Early Identification of Dying Trajectories in Emergency Department Patients: Potential Impact on Hospital Care

Introduction

Emergency departments (EDs) provide care for many patients with an advanced, life-limiting illness.^1–7 The ED clinical focus is on resuscitation and stabilization and suits the needs of the acutely ill and injured.^1,2,8,9 However, this approach may not be concordant to patient goals-of-care in the seriously ill with chronic, severe end-stage, life-limiting disease.^1,5,10,11

Initial ED management sets the trajectory for inpatient disposition/care, prompting palliative care (PC) leaders to recommend early patient-centered ED goal setting.^3,8 Challenges include a lack of buy-in from ED clinicians and a hectic ED environment with competing demands. However, a one-size-fits-all approach to ED care is not optimal in the terminally ill. A simple approach to prognostication may serve as a trigger for goals-of-care considerations (or palliative team consults).^1,8 To our knowledge, no study addresses early ED prognostication and impact on subsequent care provision and palliative outcomes.

Lunney and colleagues define functional trajectories based on disease diagnosis/progression/functional decline to serve as a prognostic guideline so patients/clinician can prepare for subsequent care and death.^12,13 In this study we retrospectively classify patients that presented to the ED and subsequently died following hospital admission into such dying trajectories. The purpose was to compare subsequent inpatient care provision and related outcomes between those trajectory groups.

Methods

A list of ED patients who presented for nontraumatic reasons and died after hospital admission from January 1 to December 31, 2002 was generated for three midwestern teaching hospitals, an urban hospital, an urban community hospital, and a suburban community hospital. We randomly selected fifty records of patients who died, from each hospital (n=150). However, ten charts were missing/incomplete, yielding a final sample of 140. Patients included ≥55 years old who survived ≥48 hours from ED presentation. Demographics, patterns of PC-related outcomes, and pain management in ED/hospital were reviewed. Two institutions have well-established PC-consult services. ED record/information readily available to ED physicians was reviewed for designating a dying trajectory.

We operationally defined 5 dying trajectories: 1) Disseminated Cancer: solid organ tumor with evidence of distant metastases; 2) Frailty: bedbound with incontinence and cognitive impairment; 3) Organ Failure: two or more hospitalizations within one year for end-stage liver-disease, end-stage renal-disease, congestive heart failure (CHF), or chronic obstructive pulmonary disease (COPD) on home oxygen; 4) Sudden Death: home-dwelling patients without a traumatic mechanism of injury and without previous stated comorbid illnesses; or 5) Other: those who did not fit into any of the above trajectories without a previous diagnosis of cancer, chronic organ failure or frailty, and without an unexpected death (for example, the diabetic patient who developed septic shock from multilobar pneumonia). This study was approved by the Wayne State University Institutional Review Board.

Results

As shown in Table 1, the patients were categorized into: [% (95%CI)] Disseminated Cancer 17.1% (10.9-23.4); Frailty 18.6% (12.1-25.0); Organ Failure 30.0% (22.4-37.6); Sudden Death 23.6% (16.5-30.6); and Other 10.7% (5.6-15.8). The overall age was 73.7 years and Organ Failure patients (68.7 years) were younger. The mean length of hospital stay (LOHS) was 11 days, with no statistical difference amongst groups, with a trend for patients with Frailty/Organ Failure to have longer LOHS compared to patients with Disseminated Cancer. This may reflect earlier consensus around goals-of-care for a well-defined entity like cancer.

The most common presenting ED chief complaint was shortness of breath/ respiratory distress (30/120, 25%), followed by neurologic events (for example, altered mental status, stroke, or seizure) (28/120, 23%). In Sudden Death, the most common ED complaints were constitutional/pain.

The overall prevalence of do-not-resuscitate (DNR) orders during hospital stay was 83%. However, most DNR orders were placed after multiple discussions in-hospital. Only 11% (15/140) had established DNR orders documented in ED. Surprisingly, Disseminated Cancer patients had a similar prevalence (4/24, 16%) of preexisting DNR.

DNR was placed an average of seven days after admission. DNR order to death time averaged five days across trajectories (three days in Sudden Death to six days in Disseminated Cancer).

Many (42%) deaths occurred in the intensive care units (ICU); most in the Sudden Death group (60%) as compared to the Disseminated Cancer group (33%). Death followed withdrawal-of-life-support interventions in 24% of dying patients; most (47%) in Sudden Death versus Organ Failure (15%). Overall, 14% of all patients died under hospice care, and utilization of hospice was similar across trajectories.

Pain was documented in half of all dying patients (49%, 68/140) with the highest burden of pain in the Disseminated Cancer group (65%). Overall, 72% of patients had opiates prescribed for pain control, uniform along trajectories. Declaration of depression documentation was 9%, highest in the Disseminated Cancer and Frailty groups.

Discussion

The ED is the gateway to inhospital care after a terminal crisis event for many patients with advanced life-limiting illness.^2,3,8,9,14 End-of-life ED care is increasingly being addressed and a recent article classifies ED deaths as “spectacular–a sudden, often traumatic, failed resuscitation death where ED care is prioritized, consuming the attention of ED staff” or “subtacular–a slow process of dying, often in the terminally ill, where care is a lower priority despite complex patient/family needs, likely due to a lower need for clinical interventions.” ¹⁴ Limited studies, however, focus on the inpatient care trajectory of ED patients who die later.^8,15.

Barriers to optimal ED end-of-life care include the lack of: advance directive documentation, access to records, primary care provider-to-patient/ED communication, and staff for family support.^2,4,16,17,18 The pressing demand to maximize ED throughput and a general consensus that ED clinicians may feel these conversations are best held in a nonchaotic setting with qualified personnel may also hinder early prognostication/goals-of-care discussions.^2,4,18 Admittedly, it is a challenge to predict which ED crisis presentation is a patient's terminal event. ¹³ Early prognostication is one step that may aid in the planning crucial for appropriate end-of-life care, and tools identified for ED assessment of end-of-life patients may assist this process.^8,13 We retrospectively assess whether a patient's dying trajectory is potentially recognizable on ED assessment and if it impacts subsequent care provision.

Although the majority of our dying patients were on known recognizable accelerated dying trajectories on ED presentation, only 11% had preexisting advance directives, similar to prior studies.^16,17 The rapid decision making required for the acute terminal crisis event leaves limited time for dialogue and adds to the challenge of having conversations with family with whom there is no prior established rapport.^8,18 This time may be optimal for PC clinician support to the ED and hence a major role of the PC in early, real-time, determinative decision making, but prompt availability will be essential. ⁸

The majority of ED patients who died in-hospital were elderly and classifiable into known accelerated dying trajectories based on information available in the ED. Most presented with advanced cancer or severely restricted activities of daily living or cognitive impairment. Respiratory distress/acute neurological event was the predominant ED complaint at end-of-life, as shown in studies that suggest immediate life-sustaining and palliative ED needs in the elderly^7,9 and those under hospice care. ¹⁹ It is not surprising that ED physicians in a resuscitative culture struggle to do the right thing and default to life prolonging interventions when faced with a patient in acute respiratory distress (at end-of-life) and unclear advance directives.^9,18

On charting the in-hospital course of dying patients, we find that it took an average of one week to place DNR orders with death following in about five days; same across trajectories, including cancer. This may reflect the efforts/time-limited trials needed in order to reach consensus, particularly if prior life prolonging measures were initiated. ²⁰ Hospice is underutilized and dying patients had pain declarations despite use of opioids across all trajectories, similar to prior studies.^15,19

The patterns of PC-related outcomes did not vary with trajectory in our study and this may be due to small sample size or certain amount of time necessary to reconcile/ reach consensus with preinstituted life prolonging interventions. Future prospective, larger studies may help delineate if early ED identification of dying trajectories allows for earlier consensus around goals or impacts subsequent inpatient PC provision and outcomes.

Limitations

This is a retrospective convenience-sample analysis, limited to easily available documentation of the distressing symptoms and assignment of trajectories. In this preliminary study we also do not study patients who may fit the trajectories but were discharged alive. We recognize that assigning a dying trajectory may be an oversimplification, and sometimes boundaries are blurred or assignment to a single trajectory may be difficult. In this study we used a hierarchical model, where Disseminated Cancer trumped Frailty, which trumped Organ Failure, which trumped Sudden Death.

Conclusions

Most ED patients that die in-hospital are on known disease/dying trajectories that can be recognized early and hence trigger prognostication which may assist ED physicians in identifying patients for early goals-of-care discussions or PC consultations. The majority of patients with a terminal event present to the ED with no advance-directives, and DNR order placement takes an average of one week, perhaps time spent to undo the effect of pre-instituted life-prolonging interventions. Further larger studies are needed to ascertain if early ED prognostication impacts future PC provision.