An Exploratory Study of Goals of Care Conversations Initiated with Seriously Ill Veterans in the Emergency Room

Introduction

Emergency department (ED) visits are common for older patients with chronic, life-limiting illnesses and may be an indicator of an accelerated rate of decline. ¹ Ouchi et al. suggest that an ED visit can be an “inflection point” in the illness trajectory and may offer a valuable opportunity for clinicians to work with patients to consider or re-consider their goals and values to inform present and future health care decision making. ² If a patient already has a documented goals of care conversation (GoCC), an ED visit offers an opportunity to review and update that discussion in light of changing medical circumstances. If a patient has not had such a discussion, it could be initiated in the ED to help ensure the ensuing episode of care is concordant with the patient's values, goals, and life-sustaining treatment (LST) preferences. The goals and preferences identified during this inflection need to be accurately captured in the electronic health record (EHR) so that any subsequent care is aligned with the patient's documented goals.

GoCC with high-risk patients (i.e., those who are considered to be at high risk for a life-threatening clinical event because they have a serious life-limiting medical condition associated with a significantly shortened lifespan) have been associated with enhanced quality of life, less aggressive medical care, fewer hospitalizations, and earlier referral to palliative care or hospice.^3–8 Despite the possible benefits of initiating GoCC with patients with advanced illness in the ED, and endorsement of their value by the 2016 Society for Emergency Medicine Consensus Work Group, ⁹ there is a paucity of studies about GoCC in the ED and the extent to which GoCC are occurring in the ED is unknown. ² The Department of Veterans Affairs (VA), Veterans Health Administration (VHA), provides a unique opportunity to study the practice of initiating GoCC in the ED. The VA is the largest integrated health care network in the United States to document the goals and LST decisions of high-risk patients with a standardized set of LST orders and a unique LST progress note template. ¹⁰

The creation of the LST template and LST orders was part of the VA's Life-Sustaining Treatment Decisions Initiative (LSTDI), a national culture and practice change initiative that aims to ensure that all Veterans at high risk for a life-threatening clinical event have their goals, values, and LST decisions proactively elicited, documented, and honored across the continuum of care. ¹⁰

Although one of the objectives of LSTDI is to encourage proactive GoCC earlier in the course of serious illness by a trusted primary care clinician, while the patient has decision-making capacity, but before a health crisis such as an ED visit or hospitalization, the reality is that GoCC will be initiated at opportune moments in multiple settings across the health care system. For instance, of the over 347,141 first-time GoCC conducted in VHA since 2017, 52% were initiated while the patient was acutely hospitalized, nearly 42% were conducted in an outpatient setting, and the remaining 6% were conducted in long-term care. Improving our understanding of GoCC that are conducted in ED settings will provide an evidence base to inform the development of context-specific organizational and communication strategies to promote high-quality GoCC with seriously ill older adults regardless of where they occur.

The purpose of this study was to assess whether GoCC are occurring with patients in VA EDs, to describe characteristics of patients whose first-time GoCC occurred in the ED, to characterize these patients' goals of care and LST decisions, and to identify the extent to which palliative or hospice consultations occur following the ED visit.

Materials and Methods

We conducted an exploratory analysis of GoCC in the ED using VHA's Corporate Data Warehouse (CDW)—an administrative health data repository that contains EHR for all VHA clinics. ¹¹ Within VA, all GoCC are documented on a standardized, national, LST template, which consists of nine distinct elements of which four are mandatory (decision-making capacity, patient goals of care, cardiopulmonary resuscitation [CPR] status, and consent for the LST plan). ¹⁰ We identified 145,425 patients who had at least one documented GoCC between July 1, 2018 (the date that VHA facilities were required to implement VA Directive 1004.03 Life-Sustaining Treatment Decisions), and July 1, 2019. From this population, 10,780 (7%) experienced their first GoCC in the ED. These patients were linked to the demographics and utilization domains of the CDW to conduct the exploratory data analysis described below. ¹¹ We extracted biological sex, age, race and ethnicity, marital status, and one-year risk of hospitalization or mortality from the care assessment need (CAN) score estimated closest to the first GoCC in ED.^12,13 The CAN score was developed to identify patients at risk of admission or mortality within one year and is used extensively by primary care teams to help identify high-risk patients. LSTDI uses CAN scores to identify patients who would benefit from GoCC. CAN scores are estimated weekly on all VA patients with a designated primary care provider and ranges from 0 (lowest risk) to 99 (highest risk). From the utilization domains of the CDW, we identified patients with a diagnosis of six common comorbid conditions. ¹⁴ These conditions included cancer, cardiovascular disease, dementia, end-stage lung and renal diseases, and frailty. Patients were considered to have the disease if they had a single diagnosis for the disease captured within the VA's CDW within one year of their first GoCC in ED. Finally, we identified any consultation in palliative or hospice care using a combination of VA clinic stop codes (i.e., location where consultation occurred) and Evaluation and Management Current Procedural Terminology codes.

We conducted exploratory analyses and described the characteristics of the 10,780 patients receiving initial GoCC in the ED. All descriptive results were generated by the tableone package ¹⁵ of the R language and environment for statistical computing. ¹⁶ The tableone package is an R package that facilitates the construction of tables of patient characteristics and can summarize both continuous and categorical variables.

As a quality improvement analysis for VA's Life Sustaining Decisions Initiative, a national quality improvement program, this work was not subject to human subject research regulations; thus, no Institutional Review Board approval was obtained. ¹⁷

Results

During our study period, there were 1,558,060 patients with at least one ED visit. Of these patients, 386,023 had a CAN score of 90 and over, indicating higher risk of hospitalization or death within a year, and potential to benefit from a GoCC. Of these higher risk patients, 77,743 had a documented GoCC, and of those with a documented GoCC, 10,780 received their first GoCC in the ED. Table 1 describes the demographics of our cohort. Nearly three-quarters (7987; 74%) were older than 65 years and most were male (10,186; 95%); nearly three-quarters (7855; 73%) were white, approximately half (4943; 46%) were married, and 6595 (60%) had CAN scores over 90. Of the six comorbid conditions we analyzed, 9089 (84%) of the patients in the cohort had at least one and 5657 (53%) had multiple conditions. Frailty (N = 5895; 55%) and cardiovascular disease (N = 5532; 51%) were the most prevalent. The majority of patients had decision-making capacity (N = 9473; 87.9%).

Table 2 describes goals of care within categories of the CAN score. Of the 10,780 patients with a first GoCC in the ED, 5772 (54%) had a CAN score over 90 and 823 (7.6%) patients were missing a CAN score. Overall, 1215 (11%) lacked decision-making capacity with 271 (7%) of these patients in the lowest CAN category and 387 (17%) in the highest risk CAN category. The most commonly indicated goals of care were to maintain function and independence, to prolong life, and comfort. The percentages of patients with goals to prolong life or to be cured were lower in the highest risk CAN score category.

Table 3 shows results related to LST decisions for CPR and other LSTs. The percentage of patients who desired CPR was lower among the highest risk CAN score patients (64% vs. 51%) (Table 3). Approximately half of the patients in the cohort had recorded information on the LST template for other LSTs (Table 4). Three-quarters of these patients indicated they wanted full scope of treatment in circumstances other than cardiopulmonary arrest. Approximately 15% indicated they wanted limits to specific LSTs in circumstances other than cardiopulmonary arrest. The percentage of patients wanting full scope of treatment was higher among the lowest risk CAN score patients; and the percentage of patients requesting limits to LSTs was highest among higher risk CAN score patients. Table 4 shows the types of limits requested by the 1098 patients in the cohort, who indicated they wanted limits to LSTs. Approximately two-thirds of these patients requested limits to invasive mechanical ventilation (Table 4).

Table 5 shows the time to a palliative care or hospice consult following the initial GoCC in the ED. A total of 1910 (18%) of patients had either a palliative care or hospice consult after their initial GoCC in the ED. Of those patients, 1232 (65%) had either a palliative care or hospice consult within 30 days following their initial GoCC in the ED, and only 46 (0.04%) patients had palliative care or a hospice consult before their initial GoCC in the ED.

Additional findings related to the characteristics of the 10,780 patients in the cohort included, 1465 (14%) died within three months of their initial GoCC in the ED. There were 1747 (16%) patients who had a follow-up GoCC within three months of their initial GoCC in the ED, with 1265 (72%) in an inpatient setting, 268 (16%) in an outpatient setting, and 214 (12%) in the ED. Over half (N = 6316; 58.6%) of the initial GoCC in the ED were conducted by residents, one-third by physicians (3821; 35.4%), and <4% by nonphysician independent practitioners (e.g., nurse practitioners).

Discussion

To our knowledge, this is the first exploratory analysis of its kind to describe patient demographics and health-related characteristics for patients receiving their first GoCC in the ED. Four findings are notable and will be discussed below. First, this study verified that GoCC are occurring in the context of emergency care and that clinicians (primarily resident physicians) are identifying older patients with one or more comorbidities for these serious illness conversations. Importantly, patients who had their first GoCC in the ED were not all at high risk of a life-threatening event based on their CAN score—suggesting that ED clinicians are initiating these serious illness conversations with patients at different stages of their illness trajectory. As Ouchi posits, clinicians may be treating the ED visit as an “inflection point” in the illness trajectory and an opportunity to help the patient consider their care preferences in light of current or future decline. Furthermore, the care intensity in the ED is likely to vary between patients—some are in the ED for acute management of a chronic illness and others are critically ill, and may need communication restricted to the current crisis. How and what clinicians and patients decide about what decisions need to be made are likely to vary based on the patient's acuity of illness and the urgency of decision making. ²

Second, 16% of patients with a first GoCC in the ED had a follow-up GoCC within 90 days. A majority of the follow-up conversations occurred during a subsequent hospitalization and 12% occurred during another ED visit. This finding is consistent with the view that GoCC are not one-time events, but a communication and shared decision-making process that should be revisited as the patient's disease progresses and opportunities to engage with the patient present themselves. However, a higher follow-up rate may be desirable given that these conversations and LST decisions are occurring in the context of a potential health crisis. Patients may wish to revisit their decisions once the crisis has passed. Optimally, follow-up would occur in primary care or an outpatient specialty clinic. Our data, however, show that strikingly few follow-up conversations are occurring in an outpatient setting, but rather in the context of another episode of acute care.

Third, patients with higher CAN scores more often stated a preference for “comfort” and “maintaining function” as goals of care, with lower proportions preferring to “prolong life.” In addition, these patients consist of lower proportions of patients who prefer CPR or full scope of treatment. This is consistent with prior literature demonstrating a relationship between increasing age and declining intensity of LST interventions.^18,19 These findings suggest that GoCC in the ED may be an important mechanism for identifying patients who would benefit from a palliative care consultation and for improving decision quality, that is, aligning treatment plans with the patient's articulated values and preferences for care.

Finally, 18% of patients who had an initial GoCC in the ED had a palliative care or hospice consult within 90 days of their ED visit, with a majority of these consults happening within 30 days of the first GoCC in ED. A goal of LSTDI is timely palliative and/or hospice care for patients with advanced illness and these findings suggest that ED-initiated GoCC may improve access to palliative and hospice resources and promote care that is aligned with the patient's goals, values, and preferences.

Although our findings are promising, in that first time GoCC are occurring within the ED, only a small fraction of high-risk patients who seek emergency care within VA are proactively offered GoCC. At least some of these encounters may represent a missed opportunity on the basis of which future studies can explore. For instance, what predicts whether a seriously ill patient (or surrogate decision maker) is offered a GoCC in the ED and what are provider and patient facing facilitators and barriers to implementing GoCC in the ED? Given that 6 out of 10 GoCC in this study were initiated by resident physicians, how do facilities ensure that residents are trained, competent, and comfortable in identifying appropriate patients and in conducting high-quality GoCC—especially given rapid turnover in trainees. Furthermore, how do contextual features specific to emergency care (e.g., increased patient volume and limited privacy) shape whether GoCC are initiated and how they are conducted? How can palliative care expertise be utilized to support GoCC in the ED? Research directed at these questions will help support context-specific interventions to promote proactive GoCC in the ED.

This study has important limitations. The CAN score is model based, and patients could be misclassified in terms of level of risk of hospitalization and death. In addition, a CAN score is only available for patients actively receiving care from a VA primary care provider, hence certain high-risk patients who sought care in the ED and had a GoCC may have been missed. Despite these limitations, there appear to be patterns across CAN scores that we would expect with increasing severity of disease and have been demonstrated in previous studies.^18,20 Furthermore, the frequency of follow-up palliative care consults may have been underestimated, given that not all palliative consults are documented as such. Finally, due to the unique characteristics of VA enrollees, the findings may not be generalizable beyond VA.

Conclusion

ED visits are common for seriously ill older adults and may provide a valued opportunity for patients and health care professionals to discuss patients' values, goals, and preferences within the context of a spectrum of declining health. In this study, we verified that GoCC are being initiated in the ED with Veterans at differing stages in their illness trajectory and that higher proportions of higher risk patients preferred to limit LST. However, only a fraction of seriously ill Veterans were proactively offered conversations and further research is needed to understand how to promote high-quality GoCC within an integrated health care system.