Now that non-invasive ventilation (NIV) is well established in clinical practice, particularly for chronic obstructive pulmonary disease (COPD),^1,2 it is likely that more patients will survive an acute exacerbation, especially in countries such as the UK where comparatively few patients with COPD are ventilated invasively. However, it is possible that some patients are now just being saved for a future life of poor quality at home, punctuated by recurrent admissions to hospital because their respiratory reserve is so marginal that even trivial exacerbations are sufficient to provoke life threatening ventilatory failure.

Before NIV was widely available, Connors et al³ showed that hypercapnia during an admission with an acute exacerbation of COPD was a poor prognostic indicator. In a prospective study of a cohort of 1016 patients who were admitted with an exacerbation of COPD and a Paco₂ of 50 mm Hg (6.6 kPa) or more, they found that 11% of the patients died during the index hospital stay. The 60 day (20%), 180 day (33%), 1 year (43%), and 2 year (49%) mortality rates were all high; 446 patients (44%) were readmitted 754 times in the following 6 months. At 6 months only 26% of the cohort were both alive and able to report a “good”, “very good”, or “excellent” quality of life. Survival time was independently related to severity of illness, body mass index (BMI), age, prior functional status, Pao₂/Fio₂, congestive heart failure, serum albumin, and the presence of cor pulmonale. Given that current recommendations state that patients with an acute respiratory acidosis (pH <7.35) after initial treatment and a Paco₂ above 6 kPa should be offered NIV,⁴ all patients who have received NIV acutely fall into this poor prognostic group.

LONG TERM OUTCOME FOLLOWING NIV IN HOSPITAL

A number of studies have looked at the longer term follow up after an admission requiring NIV. Overall the prognosis is poor, but patients receiving NIV acutely appear to fare better than those who require endotracheal intubation (ETI) and mechanical ventilation (MV). In their randomised controlled trial comparing NIV with immediate ETI and MV, Conti et al⁵ showed that, in those who could be managed successfully with NIV, there was an advantage both in the short term and also in the year after hospital discharge. There were fewer admissions to hospital and ICU, and fewer patients needed de novo long term oxygen therapy (LTOT). There was also a trend towards improved survival (74% v 54%, p = 0.43). This confirms the findings of two previous studies comparing NIV patients with historical controls who had been invasively ventilated.^6,7 Imperfect matching is one possible explanation in these studies,⁸ but patients who are intubated and mechanically ventilated may lose a considerable amount of muscle bulk rendering them susceptible to further episodes of ventilatory failure.^9,10 Longer term follow up from the study by Plant et al¹¹ failed to show any statistically significant benefit from NIV compared with conventional therapy. It may be significant that few patients in either group were intubated and ventilated and this is an important difference when compared with the studies mentioned above. The study showed a median survival of 13 and 16 months in the conventional and NIV groups, respectively. In a retrospective study of 120 patients who had received NIV acutely, Scala et al¹² found a 6 month mortality rate of 35%; this was greater in those with chronic co-morbidities (54%) than in those without (30%), and was greater in those with low activities of daily living scores.

In this issue of Thorax Chu et al¹³ report their experience on post-discharge outcomes (need for further hospitalisation, recurrent respiratory failure requiring ventilatory support, and death) and the risk factors associated with them in 110 patients ventilated non-invasively for an acute exacerbation of COPD. During the year following discharge 79.9% were readmitted, 63.3% had another life threatening event, and 49.1% died. Survivors spent a median of 12% of days in hospital in the subsequent year. The number of days in hospital in the previous year and a low Katz score (indicating difficulty with activities of everyday living) predicted early readmission; home oxygen use, APACHE II score, and a lower BMI predicted early recurrent acute hypercapnic respiratory failure or death; and the MRC dyspnoea score predicted early death.

USE OF DOMICILIARY NIV

It is clear that patients who have received NIV for an acute exacerbation of COPD are a group at high risk for subsequent hospitalisation and death; those at the most severe end of the spectrum are not surprisingly those at most risk. NIV at home for these patients might improve the long term outcome. Is there any evidence to support this? In a 1 year controlled trial Casanova et al¹⁴ randomised 52 patients with severe stable COPD to either NIV plus “standard care” or to standard care alone. One year survival was similar in both groups (78%) as was the number of acute exacerbations. The number of hospital admissions was less at 3 months in the NIV group (5% v 15%, p<0.05), but this difference was not seen at 6 months (18% v 19%, respectively). There was either no or little difference between the groups in dyspnoea scales, gas exchange, haematocrit, pulmonary function, cardiac function, and neuropsychological performance. However, the number of patients was too small to avoid a type II error. Clini et al¹⁵ conducted a prospective randomised controlled trial of 90 patients randomised to continuing LTOT alone or LTOT with the addition of NIV. There were small improvements in the NIV group in resting Paco₂ and dyspnoea and health related quality of life. There was no improvement in survival or hospital stay, although there was a trend towards less time in hospital in the NIV group compared with an increase in the LTOT group when compared with the period before the study. ICU stay was reduced in both groups, but more in the NIV group than in the LTOT group.

The suggestion of reduced hospitalisation rates (neither study was powered to address this end point) in both these studies is an interesting observation. The level of ventilatory support in both was modest and, while it has been suggested that this may not have been sufficient to control sleep related hypoventilation,¹⁶ ventilator settings similar to those used in these studies offload the respiratory muscles¹⁷ and reduce the sensation of dyspnoea^18,19 associated with an acute exacerbation. It is possible that NIV reduced the impact of exacerbations upon the patient and this may have contributed to the trend towards reduced hospitalisation. In a small group of highly selected patients admitted to hospital recurrently with exacerbations of COPD requiring NIV, Tuggey et al²⁰ showed a reduction in the need for hospital and ICU admission in the year following the introduction of home NIV compared with the year before. This was associated with a reduction in costs even when that of the ventilator, masks, etc was taken into account. However, this study was uncontrolled and the quality of life of the patients was not measured. A placebo effect of NIV cannot be discounted and has been seen with sham CPAP²¹ and sham NIV.²² The placebo effect of a “breathing machine” to a patient who is very short of breath should not be underestimated.

FUTURE ROLE OF LONG TERM DOMICILIARY NIV

There are now three potential roles for long term domiciliary NIV in patients with severe COPD. Firstly, patients who are genuinely intolerant of LTOT because of severe symptomatic hypercapnia may benefit from NIV if hypercapnia is controlled. Secondly, NIV may improve survival and quality of life in patients already established on LTOT but who are also hypercapnic. The published evidence to date does not support this role, but the existing randomised controlled trials can be criticised. A multicentre German study which aims to randomise a total of 300 patients to NIV or standard medical treatment with all cause mortality as the primary end point may help to answer this question.²³ Thirdly, and perhaps most promisingly, NIV may have a role in patients who have required it because of a severe exacerbation. As detailed above, these are patients with a poor prognosis and the study by Chu et al¹³ provides useful further data to inform the power calculation of a prospective randomised controlled trial with the important end points of survival and hospitalisation rates. The drop out rate from the NIV group should be low as patients randomised to the intervention will have had personal experience already; those who do not like it or cannot tolerate it are likely to refuse home ventilation and therefore randomisation to the study. Patients may not need to use NIV during sleep²⁴ or every day²⁵ to gain benefit. Diaz et al²⁵ in a physiological study of 36 patients comparing NIV with sham ventilation delivered during wakefulness for 3 hours per day 5 days per week over a period of 3 weeks showed impressive changes in diurnal arterial blood gas tensions. This study suggests that even short periods of NIV during the day may have an effect in patients with ventilatory failure due to COPD.

The stage is now set for a randomised controlled trial comparing domiciliary NIV with conventional treatment in patients who have been ventilated non-invasively during an acute exacerbation. As well as survival, quality of life and a detailed health economic analysis are needed. Patients should be encouraged to use the ventilator each night during sleep but, if they are unable to do this, daytime use may still be beneficial. Even those whose use is only intermittent may still derive benefit, both in terms of daytime function²⁵ and also by reducing the impact of the excacerbation. If benefit is confirmed, monitoring the pattern of use—which is now easy with microprocessor technology—will provide important insights into how this intervention might help patients.