Use of high‐flow nasal cannula oxygen therapy for patients with terminal cancer at the end of life

Abstract Background Few studies have focused on high‐flow nasal cannula (HFNC) usage in the last few weeks of life. The aim of this study was to identify the status of HFNC use in patients with cancer at the end of life and the relevant clinical factors. Methods We performed a retrospective cohort study in a tertiary hospital in the Republic of Korea. Among patients with cancer who died between 2018 and 2020, those who initiated HFNC within 14 days before death were included. Patients were categorized based on the time from HFNC initiation to death as imminent (<4 days) and non‐imminent (≥4 days). Results Among the 2191 deceased patients with terminal cancer, 329 (15.0%) were analyzed. The median age of the patients was 66 years, and 62.9% were male. The leading cause of respiratory failure was pneumonia (70.2%), followed by pleural effusion (30.7%) and aggravation of lung neoplasms (18.8%). Most patients were conscious (79.3%) and had resting dyspnea (76.3%) at HFNC initiation. Patients received HFNC therapy for a mean of 3.4 days in the last 2 weeks of life, and 62.6% initiated it within 4 days before death. Furthermore, female sex, no palliative care consultation, no advance statements in person on life‐sustaining treatment, and no resting dyspnea were independently associated with the imminent use of HFNC. Conclusions Many patients with cancer started HFNC therapy at the point of imminent death. However, efforts toward goal‐directed use of HFNC at the end‐of‐life stage are required.


| INTRODUCTION
Dyspnea is a prevalent symptom that 10%-70% of patients with cancer experience near death. 1,2 Despite the advances in unraveling the pathophysiology and diagnostic workup of dyspnea, those in treatment are unparalleled. 3 Thus, dyspnea still bothers patients with advanced cancer and remains a challenge for physicians. 4 To date, pharmacologic agents, such as opioids, and non-pharmacological approaches, such as oxygen therapy, are widely used to manage dyspnea. 3 In patients with advanced cancer, an initial approach with non-pharmacologic methods and treatment of underlying causes is recommended. 5 However, oxygen therapy is narrowly recommended for patients with hypoxemic dyspnea. 5 Conventional oxygen therapy uses a nasal cannula or facial mask to deliver low oxygen flow. Additionally, noninvasive ventilation improved gas exchange, but tolerance was poor due to synchronizing difficulty, claustrophobia, and distinctive mask-related side effects. Therefore, a high-flow nasal cannula (HFNC) was recently introduced in clinical practice. It supplies a high flow of heated and humidified oxygen via an interface with a silicon nasal cannula without occlusion, enabling patients to talk or eat while on oxygen. 6 Notably, non-invasive ventilation is superior to conventional oxygen in reducing dyspnea and opioid doses in patients with terminal cancer who only received palliative care. 7 Moreover, HFNC is superior to conventional oxygen in alleviating dyspnea and well tolerated in patients with cancer. [8][9][10] Meanwhile, the role of oxygen and its optimal delivery methods at the end of life has yet to be established. 11 Even though HFNC is a tolerable option, it is barely suggested in a time-limited manner when dyspnea is unrelieved by conventional oxygen therapy. 5 To further develop a consensus on HFNC application at the end of life, understanding the current status of HFNC use is essential. Therefore, we aimed to investigate the status of HFNC use in patients with terminal cancer at the end of life and the relevant clinical factors.

| Study design
We conducted a single-center retrospective study of patients with cancer who died at Seoul National University Hospital (SNUH) between January 2018 and December 2020. SNUH is a tertiary university hospital with 1751 beds and 1800 doctors in the Republic of Korea that does not operate in an inpatient hospice-palliative care ward. We reviewed data based on the last admissions before death to evaluate HFNC usage in the do-not-intubate setting at the end of life. Next, we excluded patients who died in the emergency department and those who underwent mechanical ventilation during admission. Patients who did not receive HFNC were excluded. Additionally, patients who did not initiate HFNC therapy 14 days before death were excluded as the "last 14 days" was one of the time criteria for end-of-life cancer care's intensity ( Figure 1). 12 Additionally, the study period was prior to the emergence of the Omicron variant of coronavirus disease 2019 (COVID-19) when infected patients received treatments at hospitals dedicated to COVID-19 in Korea. Hence, we supposed that the pandemic would not likely affect patients' care and access to HFNC in SNUH.

| Data collection and measurements
Demographic and clinical characteristics at baseline, status during HFNC application, and data regarding HFNC use patterns were obtained from electronic medical records. Symptoms and image findings were based on medical records mainly written by internists and formal readings by radiologists, respectively. We collected the data for the department where patients were initially admitted and classified them as "medical" for any subdivision of internal medicine and "non-medical" for other departments. In addition, we classified patients as recipients of palliative care consultation if they had medical records indicating that they were requested to the palliative care team of SNUH. The team comprises medical oncologists, palliative care nurses, and medical social workers with sufficient clinical experience in palliative care. The consultation was conducted when the primary attending physicians determined it was necessary for the patients' disease course and made a request. During the consultation, the team comprehensively assessed palliative care needs through interviews and underwent discussion for goals of care and advance care planning at the end of life. We evaluated the Charlson comorbidity index 13 after excluding malignancy-related conditions such as solid tumors, leukemia, and lymphoma (i.e., non-cancer CCI). Given that the last 3 days before death is the most common definition of impending death in patients with cancer, we used the time from HFNC initiation to death to categorize into the imminent (<4 days) and non-imminent (≥4 days) groups. 14,15 To evaluate short-term change in pharmacologic measures for dyspnea, we calculated opioid doses within 48 h before and after HFNC initiation using morphine equivalent daily dose (MEDD). 16 The "Act on Hospice and Palliative Care and Decisions on Life-Sustaining Treatment (LST) for Patients at the End of Life" was implemented in February 2018 in the Republic of Korea. 17 It enables patients to make advance statements in person that they do not require LST through advance directives (form number 6) or physician orders for LST (form number 1). Therefore, we considered them as having "advance statements by patients" if there were either form number 1 or 6. Furthermore, at an imminently dying state, specific preferences should also be decided and documented (hereafter, "LST documentation") for the following treatments: cardiopulmonary resuscitation, mechanical ventilation, hemodialysis, anti-cancer treatment, transfusion, inotropic agents, and extracorporeal membrane oxygenation. However, the LST document does not explicitly encompass HFNC. Besides, if the patient has no advance statements or cannot express the intention of LST, first-degree family members should decide on behalf of the patient. In this study, we reviewed the presence of advance statements, LST documentation, and documentation dates.

| Statistical analysis
Descriptive data were used to summarize the demographic and clinical characteristics. Pearson's chi-square or Fisher's exact tests and Kruskal-Wallis analysis of variance were used for categorical and numeric variables to compare groups, respectively. We performed a univariable analysis, and statistically significant variables were included in the multivariable logistic regression analysis with backward selection to identify relevant factors in the imminent use of HFNC. All statistical analyses were two-sided, and the significance level was set at p < 0.05. All analyses were conducted using STATA version 16.0 (StataCorp LP).

| RESULTS
Among the 2191 patients with cancer who died during the study, 329 initiated HFNC treatment 14 days before death and were included in the final evaluation. The percentage of patients who used HFNC during the last 14 days steadily increased from 13.3% to 17.1% annually ( Figure 2). Overall, 62.6% (206/329) and 37.4% (123/329) were in the imminent and non-imminent groups, respectively.  Table 1 shows the baseline characteristics of the imminent and non-imminent groups. Although evenly distributed, the imminent group was younger (median, 65 vs. 68 years, p = 0.031), with a higher percentage of females (41.8 vs. 29.3%, p = 0.023) than the non-imminent group. Moreover, the imminent group stayed in the hospital for a significantly shorter duration (9 vs. 13 days, p < 0.001), with a lower proportion of lung cancer (22.8 vs. 33.3%, p = 0.037) than the non-imminent group. Regarding advance care planning, a significantly lower proportion of patients in the imminent group received palliative consultation (44.7 vs. 61.0%, p = 0.004) and provided advance statements in person (45.6 vs. 63.4%, p = 0.002) than the non-imminent group.

| Clinical status at the time of application of HFNC
Pneumonia (70.2%) was the most common etiology of respiratory failure, followed by pleural effusion and lung cancer or metastasis aggravation. The reasons for applying HFNC were balanced between the two groups, with hypoxia, dyspnea, and tachypnea presented in 90.0%, 71.1%, and 49.9% of the population, respectively.
Patients who were alert (75.7 vs. 85.4%, p = 0.037) or had resting dyspnea (71.8 vs. 83.7%, p = 0.014) were substantially underrepresented in the imminent group compared with the non-imminent group. Additionally, patients in the imminent group received more oxygen before HFNC (median, 10 vs. 7 L/min, p = 0.002) and had significantly lower percutaneous oxygen saturation (SpO2; 88 vs. 89%, p = 0.003). No significant difference in the MEDD was observed between the two groups ( Table 2).

| Factors associated with imminent use of the HFNC
In the univariable logistic regression analyses, younger age (p = 0.043), female sex (p = 0.024), a diagnosis of other than lung cancer (p = 0.038), no palliative care consultation (p = 0.004), and no advance statement in person (p = 0.002) were associated with the imminent use of HFNC. Additionally, non-alert mental status (p = 0.039) and no resting dyspnea (p = 0.015) were associated with its imminent use.
For patients who completed LST documentation first, the mean interval from documentation to HFNC application was 7.4 days, and they had additional 2.7 days until death ( Figure 3B). Next, patients who started HFNC first had a mean interval of 1.6 days before completing LST documentation and an additional 2.2 days until death ( Figure 3C). Lastly, for patients who died without complete LST documentation, the mean interval from HFNC application to death was 2.2 days, which was shorter than that of the two previously mentioned groups ( Figure 3D).

| DISCUSSION
Our study focused on using HFNC in patients with terminal cancer during the last 2 weeks of life. Notably, 62.6% of HFNC users started treatment when their death was imminent. Female sex, no palliative care consultation, no advance statement in person, and no resting dyspnea were associated with the imminent use. Additionally, the average duration of HFNC use was 3.4 days, whereas its duration in the non-imminent use group was approximately Healthcare system in the Republic of Korea has two components: (1) National health insurance to provide coverage to all citizens, managed comprehensively in the form of social insurance and funded by beneficiaries' contributions; (2) Medical aid to provide support to lower income groups, funded by general revenue. Those who are neither citizens of the Republic of Korea nor have obtained health insurance qualifications through the residence for a certain period of time fall into the "None" category. c "Medical" department for any subdivision of internal medicine, and "non-medical" department included obstetrics and gynecology, general surgery, emergency department (short stay units only), orthopedics, and thoracic surgery.   [19][20][21][22] and supports what other physicians consider as indications of HFNC. 19 Then, it reflects the features of an acutecare hospital, even for patients with cancer at the end of life. We also observed that most patients maintained HFNC until death, with only a few opting for withdrawal. However, insufficient evidence shows the benefit of highflow oxygen in patients dying from respiratory failure for either substantial survival prolongation or symptom relief, regardless of the underlying reason. 21,23 In addition, prolonged use of HFNC did not significantly lower opioid usage in this real-world practice. Therefore, setting a clear goal before applying HFNC is valuable.
Several clinical signs appear in the last days of life in patients with advanced cancer. 14,15,24 A decrease in oxygen saturation is one of them, which can be taken as a natural dying process, 24 and mere correction with oxygen at this stage would do more for life-sustaining than alleviating symptoms. Moreover, when patients reach this stage, evaluating efficacy by a "time-limited trial 5 " is less valuable as patients tend to become unconscious. In a recent study at another tertiary hospital, HFNC weaning for patients with cancer at the end of life was difficult; only a minority were liberated from it, even with their protocol. 25 Thus, we T A B L E 3 Univariable and multivariable analysis for imminent use a of high-flow nasal cannula.  assume that the non-imminent use of HFNC is beneficial in securing time for a time-limited trial and discussion for goals of care at the end of life. Accordingly, the presence of palliative care consultation or advance statements by the patients may lead to more beneficial usage of HFNC. Considering prior studies that palliative care consultation helped patients receive less aggressive end-of-life care via goals-of-care discussions, 26,27 the non-imminent group may have benefitted from end-of-life discussion and not solely sustain life. In this context, no palliative care consultation could lead to HFNC initiation in the imminent dying state. In addition, no prior documentation on the advance statement in person would make it harder for patients, their families, and physicians a chance to make shared decisions to avoid imminent use or prompt withdrawal of HFNC, currently in the gray zone. These findings imply that advance care planning may help the optimal use of HFNC at the end of life.

Univariable
Despite the growing interest in the influence of sex differences in clinical studies, the relationship between sex differences and end-of-life care for patients with cancer is not well understood. In this study, we observed that female patients were likely to initiate HFNC therapy at imminent death. This finding contradicts a previous study, which showed that male patients receive more aggressive end-of-life care than females, such as intensive care units. 28 In a setting where we excluded those who applied mechanical ventilation, a core part of intensive care units, female patients may seem to be receiving more aggressive care than males, such as the imminent use of HFNC. Therefore, additional studies regarding sex disparities in HFNC use would help clarify this issue.
Our study showed that no resting dyspnea was associated with the imminent use of HFNC. Assumably, physicians would take it as they are not in the process of imminent death and apply HFNC without much concern. Although the level of consciousness may affect how patients complain of their discomfort, the multivariable analysis revealed that no dyspnea at rest was an independent factor associated with the imminent use of HFNC.
An essential aspect of using HFNC to alleviate dyspnea in patients with cancer is whether it adheres to the individual goal, as the American Society of Clinical Oncology guidelines recommended. 5 Meanwhile, the intensive care unit (ICU) admission rate of cancer patients at the end of life in Korea increased steadily, reaching 30% in the last month reported at a tertiary referral hospital by retrospective data, 29,30 and 20% in the last 6 months before death by national claim data. 31 As end-of-life care still being aggressive, we believe this is a good starting point to discuss the appropriate use of HFNC at the end of life. When used appropriately, HFNC could ease the patient's symptoms without requiring additional invasive ventilation. [32][33][34] If not, some may lose the chance of recovery by not receiving HFNC properly, while some merely extend an undesirable life by it. Since HFNC remains a scarce medical resource in hospitals, its unnecessary use may conflict with the issue of distributive justice from an ethical standpoint. 35 From the physician's perspective, however, various concerns exist in applying HFNC, such as the need to gain time for LST documentation, uncertainty about the reversibility of the patient, or demands for HFNC application by patient families who find it difficult to accept their loved one's death. [35][36][37][38][39] Hence, we suggest delicately exploring the patient's goals, values, and preferences through serial conversations with stakeholders regarding the HFNC application. Furthermore, time-limited trials within a few hours can help decide the continuation or withdrawal of HFNC. 5,40,41 Moreover, compassionate removal should also be considered to allow natural death if HFNC use does not achieve the goal and when it has only a minor practical effect. 42 As we analyzed a relatively homogeneous population of patients with terminal cancers at the end of life, we expect this study to provide fundamental data for developing guidelines for HFNC use at the end of life. However, there are some limitations in this study. First, it was performed in a single tertiary hospital primarily focused on the acute care of patients; therefore, we should be cautious in generalizing our results. Second, dyspnea and hypoxemia could not be evaluated objectively, so data regarding HFNC usage in specific populations with hypoxemic dyspnea is limited. Lastly, because of the study's retrospective nature, it was difficult to evaluate the patients stated goals of care thoroughly. Therefore, further well-designed, large prospective studies assessing objective findings and goaldirected use are warranted to overcome these limitations.

| CONCLUSION
Many patients with cancer who underwent HFNC at the end of life initiated it in an imminently dying state. Additionally, inadequate advance care planning was likely associated with the imminent use of HFNC. Therefore, it is necessary to communicate patients' stated goals of care in advance and to work toward goal-directed use of HFNC at the end of life.