Non-invasive ventilation has been the subject of numerous past paper SAQs:
The best resources for this topic are:
A Cochrane review of NIV in acute pulmonary oedema found that it dramatically improves survival (RR = 0.6) and decreases the rate of intubation and the length of ICU stay. LITFL mention "over 20 RCTs" in support of this therapy; the abovementioned Cochrane review actually found 32 eligible studies in total, all of which had something positive to say. Only the small studies had found a mortality benefit, but it was a substantial one (RR = 0.6).
A Cochrane review of NIV in the COPD patient group has demonstrated that in comparison to the invasively ventilated group, patients treated with NIV have greatly reduced mortality (by half), reduced hospital stay, and reduced risk of pneumonia. LIFL mentions "over 14 RCTs" in support of this therapy, perhaps because 14 trials met the inclusion critearia for the abovementioned review. This issue is explored to a greater depth in the chapter on ventilation strategies for COPD.
As soon as they can breathe spontaneously, they should be extubated and supported with NIV. This Cochrane review suggests this strategy can halve mortality and reduce the rate of VAP to one third.
NIV is already the mainstay of chronic maintenance for OSA; a recent review by Carrillo et al (2012) has shown that in acute respiratory failure the OSA patients benefit from NIV even more than COPD patients. A 2015 summary of recent opinions (Shirley et al) concluded that it is a essential part of intensive care for the morbidly obese patient.
A meta-analysis by Chiumillo et al (2013) found ten studies meeting inclusion criteria; of these five studies reported mortality, with startling findings: NIV reduced the relative risk of mortality (RR = 0.26) when compared with just oxygen or straight-up intubation. NIV was was associated with a significant reduction in intubation rate, in the incidence of overall complications and infections.
This issue is explored to a greater depth in the chapter on ventilation strategies for status asthmaticus. In brief, the most recent meta-analysis of evidence (Lim et al, 2012) has failed to demonstrate a clear mortality benefit. However, it does seem to prevent intubation and decrease the length of ICU stay, as well as allowing for improved delivery of nebulised drugs.
Cooperative hypercapneic high-risk patients may benefit (Ferrer et al, 2006) but not all-comers (Su et al, 2012) and it may actually be dangerous in unselected patients (Esteban et al, 2004) as waiting for NIV to work in a patient who clearly requires re-intubation is a pointless time-wasting exercise with an associated increase in mortality.
The college model answer to Question 1 from the second paper of 2015 listed this as one of the indications for NIV. Support for it rests largely on small-scale observational studies. Alonso et al (2016) have performed a good review of this practice, but it is not available for free. Instead, one may look to the Cochrane review by Bright-Thomas et al (2013). None of the examined studies were particularly large, and none looked at survival as a specific variable (though in case series prolonged survival until transplant was observed).
Though classical teaching recommends against NIV in suppurative lung disease, a recent review by Carrillo et al (2012) has demonstrated that there is a group of pneumonia patients who benefit from NIV. These are patients with pre-existing cardiac or respiratory disease, i.e. those in whom the NIV has something to work on. In contrast, patients with healthy hearts and lungs have neither COPD nor heart failure, and thus nothing for the NIV to work on; in this group mortality was actually increased. Mortality increased in proportion to the hours of NIV, suggesting that these people just need to be intubated, and that NIV is a time-wasting exercise for them. These findings were confirmed in a more recent retrospective study of CAP patients (Murad et al, 2015) which found that NIV in this setting was associated with high failure rates (76% required intubation anyway) and with no improvement in mortality among those patients in whom it was used as first-line therapy.
"Prophylactic NIV" in the post-operative setting has been promoted as a means of making more rapid the respiratoryr ecovery of at-risk patients (eg. cardiac, thoracic and abdominal surgery). Unfortunately, the data to support this is missing. A recent (2012) review by Jaber et al could only make lukewarm and nonspecific recommendations.
Its not that noninvasive mechanical ventilation enhances the survival of immunocompromised patients; rather, invasive ventilation seems to be associated with a poorer outcome, perhaps because of the increased risk of nosocomial pneumonia and poorer secretion clearance. The college model answer to Question 4 from the second paper of 2015 urges against the use of invasive ventilation in the neutropenic patient ("Attempt to avoid invasive respiratory support if possible").
Most of the data in support of this is actually in support of NIV versus standard oxygen, not invasive ventilation. As an example, a small (n=52) trial by Gilles et al (2001) compared intermittent NIV to standard supplemental oxygen, and found that the NIV group managed to avoid intubation (12 vs 20 patients). . Earlier, Antonelli et al (2000) showed an improvement in ICU length of stay and rate of intubation among neutropenic patients with solid organ malignancy. However, NIV here was used in a very bizarre manner. Gilles et al used it in 45-minute bursts, for example. Antonelli's team used continuous NIV, but aimed for high tidal volumes ( 8-10ml/kg) and used low PEEP. Even though varieous society clinical practice guidelines made recommendations for the use of NIV in these groups (eg. Keenan et al, 2011)
A more recent entry into this debate comes from Lemiale et al (2015, JAMA) who investigated the effects of NIV on 28-day mortality (n= 374). No mortality difference was found in these patients (whose PaO2/FiO2 ratios were all around 140 on average). There was no improvement in the rate of intubation. The only criticism is that the authors overestimated mortality from respiratory failure (they expected 35% but it was closer to 26%) and so the study ended up being underpowered. An excellent discussion in response to this was published in the Journal of Thoracic Disease (Barbas and Neto, 2016). In summary, those who need intubation still need intubation, and using NIV to delay intubation only increases mortality.
This is explored in greater detail in the chapter on post-extubation stridor. In brief, the role of NIV in post-extubation stridor is unclear. It may benefit a select group of patients, but in others it delays an otherwise inevitable intubation. It may also make intubation more difficult, as the cords become more oedematous with time. The college mention CPAP as one of their solutions for stridor in the model answer for Question 27 from the second paper of 2008.
if the patient requires intubation but is "not for " intubation, then NIV may a way of palliating them, or offering a sensibly non-aggressive therapy to give them a chance of survival. A review of this practice from Azoulay et al (2010) suggests that reversible causes of dyspnoea may be treated with NIV, and that it can provide comfort for transiently increased effort of breathing. However, the authors do caution against futile therapies, and describe terminal dyspnoea as one of the symptoms of the end of life. Azoulay et all then went to to perform a prospective observational study of 780 patients among 54 ICUs in 2013 which discovered an (unsurprisingly) high mortality among the patients that declined intubation. Surprisingly, among survivors 90 day quality of life was similar to the intubated group, suggesting that NIV promotes a satisfactory level of survival in this "palliative" group.
Question 24 from the second paper of 2014 specifically asked about the role of planned NIV immediately following extubation. The candidates were encouraged to "explain how this differs from therapeutically applied (rescue) NIV used in the same context". The examiners quote several studies in their model answer: