The utilization of intensive care has changed over the last two decades. Patients with AIDS and acute respiratory failure (ARF) requiring mechanical ventilation or septic shock, who had mortality rates greater than 75% in the 1990s, now have far lower mortality rates than patients with other conditions such as New York Heart Association (NYHA) III heart failure, cirrhosis of the liver, or cancer [1]. Similarly, major survival gains have been achieved in patients with hematological malignancies and ARF. Advances in treatments for hematological malignancies [24], a systematic clinical approach [5], and noninvasive management strategies [6] have decreased the mortality rate in these patients from nearly 100% [7, 8] to less than 50% [9].

The radical transformation undergone by intensive care over the last half a century has increased survival rates among patients with the most severe conditions (cancer, transplantation, advanced AIDS, multiple injuries, burns, or major surgery) to the levels achieved in poliomyelitis patients with ARF in the 1950s [10]. Over the last decade, intensivists have been increasingly asked to admit patients experiencing severe but reversible acute events during the last few months of their lives. These patients are often very elderly and have major comorbidities and/or chronic disabilities (with musculoskeletal, respiratory, cardiac, and neurological impairments). The decision to admit a patient to the intensive care unit (ICU) depends on the wishes of the patient—when they are known—and on the medical data supplied by the professionals caring for the patient [11]. Not infrequently, decisions to limit active treatments (e.g., mechanical ventilation, catecholamines, cardiac massage, and dialysis) are taken immediately upon ICU admission to protect patients from useless interventions that contradict their preferences and values and would deprive them from communicating with their families during the very last moments of their lives [12]. Although intensivists are steeped in a culture of survival, they are now striving to prevent unreasonable therapeutic interventions by introducing palliative care into the ICU, in collaboration with health-care teams that regularly provide palliative care (e.g., in internal medicine, geriatrics, neurology, oncology, hematology, pulmonology, and cardiology) [13]. More recently, palliative care has become the standard of care in patients with inaugural malignancies, limiting side effects of undue chemotherapy, and increasing quality of life as well as duration of life [14].

Outside the palliative care setting, NIV has considerably improved the survival of patients with hypercapnic exacerbations of chronic obstructive pulmonary disease (COPD) [15]. This effect is ascribed to decreased use of endotracheal intubation [16, 17], which results in lower rates of ventilator-associated infection [18, 19]. NIV has also been used successfully in many other situations [2022] including cardiogenic acute pulmonary edema [23], postoperative ARF [24], particularly in solid organ transplant recipients [25], and ARF in patients with immune deficiencies [26, 27]. However, the effects of NIV or continuous positive airway pressure (CPAP) in hypoxemic ARF remain somewhat controversial [28, 29].

Several studies provide evidence that denial of ICU admission is high, especially among older patients and those with chronic respiratory disorders and cancer [11, 3034]. Implementing palliative NIV in the ICU or in the wards remains a controversial issue. Even though it is regularly performed [35], quantitative and qualitative data are scarce.

ARF and dyspnea are extremely common in patients receiving palliative care

Together with pain, dyspnea is among the most common symptoms in palliative care patients [36, 37]. The symptomatic treatment of dyspnea related to hypoxemia in a patient with ARF relies on supplemental oxygen [38]. Two situations can be distinguished according to whether the dyspnea is related to a reversible factor (e.g., pulmonary edema, infection, or drug overdose) or is a manifestation of an irreversible process (terminal dyspnea, due for instance to progression of a malignant or neurological disease such as carcinomatous lymphangitis, nerve compression, amyotrophic lateral sclerosis, or other advanced degenerative neuromuscular diseases).

When dyspnea is related to a reversible event, symptomatic treatment is combined with interventions that target the cause (e.g., diuretics, antibiotics, antidotes, or physical therapy). NIV has been evaluated in this situation. Terminal dyspnea, in contrast, is a symptom of the end of life and requires comfort care [38]. Opiates in subanalgesic dosages relieve the subjective sensation of dyspnea without improving ventilation or oxygenation [37, 38]. Anxiolytic agents may be required [39], but this is matter of debate [40, 41]. The use of NIV in patients with terminal dyspnea is controversial. NIV for terminal dyspnea has been described both as an example of futile care and as a tool for improving patient comfort at the end of life.

Palliative NIV in patients for whom endotracheal ventilation is not an option

The appropriateness of palliative NIV in patients with decisions to forego endotracheal mechanical ventilation has been a matter of debate for nearly 15 years (Table 1). The first published study was conducted in France in 30 elderly patients [42]. NIV was successful in 60% of these patients and well tolerated in 75%. All the survivors recovered their previous level of function [42]. A 1994 report describes the results of NIV in 11 patients with ARF during terminal illnesses [43]. Among them, four patients failed to benefit from NIV and died; NIV discontinuation in these four patients was not perceived as traumatic by the patients or families. In addition, palliative NIV relieved the sensation of dyspnea while allowing patients to communicate with their loved ones. Although the authors of this study concluded that NIV improved patient comfort and dignity, the sample size was small and the potential benefits of NIV were not measured rigorously using a questionnaire or a qualitative method. At the same time, palliative NIV was challenged on ethical and economic grounds [44, 45]. In addition to the two above-mentioned studies [42, 43], ten studies evaluated palliative NIV in patients with do-not-intubate decisions. Table 1 lists the main results from these ten studies in 458 patients. Five main conclusions can be drawn: NIV is often offered to patients receiving palliative care; among patients given palliative NIV for ARF related to reversible events, nearly half survive and can return home; survival varies with the underlying condition (being higher in patients with COPD or acute pulmonary edema than in patients with cancer) and with the cause of ARF; survival is lower in patients with a coma at NIV initiation or with an ineffective cough; and patient comfort and family satisfaction, although not specifically measured, are not described as poor.

Table 1 Studies of palliative NIV (10 studies, 458 patients)

Interestingly, two practice surveys have evaluated whether palliative NIV is widely used in patients for whom endotracheal ventilation is not an option. In the UK, among 38 physicians specialized in Duchenne’s muscular dystrophy, nearly 80% reported discussing palliative NIV with the patients and families, and the same proportion had already used palliative NIV for their patients [46]. The other survey was done in Canada and the USA in intensivists, pulmonologists, and physical therapists [47]. The vast majority of respondents felt that NIV should be offered to patients with ARF and decisions to forego endotracheal ventilation. Nearly half the respondents perceived palliative NIV as a component of comfort care, and nearly 80% had used palliative NIV in patients with end-stage COPD or heart failure. Palliative NIV was less often used in patients with cancer and in those who had chosen comfort care only (60 and 50%, respectively). Table 2 lists the main advantages and disadvantages of palliative NIV. It is still controversial to consider that the ICU is the best place to offer NIV as a palliative tool or to state that the ICU is a good place to die. Along this line, a very recent paper has shown that cancer patients and their relatives considered that the ICU was a very bad place to die [48]. Studies are needed to compare palliative NIV in both places, because we know that in real life it is administrated in both the ICU and the wards [49].

Table 2 Possible advantages and disadvantages of palliative NIV in the ICU

Despite this evidence that palliative NIV is widely used, further studies are needed, for several reasons. No accurate data are available on the rate of palliative NIV use in patients for whom endotracheal ventilation is not an option. The impact of palliative NIV has been measured in terms of overall survival but not quality of life. Qualitative assessments via interviews or questionnaires have not been performed, although a qualitative approach is crucial to assess the risk/benefit ratio of palliative NIV. The impact of palliative NIV has been evaluated in patients with a variety of underlying conditions (e.g., heart failure, COPD, and cancer) but has not been measured according to the cause of the respiratory deterioration. Finally, the perceptions of the health-care professionals along the chain of care (nurses and physicians in wards and ICUs) have not been reported. To date, a single qualitative study reported beneficial effects of palliative NIV in patients with amyotrophic lateral sclerosis in the absence of acute events [50]. Promising improvements in quality of life were obtained. However, none of these patients had ARF. Instead, they experienced a gradual deterioration in their respiratory status, and their quality of life and respiratory function were evaluated every 2 months.

Position of learned societies regarding NIV use in patients with decisions to forego intubation

Palliative care specialists have expressed concern about the increasing use of palliative NIV [51], and most surveyed physicians request that learned societies issue recommendations about palliative NIV [46]. Physicians worry that patients may perceive palliative NIV as a highly technical, uncomfortable, and aggressive procedure that conflicts with their preferences and values. The term “noninvasive” may not be fully justified. The noninvasiveness of NIV has never been qualitatively established from a patient or proxy perspective. Actually, recommendations have been issued for patients with degenerative neurological diseases in the absence of acute events [52].

According to the consensus statement on NIV for ARF issued in 2001 [American Thoracic Society (ATS), the European Society of Intensive Care Medicine (ESICM), Société de Réanimation de Langue Française (SRLF), the European Respiratory Society (ERS)], palliative NIV is appropriate in selected patients for whom endotracheal ventilation is not an option provided the cause of ARF is reversible and NIV improves patient comfort [53, 54]. Palliative NIV is not discussed in the 2002 recommendations issued by the British Thoracic Society [55]. A European survey coordinated by the ERS indicated that palliative NIV was used in a large proportion (31%) of patients with decisions to limit active treatments [35]. The authors concluded that a qualitative evaluation of palliative NIV was needed. In 2007, a Society for Critical Care Medicine (SCCM) task force suggested an approach for deciding when to offer NIV to patients for whom endotracheal ventilation is not an option [56]. This approach involves a daily evaluation of NIV objectives, which are classified into three categories (Tables 3, 4), among which patients may move from day to day: NIV as life support for patients without treatment-limitation decisions; NIV as life support for patients with do-not-intubate decisions, under the rationale that the adverse effects of NIV (e.g., some degree of patient discomfort, leaks, and alarms) are acceptable given the expected survival benefits; and NIV as a palliative measure in patients receiving comfort care only. The last two categories define palliative NIV. Preserved communication between the patient and family is considered one of the main benefits of NIV. NIV for comfort care is intended to relieve the sensation of dyspnea. In this situation, recovery is not a treatment objective and adverse effects of NIV are consequently unacceptable. This classification scheme can be expected to improve communication among physicians, other health-care professionals, patients, and families. It rests on an ethical scaffold involving clarification of treatment goals, respect for patient preferences and values, and communication with and between the patient and family.

Table 3 Classification of situations for NIV use established by the SCCM (from [56])
Table 4 Data on situations for NIV use as established by the SCCM [56]

A 2006 consensus statement on NIV in ARF [Société Française d’Anesthésie et de Réanimation (SFAR), Société de Pneumologie de Langue Française (SPLF), and SRLF with contributions from the Association des Anesthésistes Réanimateurs Pédiatriques d’Expression Française (ADARPEF), Groupe Francophone de Réanimation et Urgences Pédiatriques (GFRUP), Services d’Aide Médicale Urgente (SAMU) de France, and Société Française de Médecine d’Urgence (SFMU)] describes the indications for palliative NIV as follows: “NIV can be used in patients who choose to forego endotracheal ventilation or whose prognosis is too poor to warrant endotracheal intubation (Grade 2+). In patients at the end of life, NIV is warranted only when it improves patient comfort.” This consensus conference did not specifically recommend the use of a code status such as “Do Not Perform NIV.”

Among the important issues not addressed by the current literature, studies are warranted to find answers to the three following points: (1) we do not know whether palliative ventilation increases duration of life or if it extends the dying process. Similarly, objective data are needed to guide the optimal duration of trial of palliative NIV so as to avoid extending non-beneficial care in patients who will inexorably die. (2) Qualitative observational data are also required to list actual benefits from palliative NIV, such as improvement of family experience, patient’s well-being, quality of end-of-life care, family satisfaction, global clinician’s perspective. (3) Should palliative NIV be performed in incapacitated patients in order to either get a chance of survival (SCCM 2 type) or to alleviate symptoms of respiratory distress?

Description of the OVNI study

OVNI is a longitudinal observational multicenter study performed between November 2010 and April 2011. All patients given NIV were monitored daily until hospital discharge to determine survival, treatment-limitation decisions (if any), and treatment intensity. In the ICUs, patients were classified daily according to SCCM task force criteria (Tables 3, 4) as receiving NIV as life support without treatment-limitation decisions, receiving NIV as life support with do-not-intubate orders, or receiving NIV as comfort care. The last two categories define palliative NIV. The patients and families will be followed up for 1 year, with phone calls on days 90, 180, and 365 to collect data on survival, quality of life, and self-sufficiency. Information will be obtained on family burden (symptoms of anxiety, depression, and stress). The nurses, intensivists, and referring physicians will be asked about their satisfaction with treatment-limitation decisions and with the results of palliative NIV. This study should provide qualitative and quantitative information on the results of NIV at the end-of-life.


Palliative NIV is regularly performed in the ICU as well as in the medical wards or the emergency department. The technique is perceived as feasible and provides survival benefits in hypercapnic patients and those with cardiac pulmonary edema. In cancer patients, benefit seems dismal. Current recommendations from teaching societies recognize the use of palliative NIV. However, no qualitative data are available to date. For instance, studies are needed to identify actual benefits from palliative NIV that are not related to survival. In patients who survive, quality of life data may convince clinicians that palliative NIV actually extends duration of life and does not merely extend the dying process. In patients who die, studies reporting quality of dying and death for patients and post-ICU burden in relatives (e.g., complicated grief, post-traumatic stress disorder, anxiety, and depression) are warranted. Further recommendations should help to increase the use of palliative NIV in patients who may benefit, but also highlight those patients in whom the technique provides no survival and no qualitative benefits.