Principles of mechanical ventilation

Management of hypercapnic respiratory failure

Weaning from IMV

Care planning and delivery of care

Novel therapies

Modes of mechanical ventilation

There are two basic modes of providing mechanical ventilation. In volume-targeted, the operator sets the tidal volume to be delivered and the duration of inspiration (Ti). The ventilator generates whatever pressure is necessary to deliver this volume within this time. In pressure-targeted, the operator sets the inspiratory pressure. The volume of air the patient receives is a function of the impedance to inflation of the lungs and chest wall and the inspiratory time. The Ti should be of sufficient length to achieve an adequate volume and at a frequency that allows the patient time to fully exhale.

The terminology used for pressure-targeted ventilation can cause confusion. In bi-level ventilation, one pressure is set for inspiration (IPAP) and a second pressure for expiration (EPAP). The difference between the two is the level of ventilatory assistance or PS. This mode is most commonly used for NIV. The same term, CPAP with Pressure Support’, can be used to describe a mode of invasive ventilation and/or non-invasive ventilation on some ICU ventilators. The operator sets an incremental inspiratory pressure above the CPAP setting rather than setting an absolute level of inspiratory pressure.

Pressure-targeted ventilation has a number of advantages. First, the pressure delivered is constant and this avoids the sudden and uncomfortable pressure increase that occurs with volume control. Second, pressure-targeted ventilation compensates for air leak,12 ,13 which is an inevitable consequence of the interfaces used for NIV. Third, positive pressure throughout expiration (EPAP) flushes exhaled CO₂ from the mask and distal ventilator tubing,14 ,15 aids triggering (see below) and counteracts the tendency for upper airway collapse during expiration. Pressure ventilators have been used in almost all of the randomised controlled trials (RCTs) in AHRF.16 In the UK, volume ventilators are rarely employed (outside of specialist centres) and will not be considered further in this guideline.

In the Spontaneous (S) mode (also known as assist mode), the ventilator delivers assisted breaths in response to patient inspiratory effort. If the patient fails to make adequate inspiratory effort, no ventilator support is delivered. By contrast, in the timed (T) mode (also known as control mode), the ventilator delivers breaths at a rate set by the operator regardless of patient inspiratory effort. ‘Pressure-controlled ventilation’ (PCV) is the term used to describe a mode in which the operator sets the inspiratory pressure, the length of inspiration and the inspiratory rate. In the spontaneous/timed (S/T) mode (also known as assist control), a backup rate is set by the operator. If the patient's RR is slower than the backup rate, machine-determined breaths will be delivered (ie, controlled ventilation). If the patient breathes faster than the backup rate, no machine determined breaths will be delivered and all breaths will be triggered (or assisted). The proportion of controlled and assisted breaths often varies, depending on the patient's state of alertness and respiratory drive.

Trigger sensitivity refers to the effort required by the patient to initiate, or trigger, the ventilator. The lower the trigger sensitivity, the greater effort the patient needs to make to trigger a supported breath. Different trigger settings may be required for individual causes of AHRF (see Management of hypercapnic respiratory failure section).

S/T is the NIV mode most commonly employed in treating AHRF. There have been no trials comparing PS ventilation and PCV in the treatment of AHRF. Bench studies suggest that ventilators designed specifically for NIV have superior performance over standard ICU ventilators used to deliver NIV, particularly in the presence of significant leak.17–22 The extent to which individual types of ICU ventilators (set in the NIV mode) can compensate for leak and the adequacy of patient triggering varies.23 Generally, ICU ventilators appear more prone to patient-ventilator asynchrony than home care ventilators.24

Evidence statement

Most RCTs that demonstrate an advantage to NIV in AHRF have used pressure targeted ventilators (Level 1+).

Recommendation

1. Pressure targeted ventilators are the devices of choice for acute NIV (Grade B).

Good practice points

• Both PS and pressure control modes are effective.

• Only ventilators designed specifically to deliver NIV should be used.

Choice of interface for NIV

The FFM is the most suitable interface, as mouth breathing predominates in AHRF. To accommodate the natural diversity of the human face, a range of shapes and sizes of FFM should be available. Reported studies suggest that different types of interfaces do not affect outcome, but the trials have been small and comparison of masks has been inadequately powered to detect a difference.25–37 The helmet interface, which covers the whole head, is an alternative to an FFM,38–44 but triggering is ineffective. Patients may report about noise caused by turbulence within the helmet,45 and it is not possible to provide humidified gases because of ‘rain out’ in the helmet. A mask that covers the whole of the face (including the eyes, but not the ears) is useful when air leak remains excessive or when nasal bridge ulceration develops,46 and is sometimes better tolerated by the confused or agitated patient. In those who find the FFM claustrophobic or distressing, experienced practitioners may consider using a nasal mask or nasal pillows. Mouth leak limits the effectiveness of nasal interfaces during sleep and nasal pillows are more easily dislodged than the FFM.

Ventilators designed for NIV usually employ a single lumen circuit whereas IMV ventilators use a dual lumen circuit (separate tubing for inhalation and exhalation). In the former, a mask with an integral exhalation port is commonly used. If not, an exhalation port needs to be inserted into the ventilator circuit close to the mask. A minimum EPAP of 3 cm is required to vent exhaled air.14 ,47 The website http://ersbuyersguide.org/ offers information on NIV interfaces that are currently available.

Evidence statement

An FFM is the interface of choice for general/non-specialist use (Level 4).

Recommendation

2. An FFM should usually be the first type of interface used (Grade D).

Good practice points

• A range of masks and sizes is required, and staff involved in delivering NIV need training in and experience of using them.

• NIV circuits must allow adequate clearance of exhaled air through an exhalation valve or an integral exhalation port on the mask.

Indications for and contra-indications to NIV in AHRF

The indication for NIV will vary according to the underlying cause, severity of illness and associated complicating factors. Broad criteria can be used and are summarised in figure 1, and further discussed in Management of hypercapnic respiratory failure section. Severe facial deformity, fixed upper airway obstruction or facial burns, will occasionally make NIV impossible. A number of other contra-indications have been suggested (see figure 1).48 These have most often been employed as exclusion criteria in clinical trials rather than being definitively shown to result in a worse outcome.16 Some of the criteria have been challenged. For instance, coma has been regarded as an absolute contra-indication, because of its associated loss of airway protection, but Diaz et al49 report similar outcomes with NIV in those with a Glasgow Coma Score <8 as the outcomes found in more alert patients. Similarly, confusion, agitation and cognitive impairment make NIV more difficult to apply but should not preclude its use.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Summary for providing acute non-invasive ventilation.

There is less haemodynamic compromise with NIV than with IMV, and hypotension should rarely preclude using NIV. Significant arrhythmia, especially if causing hypotension, may tip the balance towards preferring intubation as, in these circumstances, cardioversion may be indicated.

An acute pneumothorax should be drained before applying NIV. If it is too small to allow the safe placement of a chest drain (or is suspected to be chronic) NIV may proceed with careful monitoring. Using a lower inflation pressure seems theoretically sensible but is without evidence. If the patient deteriorates, NIV should be discontinued—in case it is contributing to the development of a tension pneumothorax—and an urgent chest radiograph obtained.

Vomiting has been considered a contra-indication. The key issue is whether the NIV mask can be rapidly removed, that is, an assessment of whether the patient can signal the need to vomit. Marked abdominal distension may sometimes precipitate AHRF in individuals at risk, for example in COPD or morbid obesity. Management should then address the underlying cause of abdominal distension and manage the risk of vomiting by inserting a nasogastric tube. Similarly, in the at-risk patient, hypercapnic respiratory failure may complicate the later stages of pregnancy (eg, kyphoscoliosis or muscular dystrophy). NIV is ideally suited to manage this complication. The need for NIV should be electively assessed (by nocturnal monitoring), but mask ventilation can be initiated during delivery should respiratory distress develop in an at-risk patient.

The presence of copious secretions increases the risk of treatment failure,50 but NIV may also improve the ability to clear secretions and improve alveolar ventilation.51 ,52

Respiratory arrest or peri-arrest have been considered as absolute contra-indications as NIV is intended to supplement spontaneous breathing. However, as bag and mask ventilation (itself a form of NIV) is used as a prelude to intubation, a short trial of NIV by an experienced operator, can be justified while paying special attention to the risk of glottic occlusion.

In summary, the presence of adverse features is an indication for more intense monitoring and placement within HDU/ICU rather than a contra-indication per se.

Evidence statement

There are few absolute contra-indications to a trial of NIV but some adverse features, especially when combined, require more caution and more intense monitoring (Level 4).

The presence of adverse features increases the risk of NIV failure (Level 2++).

Recommendation

3. The presence of adverse features increases the risk of NIV failure and should prompt consideration of placement in HDU/ICU (Grade C).

Good practice points

• Adverse features should not, on their own, lead to withholding a trial of NIV.

• The presence of relative contra-indications necessitates a higher level of supervision, consideration of placement in HDU/ICU and an early appraisal of whether to continue NIV or to convert to IMV.

Monitoring during NIV

Continuous monitoring of oxygen saturation is essential. Repeated measurement of ABG tensions will be required and can be assessed by capillary sampling or intermittent arterial puncture, noting that capillary sampling is less painful for the patient.53 ,54 One advantage of HDU/ICU placement may be to allow the safe use of an indwelling arterial line for blood sampling. Transcutaneous pCO₂ (TcpCO₂) monitoring is a commonly employed investigation in home ventilation units and the devices are increasingly being employed in hospitals. Small studies have reported on its use in acute respiratory acidosis.55–57 A study by van Oppen et al58 reported on 10 patients receiving acute NIV and demonstrated that TcpCO₂ monitoring is reliable over 12 h and provides an adequate estimation of pH. Further studies are needed to assess the role of transcutaneous CO₂ monitoring.

ECG monitoring is advised for all patients with a tachycardia >120 bpm, dysrhythmia or known cardiomyopathy. As in all severely ill patients, serial vital signs (and National Early Warning Scores, where implemented) should be recorded.

Good practice points

• Oxygen saturation should be continuously monitored.

• Intermittent measurement of pCO2 and pH is required.

• ECG monitoring is advised if the patient has a pulse rate >120 bpm or if there is dysrhythmia or possible cardiomyopathy.

Supplemental oxygen therapy with NIV

There are no trials to guide the use of oxygen enrichment. It is well recognised that hyperoxygenation is harmful in the self-ventilating patient with AHRF.59–61 In the absence of harm from modest hypoxaemia, and to avoid confusion that might arise from having different target saturations in different conditions, a saturation range of 88–92% is recommended in all patients with AHRF either spontaneously breathing or when receiving NIV.62 This is usually easily achieved in AECOPD, but severe hypoxaemia may complicate AHRF in other causative diseases such as CWD.

As for the best method of supplying oxygen, Padkin and Kinnear,63 in a study of patients who were not acutely unwell, reported no difference in inspired content whether delivered directly into the NIV mask or into the ventilator tubing close to the mask. Introducing oxygen at the ventilator end of the tubing was less effective. The mean FiO₂ achieved was 31% at 1 L/min, 37% at 2 L/min, 40% at 3 L/min and 44% at 4 L/min. Flow rates >4 L/min provided minimal additional increase. Kaul64 found that the higher the inspiratory pressure, the less additional benefit resulted from higher flow rates (because higher pressures increase leak). High flow rates also resulted in delay triggering the ventilator. As this risks promoting patient ventilator asynchrony, technically advanced NIV ventilators that allow precise oxygen blending (and a higher FiO₂ enrichment) are a safer and more appropriate alternative when hypoxaemia is severe.

Evidence statements

In AHRF-targeted oxygen therapy (SaO₂ 88–92%) reduces mortality (Level 1+).

When providing NIV, oxygen enrichment is best given at or near the mask (Level 3).

Recommendations

4. Oxygen enrichment should adjusted to achieve SaO₂ 88–92% in all causes of AHRF being treated by NIV (Grade A).

5. Oxygen should be entrained as close to the patient as possible (Grade C).

Good practice points

• As gas exchange will improve with increased alveolar ventilation, NIV settings should be optimised before increasing the FiO2.

• The flow rate of supplemental oxygen may need to be increased when ventilatory pressure is increased to maintain the same SaO2 target.

• Mask leak and delayed triggering may be caused by oxygen flow rates >4 L/min, which risks promoting or exacerbating patient–ventilator asynchrony. The requirement for high flow rates should prompt a careful check for patient–ventilator asynchrony.

• A ventilator with an integral oxygen blender is recommended if oxygen at 4 L/min fails to maintain SpO2 >88%.

Humidification with NIV

There is no evidence to guide the use of humidification in acute NIV. Heated humidification may reduce upper airway resistance and increase comfort when leak is high.65 In short-term studies, heated humidification reduces upper airway dryness,66 ,67 which might improve tolerance and aid secretion clearance, but this has not been proven. Humidification should only be considered when upper airway dryness is a problem or secretions are difficult to expectorate.

Evidence statement

No evidence exists to guide humidification practice in acute NIV (Level 4).

Recommendation

6. Humidification is not routinely required (Grade D).

Good practice point

• Heated humidification should be considered if the patient reports mucosal dryness or if respiratory secretions are thick and tenacious.

Bronchodilator therapy with NIV

As part of a PhD thesis, Kaul68 found that nebulised bronchodilators given concomitantly with NIV in stable patients produced less benefit than when given while patients were breathing spontaneously. Brief discontinuation of NIV for the administration of bronchodilators appears to be safe.69 Acccordingly, bronchodilator therapy is probably better given during breaks in NIV. This may also facilitate coughing and the clearing of respiratory secretions. If discontinuing NIV results in patient distress, it should be continued and a nebuliser sited proximally in the circuit.70

Good practice points

• Nebulised drugs should normally be administered during breaks from NIV.

• If the patient is dependent on NIV, bronchodilator drugs can be given via a nebuliser inserted into the ventilator tubing.

Sedation with NIV

Patient agitation and distress are common in AHRF and may be made worse by the application of NIV before gas exchange has improved and the patient has sensed a reduction in the work of breathing. Despite this, sedatives/anxiolytics and/or opiates are infrequently used due to concern about depressing respiratory drive. This is understandable if NIV is delivered in an inappropriate environment that is unable to provide continuous monitoring and that does not have the ready availability of medical staff to perform safe intubation if needed. On the contrary, relieving patient distress is an important goal and might be expected to increase comfort and the success of NIV. In a 2007 survey of members of the critical care assemblies of the American College of Chest Physicians and the European Respiratory Society, respondents reported using sedatives or opiates in only 25% of cases and 21% stated they had never used either.71 The risk of respiratory depression was given as the reason for non-use. Individual practice was highly variable and, as the response rate was poor (42% European, 14% North American), the conclusions reported are more qualitative than quantative. When treatment was given it was mostly by bolus injection and rarely according to a sedation protocol. Greater experience in the use of NIV and being a critical care clinician increased reported use of opiates/sedation.

In the 2013 BTS audit, involving 2693 cases, NIV failed to reverse AHRF in 30% of patients.5 Agitation was reported as the principal reason in 31% of these. Sedation was ‘attempted’ in 84%. No details are available on what agents were used, or outcome in those so treated. As 91% of all NIV treatments were provided outside of the HDU/ICU, it appears sedation is now more commonly employed but in a potentially unsafe environment.

In the ICU setting, case series have reported that infusions of propofol,72 dexmedetomidine73 and remifentanyl74 ,75 are safe, improve comfort and reduce the failure rate of NIV. Senoglu et al76 compared infusions of dexmedetomidine and midazolam in 45 AECOPD cases with AHRF, using a protocol aiming at a standard degree of sedation. No differences were found in effectiveness between the two agents. There were no significant adverse events and no patient failed to improve with NIV. In another report, the addition of infused dexmedetomidine to a standard protocol of ‘as needed’ bolus intravenous midazolam and fentanyl, given according to a sedation protocol, failed to show benefit, but sedation goals were readily achieved and there was good NIV tolerance and success with the standard protocol.77 A review of sedation to facilitate NIV tolerance makes the pharmacological case for preferring an opiate to a benzodiazepine (because the latter promotes upper airway obstruction through inhibiting the pharangeal dilating muscles) but concluded that studies to date have been too small, have used different drugs and therapy regimes and employed a variety of outcome measures.78 Guidance on the use of sedation within hospitals might be expected to improve patient safety when implemented.79

Evidence statements

Patient distress is common in AHRF and often made initially worse by applying NIV (Level 4).

There is inadequate evidence to guide the use of sedation/anxiolysis in acute NIV. Their use in a critical care setting is reported to improve outcome and reduce patient distress (Level 2−).

Recommendations

7. Sedation should only be used with close monitoring (Grade D).

8. Infused sedative/anxiolytic drugs should only be used in an HDU or ICU setting (Grade D).

9. If intubation is not intended should NIV fail, then sedation/anxiolysis is indicated for symptom control in the distressed or agitated patient (Grade D).

Good practice point

• In the agitated/distressed and/or tachypnoeic individual on NIV, intravenous morphine 2.5–5 mg (± benzodiazepine) may provide symptom relief and may improve tolerance of NIV.

NIV complications

The reported rate of complications varies widely. One review gives an incidence between 30% and 50%,80 the range partly depending on how a complication is defined. Extended duration of NIV, patient agitation and the frequent need to adjust mask fit are all associated with an increase in rate/severity of mask-related problems.

Nasal bridge ulceration is the most common problem (5–10%) and may be severe enough to result in NIV failure.81 Over-tightening is a common cause. NIV masks are designed to mould to the face when pressurised which over-tightening impairs. Should signs of skin trauma become apparent, a barrier dressing and a strategy of regular breaks and alternating between two interface types should be used. Latex allergy occasionally results in florid skin reactions. Some patients seem especially prone to mask-related rash even in the absence of allergy. Topical steroids may be indicated and/or antibiotics if the wound becomes infected.

NIV may cause severe gastric distension. It usually indicates poor coordination between patient and ventilator and it may be necessary to insert a nasogastric tube. Sinus or ear discomfort and nasal mucosal congestion or drying/ulceration can all occur. The value of humidification in preventing these side effects is uncertain but water-based nasal gels and topical corticosteroids or decongestants can be used. Petroleum-based emollients should not be used with supplemental oxygen.

An acute pneumothorax may be life-threatening but difficult to detect. The development of unexplained agitation/distress or chest pain requires this complication to be excluded.82 Co-existent interstitial lung disease or previous episodes of spontaneous or ventilator-induced pneumothorax increase the risk. Using a lower IPAP to avoid large tidal volumes, and a lower EPAP to avoid significantly increasing end-expiratory lung volume (EELV), are logical but not evidence based. If a pneumothorax develops, intercostal drainage is usually required.

Good practice points

• Minor complications are common but those of a serious nature are rare. Patients should be frequently assessed to identify potential complications of NIV.

• Care is needed to avoid overtightening of masks.

• Previous episodes of ventilator-associated pneumothorax warrant consideration of admission to HDU/ICU and use of NIV at lower than normal inspiratory pressures.

• The development of a pneumothorax usually requires intercostal drainage and review of whether to continue with NIV.

Sputum retention

Sputum retention can be a precipitant for AHRF, can cause NIV to fail and is a common reason for respiratory distress post-extubation in patients initially managed by IMV. Excessive sputum production characterises bronchiectasis and CF, and complicates some patients with AECOPD. Promoting sputum clearance can be particularly challenging in those with NMD and in the morbidly obese. Techniques, such as manually assisted cough and mechanical insufflation–exsufflation (MI-E), aid sputum clearance in patients with NMD.83 ,84 However, in a study including patients with either scoliosis or COPD, MI-E reportedly had no benefit.85 In another RCT, the use of MI-E reduced post-extubation respiratory failure in a mixed group of patients including some with AHRF.86 This study also provided NIV to those in respiratory distress. The reader is referred to the BTS Physiotherapy Guidelines87 for more detailed information.

Mini-tracheostomy facilitates secretion clearance in the spontaneously breathing patient88 and may have a role when sputum retention is thought to be a major determinant of AHRF, such as in CF. It is not an easy technique to perform in the anxious and breathless patient and training opportunites are rare. Clinicians who insert percutaneous tracheostomies are best placed to provide a service and the HDU/ICU is the best environment in which to perform mini-tracheostomy. In an attempt to avoid intubation, a combination of respiratory support by NIV and suctioning via a mini-tracheostomy has been described. This probably only has application if IMV is not desired by the patient as, in most such cases, IMV offers more chance of a successful outcome. In the patient initially managed by IMV, a mini-tracheostomy may be inserted at the time of endotracheal tube decannulation in patients with a high secretion load and/or a poor cough.

Evidence statements

Manual-assisted cough and MI-E are safe methods for aiding secretion clearance (Level 1+).

MI-E is more effective than manual-assisted cough in patients with stable NMD (Level 2+).

Mini-tracheostomy is a useful bedside procedure that can markedly improve secretion clearance, but requires patient cooperation and a skilled operator to be performed safely (Level 4).

Recommendations

10. In patients with NMD, mechanical insufflation and exsufflation should be used, in addition to standard physiotherapy techniques, when cough is ineffective and there is sputum retention (Grade B).

11. Mini-tracheostomy may have a role in aiding secretion clearance in cases of weak cough (NMD/CWD) or excessive amounts (COPD, CF), (Grade D).

Modes of IMV

Critical care ventilators are complex devices capable of delivering multiple modes.89 ,90 The traditional divide between pressure and volume has become blurred and hybrid modes combine aspects of both. Most patients with AHRF do not require sophisticated modes of providing IMV.

Initially, when airway resistance is high and/or compliance is low (eg, in asthma, CF or bronchiectasis) a period of mandated or ‘controlled mechanical ventilation’, often combined with deep sedation to reduce spontaneous breathing effort, allows time for bronchodilators, steroids and antibiotics to treat airway inflammation, overcome infection and for ‘bronchial toilet’ to be provided. These considerations also variably apply to the restrictive causes of AHRF. In addition, poor triggering, because of muscular weakness, is a risk in patients with NMD in whom a prolonged period of controlled mechinical ventilation may be necessary. In all patients with AHRF, allowing restorative sleep is important.91–93

Management should shift towards supporting rather than mandating the pattern of ventilation as recovery begins. If there is adequate spontaneous effort, and the RR is not excessive, a switch to PS is recommended to reduce the need for sedation and also as the risk of respiratory muscle wasting may be reduced by establishing early spontaneous breathing. The concept that suppressing spontaneous breathing is causally related to diaphragm wasting is contentious in the literature. Space constraints prevent a fuller examination. One initially compelling human study that claimed to have demonstrated ‘disuse atrophy’ was subsequently critised because the diaphragms had been denervated.94 A study in patients with adult respiratory distress syndrome (ARDS) reported that those patients allowed to breathe spontaneously had less need for sedation than patients treated with controlled IMV, a reduced requirement for vasopressors, fewer days of ventilatory support, earlier extubation and a shorter length of ICU stay.95 This strategy has not been assessed in AHRF.

Evidence statements

Establishing early spontaneous breathing reduces the need for sedation, improves cardiac function and reduces the duration of IMV in ARDS (Level 1−).

Recommendations

12. Spontaneous breathing should be established as soon as possible in all causes of AHRF (Grade C).

13. Controlled IMV may need to be continued in some patients due to severe airflow obstruction, weak muscles leading to poor triggering or to correct chronic hypercapnia (Grade C).

Good practice point

• In obstructive diseases, controlled IMV should be continued until airway resistance falls.

Invasive ventilation strategy

In obstructive causes, tidal volume (Vt) is limited by the airflow obstruction and compounded by the mechanical disadvantage of hyperinflation. The use of high inflation pressures, to achieve a ‘normal’ Vt, risks dynamic hyperinflation.96 It most dramatically occurs soon after intubation but may develop on switching ventilation mode, for example, from controlled to assisted ventilation.97

The adverse consequences of hyperinflation include barotrauma, impaired gas exchange and patient discomfort. The increased intrathoracic pressure impedes venous return and increases right ventricular afterload with a resulting fall in cardiac output and hypotension.98

Prolonging expiratory time limits gas trapping and is achieved by shortening the inspiratory time and reducing the minute volume, an approach recommended in airflow obstruction.99 ,100 If significant gas trapping still occurs, the recommendation is to use a lower than normal Vt in combination with a low RR and a more prolonged expiratory phase.99 ,100 This can often only be achieved using a controlled ventilation mode combined with deeper levels of sedation. On switching to PS (assist) during recovery, the inspiratory pressure needs to be sufficient to provide adequate tidal volume but not excessive. Settings therefore need to be individually adjusted and require regular review.

In ARDS, over-distention and repetitive recruitment/de-recruitment of lung units causes alveolar damage (so-called ventilator-induced lung injury) and may even provoke systemic inflammation.101 One explanation for improved outcome with low Vt ventilation (<6 mL/kg), compared with conventional practice, may be avoidance of ventilator-induced lung injury.102 The ARDS literature provides evidence for permissive hypercapnia, demonstrating that a pH above 7.2 is well tolerated.103 This is the consensus target when pH control is difficult.89 ,90 Allowing permissive hypercapnia will result in cerebral vasodilation and a rise in intracranial pressure and may also compromise myocardial contractility. Attempts to raise pH to >7.2 may, however, compound hyperinflation and barotrauma. In ARDS, a peak airway pressure of 30 cm is the usual trigger for employing permissive hypercapnia, a strategy that reduces mortality.104

In AECOPD, attempts to rapidly restore pO₂ and pCO₂ to normal are unnecessary. Although there is little evidence to provide guidance, it is suggested that the higher the pre-morbid pCO₂ (inferred by a high admission bicarbonate), the higher the target pCO₂ should be. Recovery from extreme levels of hypercapnia is recognised.105 Any metabolic causes of acidosis, for example, from insulin insensitivity or excessive B2 stimulated glycogenolysis, should be treated separately.

In NMD, an adequate tidal volume can be achieved with relatively low inflation pressures (eg, 10–15), but higher pressure is needed in CWD because of reduced chest wall compliance. Lung recruitment strategies (ie, increasing PEEP) should be considered when there is persisting hypoxia and/or evidence of premature small airway closure in dependent lung tissue. Controlled MV may need to be continued in NMD when triggering is likely to be inadequate or tiring.

Reducing the bicarbonate buffering capacity will require a period of relative hyperventilation when hypercapnia is chronic. The resulting urinary bicarbonate loss resets central respiratory drive. Carbonic anhydrase inhibitors can be used but caution is needed as high doses produce unpredictable effects through central stimulation of breathing.106 ,107

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Guide to initial settings and aims with invasive mechanical ventilation.

Evidence statements

In ARDS, a low Vt strategy improves survival (Level 1+).

In airflow obstruction, prolonging the expiratory time reduces dynamic hyperinflation (gas-trapping) (Level 2+).

Recommendations for IMV in obstructive disease

14. During controlled ventilation, dynamic hyperinflation should be minimised by prolonging expiratory time (I:E ratio 1:3 or greater) and setting a low frequency (10–15 breaths/min) (Grade C).

15. Permissive hypercapnia (aiming for pH 7.2–7.25) may be required to avoid high airway pressures when airflow obstruction is severe (Grade D).

16. Carbonic anhydrase inhibitors should not be used routinely in AHRF. (Grade C).

Positive end expiratory pressure

PEEP is an area of physiology that causes confusion among healthcare professionals. The best way to set optimal PEEP remains contentious. Simply stated, PEEP shifts the lungs to a more compliant portion of the pressure–volume curve. In restrictive causes of AHRF, lung volume is usually reduced and there may be dependent lung that is poorly ventilated or in which there is no effective alveolar ventilation. In these circumstances, increasing external PEEP increases Vt for a given inspiratory pressure, will reduce pCO₂ and improve oxygenation. In obstructive disease, PEEP improves expiratory airflow, limits dynamic hyperinflation and improves alveolar ventilation.108 ,109 Dynamic hyperinflation may be suspected by a progressive fall in tidal volume with constant ventilator pressure settings (or, with volume control, an increase in inflation pressure) and by signs of increasing patient distress such as tachycardia and hypotension.

The degree of intrinsic PEEP (iPEEP) can be estimated by examination of the expiratory flow curve and pressure110 or be measured invasively.111 ,112 Active expiratory muscle contraction, common in airflow obstruction, will artificially increase apparent iPEEP. Levels of iPEEP in obstructive airways disease have been reported to range from 4.6 to 13.6 cm H₂O.113

Setting the PEEP level in excess of iPEEP may be deleterious. This has led to the recommendation that PEEP be set at 50–80% of iPEEP.114 ,115 However, as the severity of airway obstruction in small airways will vary throughout the lung, a variable response to increasing the PEEP might be anticipated. If, on balance, an increase in ePEEP were to reduce overall airway resistance then EELV will fall even though ePEEP apparently exceeds iPEEP.116 ,117

Intrinsic PEEP is a pressure that must be overcome by patient effort before a breath can be triggered. It is, therefore, an inspiratory threshold load and may lead to ineffective triggering and patient discomfort. Offsetting iPEEP by increasing the ventilator PEEP will then reduce the effort of triggering and improve patient–ventilator asynchrony.118–120 It is important to appreciate that the same pathophysiological processes occur during treatment with NIV when a higher EPAP setting may improve triggering, patient comfort and oxygenation.

Evidence statement

In obstructive causes of AHRF, PEEP may increase tidal volume, improve compliance and reduce airflow obstruction (Level 2+).

Setting PEEP greater than iPEEP can be harmful (Level 2+).

In restrictive causes of AHRF, PEEP may assist in lung recruitment, improve compliance and correct hypoxaemia (Level 3).

Recommendation

17. Applied ePEEP should not normally exceed 12 cm (Grade C).

Sedation in IMV

Patients receiving IMV require sedation, especially before stability is achieved.89 Most ICUs use Propofol or a benzodiazepine, either alone or in combination with an opioid. Benzodiazepines with inactive metabolites and/or short acting synthetic opioids have been recommended to avoid over-sedation.121 ,122 Although sedation increases IMV tolerance, over-use is associated with adverse outcomes such as prolonged duration of IMV, increased ICU length of stay and delirium.123

To avoid this, withholding of further sedation until an objective degree of wakefulness develops has been investigated. In two trials, this strategy was shown to reduce duration of IMV and ICU length of stay.124 ,125 Studies employing sedation protocols targeting specific (higher) levels of alertness have also reported a reduction in duration of IMV, ICU and hospital length of stay.126–129 However, a meta-analysis of RCTs on sedation breaks demonstrated safety but failed to confirm benefit,130 and a more contemporary RCT, combining protocolised sedation with daily breaks, also found no benefit.131 No study has shown harm from sedation breaks. The effect of stopping or reducing sedation on patient experience has not been reported.

Evidence statements

Daily interruption of sedation is safe and may reduce the duration of IMV and ICU length of stay (Level 1+).

Sedation protocols that target specific levels of alertness may reduce duration of IMV and ICU length of stay (Level 1+).

Recommendation

18. Sedation should be titrated to a specific level of alertness (Grade B).

Patient–ventilator asynchrony

Patient–ventilator asynchrony is common and increases patient discomfort, the work of breathing, the need for sedation, the incidence of confusion, the need for tracheostomy and the mortality rate.132 ,133 The commonest cause is ineffective triggering due to either respiratory muscle weakness and/or excessive effort required to overcome iPEEP and trigger a breath.134 Trigger failure is more common during sleep and more likely if hypercapnia persists by day. A hybrid mode, such as PS with a mandatory backup rate is recommended in these circumstances to avoid pCO2 increasing during sleep.

Auto triggering refers to inappropriately delivered breaths being provided by the ventilator. It can be provoked by patient movement, suctioning, coughing and swallowing, and is more likely when the trigger sensitivity is set too high. Both a delay in the onset of a triggered breath or an inadequate amount of PS to sufficiently augment inspiratory flow can lead to an unpleasant sensation best described as ‘air hunger’. This can be difficult to detect or for the patient to report. Experienced NIV practitioners may trial increasing trigger sensitivity and/or PS, and monitor the effect on patient comfort and RR. If inadequate PS is given, the breathing rate will fall. The detection of the more subtle forms of patient–ventilator asynchrony requires examination of the pressure/flow waveforms.135 The most sensitive measure of patient–ventilator asynchrony is by simultaneous recordings of diaphragm electrical activity and pressure changes in the oesophagus.134 Flow rather than pressure triggers reduce the incidence of asynchrony,136 ,137 as has the move away from volume-controlled ventilation.138 ,139

Proportional assist ventilation (PAV) and neurally adjusted ventilatory assist (NAVA) are modes that are being assessed as ways to reduce patient–ventilator asynchrony. With PAV, the degree of pressure support is determined, on a breath by breath basis, by the patient's inspiratory effort.140–142 Compared with PS, PAV has been reported to reduce the probability of returning to a controlled mode and the incidence of patient–ventilator asynchrony.143 In NAVA, the ventilator attempts to match neural drive by adjusting the degree of PS (within safe limits), using the electrical activity of the diaphragm to ‘drive’ the ventilator. Studies comparing patient–ventilator interaction show a reduction in triggering delay with NAVA, reduced cycling delay and a reduction in asynchrony events.144 ,145 Uncertainties persist on how to adjust the NAVA level and this technical issue is currently frustrating efforts to demonstrate clinical benefit.

It is important to emphasise that patient ventilator asynchrony is common with NIV. While the same principles apply it has been less frequently recognised or investigated. It can critically affect the success of NIV and the patient experience (see below).

Evidence statements

Patient–ventilator asynchrony is common and deleterious, and can be minimised through informed adjustment of ventilator settings (Level 2+).

Proportional and NAVA have been shown experimentally to reduce ventilator asynchrony but have yet to improve patient outcome (Level 2+).

Recommendations

19. Ventilator asynchrony should be considered in all agitated patients (including NIV) (Grade C).

20. As patients recover from AHRF, ventilator requirements change and ventilator settings should be reviewed regularly (Grade C).

Use and timing of a tracheostomy

It is accepted that translaryngeal intubation beyond 10 days can be detrimental.146 ,147 Historically, it was believed that early tracheostomy reduced ventilator time and ICU length of stay.148 A survey of ICU physicians in 2005 found that 61% of respondents would perform a tracheostomy without first performing a trial of extubation and 50% favoured tracheostomy insertion within the first week.149 Two large multicentre studies have failed to show benefit from tracheostomy performed within 7 days of admission.150 ,151 A subsequent meta-analysis also reported no effect on the incidence of ventilator-associated pneumonia or mortality,152 although less sedation was required after a tracheostomy had been inserted. Tracheostomy carries a morbidity and mortality risk at the time of insertion153 and subsequently.154 A UK national report has highlighted the risk of critical airway incidents in patients with tracheostomies.155 Accordingly, consideration of the risk and benefit should be undertaken before proceeding to insert a tracheostomy and due consideration should be given to using NIV post-extubation to avoid a tracheostomy. This is particularly the case in progressive NMD/CWD when tracheostomy insertion carries the risk of permanence. These aspects, and the evidence summarised below, are considered further in Management of hypercapnic respiratory failure section.

Evidence statement

Early insertion of a tracheostomy does not reduce mortality, duration of IMV, or the incidence of ventilator-associated pneumonia (Level 1++).

Recommendations

21. Performing routine tracheostomy within 7 days of initiating IMV is not recommended (Grade A).

22. The need for and timing of a tracheostomy should be individualised (Grade D).

Good practice points

• In AHRF due to COPD, and in many patients with NMD or OHS, NIV-supported extubation should be employed in preference to inserting a tracheostomy.

• In AHRF due to NMD, alongside discussion with the patient and carers, the decision to perform tracheostomy should be multidisciplinary and should involve discussion with a home ventilation unit.

Obstructive lung diseases

Acute exacerbations of COPD account for 100 000 admissions annually in England. Of these, around 20% will present with or develop hypercapnia,2 ,7 an indicator of increased risk of death.2 ,59 The development of AHRF is often multifactorial. These include infection, mucosal oedema, bronchospasm, sputum retention, excessive O₂ therapy, sedation, pneumothorax, PE and left ventricular failure. Since the publication of the BTS guideline in 20021 and subsequent National Institute for Health and Care Excellence (NICE) recommendations,48 the use of NIV in AECOPD has increased and most hospitals admitting unselected medical patients are able to provide an NIV service.60

Restrictive lung disease

The causes of AHRF include severe chest wall deformity, neuromuscular conditions that affect the respiratory muscles and OHS. Presentation is often with advanced chronic hypercapnia. An insidious decline in health may not have been medically recognised as being due to the development of respiratory failure. Acute presentations, often with infection precipitating acute illness, are likely when the VC is <1 L. Unlike AECOPD, recurrent critical episodes do not preclude intervening good life quality, acceptable health status and prolonged survival. There are no RCTs to guide practice in AHRF and the recommendations presented are extrapolated from the AECOPD literature, from reports of the value of domiciliary NIV (most evidence coming from trials in the more progressive NMDs) and from expert opinion.

Introduction

Weaning is defined as the progressive reduction of ventilatory support leading up to extubation. Delayed weaning complicates 6% of patients managed by IMV but consumes 37% of ICU resources.223 In one study, up to 50% of patients who self-extubated did not require re-intubation,224 implying that many patients are treated with IMV for longer than necessary. Clinical criteria to be met before starting weaning are detailed below:225 ,226

• Adequate oxygenation: PaO₂/FiO₂ ratio >27 kPa (200 mm Hg)

• FiO₂ <0.5

• PEEP <10 cm H₂O

• Adequate alveolar ventilation (pH >7.3, pCO₂ <6.5 kPa).

Fluid balance should also be optimised. The detrimental effect of excess hydration is now recognised in sepsis227 and in acute lung228 and kidney injury.229 A positive fluid balance adversely affects alveolar ventilation, oxygenation, weaning progress and extubation outcome.224 ,230 Brain Natriuretic Peptide (BNP) has been reported to predict failure to wean and correlates with weaning duration; a BNP-directed fluid management strategy has been reported to shorten time to extubation, particularly in patients with left ventricular dysfunction.231

Evidence statements

Easily measured clinical parameters indicate when weaning can start (Level 2+).

Excess fluid administration may delay weaning or contribute to its failure (Level 2++).

In left ventricular dysfunction, a BNP-directed fluid management strategy has been shown to shorten the duration of IMV (Level 2).

Recommendations

59. Treating the precipitant cause of AHRF, normalising pH, correcting chronic hypercapnia and addressing fluid overload should all occur before starting weaning (Grade D).

60. A BNP-directed fluid management strategy should be considered in patients with known left ventricular dysfunction (Grade B).

Weaning methods

Despite several multinational studies, there is no consensus as to the optimal weaning method. Brochard et al232 reported that progressively reducing PS was better than other weaning methods. Subsequent trials have reported that daily (or multiple) T piece trials (SBTs) are as effective as PS weaning.233 ,234 It is likely that patient-specific characteristics are more important than the weaning protocol in determining the duration of weaning. There is agreement that the Synchronised Intermittent Mandatory Ventilation method is inferior to PS and T piece weaning. It is also accepted that a formalised weaning plan, and staff familiarity with the approach adopted on the ICU, are important factors to improve successful weaning.235

Evidence statement

Progressive reduction of PS and daily SBTs are satisfactory methods of weaning (Level 1+).

Recommendations

61. Assessment of the readiness for weaning should be undertaken daily (Grade B).

62. A switch from controlled to assisted IMV should be made as soon as the patient recovery allows (Grade C).

63. IMV patients should have a documented weaning plan (Grade B).

Assessing readiness for discontinuation of mechanical ventilation

SBTs are used to assess readiness to resume normal breathing. During the SBT, a patient breathes with minimal or no PS (defined as <8). A successful trial requires the absence of respiratory distress. Failure of an SBT may be defined by subjective (comfort) or objective (deterioration in gas exchange or measured ventilator parameters) criteria.232 ,233 Studies have shown that the majority of SBT failures occur within 30 min.236 ,237 Repeated failure of SBT should lead to consideration of other methods of weaning.238 ,239

It is important to note that the criteria that define success of an SBT do not necessarily reflect the likelihood of successful extubation. About 10% of patients who successfully manage an SBT will fail to maintain adequate gas exchange and/or develop signs of distress following extubation.240 An SBT assesses the balance of respiratory load to capacity of the respiratory muscles but does not take into account other factors that may affect success such as upper airway patency, bulbar function, sputum load or effectiveness of cough.240

Evidence statement

A SBT is useful in assessing load/capacity but does not predict the success of extubation (Level 1+).

Recommendation

64. A 30 min SBT should be used to assess suitability for extubation (Grade B).

65. Factors including upper airway patency, bulbar function, sputum load and cough effectiveness should be considered prior to attempted extubation (Grade D).

Outcome following extubation

Successful extubation is defined as the absence of the need for ventilatory support for 48 h. Patients receiving post-extubation NIV (see below) are classified as ‘weaning in progress’.241

Much of the evidence regarding the prediction of the risk of post-extubation failure has come from trials of relatively short duration IMV and with a mixture of underlying pathologies.235 ,242 ,243 Several risk factors have been identified. The more adverse factors present, the greater the risk of extubation failure. Risk factors for extubation failure are shown in box 2.235 ,242 ,243

Respiratory distress may occur early or develop later on after extubation. Early failure commonly results from loss of airway patency, for example, from upper airway oedema that becomes evident following removal of the endotracheal tube.244 Patients with NMD are at risk of early extubation failure due to bulbar dysfunction and/or ineffective cough despite a successful SBT. The planned use of NIV and an MI-E following extubation reduces the risk of early failure. Late extubation failure is more complex in aetiology and more than one cause may be present. The causes are summarised below:244

• Capacity–load imbalance: patients with severe airflow obstruction or neuromuscular weakness;

• Impaired bulbar function: aspiration of upper airway secretions, impaired gas exchange and/or obstructed breathing;

• Ineffective cough: typically in NMD/CWD but also in other patients with AHRF;

• Non-respiratory issues—myocardial ischaemia/left ventricular dysfunction, encephalopathy/delirium or severe abdominal distension.

Evidence statement

Patient, clinical and ventilatory factors aid the identification of patients at increased risk of extubation failure (Level 2+).

Recommendation

66. Care is needed to identify factors that increase the risk of extubation failure so that additional support, such as NIV or cough assist, can be provided (Grade B).

Weaning protocols

Weaning protocols that specify the steps to follow during weaning have been claimed to reduce the duration of IMV, increase the success of extubation, reduce unplanned or accidental extubation and reduce the tracheostomy rate, ventilator-associated complications and costs, compared with usual care.245 The studies summarised in this review were, however, not specific to AHRF. Most were performed in the USA, where differences in supervision of patient management exist compared with the UK. There is also marked variation in the weaning methods and protocols between the studies reported. A European study reported that a weaning protocol did not reduce ventilation time.246 Computer-automated weaning, in which adjustment in pressure settings occurs in response to changes in patient parameters, has been compared to professional-led weaning. One multicentre RCT found that duration of weaning was reduced.247 A second study reported no difference in weaning duration between automated weaning and weaning by an experienced nurse.248 There is currently insufficient evidence to support the use of automated weaning over clinical/nurse-led protocols.

Evidence statements

Weaning protocols may reduce the duration of IMV and ventilator associated pneumonia (Level 1+).

There is conflicting evidence regarding the value of computer-automated weaning (Level 1−).

Recommendations

67. Although an organised and systematic approach to weaning is desirable, protocols should be used with caution in patients with AHRF (Grade B).

68. The use of computerised weaning cannot be recommended in AHRF (Grade D).

Use of NIV in the ICU

Appropriate care environments for the delivery of NIV

A study by Roessler and colleagues from Germany randomised 51 patients to either out-of-hospital NIV or standard medical treatment. Out-of-hospital NIV was reported to be feasible, safe and effective.255 A survey of French mobile ICUs also suggests that NIV and CPAP can be safely employed pre-hospital in acute cardiogenic pulmonary oedema but not in other causes of respiratory failure.256 Further evaluation of out-of-hospital NIV in AHRF is required.

NIV is commonly initiated in the ED, but given the other priorities and pressures on emergency resuscitation areas patients should be transferred as soon as practicable to an environment appropriately staffed and equipped to provide on going care. A prospective observational study of 245 patients attending 24 hospital EDs in Australia identified the staff responsible for NIV set-up.257 This was equally distributed between nursing and medical personnel. Hess et al258 conducted a survey of 132 academic EDs in the USA, and concluded that, although NIV was widely available, physician confidence/competence was a barrier to optimal application. A survey of NIV use in UK EDs found a wide variety of practice and suggested the need for a specific ED guide for NIV.259 A pro forma-based COPD management tool, supported by targeted education, was reported to improve ED care including the use of NIV.260

Previous guidelines have recommended limiting the number of areas providing NIV to ensure that staff perform it sufficiently regularly.48 Suitable sites need to be able to provide an NIV service 24/7 and integration with critical care services is essential. If NIV is provided in more than one area within a hospital, protocols and guidelines should be shared.261 Current NICE Quality Standards for COPD include guidance on how to benchmark NIV provision.262 The requirements for an NIV service are summarised in box 3.

For all but the mildest cases, Nava and Hill263 recommend that NIV be delivered in a level 2 facility with enhanced staffing levels. A survey carried out in 1999 found that NIV was provided in level 2/3 facilities in most western European countries.264 In contrast, NIV has been delivered in admission or respiratory ward settings in the UK. This may partially account for the poor performance and high mortality rates associated with use of NIV reported by audits.3–5 The 2009 ICS recommendations265 reiterate level 2 as the appropriate clinical environment for NIV and the 2008 joint BTS, ICS and RCP guide on the use of NIV in COPD with AHRF266 recommends one nurse for every 2 NIV cases, especially during the first 24 h of treatment.

Despite this, in the 2013 BTS NIV audit, 91% of patients were treated on general medical wards despite 43% having pH<7.25.5 This was associated with a low intubation rate and excess mortality. Although the care plan in 21% of cases included IMV should NIV fail, only 3% were intubated (versus an expected rate of 7%). Overall, the audit found that mortality in the AECOPD group was 28% if NIV was delivered in HDU/ICU and 40% if not. With a median pH of 7.24 for the whole patient population (2693 cases), this suggests that some were not being treated for lesser degrees of acidosis, where NIV success is more guaranteed, and that those receiving NIV were not placed in the appropriate care environment given the severity of acidosis.

A number of strategies have been explored to support the effective use of NIV outside the HDU/ICU. Sala et al267 described the practicalities of creating a respiratory intermediate care unit. Paus-Jenssen et al,268 in a Canadian prospective study, used an expert respiratory therapist team to implement NIV across a number of clinical environments. In a similar study, critical care outreach nurses supported NIV delivery elsewhere in the hospital. As a result, mortality was reduced from 57% to 35%.183 A greater number of patients were also identified as suitable for IMV when failing NIV. Some of the challenges of care delivery in this field are highlighted in the National COPD Audit Programme 2014 findings on resources and organisation of care in acute National health Service (NHS) units,269 where only 30% of outreach programmes operated out of hours during weekdays and 59% of respiratory wards reported no level 2 capability. Cabrini reported an Italian prospective study of NIV administered in a non-ICU setting but managed by an anaesthetist-led medical team.270 In 129 consecutive treatments, 10% required intubation and there was a low mortality rate of 12.4%. These reports together suggest that collaboration between admitting teams and the ICU can improve the delivery of care in AHRF.

Hospitalisation with AHRF involves 3 phases—immediate clinical assessment, an assisted ventilation plan when appropriate and the formulation of a future care plan (short term in the event of NIV failure and long term on recovery and discharge or, depending on progress, the provision of end of life care). Figure 3 details key elements and box 4 provides a discharge checklist.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The three phases of patient management in acute hypercapnic respiratory failure.

It has been estimated that an average-sized district general hospital, serving a population of 250 000, should anticipate, depending on local COPD prevalence, up to 100 AECOPD cases requiring ventilatory support per annum. Given the additional causes of AHRF, this probably equates to 150 NIV/IMV cases in most hospitals, and considerably more in areas with high COPD and/or OHS prevalence or those hospitals serving larger populations. NIV facilities should be able to cope with seasonal variation and the increased demand that may occur during influenza epidemics.271

As discussed in the Management section, patient outcomes reported in UK national audits are notably worse than would be expected from trial data and facilities for provision of NIV, and evidence of consultation with the ICU, are frequently limited or inadequate.3 ,4 ,7 Important deficiences that have been identified include delays in commencing ventilatory support, under-recognition of more complex acid–base disturbances, use of inadequate ventilatory pressures, rare use of a different mask when NIV is failing, lack of progression from NIV to IMV and lack of consultation in decision-making. The preponderance for application of NIV in lower level facilities than elsewhere in the world (outlined above) along with evidence of a lack of integration with the ICU,3 ,4 indicates that attention directed at organisational factors are needed and are highly likely to improve patient outcome and experience in AHRF.

NIV facilities need to encompass adequate capacity, and the expertise and associated staffing levels, to deal with complex critically ill patients who have a significant risk of death. To be effective, the NIV service needs to have good operational links to the ICU in the expectation that 10% to 20% of NIV-treated patients should be managed in HDU/ICU and that many will be potential candidates for IMV. The case for a specifically identified and appropriately staffed and equipped area for providing NIV is strongly supported by the evidence. In some European countries, NIV services are provided in a Respiratory Intermediate Care Unit.264

Evidence statements

A care environment with either level 2 or 3 staffing favours a successful outcome from NIV therapy (Level 2+).

Coordination between the ICU and ward areas improves outcome in AHRF (Level 3).

Organisational aspects are pivotal in achieving best outcomes (Level 4).

Recommendations

74. NIV services should operate under a single clinical lead with formal working links with the ICU (Grade D).

75. The severity of AHRF, and evidence of other organ dysfunction, should influence the choice of care environment (Grade C) .

76. NIV should take place in a clinical environment with enhanced nursing and monitoring facilities beyond those of a general medical ward (Grade C).

77. Initial care plans should should include robust arrangements for escalation, anticipating that up to 20% of AHRF cases should be managed in a level 2 or 3 environment (Grade C).

Good practice points

• A 2–4 bedded designated NIV unit (located within a medical high dependency area or within a respiratory ward with enhanced staffing levels) provides a robust basis for the provision of NIV in a DGH serving a population of 250 000 and with an average prevalence of COPD.

• Areas providing NIV should have a process for audit and interdisciplinary communication.

Palliative care and advanced care planning

It is recognised that palliative interventions may be appropriate and yet be provided at the same time as therapies intended to prolong life.272 Accordingly, employing NIV as part of care that aims to relieve distress and has escalation limits may be entirely justified.

Effort is needed to establish patient preferences with respect to intubation or resuscitation status. Momen et al,273 in a systematic review of end-of-life conversations in COPD, found considerable variation among patients in the desire to discuss end of life. Almost 50% of patients did not wish to have such a conversation and there was a preference to wait until the disease was ‘advanced’, with patient perception that this implied the last few days of life. Advance directives/living wills assist healthcare providers in tailoring clinical response and support.274 The importance of actively involving patient/family, especially regarding ‘do not attempt cardiopulmonary resuscitation’ (DNACPR) orders, are highlighted in revised recommendations following a judicial ruling.275 The essential element is that, while patients cannot insist on CPR being performed, the matter should be discussed. Perceived patient ‘distress’—which might be exacerbated by such discussion—is no longer regarded sufficient grounds for not raising the issue. When the risk of causing physical or psychological harm is present, attempts should be made to talk to a healthcare advocate. The enormous challenges in this serve to emphasise the crucial nature of active and ongoing communication strategies. Chakrabarti reported interviews with 50 patients with stable COPD and found that discussion and demonstration of NIV equipment altered future treatment perceptions and willingness to consider an advance directive.276

Sinuff et al277 reported clinician attitudes to NIV in patients with acute respiratory failure and do not intubate/do not resuscitate instructions. While about 60% of physicians considered that NIV should be discussed in this context, 85% of respiratory therapists (those actually administering NIV) felt NIV should be actively promoted. This may reflect a lack of confidence and understanding, among physicians, of the potential for NIV to relieve distress and be effective even in advanced disease. In Denmark, 15% of patients with do not intubate instructions, and who received NIV, survived at least a year with COPD and congestive heart failure the most favourable underlying diagnoses.278

Evidence statements

In advanced disease, care planning should ideally predate acute presentation or commence as early as possible on presentation with AHRF (Level 4).

Health professionals experienced in NIV delivery have a more positive view of the benefit of NIV and perceive patient treatment wishes more postively than do clinicians with less experience of NIV (Level 4).

Recommendations

78. Clinicians delivering NIV or IMV should have ready access to palliative medicine (Grade D).

79. Multidisciplinary advance care planning should be an integral part of the routine outpatient management of progressive or advanced disease and care plans should be reviewed on presentation during an episode of AHRF (Grade D).

80. The use of NIV may allow time to establish patient preference with regard to escalation to IMV (Grade D).

End-of-life care

A questionnaire study of 118 patients with COPD, carried out in Canadian teaching hospitals,279 reported that patients with COPD were less interested in prognosis, CPR, IMV or referral to palliative care than were patients with metastatic cancer. In another study, comparing QoL between patients with advanced COPD and patients with cancer, patients with COPD reported higher levels of physical discomfort with uncontrolled shortness of breath in 78%.280 A recent review of the 4 RCTs that have explored whether NIV relieves dyspnoea in AECOPD concluded that benefit was likely but that study limitations constrained a confident conclusion.281 With regard to physical symptoms, breathlessness and fatigue are dominant in AECOPD. Attention to secretion clearance is an additional major concern in bronchiectasis and CF and for many with NMD. Ability to communicate, to feel safe, to be individually respected and enabled to retain control are common psychological needs.

Patients receiving NIV as ‘ceiling care’, who fail to improve will need appropriate end-of-life attention, including appropriate sedation/relief of distress. It is important that if withdrawal of NIV is decided on, that this is achieved with minimum distress to the patient and their relatives. The BMA guidance on end-of-life care in 2007 did not address withdrawal of assisted ventilation.282 Although withholding and withdrawing are considered ethically equivalent,283 for many individuals, including clinicians, discontinuing mechanical ventilation is felt be emotionally different to, for instance, stopping haemodialysis. This may be because of the immediacy of the consequence.284 A Japanese study reported interviews with 35 critical care physicians and found withdrawing ventilation was regarded differently to stopping other life-sustaining measures because of concern over an abrupt and distressing demise.285 There was a desire for a ‘soft landing’, with a slow and gradual death perceived as ‘natural’. The ATS Clinical Policy Statement of 2007 provides comprehensive guidance on withdrawal of mechanical ventilation, including symptom management of the dying patient. It emphasises that decision-making is a process requiring frequent discussion with patient, family, health professionals and others.286 Pro-active family-centred conferences allow time for families to adjust and provision of literature on bereavement reduces the risk of subsequent emotional morbidity.287

In practical terms, progressive reduction of pressure/backup rate to achieve CO₂ narcosis/coma and an alternative strategy of extubation when intubated or removal of NIV have both been described. In the former scenario, Cox et al288 suggest initial weaning of oxygen over 10 min with appropriate adjustment to opiate or anxiolytic medication. Once patient comfort is assured, it is suggested that mandatory ventilation is withdrawn and PS reduced to zero over 5–10 min. Kuhnlein et al289 conducted structured interviews with 29 families regarding the circumstances of dying in MND patients receiving NIV. Seventeen caregivers described the final stages and eventual death as ‘peaceful’. Eleven of the patients died peacefully while using NIV. Choking sensation was evident in some bulbar patients. The authors indicated that the use of sedatives, anxiolytics and opiates could have been improved, emphasising that palliative care training or support is needed to achieve best practice.

In conclusion, the role of NIV in achieving a ‘good death’ may currently be underutilised and there may be a lack of appreciation that a peaceful death can occur while receiving supportive ventilation.

Evidence statements

The concerns of patients with COPD towards their end of life, centre on high levels of physical symptoms, especially breathlessness (Level 3).

Clinicians often consider withdrawal of assisted ventilation (NIV/IMV) as more challenging than removal of other life support techniques (Level 4).

Good practice points

• Although removal of the NIV mask may be deemed as preferable, a dignified and comfortable death is possible with it in place.

• Clinicians delivering NIV or IMV should have training in end-of-life care and the support of palliative care teams.

Extracorporeal CO₂ removal

The technical aspects of providing prolonged extracorporeal membrane oxygenation (ECMO) or CO₂ removal (ECCO₂R) have advanced in recent years. Both are being increasingly investigated in refractory respiratory failure including AHRF. NICE has issued general guidance on the use of ECCO₂R advising that it should only be used in patients with potentially reversible hypercapnic respiratory failure or those being considered for lung transplantation.290

ECCO₂R uses a gas exchange membrane to provide partial CO₂ clearance, from 30% to 50% of the body's production, depending on blood flow and membrane efficiency. Removing carbon dioxide extracorporeally reduces the native pulmonary minute ventilation required to maintain an acceptable PaCO₂. This offers the potential benefits of either enabling protective mechanical ventilation or providing an alternative to mechanical ventilation in selected patients (such as those with COPD). There is little evidence of clear benefits to patients of ECCO₂R at present. In moderately hypoxaemic ARDS, one RCT demonstrated that lower tidal volumes and ventilatory pressures could be achieved, but this failed to translate into a meaningful improvement in patient outcome.291 Larger studies are planned in the UK and Europe. In patients with COPD, there are no RCTs exploring the role of ECCO₂R. One retrospectively matched cohort study compared outcomes between groups of patients with AECOPD who had an inadequate response to NIV. Twenty five potential candidates who had failed to improve with NIV were compared with historical controls treated in the same hospitals matched by the GenMatch process. Despite significant improvements in acidosis and respiratory distress, the trial failed to show benefit in the primary outcome of need for intubation. The complication rate with ECCO₂R was high (52%) and this contributed to the need for intubation.292

The devices available for ECCO₂R have evolved over time. Early CO₂ removal membranes were pumpless, required arterial and venous cannulation and used the patients own cardiac output to drive blood through the membrane. This resulted in significant shunting of cardiac output and the danger of limb ischaemia. An alternative approach is to take blood from a dual-lumen large bore cannula sited in a central vein and pump it through the membrane. The advantages of the veno-venous technique are principally lack of effect on cardiac output and reduced complications, particularly limb ischaemia.

Evidence statements

Extra-corporeal CO₂ removal devices can reduce PaCO₂ and minute volume (Level 2−).

Veno-venous extra-corporeal CO₂ removal in patients with AECOPD and an inadequate response to NIV has not been shown to reduce intubation rate and is associated with a 52% complication rate (Level 2−).

Recommendations

81. If local expertise exists, ECCO₂R might be considered:

• If, despite attempts to optimise IMV using lung protective strategies, severe hypercapnic acidosis (pH <7.15) persists (Grade D);

• When ‘lung protective ventilation’ is needed but hypercapnia is contraindicated, for example, in patients with coexistent brain injury (Grade D);

• For IMV patients awaiting a lung transplant (Grade D).

Good practice point

• ECCO₂R is an experimental therapy and should only be used by specialist intensive care teams trained in its use and where additional governance arrangements are in place or in the setting of a research trial.

Helium/oxygen ventilation

When mixed with oxygen (Heliox), the lower density of helium reduces resistance in the large airways where flow is predominantly turbulent compared to air/oxygen ventilation. It therefore has a theoretical advantage in obstructive causes of AHRF.293 Heliox has been studied in combination with both NIV and IMV. It increases the delivered dose of bronchodilators and has been reported to improve symptoms and physiological variables in spontaneously breathing asthmatics.294 ,295 At oxygen concentrations >40%, Heliox has no benefit compared with oxygen–air mixtures.296 A large RCT in AECOPD found that Heliox in combination with NIV did not reduce rates of intubation, duration of ventilatory support or mortality.297 Heliox has been reported to reduce pCO₂ and airway pressures in intubated patients with severe asthma298 but a subsequent meta-analysis concluded that it did not affect outcome.299 An uncontrolled study reported that Heliox improved patient comfort in the presence of post-extubation respiratory distress when stridor was present.300

Evidence statement

The use of Heliox does not reduce rates of intubation and length of IMV, nor does it reduce mortality in patients of AECOPD or asthma (Level 1+).

Recommendation

82. Heliox should not be used routinely in the management of AHRF (Grade B).