based education and a path to procure early outpatient follow up, guided by a multi-stakeholder team, is a solid foundation for systems considering our CodeHF approach.

SUMMARY, FUTURE DIRECTIONS AND CHALLENGES

In this ED-based quality improvement feasibility study, we successfully created and implemented, with modest success, an efficient, evidence-based risk stratification pathway for acute HF patients. Our preliminary results suggest that more efficient (reduced time in the ED and improved timely outpatient follow up) and efficacious (increased safe ED discharge rates and fewer short hospitalizations) care can be delivered by early identification and risk stratification of patients with acute HF presenting to the ED with shortness of breath. The reduced ED stay can be attributed to the early identification of eligible patients, as demanded by the pathway, compounded by the availability of an evidence-based risk score which facilitated timely ED disposition. The reduction in unnecessary admissions has the potential to improve patient’s quality of life and make beds available for those warranting hospital admission.

Despite multiple interventions, early HF readmission continues to be a national healthcare problem⁵ necessitating a safe method of evidence-based triage to reduce readmissions. We demonstrate promising results of a unique program, the CodeHF pathway, to standardize care in the ED. While the risk stratification score does not mandate each patient’s disposition, it serves as an evidence-based guide to ED physicians and provides an opportunity to standardize care for all patients.

Unfortunately, the CodeHF pathway was activated in only 13.5% of cases with large physician-level variability. With the results of these preliminary data, we continue to iteratively work on improving its use and are now implementing it as a standing order (similar to Code STEMI). Triage nurse education has begun and after administrative approval, the standing order will be introduced in March 2020. Whether this improves the consistency of its use and whether the preliminary benefits are sustained or improved will require future re-evaluation.

In conclusion, to develop a more efficient and standardized means of managing patients with decompensated HF, we developed and implemented the CodeHF pathway and have provided preliminary data demonstrating improved processes of care and outcomes. Other hospitals may choose to build upon these experiences in identifying novel strategies to improve the care and outcomes of patients with acute heart failure.