Factors Associated With Failure of High-Flow Nasal Cannula

Discussion

This study evaluated the general clinical practice of HFNC and factors associated with HFNC failure. HFNC was used in various medical subjects, most of whom were hypoxemic due to pulmonary diseases (including pneumonia). Subjects with DNR/DNI orders comprised a third of all HFNC applications. About half the subjects were changed directly to HFNC from low-flow devices without application of a reservoir or other high-flow devices, and a substantial proportion of the subjects were not appropriately monitored with ABG before and during HFNC application. Application of HFNC was most commonly decided by residents, resulting in significantly higher HFNC failure rates compared to cases decided by pulmonologists. In non-DNR/DNI subjects, the HFNC failure rate was 27.0% and significantly associated with breathing frequency, ≤ 6 h before HFNC application, ≤ 6 h before HFNC application, and the ROX index ≤ 6 h after HFNC application.

HFNC was devised to overcome the problems of conventional supplemental oxygen devices. The technique has unique mechanisms and can provide many clinical benefits to patients with hypoxemia. The efficacy of HFNC has been reported in a number of clinical settings,^22–27 but the quality of evidence provided was limited in its ability to draw conclusions or make recommendations. Although recent large-scale studies and systematic reviews have reported more evidence regarding HFNC,^28,29 the clinical impact on mortality or advantages over NIV are not clear.^30–32 Despite these controversies, the use of HFNC is increasing rapidly, resulting in greater risks of adverse events due to overuse or misuse. Of these risks, the most serious is a delay in escalating to NIV or invasive ventilation.³³ To prevent or minimize these adverse events, it is important to evaluate current clinical practices with HFNC. However, studies on how HFNC is actually used are limited.

According to our results, the clinical conditions in which HFNC were applied were similar to those of other supplemental oxygen devices. Although hypoxemia due to pulmonary disease (including pneumonia) was the main cause for applying HFNC, DNR/DNI subjects constituted a substantial proportion of subjects. It has been reported that, in DNR/DNI subjects with cancer, idiopathic pulmonary fibrosis, pneumonia, or COPD, HFNC improves oxygenation and respiratory mechanics.^34,35 HFNC was likely applied for this reason in the DNR/DNI subjects, who represented about one third of the subjects of this study. However, contrary to general assumptions, pulmonary diseases (including pneumonia) were about 3 times more highly represented than oncologic diseases, including lung cancer. It is difficult to evaluate clinical practice and factors associated with outcomes for HFNC in DNR/DNI patients because these patients are commonly maintained in the ICU without active monitoring, modification, or intervention.

HFNC is the most advanced device in supplemental oxygen therapy and has high cost. Therefore, HFNC should be considered in clinical conditions in which other supplemental oxygen devices fail or are expected to be ineffective in improving oxygenation. However, about half of our subjects (48.5%) were directly escalated from low-flow devices to HFNC without use of reservoir or other high-flow devices. This fact reflects the clinically important concern that has arisen regarding the overuse and misuse of HFNC. Among the controversies in clinical effectiveness,^16,17 such overuse may lead to misuse of HFNC, which can result in potentially dangerous delays in escalation to NIV or invasive ventilation.¹⁸ Furthermore, considering that cost-effectiveness studies in adult patients are very limited, the use of reservoir or other high-flow devices, which are much less expensive, should be considered or tried before HFNC application.

HFNC is sometimes advantageous and is convenient to apply and use. However, this new modality is relatively unfamiliar to many clinicians, and most patients treated with it are severely hypoxemic, which raises an important question regarding decisions to use HFNC. Deciding appropriate application times and clinical conditions is critical in clinical effectiveness, therefore it matters who decides to apply HFNC. In this study, we noted that the resident decided on HFNC use in 59.7% of cases. In only about a quarter of the subjects was the pulmonologist the one who decided on HFNC application. In multivariate analysis, the decision by residents to begin HFNC significantly increased the risk of HFNC failure compared to when pulmonologists made this decision. This result indicates several possibilities: (1) that the time at which the decision was made by the resident was too late to achieve much effectiveness; 2) that the subject's condition was too severe to be managed with HFNC; or 3) that a pulmonologist was not available at the time of decision to apply HFNC. Therefore, having pulmonologists make such decision is essential to ensuring clinical improvement; our data also indicate that it will be important to adopt educational programs that properly qualify residents and other specialists who are likely to use HFNC.

Monitoring of oxygenation and ventilation is very important in patients with respiratory distress. Most subjects in this study were monitored for breathing frequency and pulse oximetry before and during HFNC. However, ABG was not performed in a substantial proportion of subjects. Patients who need HFNC are more severely compromised and at risk of deterioration in ventilation at any time. Therefore, intensive monitoring is required. However, breathing frequency cannot reflect ventilation appropriately, and pulse oximetry indicates only oxygenation, not ventilation. ABG's are the usual measure for monitoring ventilation along with oxygenation. According to the results of this study, even in non-DNR/DNI subjects who needed proper monitoring of oxygenation and ventilation, ABG was not monitored in 15.2%, 14.8%, and 28.0% of subjects ≤ 6 h before and after HFNC application and ≤ 6 h before HFNC weaning, respectively. Insufficient analysis of ABG could contribute to HFNC failure secondary to inadequate monitoring of ventilation.

The HFNC failure rate in non-DNR/DNI subjects was 27.0%. Although this rate cannot be compared because of an absence of comparable studies, the failure rate of more than a quarter of subjects was above expectations. Predicting HFNC failure with a subsequent decision to escalate to NIV or invasive ventilation is very important to clinicians. Some clinical or respiratory parameters, including high requirement, higher , history of intubation, cardiac comorbidity, and / ratio, have been associated with HFNC failure.^36–38 Also, in multivariate analysis from this study, among the respiratory parameters before application, increases in breathing frequency and decreases in and were significantly associated with an increase in HFNC failure, indicating the importance of careful patient selection with monitoring before HFNC application.

The ROX index, defined as the ratio of / to breathing frequency, was recently described and can predict outcomes of HFNC.^19,39 In our results, the ROX index just after HFNC application was higher in subjects with HFNC success and was significantly associated with HFNC failure in univariate and multivariate analyses, indicating the predictive power of the ROX index for HFNC outcomes in the initial phase of HFNC application. Furthermore, in comparing the ROX index ≤ 6 h before HFNC weaning and the ROX index ≤ 6 h after HFNC application, the ROX index was significantly increased and decreased in the HFNC success and failure subjects, respectively. These results suggest the necessity of the assess and the close monitoring of the ROX index during use of HFNC.

Analysis of HFNC settings and changes of respiratory parameters yielded results that were in line with expectations. In the HFNC failure subjects, the and flow were higher than in the HFNC success subjects, and the last and flow were higher than initial values. Respiratory parameters, excluding , improved after HFNC application in the HFNC success subjects. The development of complications was very low.

Although there were severe limitations in subgroup analyses due to the retrospective nature of the study with insufficient cases, the HFNC failure rates were evaluated in terms of postextubation and acute hypercapnic respiratory failure. In the postextubation subjects without DNR/DNI, the HFNC failure rate was 27.1%, which was higher than the result from a prospective study with low-risk subjects (4.9%)⁴⁰ and was comparable with high-risk subjects (22.8%).⁴¹ Despite the fact that HFNC is not a ventilator and may be limited regarding the ability to eliminate CO₂, HFNC has been attempted as an alternative to NIV in hypercapnia^42–44 because NIV is uncomfortable and interferes with speaking and eating. In non-DNR/DNI subjects with acute hypercapnic respiratory failure, the HFNC failure rate was higher than in all non-DNR/DNI subjects (35.6% vs 27.1%), indicating cautious HFNC use in hypercapnia.

Due to its retrospective nature, there were important limitations in this study. First, there were missing values in each subject's data, which could lessen the accuracy of the analysis. In particular, the comparisons or analyses of respiratory parameters according to time after HFNC application could not be performed because of inconsistencies in the records of respiratory parameters. However, we collected data about respiratory parameters ≤ 6 h after HFNC application to examine the immediate response for HFNC. Second, the proportion of medical subjects was very high. Therefore, current practice related to HFNC among surgical patients could not be reflected in this evaluation. Third, the results of this study are difficult to generalize because of an insufficient number of centers and cases, as well as regional restriction. Fourth, the presence of ARDS and septic shock are important triggers for HFNC use. The onset time of septic shock or ARDS could be before or after HFNC application. However, use of administration code(s) did not allow us to identify the time between HFNC application and the development of ARDS and septic shock, resulting in flaws in examining the relationship between the intensive oxygen therapy and development of these complications. Fifth, in the Republic of Korea, HFNC is usually initiated and monitored by nurses according to prescription by physicians. Therefore, there is a limitation in generalizing the findings of this study to the clinical situations in which respiratory therapists initiate and monitor HFNC. Despite these limitations, this evaluation revealed important aspects of current clinical practice of HFNC.