Exploring clinicians' beliefs and practices regarding Non-Invasive Ventilation devices: An international survey study

Exploring clinicians'


Introduction
The use of Non-Invasive Ventilation (NIV) in the acute care setting has expanded over several decades and has been the standard of care for many conditions with acute respiratory failure such as COPD exacerbations and acute cardiogenic pulmonary edema. 1 The indications have further expanded to include postoperative hypoxia, post extubation respiratory failure and reducing post extubation failure in high-risk patients. 2,3 NIV has some attractive features compared to invasive ventilation with artificial airways such as the reduced need for sedation, reduction in ventilator associated pneumonia, and reduced need for invasive ventilation. 3 Traditionally, NIV has been applied with the use of dedicated NIV ventilators as they were documented in studies to perform better than critical care ventilators especially regarding leak compensation and patient-ventilator synchronies. 4 However, many new generation critical care ventilators are now equipped with NIV software capabilities that challenge this notion of inferiority to dedicated non NIV specific devices. 4 The development of sophisticated software in NIV and high flow oxygen therapy (HFOT) modes make the use of critical care ventilators a more attractive option as an all-in-one machine, on the other hand, critical care ventilators are larger, more sophisticated, and more expensive and might be difficult to use outside of the intensive care settings.
Most studies that compare different ventilators used for NIV are bench lung models simulator studies that utilize different model criteria, and the lack of standard references or consistent terminology may have influenced the understanding and assessment of NIV devices. 5 Given the conflicting literature results, we aimed to assess clinicians attitudes and perceptions toward the use of NIV and how they differ between dedicated NIV and critical care ventilators.

Methods
An international online survey to clinicians with seven questions (Table 1) regarding their thoughts and experience in using NIV in their acute care settings was posted on the Society of Mechanical Ventilation website and promoted via emails and social media. The study dates were from November 2022 to March 2023. The survey was anonymous, and an exemption of consent was obtained from the Institutional Review Board. Analysis of variants (ANOVA) was performed for the total responses in each question, followed by Multivariate analysis of variants (MANOVA) for responses per occupation. Statistics was performed using R software (version 4.3.0, R Core Team 2022, Vienna, Austria) to analyze the differences in results, using the null hypothesis that there is no significant difference between the groups in terms of their average answers to the different questions, P value of < 0.05 was considered significant.

Results
A total of 514 respondents participated in the survey. Respondents were comprised of 218 physicians, 218 respiratory therapists, 28 nurses, and 50 other clinical roles involving respiratory management (engineering and technicians). The responders were from 28 countries ( Table 2).
The majority of respondents (67.3%) reported using both critical care ventilators and dedicated noninvasive ventilators for patients requiring non-invasive positive pressure support. This was maintained across all professional designations, with greater than 67% of nurses, respiratory therapists, physicians, and other practitioners using both types of ventilators in their practice. In general, those differences were not statistically significant (P 0.09). However, with multivariate analysis per occupation, the results were statistically significant (P < 0.001) ( Table 3).
A majority (45.3%) of participants reported the dedicated non-invasive ventilator systems provided the best synchrony with patient respirations, with 50% of nurses and 58.3% of respiratory therapists attesting to its superiority. A plurality of both physicians (40.8%) and the other respiratory specialist participants (40%) favored critical care ventilators for synchrony. Those were not statistically significant (P 0.59) but were significant with multivariate analysis per occupation (P 0.04).
Non-significant results in general (P 0.6) and in multivariate analysis (P 0.07) were obtained in evaluating for leak compensation, with 57.1% of nurses and 63.8% of respiratory therapists reporting better results with dedicated non-invasive ventilators. Physicians were relatively evenly split, with approximately one-third of participants split among all categories. Non-denomination participants were largely in favor of critical care machines (52%). Overall, 53.1% of participants expressed that dedicated non-invasive ventilators to be superior in providing non-invasive positive pressure support, with a majority of participants expressing similar preferences among nurses (67.9%), respiratory therapists (70.6%), and other respiratory practitioners (40%). However, among physicians, a majority (42.2%) favored critical care ventilators over the dedicated devices in providing non-invasive support. The general results were not statistically significant (P 0.42) even with multivariate analysis per occupation (P 0.098).
56.4% of the participants indicated that they have automatic synchronization software on the ventilators they use, 22% that they don't have such software, and 21.6% were not sure if they do or not. The results were not statistically significant in all responders (P 0.22) but significant in multivariate analysis per occupation (P 0.008).

Discussion
Our results show that the majority of clinicians use both dedicated and critical care ventilators compared to exclusively using either one alone. Historically, the use of NIV has been limited to dedicated non invasive ventilators and many studies have documented that dedicated non invasive ventilators operate better than critical care ventilators for NIV. 6,7 Newer generation critical care ventilators have added new software with different algorithms that challenge this hierarchy. 8 With recent recalls of some of the dedicated non invasive ventilators, it seems that the use of critical care ventilators for NIV is on the rise. 9 Additionally, having HFOT, NIV, as well as invasive modes available make critical care ventilators an attractive option as all features are in one device.
Our results showed a non-significant difference in opinions regarding patient-ventilator synchrony between NIV and critical care ventilators, however in the multivariate analysis, there was a significant difference largely favoring NIV influenced by RT favoring NIV. The majority of clinicians now have some kind of automatic synchronization software on the ventilators they use.
Patient-ventilator asynchronous events (AEs) have long been an area of concern in ventilated patients, leading researchers, and manufacturers to investigate methods of detecting and reacting to AEs utilizing algorithmic software. Early efforts included algorithms designed to detect ineffective respiratory efforts in ICU ventilators. 10 However, ICU ventilators were previously designed without consideration of significant circuit leaks, given that patients were 88 typically ventilated via endotracheal tubes. As these devices added additional ventilator modes, including non-invasive ventilation (NIV), leaks became a more commonly encountered issue, as NIV masks have far more frequent leaks than do endotracheal tubes.
Leaks were shown to be a major source of AEs in ICU ventilators in a multi-center study in France. 11 Shortly thereafter, efforts for algorithmic detection of AEs during ICU ventilator NIV were started. 12,13 In contrast, dedicated NIV machines, such as those used at home by patients with chronic respiratory failure, have long had to deal with leaks. However, in hospitalized patients with acute or acute-on-chronic respiratory failure, the comparative efficacy of dedicated NIV machines versus ICU ventilators' NIV modes was previously unknown.
Carteaux and colleagues 14 recognized that dedicated NIV machines did better managing asynchronous events (AEs) especially when related to leaks than did ICU ventilators in NIV mode. Leaks can be mistaken for inspiratory efforts, thus causing an auto-trigger (AT) event. This finding was true even with ventilators utilizing algorithms to detect and correct AEs. The same authors also found significant variation among tested ventilators at the time, such that each machine needed to be individually assessed, rather than being able to draw conclusions about ICU ventilators as a whole. Later studies showed ICU ventilator NIV algorithms equaling or surpassing dedicated NIV machines, with some systems incorporating diaphragm electrical activity. 15 As NIV software evolved, there was particular interest in utilizing ventilator waveforms alone for detecting all varieties of AEs during NIV. To establish a "gold standard" to compare software algorithms against. Longhini and colleagues 16 tested the sensitivity, specificity, PPV, and NPV of ICU clinicians at recognizing AEs during NIV. Prior software algorithms had been successful at detecting NIV ineffective efforts, 7,9,17 but Lettelier and colleagues 18 were the first to algorithmically autodetect and categorize a variety of NIV AEs (including autotriggering, double-triggering, and ineffective efforts). Notably, their model made specific corrections for leak detection so that when leaks were too large ventilatory cycles were discarded, rather than being mistakenly categorized as auto-triggers.
To date, the overwhelming majority of software algorithms utilize explicitly programmed mathematical models to detect and categorize AEs. With the rise of machine learning (ML) methodologies, especially supervised ML where pre-categorized training datasets are fed into ML models, new forms of AE detection are being assessed. Similar to older modeling efforts using explicitly defined algorithms, ineffective efforts were some of the first AEs modeled using ML. 19 However, invasive mechanical ventilation was a more common target of ML research, and several studies have attempted to broaden the detection and classification of AE varieties. 20,21,22,23,24 To date, no ML models have been published that directly assess NIV modes of ventilation, either with dedicated NIV machines or ICU ventilators.
One significant limiting factor in assessing the clinical utility of automatic synchrony software's and software-driven AE detection is the lack of head-tohead comparisons of software varieties. Typically, studied synchrony software is compared to a gold standard measure-expert clinician detection and classification of AEs. AEs. No direct comparisons have yet been published that of either compare different between-group differences in explicitly designed algorithms, or different competing ML models, or that assess the performance difference between designed algorithms versus and ML models have yet been published. In summary, while software utilizing either algorithm-or ML-based detection of AEs shows significant promise to improve patientventilator synchrony in both NIV and other ventilator modes, there is marked heterogeneity in software between ventilator machine types, manufacturers, and the underlying methodology used to detect and intervene on AEs.
Circuit leaks in both invasive and non-invasive mechanical ventilators continue to constitute a major challenge as it limits the ability of mechanical ventilators to deliver a synchronized breath. Air leak in non-invasive ventilation is a major concern as it not only impacts arterial oxygenation but also causes mouth and throat dryness, eye irritation but also impact on sleep quality. Alternative mechanisms to address this by tightening the mask would lead to skin and nasal lesions in a significant percentage of the patients. 25 Most mechanical ventilators have leak compensation software that allows them to adjust and respond to increase or decrease in leaks accordingly. Such software monitors the percent leak and responds to leaks by correct quantification within a few breaths, but it also attempts to reduce auto trigger. 26,27 Opinions on leak compensation between ventilator types also did not differ significantly. As noted by Scott, 4 these findings may be explained by the wide In addition to leak compensation and synchrony, a survey by Crimi and colleagues 30 reported the presence of a dual circuit, inspiratory oxygen fraction control, transportability, monitoring capabilities, drug delivery and alarm control as additional factors influencing ventilator choice, in order of decreasing power. In contrast, they also demonstrated that geographic location of the respondents, clinical scenario (i.e. cardiogenic edema vs hypercapnic failure), familiarity with the devices, availability of the machines, and difficulty in setting the ventilator were factors not significantly associated with influencing ventilator preference, however the external validity of these results to hospitals outside of Europe has yet to be proven. The training and experience of those who utilize NIV also likely has a prominent role in influencing the preferred mode of ventilation, and perhaps accounts for the significant difference in opinion demonstrated in the multivariate analysis of superior patient synchrony. The same study 27 highlight that pulmonologists and those who practice more frequently outside of the ICU have more familiarity with dedicated NIV machines and thus may be more inclined to favor these machines, when compared to intensivists who may be more familiar with the opposite.
One belief is that dual circuits provided by critical care ventilators minimize the risk of CO2 rebreathing, making them preferable for cases of hypercapnic respiratory failure. Non-rebreather valves lower this risk in single-limbed circuits, at the cost however of adding expiratory resistance to the circuit. 31 Given the lack of consensus on the overall best ventilator type to use for NIV, it's likely that the decision is influenced by multiple factors, with cost and availability of modes and settings also influencing personal preferences. 4,5 Our study has limitations. Although surveys can be a valuable tool for medical research, the results should be viewed with caution. The purpose of this study was to gain insight into device preferences and perceptions in the delivery NIV. Practice variation, participant bias, variable availability of different devices, level of training, and technical education should all be considered.

Conclusion
Although multiple devices can provide NIV, clinicians have adapted to evolution in technology. Despite the lack of evidence, there is wide variability in opinion with no clear consensus regarding the clinicians' attitude towards which ventilators are superior to use during NIV, especially according to surveyed occupation. Well conducted comparative studies are needed to answer the question of which ventilators are superior in NIV.