The Progression of End-of-Life Wishes and Concordance with End-of-Life Care

Introduction

Since 2013, Kaiser Permanente Northern California (KPNC) has engaged in a systematic effort to elicit, document, and honor the care preferences of patients as they near the end of life. For patients of advanced age or with a serious illness, this is achieved through clinician-facilitated discussions about patients' values, beliefs, and goals in the context of their illnesses. These preferences are translated into actionable medical orders using the Physician Orders for Life-Sustaining Treatment (POLST) form, which is designed to remain with the patient wherever he or she resides and is respected across all healthcare settings statewide. As per the National POLST Paradigm organization, the POLST form is becoming increasingly standard across the country, with 50 states actively developing or having fully implemented a program. ¹

We reviewed the wishes of patients who participated in a POLST Paradigm advance care planning program and whether those wishes were respected at the end of life. We hypothesized that patients' recorded wishes on their POLST would be their ultimate care preferences if the conversation occurred within about 12 months of their death and that patients would receive care concordant with their documented care preferences ∼95% to 98% of the time. This estimate was based on two studies examining the concordance rate of the Respecting Choices' advance care planning program of La Crosse, Wisconsin.^2,3

Methods

We focused the study on patients who had participated in KPNC's Life Care Planning (LCP) Advanced Steps (AS) program, which is offered to patients if their physician estimates that they are in the last year of life. The referral-based program—which is a POLST Paradigm Program modeled on Respecting Choices' Last Steps program—helps patients discuss their care preferences in the presence of their designated decision maker (DDM) during a detailed conversation with an AS facilitator.

We randomly selected 300 of the 3701 patients who had participated in the AS program and died in 2015, creating a geographically representative sample. Some of these patients had an AS discussion only one day before their death, whereas others did so more than two years before death. The average time between an AS conversation and death among patients in the sample was ∼10 months.

Three LCP experts at KPNC conducted an in-depth chart review of the patients' electronic health records (EHR) using a standardized questionnaire addressing patients' wishes, any changes of those wishes before death, the role of DDMs, and the care that patients received in the last care setting. The experts were not involved in the care of or LCP conversations with the patients in the study. Concordance was defined as documentation that care received in setting before death was either consistent with documented wishes on patient's POLST form or inconsistent with POLST selections, but consistent with the DDM's or patient's verbal guidance.

One notable strength of this study was the breadth of information the researchers could access to determine care preferences and final care received. Kaiser Permanente is an integrated health system with an EHR that reflects care in a variety of care settings. The study scope was limited to only those patients who participated in the AS program, which restricts the study findings to patients who had a conversation with a specialized facilitator.

Results

The chart review revealed a high degree of concordance between care preferences and care received. Of the 293 cases where we could assess care preferences and care received, we found that 290 cases—or ∼99%—were concordant. (For seven patients, there was not sufficient information to determine concordance.) (Table 1).

Although all 300 patients in the study had an AS conversation before death, we found that some had not filed POLST forms. Among patients who filed a POLST, many revised their preferences at some point between their AS conversation and their terminal care setting. Specifically, 19% of patients (48 out of 253 patients who filed a POLST during their AS conversation) revised their care preferences in some way. Care preferences were more likely to change if the AS conversation was not recent (>12 months ago) or if the patient returned to the hospital for care. Those revisions generally were not made using advance healthcare documents; they most often happened during a conversation between the care team and the patient or DMM. In 85% of the cases, where care preferences changed, the patient or DDM opted for less intensive care. In most such cases, a DDM or family member changed the preference, generally when the patient experienced a medical emergency or a significant deterioration of condition in which they could no longer speak for themselves (Table 2).

We often found documentation of real-time goals of care discussions between the treating physician and the patient or DDM that took place during the final care encounter. This practice demonstrated a consistent effort from physicians to ascertain and honor patient's current care preferences. In three of the 293 cases, where we could assess concordance, a physician or a clinician made a care decision at odds with the patient's or DDM's care preferences. One case occurred in a hospital emergency department, one in a community dialysis clinic, and one in a community skilled nursing facility.

Discussion

We found a high level of respect for patients' wishes across care settings; the three exceptions we found were not the norm. In each of these discordant cases, the DDM was not immediately available, a circumstance that leaves clinicians to make their own determination about how best to proceed.

Although our study was designed to determine care concordance, it highlighted the progression of patients' wishes over time—even for patients who have detailed, facilitated conversations about their choices—and the myriad formats for expressing and re-expressing those wishes. Most of the changes occurred verbally. Although the POLST serves as an important document to communicate patient preferences when they are unable to do so themselves, providers should not rely on the document to replace ongoing goals-of-care conversations with patients or their DDMs, especially in the hospital setting. Instead, the POLST ought to be easily accessible to hospital physicians and staff so it may serve as a springboard for the real-time goals of care conversation that remains essential.

Conclusion

A skillful, facilitated advance care planning conversation is a worthwhile approach for eliciting patients' wishes, and the POLST form is a valuable tool. When patients' preferences about medical care are made known in this way, they overwhelmingly receive concordant care. However, our findings highlight the need for comprehensive, continuous conversations across all care settings, even after a POLST is completed. As the U.S. healthcare system continues to improve end-of-life care, more research on the documentation and progression of care preferences is needed to fully understand how healthcare providers can best identify and act on patients' wishes, especially when a DDM is unavailable.