Guideline on the management of acute chest syndrome in sickle cell disease.

Abstract

Sickle cell disease (SCD) affects 12 000–15 000 individuals in the UK. Whilst homozygous SCD (sickle cell anaemia — HbSS) is the most common and severe genotype, and is where most of the evidence exists, this guidance should be used for all genotypes of SCD. Acute chest syndrome (ACS) is defined as an acute illness characterized by fever and/or respiratory symptoms, accompanied by a new pulmonary infiltrate on chest X-ray (Charache et al, 1979; Ballas et al, 2010). This definition encompasses cases both where an infective organism is isolated and where no infective cause is identified. It is unique to SCD but in some cases ACS may appear to be similar to bacterial pneumonia in a patient without SCD. ACS may have a severe clinical course and can progress rapidly from mild hypoxia to respiratory failure and death. The presence of hypoxia is not included in the definition, but in clinical practice, hypoxia is a useful predictor of severity and outcome (Vichinsky et al, 1997, 2000). Historically, ACS is one of the most common causes of death in patients with SCD (Platt et al, 1994; Lucas et al, 2008), although mortality is improving with improved medical management (Fitzhugh et al, 2010). ACS can also be associated with significant morbidity, including long-term parenchymal lung damage, pulmonary vascular abnormality and neurological sequelae. Patients may present to hospital acutely unwell with ACS or ACS may develop during a hospital admission following a painful crisis or post-operatively. This guideline will be useful for clinicians working in both high and low prevalence areas and may serve as the foundation for clinical governance measures that can be introduced in individual units with the aim of improving care for patients with SCD. It can be used to inform local pathways and protocols, and we recommend that all acute healthcare organizations should have a protocol for management of this condition (Sickle Cell Society, 2008). In specialist centres such pathways and protocols will be developed locally, but smaller centres should consider linking with a nearby specialist centre and using a shared protocol. Units in acute healthcare organizations in low prevalence areas who may only rarely be confronted with ACS should have clear guidance on: (i) which specialist centre to seek advice from, (ii) criteria for transfer to the specialist centre and (iii) criteria for transfer to a critical care intensive care unit (ICU) (level 3) or high dependency unit (HDU) (level 2). The guideline group was selected by the British Committee for Standards in Haematology (BCSH) to be representative of UK-based medical experts. MEDLINE and EMBASE were searched systematically for publications from 1966 until April 2013 using a variety of key words. The writing group produced the draft guideline, which was subsequently considered by the members of the General Haematology Task Force of the BCSH. The guideline was then reviewed by a sounding board of approximately 50 UK haematologists and members of the BCSH and British Society of Haematology Committee, patient representatives and other interested parties. Comments were incorporated where appropriate. The ‘GRADE’ system was used to quote levels and grades of evidence, details of which can be found in Appendix I. Criteria used to quote levels and grades of evidence are as outlined in the Procedure for Guidelines Commissioned by the BCSH (Appendix I). Acute chest syndrome refers to a spectrum of disease from a mild pneumonic illness to acute respiratory distress syndrome and multi-organ failure. The initial insult, which may be pulmonary infection, fat embolism and/or pulmonary infarction, causes a fall in alveolar oxygenation tension, which causes HbS polymerization. This, in turn, leads to decreased pulmonary blood flow that exacerbates vaso-occlusion, producing more severe hypoxia such that a vicious cycle of hypoxia, HbS polymerization, vaso-occlusion and altered pulmonary blood flow ensues. There is evidence that VCAM1, which is up-regulated by hypoxia and fat embolism, has a role in the development of ACS as an inducer of red cell adhesion It contributes to increased adherence of red cells to the respiratory endothelium and increased vaso-occlusion within the pulmonary circulation, leading to increased hypoxia (Stuart & Setty, 1999). Acute chest syndrome may occur secondary to infection, and infectious organisms were identified in 38% of cases who underwent detailed investigations including blood culture, nasopharyngeal sampling for viral culture, sputum culture and serum samples for antibody response and bronchoscopy (Vichinsky et al, 2000). The identification of a specific infectious organism is less likely with standard investigations alone. An infective aetiology is more common in children than in adults and shows seasonal variation in children, being three times more common in winter (Vichinsky et al, 1997). Viral infection is the commonest cause of ACS in children under 10 years of age. The most common bacterial organism identified in adults is Chlamydophila pneumoniae and in children is Mycoplasma pneumoniae and the commonest virus identified is the respiratory syncytical virus (RSV). Mycoplasma has been isolated in 12% of episodes in children under 5 years of age, 14% of episodes in children aged 5–9·9 years but only 3% of episodes in the 15 years and over age group (Neumayr et al, 2003). Staphylococcus aureus, Streptococcus pneumoniae, Haemophilus influenzae and respiratory viruses other than RSV are also seen (Vichinsky et al, 2000; Dean et al, 2003; Neumayr et al, 2003). The prevalence of both viruses and atypical bacteria as common causes of infection suggest that the clinician must carefully consider the antimicrobial agents prescribed, which should provide coverage against atypical bacteria. The clinical course of ACS is significantly different to that of infectious pneumonia in non-sickle individuals, which is probably because of the damaged microvasculature of the lung in individuals with SCD (Ballas et al, 2010). Acute chest syndrome may also be caused by fat embolism. During a painful crisis, vaso-occlusion within the bones leads to bone marrow necrosis and the release of fat emboli. These enter the blood stream and lodge in the pulmonary vasculature causing acute hypoxia. Evidence of fat emboli has been shown on autopsy studies, and fat-laden macrophages have also been found in bronchoalveolar fluid and in induced sputum (Vichinsky et al, 1994). In situ microvascular occlusion and pulmonary infarction can also be associated with ACS and it can be secondary to hypoventilation, causing pulmonary atelectasis, hypoxia and pulmonary intravascular sickling. Microvascular pulmonary infarction must be distinguished from pulmonary embolism, which can present with chest pain and tachypnoea but without a new infiltrate on chest X-ray. Whilst this group of patients have a hypercoagulable state and are at an increased risk of pulmonary embolism, the clinical picture is usually distinct from ACS. Severe bony pain from rib infarcts can lead to splinting and regional hypoventilation in the areas of pain (Rucknagel et al, 1991; Gelfand et al, 1993). Alveolar hypoventilation can also occur secondary to opiate narcosis or in the post-operative period, following a general anaesthetic (Vichinsky et al, 2000). A pre-existing diagnosis of asthma has been shown to be associated with an increased incidence of ACS in children (Boyd et al, 2006). A cohort of 291 infants followed by the Cooperative Study of Sickle Cell Disease for up to 20 years found that 16·8% had asthma and those with asthma has almost twice as many episodes of ACS (0·39 episodes per patient year vs. 0·20 episodes per patient year; P < 0·001) (Boyd et al, 2006). Those with SCD and asthma were younger at their first presentation with ACS with a median age of 2·4 years (Boyd et al, 2006). Bronchospasm and wheeze are common in the context of ACS, and may contribute to local hypoxia. They are most frequently seen in children and commonly have a viral aetiology. Acute chest syndrome is defined as an acute illness characterized by fever and/or respiratory symptoms, accompanied by a new pulmonary infiltrate on chest X-ray. Severe hypoxia is a useful predictor of severity and outcome (1B). Acute chest syndrome has a multifactorial aetiology and an infective cause is common and this should be considered in treatment algorithms (1B). Acute chest syndrome is the second most common reason for hospitalization in SCD and is a leading cause of morbidity and mortality. The clinical features of ACS may not be evident at the time of admission. Nearly half of patients present initially with a painful vaso-occlusive crisis and then develop this complication whilst in hospital. ACS will often develop 24–72 h after the onset of severe pain (Gladwin & Vichinsky, 2008). All patients admitted with painful crisis should be considered to be potentially in the prodromal phase of ACS. Additionally, ACS can develop post-operatively, especially following abdominal surgery and in patients not given a pre-operative blood transfusion (Howard et al, 2013). Patients with minimal clinical features may become critically unwell rapidly and vigilance throughout admission for the development of ACS is mandatory, particularly in patients presenting with rib or chest pain. Regular monitoring of vital signs and at least daily chest examination are essential. Often, the clinical diagnosis is sought when a patient is found to be hypoxic. The most common respiratory symptoms of ACS are cough, chest pain and shortness of breath. Chest pain may be pleuritic in nature and cough productive. Rib and sternal pain, chills, wheezing and haemoptysis may also occur. The clinical features vary depending on age. Young children more often present with fever, cough and wheeze whereas pain and dyspnoea are predominant features in adults. Pain is a less prominent feature in children (Table 1). Clinical signs often precede the chest X-ray findings. In addition to the rather general nature of these signs, which could indicate a primary bacterial or viral pneumonic process, the chest examination may be normal. It is important if this is the case, that the diagnosis is not excluded at this stage. If there is strong clinical suspicion, close monitoring should be undertaken and, perhaps, therapy initiated. A normal chest examination is more likely to be encountered in children. Clinical features overlap with those of pneumonia in a patient without SCD. ACS is often precipitated by infection but treating ACS as a purely infective episode may lead to progression and rapid clinical deterioration. The severity of ACS is variable, ranging from a mild illness to a severe life-threatening condition. In addition to the age-dependent variation of clinical presentation, the age of the patient also influences severity. Although ACS is more common in children (21 events per 100 person years in children vs. 8·7 events per 100 person years in adults) it tends to follow a milder course with infection frequently implicated in the aetiology. In contrast, ACS in adults tends to be a more severe illness marked by severe hypoxia, a higher requirement for transfusion and higher mortality. It can be considered as a form of acute lung injury that can progress to acute respiratory distress syndrome progressing, albeit infrequently, to acute multi-organ failure. Acute chest syndrome can be severe in all sickle genotypes, with similar death rates per event in HbSS and HbSC (Vichinsky et al, 1997). All patients should be treated aggressively, irrespective of their sickle genotype. Hospital stay is often shorter in children (5 d vs. 10 d in adults) (Vichinsky et al, 1997). Recurrence is a feature and some patients have multiple episodes with previous episodes increasing the likelihood of further similar events. All patients, in particular, those with chest signs and symptoms can progress rapidly during ACS to acute hypoxic respiratory failure and therefore regular SpO2 monitoring is essential. Predictors of acute respiratory failure include extensive lobar involvement and a history of cardiac disease (Vichinsky et al, 2000). Mechanical ventilation has been reported as necessary in up to 20% of cases and it is more likely to be required in older patients. In the large multicentre National Acute Chest Syndrome Study, mechanical ventilation was provided for 13% of patients, of whom 81% recovered (Vichinsky et al, 2000). Neurological features, such as altered mental status, seizures and strokes, may be associated with ACS. Patients with neurological symptoms more often progress to acute respiratory failure and have a significantly higher mortality compared to those without neurological features (Vichinsky et al, 2000). A recent history of ACS is a risk factor for overt stroke, silent stroke and posterior reversible encephalopathy syndrome in children (Ohene-Frempong et al, 1998; Henderson et al, 2003). An acute drop in haemoglobin concentration with an associated increase in markers of haemolysis prior to the onset of ACS is common. Reported falls from steady state haemoglobin values have varied from 7 g/l for all genotypes (Vichinsky et al, 1997), to 16–22·5 g/l depending on genotype (Maitre et al, 2000). A greater fall in haemoglobin values has been documented for cases with bilateral lung involvement than for unilateral disease (28·3 g/l vs. 13·3 g/l) (Davies et al, 1984). Acute chest syndrome remains a leading cause of premature mortality in SCD. In a recent national survey in the United Kingdom, ACS was the third most common cause of death reported in adults (Lucas et al, 2008). It is a recognized risk factor for early death in HbSS patients above the age of 20 years (Platt et al, 1994). Respiratory failure is the most common cause of death. Other causes of death in these patients included pulmonary haemorrhage, cor pulmonale, overwhelming sepsis and cerebrovascular events. Mortality rates in ACS will be dependent in part on the appropriateness of medical management. Even with good medical treatment, overall mortality rates of up to 3% are reported, with the overall death rate in adults being four times higher than in children (Vichinsky et al, 1997, 2000). The diagnosis of ACS is typically straightforward when a high level of clinical suspicion is combined with the usual clinical features. In this scenario, the appearance of consolidation on a plain chest X-ray in a sickle cell patient with recent onset hypoxia, tachypnoea, chest signs, fever and chest pain is diagnostic (Vichinsky et al, 2000). However, the diagnosis can be difficult because (i) adult patients are often afebrile (Vichinsky et al, 1997); (ii) clinical features may be few (Davies et al, 1984); (iii) hypoxia is difficult to determine on clinical examination unless severe (Vichinsky et al, 1997) and (iv) the radiological signs often lag behind the physical signs (Charache et al, 1979; Davies et al, 1984; van Agtmael et al, 1994). Standard patient monitoring of patients with SCD admitted with acute pain crisis should include at least 4-hourly SpO2 monitoring (on air), heart rate monitoring and blood pressure monitoring. SpO2 should be measured on room air to enhance the sensitivity and specificity of the test to detect significant hypoxia and this is particularly helpful if the patient is located in a general ward environment. Daily clinical examination is also required, which should be more frequent if there is clinical concern. The investigations required for the diagnosis and management of ACS are standard tests available in acute district general hospitals. It should be stressed that clinical suspicion is vital to early diagnosis. Typical findings on plain chest radiography in ACS are segmental, lobar or multilobar consolidation usually involving the lower lobes, or collapse with or without pleural effusion (Leong & Stark, 1998; Madani et al, 2007). Children are less likely to have pleural effusions, but are more likely to have upper or middle lobe disease (Davies et al, 1984; Vichinsky et al, 1997, 2000; Maitre et al, 2000). Radiological signs are often absent early in the illness (Charache et al, 1979; Davies et al, 1984; van Agtmael et al, 1994), or if present, may underestimate the severity of hypoxia (Bhalla et al, 1993), and treatment must not be delayed because the changes on chest X-ray are unremarkable. If initial chest X-ray is normal, a repeat chest X-ray should be performed if there is ongoing clinical suspicion. In the situation of unexplained hypoxia (no clinical features and no radiological signs on chest X-ray), a computerized tomography (CT) scan of the pulmonary arteries should be considered. This will provide clinical data on the pulmonary vasculature as well as the lung parenchyma. An acute fall in haemoglobin concentration and platelet count are often seen in ACS. These are markers of disease severity. A decreasing platelet count to less than 200 × 109/l is an independent risk factor for neurological complications and the need for mechanical ventilation (Vichinsky et al, 2000; Velasquez et al, 2009). Measurement of reticulocytes will confirm adequate bone marrow function and exclude red cell aplasia (erythrovirus B19 infection). Patients with ACS are at risk of developing multi-organ failure as a result of systemic fat emboli and require monitoring of their renal and liver function. C-reactive protein (CRP) levels, if elevated, may be used to follow the clinical progress of patients with ACS. If there is evidence of co-existent infection, the sensitivity and specificity of CRP levels as a biomarker of vaso-occlusion is reduced, but an increasing CRP is still a useful marker of a worsening clinical condition. Blood must be urgently ABO-, Rh- (C, D and E) and Kell-typed with antibody screening for all sickle cell patients at increased risk of ACS (severe vaso-occlusive crisis, previous ACS), and for those with mild ACS as their clinical condition may suddenly deteriorate and require emergency transfusion. Red cell alloimmunization is a common problem in SCD patients and can pose significant difficulties in their management (Vichinsky, 2001). Blood for transfusion should be fully matched for Rh (C, D and E) and Kell type and should be sickle-negative. If the patient is not known to the hospital, they should be asked if they are carrying an antibody card with details of their latest alloantibody profile and any previous alloantibodies detected. In addition every effort must be made to obtain the red cell phenotype and latest red cell alloantibody profile from the hospital where their usual care is provided. This is especially important because red cell alloantibodies often become undetectable over time in sickle cell patients (Vichinsky, 2001), thereby increasing the risk of delayed haemolytic transfusion reaction if antigen positive blood is inadvertently transfused. It must be remembered that in the UK red cell units for sickle cell patients will usually have to be ordered in from the National Blood Service, which will increase the time it takes to make the blood available for the transfusion. Therefore in requesting blood for emergency transfusion, it is best to liaise directly with the blood transfusion staff, informing them that the blood is required for a sickle cell patient in addition to providing relevant details on the crossmatch request form. This must be performed in all patients with ACS who are febrile although the yield is low (Vichinsky et al, 1997). Although pulse oximetry can be used to provide a reliable estimate of arterial SpO2 in patients with sickle cell anaemia (Ortiz et al, 1999; Fitzgerald & Johnson, 2001), ABG measurement is the gold standard in determining the partial pressure of oxygen and partial pressure of carbon dioxide in adults with suspected ACS. An ABG on room air is useful for making the diagnosis of ACS, assessing severity and for guiding decisions about the need for HDU/ITU admission and for urgent blood transfusion and is the investigation of choice for this reason in patients with SpO2 of ≤94% on room air, even if they have higher SpO2 with oxygen therapy. Patients in clear respiratory distress or in whom SpO2 falls rapidly to <85% when oxygen is removed need immediate medical review and escalation of therapy and ABGs on room air are unlikely to be needed to confirm a diagnosis or influence therapy. If samples for ABG measurement are taken whilst the patient is breathing oxygen then this must be acknowledged and the arterial-alveolar (A-a) gradient calculated to determine the degree of shunting and alveolar hypoventilation. This is unlikely to be feasible outside the HDU/ITU setting, but may be useful to guide the need for escalation of respiratory support. Arterial puncture is often very distressing to children and thus pulse oximetry is the mainstay of monitoring oxygenation. Venous or capillary samples give similar values to ABG in terms of pH, PaCO2, base excess and bicarbonate but poor correlation with PaO2 (Yildizdaş et al, 2004). A lower than normal PaO2 (11 kPa) on room air is common in clinical studies, but the prevalence varies with some studies reporting a prevalence of around 70% of cases (van Agtmael et al, 1994; Bernini et al, 1998). The mean PaO2 of patients with ACS was 9·3 kPa (70 mmHg) and 9·2 kPa (69 mmHg) in two large studies (Vichinsky et al, 2000; Fartoukh et al, 2010), and in another study, nearly one-fifth of patients had a PaO2 less than 8·0 kPa (Vichinsky et al, 1997). In the majority of studies PaCO2 was in the normal range with a mean value of 4·7 kPa being recorded in 141 of 252 of first ACS events in adults (Vichinsky et al, 1997). However, an unexpectedly high prevalence of hypercapnia (46%) was reported in one study among adult patients with no evidence of opioid abuse (Maitre et al, 2000). We would suggest that a PaO2 less than 8 kPa on room air should be accepted as severe hypoxia and a PaCO2 of greater than 6 kPa as hypercapnia. Acute and convalescent samples for antibodies against atypical respiratory organisms, including M. pneumoniae, Chlamyophila pneumoniae and Legionella should be sent, if feasible, and for erythrovirus B19 if indicated by the clinical and haematological features (Vichinsky et al, 2000). Mycoplasma infection may be suggested by red cell agglutination on a stained blood film and the presence of cold agglutinins in serum. The usual screening test for Legionella is by testing for the urinary antigen. PCR/Immunofluorescence testing for viruses in sputum and nasopharyngeal aspirate are generally performed as a virus panel for a range of viruses. This may include testing for influenza A (including H1N1 subtype), influenza B, metapneumovirus, adenovirus, parainfluenza and RSV but will be dictated by the clinical picture and local microbiological advice. Identification of specific infective organisms will help to guide antibiotic prescription and appropriate use of isolation facilities. Although a number of other investigations have been reported in the literature, these are not routinely used in the diagnosis and management of ACS in the UK. In a study of ten children with moderate to severe ACS, high-resolution computerized tomography (HRCT) performed within 48 h of presentation accurately detected microvascular occlusion in areas of the lungs that looked normal on chest X-ray, with an average sensitivity and specificity of 84% and 97% respectively (Bhalla et al, 1993). In contrast to chest X-ray, the extent of microvascular occlusion on HRCT correlated with the clinical severity and degree of hypoxia (Bhalla et al, 1993). However, in view of the high radiation dose delivered by CT, the use of this modality for diagnosis of ACS is not recommended. The routine use of CT is also to be discouraged because of the tendency of ACS to recur and also because people with SCD frequently require other radiological tests for a range of other indications. These considerations should not, however, detract from the use of HRCT where it is judged to be clinically appropriate. Likewise, a CTPA should be done if there is a clinical suspicion of pulmonary embolism. This usually reveals widespread perfusion defects with normal ventilation (Noto, 1999; Feldman et al, 2003; Kaur et al, 2004). However, the appearances may be confused with pulmonary emboli or diminished by the presence of other pathology, such as pleural effusions. The role of V/Q lung scanning in the diagnosis of ACS has not been formally investigated in prospective studies and therefore it has a limited role in the investigation of this syndrome. The complication rate of bronchoscopy in patients with ACS has been reported as 13%. The commonest reported complication was a transient fall on SpO2, but there was a small risk of mechanical ventilation following the procedure (Vichinsky et al, 2000). In practice, bronchoalveolar lavage is only ever likely to be done in patients who have progressed to intubation and ventilation. These data, combined with the lack of immediate availability of bronchoscopy in most hospitals, support the use of bronchoscopy only to answer specific clinical questions based on the clinical features and first line investigations. Secretory phospholipase A2 (sPLA2) is an enzyme that is thought to release inflammatory free fatty acids from bone marrow lipid (Styles et al, 1996) and it was initially shown to be elevated in ACS with the levels of phospholipase A2 correlating with the severity of ACS. It has been suggested that measurement of levels of this enzyme may be useful in diagnosis of ACS, but the test is not widely available (Styles et al, 1996, 2000, 2007) and more recent studies have shown a positive predictive value of only 24% in ACS (Styles et al, 2012). Oil-Red-O staining of bronchoalveolar lavage samples has been used to diagnose pulmonary fat embolism (Vichinsky et al, 1994, 2000; Maitre et al, 2000). Although ACS patients with pulmonary fat embolism have a more severe clinical course (Vichinsky et al, 1994), this is not standard practice as it is not necessary to demonstrate the occurrence of fat embolism in routine clinical practice. The immediate aim of treatment in ACS is to prevent or reverse acute respiratory failure. Rapid resolution in the majority of patients will occur with application of the simple measures described below. The key to success is early recognition of ACS and institution of treatment without delay. Prompt and effective treatment will potentially minimize the occurrence of irreversible lung damage and its associated long-term sequelae. These aims are best achieved if haematological, acute medical and critical care support are provided at an early stage. Acute chest syndrome may present as a medical emergency. Sickle cell patients with ACS are often very ill and require close monitoring and management by a multi-disciplinary team. Close cooperation within and between clinical teams (haematology, acute medicine and critical care) is essential to ensure the optimal delivery of care. It is not clear what the optimum SpO2 is and a pragmatic approach is that oxygen should be given to maintain SpO2 ≥ 95% or within 3% of the patient's baseline. SpO2 should be monitored at least 4-hourly so that any deterioration indicated by increasing oxygen requirement can be easily detected. If the patient is becoming increasingly dependent on oxygen a senior member of the clinical team should be informed. More frequent or continuous monitoring is required when there is clinical concern. Patients with ACS are usually too unwell to maintain adequate hydration orally. Intravenous crystalloid infusion should be given until the patient is able to drink adequate amounts of fluid. Fluid requirements should be individualized and be guided by the patient's fluid balance and cardiopulmonary status. Clinical consideration must be given to avoiding fluid overload and the development of acute pulmonary oedema. Daily fluid balance should be monitored using a fluid balance chart of input and output with a daily target for fluid balance. Vaso-occlusive sickle cell crisis affecting the thorax (ribs, sternum and thoracic spine) causes chest splinting and alveolar hypoventilation, which can be complicated by lung atelectasis. There is a high correlation between thoracic bone infarction and ACS (Rucknagel et al, 1991; Gelfand et al, 1993). Therefore effective pain relief, using the World Health Organization analgesic ladder is an important aspect of the management of ACS (Rees et al, 2003). However, care must be taken to avoid alveolar hypoventilation from opioid overdose, as this would heighten the risk of ACS. Two studies have found that the use of morphine may increase the likelihood of developing ACS (Kopecky et al, 2004; Buchanan et al, 2005). However, a small case-crossover study did not support these findings (Finkelstein et al, 2007). Adequate analgesia allows deeper breathing (Needleman et al, 2002), therefore careful monitoring with frequent review and assessment of pain and sedation scores, in addition to cardiorespiratory monitoring, is required. Incentive spirometry coupled with effective pain relief helps to reduce the risk of ACS in children and young adults admitted with thoracic bone ischaemia and infarction by reducing chest splinting and atelectasis (Bellet et al, 1995). Whilst there are no studies on its specific use in prevention of ACS in adults or as treatment for ACS it is likely to be an important adjunct to other forms of therapy. The use of the incentive spirometer should be tailored to the patient's needs as determined by the physiotherapist, but as a guide, 10 maximum inspirations every 2 h during the day and while the patient is awake during the night is a reasonable starting point (Bellet et al, 1995). Incentive spirometry may also be useful in the p