A measured dose of optimism for the evolution of ICU-based palliative care

Ideally, palliative care is involved early in the course of illness, and usually in the outpatient setting. With this model of care, the treatment provided is guided by shared decision making that includes the patient’s voice and a trusted clinician’s input. As the patient’s health status declines over months to years, a transition is made from treatments with curative intent toward therapies designed to optimize comfort. Throughout the entirety of the illness, the clinician also supports the patient and family, managing distress and optimizing quality of life.

However, many patients spend their final days in an intensive care unit (ICU) where palliative care is more difficult to provide. Because of the challenges that critical illness and the ICU environment bring, this setting also needs to be a focus of thoughtful palliative approaches. Critical illness is usually difficult to anticipate, acute, often multi-systemic, and may change course rapidly and unexpectedly. Critically ill patients frequently experience physical symptoms (pain, breathlessness) and psychological symptoms (anxiety, delirium) that are under-recognized and undertreated because life support machines limit patients’ their ability to report them. Although most of the millions of patients managed in ICUs worldwide each year want to survive, (1) nearly 20% will die, (2) most commonly after a decision to limit or withdraw life support. (3) These intense, emotional decisions are often made by surrogates who must rely on clinicians whom they are unlikely to have known longer than hours to days in the midst of a technology-focused environment that can feel chaotic, impersonal, and overwhelming. Unsurprisingly, much research in recent years has focused on ways to improve the delivery of palliative and end-of-life care in the ICU setting.

In this issue of JAMA, the report by Sprung et al (4) provides a unique evolutionary insight into the provision of end-of-life care to critically ill patients over a 15-year period. This observational study (ETHICUS-2) was conducted in 2015–2016 in 22 medical and surgical ICUs from 14 European countries that participated in a similar study conducted in 1999–2000 (ETHICUS-1). (5) In order to facilitate comparability between the two studies, the authors studied only ICUs that had participated in ETHICUS-1, used the same key definitions for end-of-life outcomes, and used similar methods for data collection and validation. To better understand changes in care delivery that may have influenced any observed changes over time, post-hoc, the authors asked senior ICU representatives to provide information about their ICU’s organizational structure, as well as the presence of 12 different ‘ethical practices.’

In the 1999–2000 study, the 22 participating centers admitted 22081 patients to the ICU, including 2807 (12.7%) who died or had limitations of life-sustaining therapies, whereas in the 2015–2016 study, these 22 centers admitted 13,625 patients to the ICU, including 1785 (13.1%) who died or had limitations of life-sustaining therapies. In the more recent cohort, decisions to withhold or withdraw treatment were more common (1601 (89.7% in 2015–2016) vs. 1918 (68.3% in 1999–2000), difference, 21.4% (95% CI, 19.2–23.6%), and were also more quickly implemented (median time from ICU admission until the first limitation of care was 2.1 days in the 2015–2016 cohort and 4.0 days in the 1999–2000 cohort), whereas the unsuccessful use of cardiopulmonary resuscitation (CPR) was less common in the more recent cohort (110 (6.2%) in 2015–2016 vs. 628 (22.4%) in 1999–2000, difference, −16.2% (95% CI, −18.1 – −283 (14.3%)). Similar to the findings reported in the earlier study, (5) the magnitude of these changes varied by region.

Consistent with other studies, (3) the majority (87.3%) of decedents had treatment limitations in place, which suggests that the need for complex medical decision-making at the end of life has become more widely accepted. In 2015, participating ICUs were more likely to have foundational elements of palliative and end-of-life care described by the authors as ‘ethical practices’ such as routine family meetings, daily deliberation about the appropriateness of care, regular end-of-life discussions, availability of palliative care specialist and ethics consultations, and training in communication and bioethics for ICU clinicians. While not all trends were statistically significant, increases in these potentially explanatory factors were consistent across ICUs in all regions. In broad strokes, end-of-life care practices appear to have changed for the better over the past 15 years.

Although this study is expansive and ambitious, several limitations should temper the interpretation of the results. First, it is possible that bias related to convenience sampling affected some of the key findings. Given that only 22 of the original 37 (59%) ICUs from 14 of 17 nations participated, it is unclear how similar the ICUs and patients included in 2015–2016 were to those included in 1999–2000. Also, a significantly smaller proportion of the patient population in the 2015–2016 study was derived from Southern Europe, a region found in the 1999–2000 study to have comparatively lower rates of withdrawing life-sustaining treatment and higher rates of unsuccessful CPR. This begs the question of whether ICUs that were less focused on improving end-of-life care practices were also less likely to have participated in the more recent study.

Second, each of the ‘ethical practices’ was self-reported by a representative from each ICU as either present or absent. Even though the number of ICUs that encouraged routine family meetings or had access to palliative care specialist consultation is known, it is unclear how those processes were implemented and how often they were used. Moreover, while the summary score of ethical practices was significantly associated with the placement of a treatment limitation, the observational study design makes it impossible to determine if any the 12 ethical practices were causally related to the observed outcomes. In fact, given the 15 years that elapsed between the two cohorts, several other unmeasured factors may have explained much of the observed change in end-of-life care practices, such as greater use of palliative care consultants, intensivists’ enhanced awareness of palliative care principles, and changing patient and family views on end-of-life care.

Despite the welcome general trends reported, the wide range in practice across ICUs reported also illustrates opportunities for further improvement. These findings are consistent with existing evidence describing the high variability present in ICU-based palliative and end-of-life care across countries, regions, and hospitals; (5) among physicians in the same ICU; (6) and within individual ICUs based on daily changes in environmental factors such as the census. (7) This variation is not surprising because collaboration between palliative care teams and ICU teams is challenging. (8) Palliative care specialists are a limited workforce (n=7,600 physicians in the US), are unevenly distributed, and poorly prepared to expand rapidly given continued loss of clinicians to burnout and retirement. (9, 10) ICU clinicians are greater in number, but may lack the skills and confidence to address many complex end-of-life issues. They may often fail to recognize when additional help is needed, as specialist palliative care is initiated infrequently and often so late that the presence of life support technologies acquired during the ICU course precludes hospice transfer.(11)

To reduce practice variation and improve the quality of patient-centered end-of-life care in ICUs, innovative approaches to collaborative care delivery are necessary to expand the functional capacity of the workforce. However, these approaches will require better methods to identify the critically ill patients and their family members who could benefit from specialist palliative care. Increasingly, such consults are triggered by specific diagnoses or arbitrary lengths of stay rather than what palliative care needs patients and their families have. (12) Identifying actual unmet needs would establish a patient-centered language beyond APACHE scores, PEEP levels, and norepinephrine doses that would be understandable by all and serve as a more direct metric for a field that has struggled to find meaningful, movable outcomes. Need-focused care models could provide durable infrastructure to support the consistency of ICU clinicians’ daily practice and help ICU teams to quickly understand which primary palliative care issues are most pressing for patients. Moreover, by identifying the presence, type, and severity of needs, hospitals could more effectively deploy available physician, nurse, social worker, or spiritual support staff, and more efficiently target constrained specialist resources to those with needs that are particularly serious or remain unmet after the primary team has attempted to address them.

Although there is relatively limited evidence for ICU-based palliative care interventions that improve outcomes that matter to patients and families, (13) a growing toolset of methods is becoming available through which novel care models could facilitate the delivery of primary palliative care. Three recent ICU clinician provider interventions have shown potential for improving communication, patient-centeredness care, and decreasing length of stay for patients who die in the ICU. (14–16) Nonetheless, other primary palliative care interventions designed to address end-of-life decision making are a reminder for just how difficult it can be to balance emotions, facts, and uncertainties in an ICU setting. (17)

Similar to its predecessor, the ETHICUS-2 study reported by Sprung et al represents the culmination of a tremendous effort to discern the contemporary pulse of palliative and end-of-life care in the ICU. Although the practices described in ETHICUS-2 don’t define a minimal standard for high quality ICU-based end-of-life and palliative care relevant to all clinicians worldwide, they can be helpful in showing where to start the journey to better it. While the results suggest that trajectories of end-of-life care in the ICU setting may be changing, the current use of life support is common and rising for patients with high risk-benefit ratios such as older adults, nursing home residents, and persons with dementia. (18) Thus, the true picture of care at the end of life may be worse for patients, families, and clinicians, or at least may be more complicated than a strict dichotomy between aggressive and comfort-focused therapies. While early outpatient palliative care before the onset of critical illness is aspirational, the difficulty of achieving this goal within current systems of care highlights the need to improve collaboration between palliative care specialists and ICU teams. The report by Sprung et al (4) provides some much needed optimism, as well as motivation and direction for the work that has yet to be done.