Discontinuation of ventilation is estimated to take up to 40% of the total duration of ventilatory support, and around 3–6% of patients admitted to the intensive care unit (ICU) require a prolonged course of mechanical ventilation (MV).¹ Patients being liberated from ventilatory support therefore occupy a significant number of ICU beds and have a major impact on healthcare resources. There have been several recent key developments in the field of weaning—the use of weaning protocols, ventilatory strategies to reduce the need for invasive ventilation and facilitate successful extubation, and the creation of regional long term ventilator units. All have the potential to affect weaning outcome, but how valuable are they in practice?

WEANING PROTOCOLS

In 1996 Ely and colleagues² showed that the implementation of a standardised protocol of daily trials of spontaneous breathing performed by nursing staff reduced the total duration of MV from 6 to 4.5 days, and complications such as need for reintubation, tracheostomy, and duration of MV >21 days were also decreased, resulting in a reduction in ICU costs. Similar protocols have reduced the duration of MV, although not necessarily ICU stay.³ Smyrnios et al⁴ implemented a hospital-wide weaning protocol and found a decrease in the need for tracheostomy by a third and a reduction in mean hospital stay from 37.5 to 24.7 days, resulting in a 30% fall in cost per case.

Despite these findings, weaning protocols have not been taken up universally. For example, in a survey of ICUs in England commissioned by the Department of Health, protocol directed weaning was reported in less than one in five units.⁵ This may be due to a variety of reasons including differences in healthcare practice and cultures, and the fact that the findings may not be universally applicable. Randolph et al⁶ compared a weaning protocol with standard care (no defined protocol) in infants and children with acute illnesses requiring MV and found that, in contrast to adult patients, the majority of children were weaned within 2 days and the weaning protocol did not influence the duration of MV. Furthermore, a recent controlled trial⁷ of adults requiring MV for >24 hours showed no difference in duration of MV, ICU stay, need for re-institution of MV, or hospital mortality in the group treated with a nursing/respiratory therapist driven protocol compared with those in whom weaning was managed off protocol by the supervising physician.

Does this latest study mean that weaning protocols are unnecessary and unhelpful? Almost certainly not. Standard practice evolves by physicians incorporating examples of best practice and research findings into their day to day care—as Tobin has noted,⁸ the issue is not what is wrong with protocol directed care but what is right with standard management. It is also important to note that this study⁷ with a negative outcome was carried out in a closed intensivist run ICU with high levels of staffing, and a routine management template was used to encourage staff to address weaning issues each day. So a fairly comprehensive “protocol” was in place anyway. It follows that protocols and guidelines may drive up the standard of routine care, especially in open ICUs, and should be evaluated and adapted to the site they are operating in to improve and update current management pathways.

NEW VENTILATORY AND EXTUBATION STRATEGIES

There is now a substantial body of evidence confirming that the application of non-invasive ventilation (NIV) in acute excerbations of COPD can prevent the need for intubation, such that NIV should be available on a 24 hour basis in units managing patients with acute respiratory failure.^9–12 For intubated patients who have failed a spontaneous breathing trial for 30 minutes to 2 hours, there is little point in repeating the trial before 24 hours has elapsed. During this period patients should receive optimum ventilatory support in assist mode to allow some muscle activity and patient control over the ventilation delivered.¹ Brochard et al¹³ have shown that gradually reducing pressure support ventilation is superior to intermittent mandatory ventilation or progressive spontaneous breathing trials in reducing weaning time. Early extubation onto NIV has been shown to be effective in patients with COPD. Nava et al¹⁴ found that non-invasive pressure support ventilation reduced weaning time, shortened ICU stay, decreased the incidence of nosocomial pneumonia, and improved 60 day survival in COPD patients extubated on to NIV 48 hours after admission with a severe hypercapnic exacerbation compared with continued standard invasive ventilation. In addition, in a heterogeneous group of patients with persistent weaning failure randomised to either NIV or continued invasive ventilation, the NIV group had a shorter total period of ventilation, reduced ICU stay (14.1 v 25 days), and reduced hospital stay (27.8 v 40.8 days) (p = 0.026).¹⁵ The need for a tracheostomy was markedly reduced (1.5% v 13.5%) and the conventional (invasive) weaning approach proved to be an independent risk factor for decreased ICU and 90 day survival (odds ratio 6.6, p = 0.035). Indeed, the results were so clear cut the trial was halted after planned interim analysis.¹⁵ These findings suggest that NIV should be considered early in the process of weaning, before a tracheostomy is performed unless there are specific indications for this (such as upper airway obstruction, severe bulbar weakness). By the same token, NIV has increasingly been used in patients who develop post extubation respiratory failure. Here one should apply a note of caution—while a historical case control study in patients with COPD showed that use of NIV reduced the need for reintubation from 67% to 20%,¹⁶ two recent prospective randomised trials^17,18 have shown no advantage to the use of NIV in post extubation respiratory failure. In fact, Esteban et al¹⁸ showed an increase in mortality in patients treated with NIV compared with those receiving standard management. There are several possible explanations for this discrepancy. Firstly, in the trial by Esteban et al¹⁸ only 10% of patients had COPD while the remainder had a variety of conditions including pneumonia, postoperative respiratory failure, trauma, cardiac failure, and ARDS. Most work suggests that NIV is more effective in patients with COPD than in those with acute hypoxaemic normocapnic respiratory failure. Secondly, the median time from extubation until reintubation was longer in the NIV group (12 v 2.5 hours), suggesting that use of NIV may delay reintubation, thereby affecting outcome adversely.

Furthermore, spontaneous breathing trials provide an overall guide to ventilatory capacity but signify little regarding cough efficacy and the ability to clear bronchial secretions if an endotracheal tube or tracheostomy is in situ. Salam et al¹⁹ evaluated the extent to which cough efficiency (measured by cough peak flow), neurological function, and the volume of endotracheal secretions affects the outcome of extubation in patients who had passed a spontaneous breathing trial. A cough peak flow of less than 60 l/min increased the likelihood of extubation failure nearly fivefold, and the combination of low cough peak flow, volume of secretions >2.5 ml/hour, and failure to respond to simple commands produced an extubation failure rate of 100% compared with only 3% in those without these risk factors. This study¹⁹ was performed in 88 routine ICU admissions with diagnoses including pneumonia, COPD, congestive heart failure, asthma and sepsis, and there was no particular emphasis on neurological or neuromuscular patients. Cough efficacy, neurological status, and the ability to clear secretions are likely to assume even greater importance in patients with neuromuscular weakness, and the results suggest these assessments (including evaluation of bulbar and swallowing function) should be added to spontaneous breathing trials in neuromuscular and neurological groups. Additional strategies such as a combination of NIV and cough insufflator/exsufflator devices^20,21 may be valuable in these patients and reduce the need for tracheostomy ventilation.²² Clearly, in neuromuscular patients with a history of gradual decline before acute ventilatory decompensation, or a progressive condition such as amyotrophic lateral sclerosis/motor neurone disease, failed trials of spontaneous breathing should not be fruitlessly and demoralisingly repeated but should prompt early referral for consideration of long term ventilatory support.

WEANING/LONG TERM VENTILATOR UNITS

Having applied strategies to optimise the probability of weaning success, what can be done for patients who fail to wean simply or who are likely to need long term ventilatory support, and how many individuals fall into this category? Various definitions have been used, but weaning delay can be considered to be the need for ventilatory support for more than 2 weeks in the absence of any non-respiratory factor preventing weaning, and the term weaning failure is used if this persists for 3 weeks or more. In the USA it has been estimated that there are over 11 000 ventilator dependent patients in acute care facilities, costing over $9 million a day.²³ A 1 year survey²⁴ in the Northern region of England identified 161 patients with weaning delay; these patients comprised 2.5% of ICU admissions and occupied 6% of ICU beds in the region. A subsequent NHS Modernisation Agency point prevalence survey of critical care facilities in England published in 2002 showed that approximately 8% of ICU patients had weaning delay and 7% weaning failure.⁵ The most common reasons for weaning failure were chronic lung disease, cardiac impairment, neurological disease, or neuromuscular disease. As a result of these findings, the NHS Modernisation Agency has recommended the creation of a specialist NIV service integrated into the critical care network and the provision of a UK-wide service for long term invasive and non-invasive respiratory support for patients who have failed to wean.⁵ The activity of one such unit is described comprehensively in this issue of Thorax by Pilcher et al²⁵ who present the outcome from a specialised weaning programme over 4 years. Of approximately 150 weaning delay patients who had received MV for around 20 days before transfer to the unit, 38% were weaned completely from ventilatory support, 35% required home ventilation, and 27% died before leaving hospital. Survival was best in patients with neuromuscular disease (who, conversely, were most likely to remain ventilator dependent) and worst in postoperative patients. Length of ICU stay before transfer, age, and APACHE II score on admission were key predictors of outcome. Female sex was associated with an increased probability of weaning, but this finding may be partly a consequence of the high likelihood of Duchenne patients (male) requiring home ventilation. The economic analysis is helpful, but further detailed work is required in this area.

In an earlier study, Smith and colleagues²⁶ found a survival rate of 90% compared with a predicted survival from APACHE II score of 53% in 40 consecutive admissions to a regional weaning unit. Nearly 30% of these patients required home ventilation—but only three via tracheostomy. To set this within a European perspective, outcome data²⁷ from a German regional unit accepting patients with weaning delay showed 60% of referrals had COPD, mortality was 24%, and 31% were discharged using NIV. Here, too, the survival rate was related to the underlying diagnosis—patients with thoracic cage and neuromuscular disorders faring better than those with COPD.

In the USA there is a longer tradition of post ICU care delivered in long term facilities, and this has been substantially finance driven. A growing number of case series reports of patients with weaning failure has shown a common pattern—average age on admission of around 70 years, predominant diagnoses of COPD or postoperative cases, with approximately 60% surviving to discharge from the facility, 30% surviving at 1 year, and around half weaned completely from ventilatory support.^28–31 A multicentre study of 23 units commissioned by the US National Association of Long Term Hospitals has recently been set up to identify characteristics of the population at risk, treatment outcome, and estimate costs of care.

The advantages and disadvantages of weaning/long term ventilator units should be considered. Results from the European and US case series would suggest that patients whose primary problem is ventilatory dependence can be managed in a less intensive step down unit, thereby reducing costs and facilitating a focus on ventilatory care and rehabilitation. ICU facilities are freed up for patients requiring more complex care such as those with multisystem failure. Survival improves and favourable 1 and 3 year results can be obtained in some subgroups, particularly patients with neuromuscular disease, although outcome remains relatively poor in the elderly and those with COPD, and families have to travel longer distances to visit patients. These centres tend to have greater familiarity with NIV techniques and are able to provide families and carers with competency training in tracheostomy and ventilatory care, and to develop comprehensive home care packages to speed discharge. However, the outcome from weaning units critically depends on the selection criteria used for admission and can be biased if high risk admissions are refused. In many reports acceptance criteria are not made explicit. In addition, results may be influenced by the fact that, as in the study by Pilcher et al, long term units are likely to accept patients from their own or local ICUs more swiftly, thereby improving outcome.

Assuming, however, that targeting resources in this manner is rational, how many long term ventilatory places are required? The Northern region of England survey²⁴ included 112 ICU (level 3) beds and suggested that a seven-bed weaning unit would be able to cater for 93% of patients with weaning delay, while a five-bedded unit would cope with 75% of cases of weaning delay—that is, a ratio of around one bed per 20 ICU beds. When planning the allocation of beds it should be recognised that in some countries many weaning failure patients are colonised or infected with methicillin resistant Staphylococcus aureus (MRSA) as a result of prolonged hospital stay and so will require single cubicle areas. These units are often most usefully placed alongside or combined with high dependency/respiratory intermediate care facilities,^32,33 allowing flexible bed use, and can act as a focus for training in NIV and long term ventilation—not least as subspecialty training in ventilatory support is being considered as part of standard respiratory medicine/pneumology training programmes in Europe.