Monitoring, precautions and complications

• All patients undergoing bronchoscopy should have heart rate, respiratory rate, blood pressure and oxygen saturation recorded repeatedly, including before, during and after the procedure. (Grade D)

• All bronchoscopy units should undertake periodic audit of bronchoscopic performance, including efficacy, complications and patient satisfaction surveys. (Good practice point (√))

• All Trusts should have a ‘safe sedation policy’, and ensure all bronchoscopy unit staff, including trainees, receive appropriate training. (√)

Hypoxaemia

• Patients should be monitored by continuous pulse oximetry during bronchoscopy. (Grade C)

• Oxygen supplementation should be used when desaturation is significant (pulse oximeter oxygen saturation (SpO₂)>4% change, or SpO₂<90%) and prolonged (>1 min) to reduce the risk of hypoxaemia-related complications. (Grade D)

• The risks of hypoxaemia-related complications are associated with baseline arterial oxygen saturation (SaO₂) and lung function, comorbidity, sedation and procedural sampling. Fitness for bronchoscopy should incorporate an assessment of these elements, and appropriate monitoring and preprocedure optimisation. (Grade D)

Cardiac arrhythmias

• Continuous ECG monitoring should be used when there is a high clinical risk of arrhythmia. (Grade D)

• When there is a high risk of arrhythmia, oxygen saturations, pulse rate and blood pressure should be optimised. Appropriate aftercare monitoring and instructions should be given. (Grade D)

• Resuscitation equipment should be readily available. (√)

• Intravenous access should be established before sedation is given and maintained until discharge. (√)

Bleeding complications

• Perform coagulation studies, platelet count and haemoglobin concentration when there are clinical risk factors for abnormal coagulation. (Grade D)

• Bronchoscopy with lavage can be performed with platelet counts >20 000 per μL. Liaise with the local haematology team regarding the need for platelet transfusion before bronchoscopy if endobronchial biopsy (EBB) or transbronchial lung biopsy (TBLB) is planned. (Grade D)

• Discontinue clopidogrel 7 days prior to consideration of EBB and TBLB. Low-dose aspirin alone can be continued. (Grade C)

• Anticoagulants should be managed according to published guidelines as set out in appendix 7 of this guideline. (√)

• The risk of biopsy needs to be weighed against the potential for benefit and appropriate informed consent obtained. (√)

Pneumothorax

• A chest radiograph should be obtained if a patient is symptomatic or there is a clinical suspicion of possible pneumothorax after TBLB. (Grade D)

• Fluoroscopic screening may improve diagnostic yield of TBLB in focal but not diffuse lung disease. (Grade D)

• Patients should be advised of the potential for delayed complications following TBLB and provided with written information regarding likely symptoms and action required. (Grade D)

Fever and infection

• Patients should receive written information regarding post-bronchoscopy fever (PBF) and appropriate management advice. (Grade C)

• Antibiotic prophylaxis is not warranted before bronchoscopy for the prevention of endocarditis, fever or pneumonia. (Grade B)

Asthma

• Patients’ asthma control should be optimised prior to bronchoscopy, especially when bronchoalveolar lavage (BAL) is likely to be performed. (Grade C)

• Nebulised bronchodilators should be considered before bronchoscopy in patients with asthma. (√)

Chronic obstructive pulmonary disease

• Chronic obstructive pulmonary disease (COPD) treatment should be optimised prior to bronchoscopy when possible. (Grade D)

• Bronchoscopists should be cautious when sedating patients with COPD. (Grade D)

Ischaemic heart disease

• Liaison with cardiologists should be considered in high-risk patients with cardiac disease and if flexible bronchoscopy (FB) is indicated within 4–6 weeks after myocardial infarction (MI). (Grade D)

• FB should ideally be delayed for 4 weeks after MI. (Grade D)

Haemoptysis

• Consider bronchoscopy after a normal CT if the patient is high risk for lung carcinoma or if the haemoptysis continues. (Grade D)

Older patients

• Age alone should not be a contraindication for bronchoscopy. (Grade D)

• The older patient may require reduced doses of benzodiazepines/opioids sedation. (Grade D)

Patients who are immunosuppressed

• When a diagnosis is not likely to be obtained through non-invasive measures, bronchoscopy with BAL can be considered to provide diagnostic information. (Grade C)

• TBLB is helpful in lung transplant recipients when rejection is a possibility. (Grade C)

Premedication

• Anticholinergics (glycopyrrolate or atropine) should not routinely be used prior to bronchoscopy due to a lack of clinical benefit and a possible increased risk of haemodynamic changes. (Grade A)

• Premedication for bronchoscopy is not routinely indicated. (Grade C)

Sedation

• Intravenous sedation should be offered to patients undergoing bronchoscopy, provided there are no contraindications. (Grade B)

• Some patients will tolerate unsedated bronchoscopy well, and patient preference should be sought. (Grade B)

• Sedative drugs should be titrated to provide the desired depth of sedation, given significant inter-patient variability in required doses. (Grade B)

• The desired depth of sedation is one in which verbal contact is possible at all times. (Grade D)

• Bronchoscopists are encouraged to document an assessment of sedation depth as part of the procedural report. (√)

Benzodiazepines

• Intravenous midazolam is the preferred drug for sedation, having a rapid onset of action, being titratable to provide the required depth of sedation, and being reversible. (Grade B)

• No more than 5 mg midazolam should be initially drawn up into any syringe prior to bronchoscopy for patients under the age of 70 (2 mg midazolam for patients over 70) to prevent potential inadvertent oversedation associated with the practice of routinely drawing up 10 mg midazolam. (Grade D)

• Only low-strength midazolam (1 mg/mL) should be available within bronchoscopy suites. High-strength midazolam (2 mg/mL or 5 mg/mL) should be restricted to general anaesthesia, intensive care and other areas where its use has been formally risk assessed. (Grade D)

Propofol

• While propofol has similar efficacy to midazolam, it should only be used when administered by practitioners formally trained in its administration (eg, anaesthetists) since it has a narrow therapeutic window beyond which general anaesthesia is achieved. (Grade B)

Opioids

• Combination opioid and midazolam sedation should be considered in patients to improve bronchoscopic tolerance. (Grade B)

• When opioids are used, short-acting agents (such as fentanyl or alfentanil) should be used to minimise post-procedural sedation. (Grade D)

• When combination sedatives are used, opioids should be administered first and allowed time to become maximally effective before administration of any other agent. (Grade D)

Topical anaesthesia

• Lidocaine should be used for topical anaesthesia during bronchoscopy, unless contraindicated. (Grade A)

• Nasal topical anaesthesia is most effectively provided using 2% lidocaine gel. (Grade A)

• Both cricothyroid and spray-as-you-go techniques are effective in delivering lidocaine to the vocal cords and trachea. (Grade B)

• Nebulisation is not recommended as a technique for delivering lidocaine to the airways. (Grade B)

• 1% lidocaine solution should be used for spray-as-you-go administration. (Grade A)

• To reduce the risk of lidocaine toxicity, bronchoscopists should use the lowest dose of lidocaine sufficient to prevent excessive coughing and provide patient comfort. (Grade D)

• Bronchoscopists should remain vigilant for objective and subjective symptoms of lidocaine toxicity, particularly given significant inter-patient variability in lidocaine absorption and metabolism. (Grade B)

• Bronchoscopists should monitor and document the total lidocaine dose delivered at all sites during bronchoscopy. (√)

Sampling and diagnostic accuracy

• Bronchoscopists should maintain a record of their personal diagnostic accuracy for FB. (√)

Lung cancer

• A diagnostic level of 85% should be attainable when definite endobronchial tumour is visible. (Grade B)

• At least five biopsy samples should be taken when endobronchial tumour is visible to maximise diagnostic yield and the volume of biopsy tissue and to allow for tumour phenotyping and genotyping. (Grade D)

• When endobronchial tumour is visible, brushings and washings can increase the diagnostic yield of the procedure. (Grade D)

• A chest CT scan should be performed prior to a diagnostic bronchoscopy in patients with suspected lung cancer. (Grade D)

Interstitial lung disease

• In suspected sarcoidosis, EBBs should be considered to increase the diagnostic yield. (Grade C)

• TBLB is recommended for the diagnosis of stage II–IV sarcoidosis. (Grade C)

• In patients with diffuse interstitial lung disease (ILD), five to six TBLBs should be taken from the same lung. (Grade D)

• Fluoroscopy should be considered for TBLB in patients with localised or focal parenchymal lung disease. (Grade D)

Diagnosis of infection

Patients who are immunocompromised

• In patients with pulmonary infiltrates who are immunocompromised and in whom tuberculosis (TB) is considered unlikely, BAL alone is usually sufficient to achieve a diagnosis. In areas or populations with high prevalence of TB, TBLB may be considered in addition. (Grade C)

• BAL or bronchial washings should be sent for microscopy for acid fast bacteria (AFB) and for mycobacterial culture in patients with pneumonia who are immunocompromised. (Grade C)

• Post-bronchoscopy sputum could be collected in patients who are immunocompromised and suspected to have TB. (Grade D)

• TBLB and EBB for invasive aspergillosis may be avoided if BAL galactomannan test is available due to the high sensitivity and specificity of the latter and inherent risks with the biopsies. (Grade C)

• In patients suspected to have invasive aspergillosis, BAL should be sent for microscopy for hyphae and fungal culture; a BAL galactomannan test should be considered to further improve diagnostic yield. (Grade C)

Patients who are immunocompetent

• Bronchoscopy may be considered in patients with non-resolving or slowly resolving pneumonia, especially if they are current or ex smokers and older than 50 years. (Grade C)

• If bronchoscopy is performed for community-acquired pneumonia, BAL specimens should be sent for legionella PCR and atypical pathogens. (Grade C)

• Bronchoscopy may be considered if the patient is suspected to have TB when sputum smear is negative. (Grade C)

• In cases of suspected TB, BAL, bronchial aspirates and post-bronchoscopy sputum appear to be complementary and should all be analysed. (Grade C)

• In areas with high or intermediate prevalence of TB, patients undergoing bronchoscopy for another indication should have samples sent routinely for cultures for TB. (Grade C)

Training

Training in FB does not fall in the scope of this guideline.

Audit standards

The following standards provide criteria which may form the basis of future audits:

• A serious adverse event rate of <1% (box 1).

• A ‘safe sedation policy’ and appropriate training in sedation for all bronchoscopy unit staff, including trainees and interval audit of sedation practice.

• Utilisation of the bronchoscopy safety checklist (see appendix 5).

• An 85% diagnostic rate for FB with visible endobronchial tumour.

• Periodic patient feedback to inform improvement and revision of the endoscopy service.

Hypoxaemia

Monitoring patients with pulse oximetry during bronchoscopy is an accurate non-invasive method for assessing hypoxaemia.9^–11 Significant decreases in oxygen saturation are commonly seen during bronchoscopy, commencing with administration of sedation and worsening on passage through the vocal cords.8 ,9 ,12^–16 Patient positioning8 ,10 ,16 and intra-procedural sampling may also influence oxygen saturations as may airway suctioning.17 ,18 Van Zwam et al8 described a twofold greater incidence of desaturation >4% and SpO₂<90% in the sitting position compared with supine. The use of preprocedure oxygen and the specific sampling procedure (BAL vs wash vs brush vs biopsy) was predictive of a higher rate of desaturation episodes (<90%), but baseline saturations were not (desaturation to <90% was seen in: BAL 89%, wash 44%, brush 15%, biopsy 10%).19 Milman et al11 (using benzodiazepine premedication without oxygen supplementation) demonstrated that 38% of patients desaturate (SpO₂<90%) before bronchoscopy, increasing to 80% of patients during the procedure. The proportion remains high after the procedure, with 60% of patients desaturating. Similarly, more severe desaturation (SpO₂<85%) was seen in 10%, 35% and 15%, respectively. No differences in oxygen saturation were described in a comparison of trans-nasal or trans-oral bronchoscopic approaches.20

The majority of desaturations are transient and do not require specific intervention.21 Supplemental oxygen given via nasal or pharyngeal catheter can reduce the incidence, degree or duration of desaturation.11 ,16 ,22 In patients undergoing BAL and TBLBs for diffuse ILD, significant hypoxaemia during bronchoscopy can be avoided with routine supplemental oxygen compared with breathing room air alone, but the proportion of patients requiring supplemental oxygen in general bronchoscopic practice is variable, ranging between 5% and 32%, and is dependent on forced expiratory volume in 1 s (FEV₁) (or peak expiratory flow rate (PEFR)).12 ,21

Patients with abnormal PEFR demonstrate an increased requirement for oxygen supplementation (PEFR<60% predicted: 58%, <45% predicted: 83%)12 and Jones and O'Driscoll21 demonstrated a greater risk of desaturation to <90% and need for oxygen supplementation with declining FEV₁ (table 4).

In studies examining the effect of BAL volume on oxygen saturation, inconsistent results have been reported.15 ,23

There are no studies that address a safe saturation threshold. Schiffman et al22 noted that significant desaturation could be prevented with 4 L/min oxygen supplementation but this did not impact on the rate of cardiac arrhythmia (sinus tachycardia, sinus bradycardia, ventricular or atrial premature contractions). Similarly, Lundgren et al18 did not demonstrate an increase in cardiac arrhythmia, despite hypoxaemia and increased cardiac work in 30% of patients. Two further studies demonstrated no significant increase in the frequency of bradycardia/tachycardia or premature atrial/ventricular activity with or without oxygen supplementation.11 ,20

Hypoxaemia can be effectively minimised by using a nasal or pharyngeal catheter for oxygen supplementation at either 2 or 3 L/min.11 In hypoxaemic respiratory failure (RR>35, arterial oxygen pressure (PaO₂)/fractional inspired oxygen (FiO₂)<200) non-invasive positive pressure ventilation (FiO₂ 0.5, adjusted to maintain SaO₂>92%) is superior to a high-flow venturi mask (constant FiO₂ 0.9) during bronchoscopy for nosocomial pneumonia.24 Several studies, including those post BAL, suggest that hypoxaemia may persist for at least 2 h after the procedure but there is no evidence relating to the duration for oxygen supplementation post procedure.25^–27

Oxygen supplementation should be used in patients with persistent significant desaturation (SaO₂ change > 4% or SaO₂<90%). Target oxygen saturations should be consistent with the principles of published guidance on oxygen therapy.

Evidence statements

• Hypoxaemia is common during bronchoscopy, it is frequently transient, and only considered significant if prolonged (>1 min). Hypoxaemia is more common in the sitting position, with oral or intravenous sedation, with decreasing FEV₁ or PEFR, or in patients who require supplemental oxygen before the procedure. The use of suction can exacerbate hypoxaemia. (Evidence level 2++)

• Pulse oximetry demonstrates excellent correlation with arterial gas analysis. (Evidence level 3)

• The use of supplemental oxygen can reduce the severity of hypoxaemia, either by nasal or pharyngeal catheter, at flow rates of at least 2 L/min. (Evidence level 2+).

• Oxygen supplementation should be targeted to patients with persistent significant desaturation (SpO₂ change > 4% or SpO₂<90%). Target oxygen saturations should be consistent with the principles of published guidance on oxygen therapy. (Evidence level 3)

Recommendations

• Patients should be monitored by continuous pulse oximetry during bronchoscopy. (Grade C)

• Oxygen supplementation should be used when desaturation is significant (SpO₂>4% change, or SpO₂<90%) and prolonged (>1 min) to reduce the risk of hypoxaemia-related complications. (Grade D)

• The risks of hypoxaemia-related complications are associated with baseline SaO₂ and lung function, comorbidity, sedation and procedural sampling. Fitness for bronchoscopy should incorporate an assessment of these elements, and appropriate monitoring and preprocedure optimisation. (Grade D)

Cardiac complications

Hypoxaemia related to FB is commonly associated with an increase in cardiac workload with elevations of heart rate (approximately 40% above baseline), blood pressure (a rise of 30% above baseline) and cardiac index (approximately 17–32% of baseline).14 ,18 ,28 ,29 Despite this, major arrhythmias are rare during bronchoscopy but believed to be related to myocardial ischaemia. The rate–pressure product (heart rate×systolic blood pressure) during bronchoscopy can approach or exceed the level associated with silent myocardial ischaemia, particularly in patients with hypertension.14 ,18 ,28 Increases in systolic blood pressure and heart rate during the bronchoscopy are associated with ECG change in 15% (ST-T change in 4%, transient right bundle branch block in 3%).29 In addition, ECG changes correlate with older age, higher pack-year smoking history but not lung function or changes in oxygen saturation. Cardiac strain has been reported to occur in 21% of patients over the age of 60 years.29

Arrhythmia

Schiffman et al22 demonstrated that bronchoscopy was associated with sinus tachycardia in 55–58%, sinus bradycardia in 5–8%, premature ventricular contraction in 8% and atrial premature contraction in 3–5%, with no significant difference according to oxygen supplementation. Payne et al20 demonstrated a 60% prevalence of baseline minor arrhythmia and 5% incidence of new minor arrhythmia with bronchoscopy.

Katz et al30 documented 12% of patients with at least an occasional atrial or ventricular ectopic beat before the procedure compared with 80% of patients during or after bronchoscopy. Major cardiac arrhythmias (defined as five or more atrial ectopics/min, supraventricular tachycardia, or five or more ventricular ectopics/min, multiform ectopic beats, couplets or ventricular tachycardia) increased from 4% before the procedure to 40% during/after the procedure. Atrial arrhythmias occur at widely differing stages of the procedure, but ventricular arrhythmias occur mainly on passage through the vocal cords. Maximum ventricular arrhythmia is correlated with minimum oxygen saturations. Asymptomatic ST-T wave changes occur in 6% of patients, typically correlating with maximum heart rates. Oxygen saturations can remain lower than preprocedure levels 3 h after bronchoscopy in 30% of patients.

Myocardial infarction

Acute MI is considered a contraindication to bronchoscopy within 4–6 weeks. Dweik et al31 retrospectively analysed the safety of bronchoscopies within 30 days of an acute MI and noted that mortality (5%) was limited to patients with active ischaemia at the time of bronchoscopy.

Evidence statements

• Bronchoscopy increases cardiac rate, blood pressure and cardiac index. The rate pressure product is often sufficient to cause myocardial ischaemia. (Evidence level 2+)

• Sinus tachycardia and atrial or ventricular premature contractions are the commonest arrhythmia noted before, during and after bronchoscopy. (Evidence level 3)

• Ventricular arrhythmia (mostly premature contraction, bi and trigeminy) occurs most commonly on passage through the cords and is associated with low oxygen saturations. Oxygen saturations may remain below preprocedure levels for over 3 h in a third of patients. (Evidence level 3)

• Myocardial ischaemia during bronchoscopy is related to heart rate and blood pressure, rather than oxygen saturation per se. It also correlates with increasing age and smoking history. (Evidence level 3)

• Bronchoscopy within 30-days of acute MI is associated with a 5% mortality (related to active ischaemia). In the absence of active ischaemia, when good clinical justification is made, bronchoscopy can be performed. (Evidence level 3)

• Incidence of cardiac arrhythmia is not affected by oxygen supplementation. (Evidence level 3)

Recommendations

• Continuous ECG monitoring should be used when there is a high clinical risk of arrhythmia. (Grade D)

• When there is a high risk of arrhythmia, oxygen saturations, pulse rate and blood pressure should be optimised. Appropriate aftercare monitoring and instructions should be given. (Grade D)

Good practice points

• Resuscitation equipment should be readily available. (√)

• Intravenous access should be established before sedation is given and maintained until discharge. (√)

Bleeding complications

It is difficult to subclassify bleeding during FB into minor, moderate or severe based on an estimate or measures of blood loss during the procedure. Aspirated blood is collected and mixed with saline, adrenaline and suctioned secretions and an accurate measure is not possible. Classification by the type of clinical intervention necessary to stop the bleeding and stabilise the patient is an easier and reproducible measure of bleeding (see table 5, adapted from Ernst et al32).

Minor bleeding occurs in 0.19% and severe bleeding in 0.26% of bronchoscopies.4 Clinical risk factors for bleeding correlate with abnormal coagulation but the rate of biopsy-related bleeding in patients with clinical risk factors is only 11% (box 2). In patients with known abnormal coagulation, bleeding occurred in a similar 11%.33 The type of biopsy, abnormal coagulation, platelets, haemoglobin or creatinine does not reliably or consistently predict bleeding risk for bronchoscopy. Over two-thirds of patients who develop bleeding have normal coagulation and no clinical risk factors for bleeding.

The majority of bleeding is mild to moderate with only 3% estimated at being more than 100 mL. Approximately 90% of bleeding stops spontaneously or requires local vasoconstrictor therapy only (adrenaline/cocaine).33 Appendix 6 provides a management approach to bleeding complications during FB.

At bronchoscopy, clinically significant bleeding was seen in 0.83%, increasing to 1.9% with biopsy, including TBLB.34 TBLB caused mild–moderate bleeding in 0.8% whereas EBB bleeding occurs in only 0.45%.34 Severe bleeding is more common in TBLB than EBB but remains <20 mL in the majority of cases of TBLB (92%).35 Spontaneous resolution of bleeding occurred in two-thirds and was treated with local instillation of adrenaline in the remainder.35 ,34 Platelet levels or coagulation studies prior to TBLB do not predict bleeding in procedures in which bleeding occurs.35

Diette et al36 prospectively studied 720 procedures, including 38 lung transplants. Transplant recipients are more likely to have TBLB and to receive aspirin therapy, to have blood loss >25 mL and to have the procedure terminated early for bleeding. In multivariate analysis, independent predictors of greater blood loss included lung transplant, performance of TBLB, longer procedure time and older patient age. A smaller prospective study found that bleeding was quantitatively similar between patients with and without lung transplant and typically minor.35

Weiss et al37 prospectively reported 66 bronchoscopies in 47 bone marrow transplant recipients with thrombocytopenia (20% had platelets <20 000/mL; 67% platelets <50 000/mL; 88% platelets <100 000/mL). Bleeding-related complications were reported in 6.9% and were usually minor.

In renal failure, including haemodialysis and non-dialysis patients, bronchoscopic biopsy and transbronchial needle aspiration (TBNA) were associated with an overall complication rate of 8%.38

The concomitant use of clopidogrel with transbronchial biopsy leads to excessive (>100 mL) bleeding in patients taking clopidogrel alone (89% vs 3.4%) and clopidogrel with aspirin (100% vs 3.4%).32 Bleeding rates are significantly higher in all categories of bleeding (minor/moderate/severe) but can be controlled in most instances by bronchoscopic means. Need for transfusion or death following haemorrhage secondary to clopidogrel is rare.32 See appendix 7 for an algorithm for the management of patients on warfarin or clopidogrel undergoing FB.

Evidence statements

• Minor bleeding occurs in 0.19% and severe bleeding in 0.26% of bronchoscopies. (Evidence level 3)

• The routine performance of coagulation studies, platelet or haemoglobin counts are of no value in predicting the risk or severity of bleeding. (Evidence level 3)

• Coagulation studies, platelet count and haemoglobin values should be estimated when clinical risk factors indicate a likelihood of abnormal coagulation. However, over two-thirds of patients with significant bleeding possess normal coagulation and no clinical risk factors for bleeding. (Evidence level 3)

• Bleeding complications in patients with thrombocytopenia undergoing bronchoscopy and lavage are approximately 7%. No data are available regarding the safety of TBLB or EBB in thrombocytopenia but the majority of bleeding complications relate to epistaxis. (Evidence level 3)

• Clopidogrel causes bleeding, ranging from mild to severe, when performing TBLB. (Evidence level 2+)

• TBLB causes a twofold increase in the risk of mild–moderate bleeding and a threefold increase in the risk of severe bleeding compared with EBB. However, the overall risk remains small and TBLB rarely causes significant blood loss (92% of patients experience blood loss <20 mL) and typically resolves spontaneously or with endoscopic instillation of cocaine/adrenaline. (Evidence level 3)

• Bronchoscopic biopsy and TBNA in patients receiving haemodialysis, or in patients with renal failure without dialysis, result in a higher rate of bleeding complications (∼8%; 4% major, 4% minor) than the general population. (Evidence level 3)

• Lung transplantation may predispose patients to greater blood loss at bronchoscopic biopsy, including TBLB. (Evidence level 2−)

Recommendations

• Perform coagulation studies, platelet count and haemoglobin concentration when there are clinical risk factors for abnormal coagulation. (Grade D)

• Bronchoscopy with lavage can be performed with platelet counts >20 000 per μL. Liaise with the local haematology team regarding the need for platelet transfusion before bronchoscopy if EBB or TBLB is planned. (Grade D)

• Discontinue clopidogrel 7 days prior to consideration of EBB and TBLB. Low-dose aspirin alone can be continued. (Grade C)

Good practice points

• Anticoagulants should be managed according to published guidelines as set out in appendix 7 of this guideline. (√)

• The risk of biopsy needs to be weighed against the potential for benefit and appropriate informed consent obtained. (√)

Pneumothorax

Pneumothorax following bronchoscopy for any indication occurs at a rate of 1 in 1000 (0.1–0.16%)4 ,39 but was as high as 0.4–0.8% in some smaller series.40 ,41 In contrast the rate of pneumothorax in TBLB has been reported to be significantly higher between 1% and 6%4 ,39 ,42 ,43 or higher still in TBLB of diffuse abnormality (9%).43^–46 TBLB remains a safe outpatient procedure with a low incidence of delayed complications.47

In relation to all adverse events, pneumothorax represents approximately 10% of all complications but rarely complicates bronchoscopy without TBLB or therapeutic bronchoscopy.4 Pneumothorax is rarely total and often delayed (∼40% of pneumothoraces)4 and when present may require intercostal tube drainage (40–70% of cases).39 ,42 ,43 ,47 The frequency of pneumothorax is related to age and the number of TBLBs.5 ,44 Routine performance of a chest x-ray after TBLB rarely provides useful clinical information in the absence of symptoms43 ,44 ,48 and may not be required. In the absence of a routine chest x-ray, monitoring for the development of symptoms associated with pneumothorax should continue for 2 h. The rate of pneumothorax following TBLB does not appear significantly different with or without fluoroscopy,45 but fluoroscopy may be of value to improve the diagnostic yield in focal rather than diffuse lung disease (focal 4.3%, diffuse 9%).43

Evidence statements

• Risk of pneumothorax from all bronchoscopic procedures is 1 in 1000 (0.1%) but increases to between 1 in 100 to 1 in 16 (1–6%) following TBLB. (Evidence level 3)

• Pneumothorax may be delayed (up to 2 h in 40% of cases) and may require intercostal tube drainage (Evidence level 3)

• Routine performance of a chest x-ray after TBLB rarely provides useful clinical information in the absence of symptoms. (Evidence level 3)

• Fluoroscopy may reduce the rate of pneumothorax in focal lung disease. (Evidence level 3)

Recommendations

• A chest radiograph should be obtained if a patient is symptomatic or there is a clinical suspicion of possible pneumothorax after TBLB. (Grade D)

• Fluoroscopic screening may improve diagnostic yield of TBLB in focal but not diffuse lung disease. (Grade D)

• Patients should be advised of the potential for delayed complications following TBLB and provided with written information regarding likely symptoms and action required. (Grade D)

Fever and infection

Post bronchoscopy fever (PBF) is not reported in a large prospective study of complications in over 20 000 patients4 but appears relatively common in other smaller prospective studies focusing on PBF (5–10%).49 ,50 PBF is most typically seen approximately 8 h (range 4–24 h) following BAL (13%) when it is associated with an acute inflammatory response characterised by fever >38°C, neutrophilic leucocytosis, elevated C-reactive protein, fibrinogen and proinflammatory cytokines,50^–52 and an absence of bacteraemia.49 ,50 ,53 Fever is typically less than 40°C, lasts on average 14 h but is rarely accompanied by a chest x-ray infiltrate.50 Antibiotic prophylaxis does not prevent PBF, pneumonia or the proinflammatory cytokine response.53 ,54

A true bacteraemia post bronchoscopy occurs in 6–8% of patients,49 ,55 most commonly coagulase negative or positive staphylococci, non-haemolytic or β-haemolytic streptococci, Citrobacter or Klebsiella species.49 ,54 ,55 National Institute for Health and Clinical Excellence guidance in March 2008 also concluded that ‘Antibacterial prophylaxis is not recommended for the prevention of endocarditis in patients undergoing procedures of the upper and lower respiratory tract (including bronchoscopy)’.56

Evidence statements

• PBF is common, particularly after BAL, and associated with a non-infective acute inflammatory response, typically starting after discharge from hospital. (Evidence level 2++)

• Antibiotic prophylaxis does not prevent PBF or pneumonia. (Evidence level 1++)

• Bacteraemia occurs in 6–8% of patients undergoing bronchoscopy. (Evidence level 3)

Recommendations

• Patients should receive written information regarding PBF and appropriate management advice. (Grade C)

• Antibiotic prophylaxis is not warranted before bronchoscopy for the prevention of endocarditis, fever or pneumonia. (Grade B)

Patients with asthma

The safety of bronchoscopy in asthma has been studied in research and clinical settings. Bronchoscopy often causes a fall in FEV₁. In healthy volunteers the mean fall in FEV₁ has been reported on average to be between 9% and 17%,15 ,57 however it can be over 20%.58 In patients with asthma the mean fall has been reported to be 10–26%.15 ,57 ,59 The majority of studies comparing the fall in FEV₁ between patients with asthma and healthy volunteers found no difference between the groups.15 ,58 ,60 Patients with increased bronchial hyper-reactivity may have a greater fall in FEV₁57 ,61 however this has not been universally reported.58 In particular, BAL can cause a fall in the FEV₁.58 ,59

Complication rates post bronchoscopy range between 3.5% and 12% depending on how the complications are reported and what procedures were performed.15 ,60^–63 In addition, patients with severe asthma are more likely to require oral corticosteroids post bronchoscopy.58 Many of the studies use bronchodilators prior to bronchoscopy to increase the FEV₁.57 ,58 ,60 ,62 Although this does not seem to reduce the percentage fall in FEV₁ it may increase the absolute FEV₁ at the end of the procedure, as the patient starts from a higher baseline.

Evidence statements

• Up to 10% of patients with asthma may develop respiratory symptoms post bronchoscopy. (Evidence level 2−)

• BAL is associated with more symptoms in patients with asthma compared with bronchoscopy alone. (Evidence level 2−)

Recommendation

• Patients’ asthma control should be optimised prior to bronchoscopy, especially when BAL is likely to be performed. (Grade C)

Good practice point

• Nebulised bronchodilators should be considered before bronchoscopy in patients with asthma. (√)

Patients with COPD

Bronchoscopy in patients with COPD appears to carry a greater risk compared with those with normal lung function. An increased risk of complications has been reported in severe COPD (defined as FEV₁<50% predicted or FEV₁<1 L and with FEV₁/forced vital capacity < 69%).64 Five percent of patients with COPD compared with 0.6% of controls experienced complication: pneumonia, respiratory failure and desaturation.64 In a separate study, bronchoscopy safety was studied in patients with hypercapnia, 77% of whom had COPD.65 Desaturation occurred in 30% of the study population, 55% of patients developed wheezing and the procedure was terminated early in 20% of patients.65 A randomised control trial studied the effect of inhaled salbutamol (200 µg) on FEV₁ after bronchoscopy in moderate to severe COPD.66 No difference was observed (change in FEV₁) in the group receiving salbutamol compared with placebo. Nine percent of patients experienced desaturation during bronchoscopy, with no difference again being observed between treatment groups.

In research bronchoscopy comparable complication rates have been reported. Hattotuwa et al67 reported a complication rate of 9%, including haemoptysis (5%), pneumothorax (2%) and bronchospasm (2%). All patients were given 2.5 mg nebulised salbutamol prior to the procedure.

Evidence statements

• Patients with COPD have a higher risk of desaturation and bronchoconstriction compared with controls, however this is not a universal finding. (Evidence level 2−)

• Nebulised salbutamol administered prior to bronchoscopy does not alter the post-bronchoscopy complication rate in patients with COPD. (Evidence level 1−)

Recommendations

• COPD treatment should be optimised prior to bronchoscopy when possible. (Grade D)

• Bronchoscopists should be cautious when sedating patients with COPD. (Grade D)

Patients with ischaemic heart disease

The haemodynamic changes during FB might increase the risk of myocardial damage during the procedure. One study has shown an increased risk of ischaemic ECG changes during the procedure in patients over 60.29 A retrospective study investigated the safety of bronchoscopy in 20 patients after acute MI. The procedure was performed an average of 12 days after MI.31 One death occurred in this study—a patient with acute myocardial ischaemia before and during the procedure. No other complications were reported. A retrospective study of patients undergoing bronchoscopy on a coronary care unit reported no difference in complications in subjects after MI compared with those without MI.68

The American perioperative guidelines recommend that elective surgery is avoided for 4–6 weeks after an acute event resulting in myocardial damage.69

Evidence statements

• Active myocardial ischaemia is a contraindication to bronchoscopy. (Evidence level 3)

• FB can increase the risk of active ischaemia, haemodynamic compromise, arrhythmia and further ischaemic events after MI. (Evidence level 3)

Recommendations

• Liaison with cardiologists should be considered in high-risk patients with cardiac disease and if FB is indicated within 4–6 weeks after MI. (Grade D)

• FB should ideally be delayed for 4 weeks after MI. (Grade D)

Bronchoscopy for haemoptysis

Bronchoscopy can be used to investigate haemoptysis. CT scans have changed the diagnostic pathway and should always be considered prior to bronchoscopy. Two studies have investigated the role of bronchoscopy following a thoracic CT scan. One study in Turkey found a bleeding site in 80% of 203 individuals even if the CT and chest x-ray were normal.70 A different study (n=200) found an endobronchial diagnosis in 0.5% of individuals when the CT was normal.71

Recommendation

• Consider bronchoscopy after a normal CT if the patient is high risk for lung carcinoma or if the haemoptysis continues.(Grade D)

Bronchoscopy in the older patient

Comorbid disease is more likely in the older patient with a potential increase in bronchoscopy risk. Several studies have shown older patients tolerate the procedure well, with no increase in complications.72^–74 A prospective cohort study5 suggested that complication rates for pneumothorax and transient hypotension increased with age. When the safety of bronchoscopy was investigated in patients over 80 years old, higher complication rates and mortality rates were reported and these patients were also more likely to be mechanically ventilated after bronchoscopy.75

Recommendations

• Age alone should not be a contraindication for bronchoscopy. (Grade D)

• The older patient may require reduced doses of benzodiazepines/opioids for sedation. (Grade D)

Bronchoscopy in patients who are immunosuppressed

FB can provide useful diagnostic information when treating patients with respiratory problems who are immunosuppressed (see ‘Diagnosis of infection’ in this guideline for details of diagnostic value). Bronchoscopy and associated diagnostic techniques are not without risk in this patient population and this should be taken into account before performing a bronchoscopy. BAL has a reported complication rate of 0–49%, depending on how the complications were defined, and deaths associated with BAL have been reported.76^–82 The main complications associated with BAL are desaturation, a drop in FEV₁ and haemorrhage.

Protected specimen brushes (PSBs) can sometimes be used for similar indications to BAL. One study involved patients who were immunocompromised following bone marrow transplant78 and showed that the complication rates for PSBs were significantly higher than for BAL (36% vs 14%).

TBLB can carry a significant higher complication risk (30%) compared with BAL alone.79 ,81 Pneumothorax, haemorrhage and desaturation were reported to be complications associated with TBLB.79 ,81

Three studies comparing the overall mortality in patients investigated with FB and patients investigated with non-invasive techniques for obtaining diagnostic samples failed to demonstrate that obtaining samples by bronchoscopy significantly reduces mortality.77 ,78 One study suggested that early FB to obtain diagnostic samples is associated with a lower mortality compared with delayed bronchoscopy for the same indication.82

In patients who have had lung transplantation, TBLB may be the only means of diagnosing allograft rejection. It has previously been reported that lung transplant patients have a higher risk of bleeding (>25 mL blood) (44%) compared to other patients undergoing bronchoscopy, resulting in 5.4% of procedures being terminated early.36 Larger studies have suggested that significant bleeding occurs in 25%83 or 13% of patients84 depending on how bleeding is diagnosed. The largest study85 suggested that bronchoscopy in lung transplant patients carried no increased risk compared with other patient groups and reported a complication risk of 0.7%. It should be noted that only 57% of procedures in this study included TBLB.85 Other groups have reported higher complication rates of between 4.8% and 22%.83 ,84 ,86 The main complications associated with TBLB are oversedation, pneumothorax and haemorrhage.

Evidence statements

• PSB has a low diagnostic yield, is rarely positive when BAL is negative and has a higher complication rate than BAL. (Evidence level 2+)

• BAL in patients who are immunocompromised has a diagnostic rate of 45–62% for infection with an associated complication rate of 0–49%. (Evidence level 2++)

• TBLB carries an increased complication rate compared with BAL of around 30%, which is mainly explained by the increased risk of pneumothorax. (Evidence level 2++)

• Lung transplant patients may be at higher risk of bleeding than other patients who are immunosuppressed. (Evidence level 2−)

• TBLB has a reported complication rate of between 0.7% and 22%, with the main complications being oversedation, pneumothorax and haemorrhage. (Evidence level 2++)

Recommendations

• When a diagnosis is not likely to be obtained through non-invasive measures, bronchoscopy with BAL can be considered to provide diagnostic information. (Grade C)

• TBLB is helpful in lung transplant recipients when rejection is a possibility. (Grade C)

Premedication

Various drugs have been considered to have a potential role as premedication, including anticholinergics (atropine and glycopyrrolate), other drugs with cardiovascular activity (eg, clonidine and labetalol), fenoterol, benzodiazepines and opioids.

Sedation

The majority of bronchoscopists in the UK use sedation for bronchoscopy, with only 5–10% routinely performing bronchoscopy on unsedated patients.96 ,97 Five randomised controlled trials,95 ,98^–101 one cohort study102 and one qualitative study103 examined patient and physician preference for sedation during bronchoscopy. These studies provided a broad consensus that patients and physicians usually preferred sedation for bronchoscopy, examining domains such as comfort, tolerance, bronchoscopic ease and willingness to undergo a repeat procedure. Some studies96 ,100 ,104 suggested that a subset of patients tolerate unsedated bronchoscopy well, suggesting that sedation should remain an option dependent on patient preference and comorbidities (which may limit suitability for sedation).

Evidence statements

• Patients and physicians usually prefer the use of sedation for bronchoscopy. (Evidence level 1+)

Recommendations

• Intravenous sedation should be offered to patients undergoing bronchoscopy, provided there are no contraindications. (Grade B)

• Some patients will tolerate unsedated bronchoscopy well, and patient preference should be sought. (Grade B)

Topical anaesthesia

Lung cancer

In patients with lung cancer, a chest CT scan should be performed prior to a bronchoscopy and patients with central lesions on the CT scan should be offered bronchoscopy if the nodal status does not influence the management decision. The diagnostic yield from bronchoscopy depends on whether the tumour is visible within the bronchial tree. Many studies have demonstrated the sensitivity of bronchial biopsy with forceps, which ranges from 48% to 93% (American College of Chest Physicians guidelines 2007, table 2).145

Studies have shown that the addition of bronchial brush specimens and washings increases the diagnostic yield of bronchoscopy in patients with lung cancer. The systematic review and meta-analysis carried out by the American College of Chest Physicians145 identified 35 studies of patients with central disease undergoing bronchoscopy. The pooled data in 4507 patients demonstrated an overall combined sensitivity of 88% for bronchoscopic techniques. Biopsy of visible lesions alone had a yield of 74% while washings had a sensitivity of 48% and yield from brushings was 59%. McLean et al146 analysed 2238 bronchoscopies carried out between January 1991 and December 1992 in five centres in Scotland. When endobronchial tumour was seen, sensitivity for brushings and washings combined was 47%. The yield from biopsy alone was 82% and when combined with brushings and washings the yield increased to 87%. When the tumour was not visible the sensitivity of brushings and washings combined was 9%.146

The optimal number of bronchial biopsy specimens in cases of visible endobronchial tumour has also been studied., Gellert147 demonstrated that at least five samples are required to achieve a 90% probability of a positive malignant biopsy for a visible tumour.147 Popovich et al148 demonstrated in peripheral lung cancer lesions biopsied using TBLB that only a 45% diagnostic yield was achieved with one biopsy and in some cases the diagnosis was not achieved before the fifth or sixth biopsy. They suggested that up to 10 biopsies may be required.

Based on recent evidence, it is reasonable to expect a diagnostic yield of 85% when the tumour is visible at bronchoscopy using bronchial biopsy.149 ,150 Slade et al149 examined the diagnostic yield from FB at a single centre between 2001 and 2007. In patients with tumour visible at bronchoscopy, the yield increased from 75% in 2001 to 94.5% in 2007.

The management of patients with advanced NSCLC now depends on the phenotyping and genotyping of lung cancer.151 However, the currently available evidence does not clarify whether this is routinely possible with bronchoscopic samples and in particular whether bronchial brushings and washings are suitable for subclassification of NSCLC. Studies of the sequence of biopsies, brushings and washings suggest that the order of sample acquisition is not important.152 However, to maximise the volume of tissue to allow the phenotyping and genotyping of advanced NSCLC, bronchial brushings and washings may be taken after sufficient biopsy specimens have been taken.

In addition to providing pathological information, bronchoscopy enables the location and extent of an endobronchial lesion to be determined. Careful assessment of the vocal cords is mandatory, since the presence of vocal cord palsy may provide evidence of mediastinal invasion (and thus inoperability). Bronchoscopy also allows for accurate assessment of the location and therefore T staging of an endobronchial tumour which may have implications for surgical treatment.

Recommendations for FB in performing conventional TBNA and endobronchial ultrasound (EBUS) TBNA as a diagnostic technique for lung cancer are discussed in the BTS guideline for advanced diagnostic and therapeutic FB in adults, which also includes a guide on how to perform TBNA.3

Evidence statements

• In patients with visible endobronchial tumour a diagnostic yield of at least 85% is achievable. (Evidence level 2+)

• A minimum of five biopsies should be taken when endobronchial tumour is visible. (Evidence level 3)

• The diagnostic yield from bronchoscopy can be increased by undertaking brush and lavage specimens in addition to bronchial biopsy. (Evidence level 3)

Recommendations

• A diagnostic level of 85% should be attainable when definite endobronchial tumour is visible. (Grade B)

• At least five biopsy samples should be taken when endobronchial tumour is visible to maximise diagnostic yield and the volume of biopsy tissue and to allow for tumour phenotyping and genotyping. (Grade D)

• When endobronchial tumour is visible, brushings and washings can increase the diagnostic yield of the procedure. (Grade D)

• A chest CT scan should be performed prior to a diagnostic bronchoscopy in patients with suspected lung cancer. (Grade D)

Interstitial lung disease

Bronchoscopy is a useful diagnostic tool in some ILDs, particularly sarcoidosis, hypersensitivity pneumonitis and organising pneumonia.153 BAL and TBLB are frequently performed in suspected cases of ILD.

TBLB is a technique which uses biopsy forceps to obtain alveolar tissue; it is particularly useful in obtaining histological specimens for suspected cases of ILD, although crush artefact during the procedure may render the samples inadequate for histological diagnosis, particularly in idiopathic pulmonary fibrosis. The size of forceps used does not appear to affect the diagnostic yield (see appendix 10 for a step-by-step guide on how to perform TBLB).153 A retrospective review of 452 bronchoscopies in localised and diffuse lesions examined by TBLB demonstrated a diagnostic yield in 55.2% of patients with a localised lesion and 67% in those with diffuse lesions. The study also demonstrated an increase in the diagnostic yield with an increasing number of specimens, so that taking more than four biopsies increased the yield from 52% to more than 70%.43 The complication rate in this series was 6%; 5.8% developed a pneumothorax (3.8% required intercostal drainage) and significant bleeding was noted in 0.2%, suggesting that TBLB is a useful diagnostic procedure with a low complication rate43

The use of BAL may be controversial, since its use in diagnosis and monitoring disease activity is limited in ILD, and its current strength is in research in understanding the underlying pathophysiological mechanisms in ILD.154 There is a suggestion that combining techniques increases diagnostic yield. In a study investigating patients with sarcoidosis, TBLB alone was diagnostic in 58%, TBNA alone was successful in diagnosing 17% of cases as was BAL alone; combining all three procedures led to a 100% diagnostic rate.155

No evidence currently exists for the coexistent use of BAL with TBLB, however it is reasonable to perform both these procedures in cases of suspected ILD. BAL and TBLB should be performed in areas shown to be affected on high-resolution CT (HRCT)156 and both procedures should be performed on the same lung. In those with typical HRCT features of idiopathic pulmonary fibrosis, BAL and TBLB are not required for diagnosis.

Diagnostic bronchoscopic procedures have moderate (67–74%) diagnostic sensitivity in patients with diffuse ILD,157 ,158 but have high diagnostic sensitivity in cases of sarcoidosis. In a study of 530 consecutive TBLBs in patients with diffuse infiltrates, the overall diagnostic yield was 50% and 75% for stage II–III sarcoidosis. The diagnostic yield of hypersensitivity pneumonitis in this series was 92%. In a case series of 183 patients with bilateral diffuse shadowing, Mitchell et al159 demonstrated that TBLB had a diagnostic yield of 77% for patients with sarcoidosis.

The diagnostic sensitivity of TBLB and BAL in sarcoidosis is estimated to be between 56% and 77%, increasing with the severity of the disease.157 ,159^–164 One study demonstrated the sensitivity of TBLB to be 55% in stage I disease, 80% in stage II disease and 82% in stage III disease.162 In sarcoidosis, EBB may also prove useful, even without visible endobronchial abnormality, particularly when parenchymal disease exists.163 Combining techniques will improve diagnostic yield.160 ,161 Mediastinal lymphadenopathy can be sampled by conventional or ultrasound guided TBNA—please refer to the BTS guideline for advanced diagnostic and therapeutic FB in adults for specific recommendations and evidence.3

A minimum of five to six TBLB samples should be taken and increasing the number of samples will increase the diagnostic yield, particularly in those with a localised lesion.164 A direct correlation has been shown between the number of TBLBs taken and diagnostic yield, with a 35% diagnostic yield for one to three biopsies, increasing to 69% with 6–10 biopsies.164 Furthermore, Popovich et al148 demonstrated in peripheral lung cancer lesions biopsied using TBLB that only a 45% diagnostic yield was achieved with one biopsy and in some cases the diagnosis was not achieved before the fifth or sixth biopsy. They suggested that up to 10 biopsies may be required.

There is conflicting evidence regarding the use of fluoroscopy during TBLB. A retrospective review by Anders et al of 112 TBLBs performed with fluoroscopy and 135 without fluoroscopy found no significant difference in the yield (72.3% vs 67.4%); however the yield for localised lesions was lower without fluoroscopy.165 Fluoroscopy may be useful when performing TBLB on localised peripheral lesions. The diagnostic rate of TBLB without fluoroscopy varies between 60% and 65%,159 ,166 and with fluoroscopy 55.2^–73%43 ,165 for localised lesions. It has therefore been suggested that fluoroscopy may be useful when performing TBLB on localised peripheral lesions. Two studies have demonstrated that knowledge of the location of the lesion from chest radiograph aids diagnostic yield. To ensure that peripheral localised lesions are being accessed correctly, fluoroscopy may therefore be required.43 ,167 The incidence of pneumothorax following TBLB without fluoroscopy is between 3% and 5%,166 ,168 although one study demonstrated no increase in the incidence of pneumothoraces (0.015%).157 A chest x-ray may be performed immediately if pneumothorax is suspected or the patient is symptomatic, or be done at least 2 h after TBLB to exclude pneumothorax.47

Evidence statements

• EBBs should be performed in patients with suspected sarcoidosis. Yield increases in patients with radiographic evidence of parenchymal disease. (Evidence level 2+)

• TBLBs should be performed in patients with radiographic stage II—IV sarcoidosis. (Evidence level 2+)

• At least five to six TBLBs should be performed in patients with ILD. (Evidence level 2−)

• There is no convincing evidence that the use of fluoroscopy reduces pneumothorax rate or diagnostic yield in non-focal lung disease. (Evidence level 3)

Recommendations

• In suspected sarcoidosis, EBBs should be considered to increase the diagnostic yield. (Grade C)

• TBLB is recommended for the diagnosis of stage II–IV sarcoidosis. (Grade C)

• In patients with diffuse ILD, five to six TBLBs should be taken from the same lung. (Grade D)

• Fluoroscopy should be considered for TBLB in patients with localised or focal parenchymal lung disease. (Grade D)

Good practice point

• Bronchoscopists should keep a record of their personal diagnostic accuracy for FB. (√)

Diagnosis of infection

Bronchoscopy is a useful procedure to diagnose infection when less invasive methods to diagnose the underlying aetiology are not suitable.

This section includes basic sampling methods (bronchial wash, BAL, TBLB, EBB and bronchial brush) but does not include advanced techniques such as EBUS TBNA, which are covered in the advanced bronchoscopy guidelines.3

The spectrum of infections encountered in immunocompromised hosts differs from that seen in immunocompetent individuals. The role of FB in obtaining diagnostic samples in these two patient groups also differs

Bronchoscopy in the diagnosis of infection in the immunocompromised host

The underlying aetiology for pneumonia in patients who are immunocompromised varies with the prevalence of disease in the community and with geographical area. Studies performed in Africa where there is a high prevalence of TB have shown the most common organism causing pneumonia in patients with HIV-related immunosuppression is mycobacterium TB.169 ,170 In Western countries with a lower prevalence of TB, commonly encountered organisms are Pneumocystis jiroveci, fungal infections, bacterial infections and viral infections such as cytomegalovirs and Cryptococcus.171^–173 A variety of bronchoscopic techniques can be used depending on the local population, disease prevalence and host defences (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Diagnostic algorithm: bronchoscopy in the immunocompromised host.

Bronchoalveolar lavage

BAL has been shown to have high sensitivity in P jiroveci pneumonia. In the absence of prior antibiotic use for treatment or prophylaxis, BAL is reported to have a sensitivity between 90% and 98% for P jiroveci.174^–179 Prior use of empiric antibiotic treatment reduces the sensitivity to around 64%.178 Bilateral BAL sampling has a higher diagnostic yield compared with unilateral BAL179 and BAL from upper lobes has higher sensitivity compared with lower or middle lobe sampling techniques.180 ,181 The comparable sensitivity of BAL to transbronchial biopsy,176 but with fewer inherent risks, means that BAL is regarded as the gold standard investigation for suspected P jiroveci infection.

BAL in patients who are immunocompromised with smear-negative pulmonary TB has been reported to have a sensitivity of 10–30% based on microscopy182^–184 and a sensitivity of 52–95% on culture.182^–184 When PCR techniques are used on BAL material, a rapid diagnosis of TB with reported sensitivity of 85.7% and specificity of 90.9%185 have been reported. Serological tests on BAL in patients with HIV (unlike those without HIV) with suspected pulmonary TB have shown a lower sensitivity, but the specificity remains high.185

In invasive aspergillosis, BAL is reported to show characteristic hyphae in 34–64%186^–188 and to be positive on cultures for Aspergillus in 23–85% of cases.186^–188 If available, galactomannan antigen detection on BAL is useful. A meta-analysis looking at the performance of BAL galactomannan testing in detection of proven invasive aspergillosis suggested that the test has a sensitivity of 94% and a specificity of 79%.189 Unlike the serum galactomannan test, BAL galactomannan is equally sensitive in patients with haematological or other causes of immunosuppression.190 There have been reports of false-positive results when using the serum galactomannan test, especially in patients who are receiving certain β-lactam antibiotics.191^–194 The reported sensitivity for PCR on BAL in cases of invasive aspergillosis is 67–100% and the specificity 96–100%.195 ,196 The PCR techniques have some limitations, including lack of standardisation for processing of samples.

BAL and bronchial washings in patients who are immunocompromised with bacterial pneumonia have been shown to change the diagnosis in 50% and change antibiotic therapy in 62% of patients who are HIV positive.197 In patients with AIDS and suspected cryptococcal disease, bronchoscopies with BAL and bronchial washings have a combined sensitivity on smear equal to that of TBLB and superior to that of TBLB fungal culture.198

TBLB/EBB and brush biopsies

TBLB has been reported to have a comparable yield to BAL in cases of P jiroveci pneumonia in some studies,171 ,199 ,200 although other studies have shown it to be inferior175 ,176 and some have shown it to have higher sensitivity than BAL.177 Since TBLB carries the risk of pneumothorax of between 2.5% and 3.7%,176 ,177 it is not recommended in patients with suspected P jiroveci pneumonia. In areas of low TB prevalence where Pneumocystis is more likely to be responsible for infection,173 TBLB may not be required during the initial bronchoscopy.

In cases of pulmonary TB in immunocompromised hosts, the sensitivity of TBLB based on positive microscopic examination is reported to be between 19%183 and 39%.182 Although fewer granulomata are seen on TBLB in patients who are immunocompromised compared with those who are immunocompetent, TBLB was the reported exclusive method of diagnosis in around 13% of patients in a study.183 The overall reported sensitivity for cultures of TBLB is reported to be 42–52%.182^–184

EBB and TBLB have a lower diagnostic yield compared with BAL in the diagnosis of invasive aspergillosis.197 ,201 Biopsy may confirm the presence of hyphae and demonstrate pathognomonic tissue damage due to hyphae. Careful consideration must therefore be given to the risk versus benefit, especially the risk of bleeding if concurrent thrombocytopenia is present, or of pneumothorax in these critically ill patients.

Brush biopsies to diagnose P jiroveci pneumonia have been consistently found to have lower sensitivity compared with BAL, with reported sensitivity between 29% and 52%.171 ,175 ,176 ,199 Due to its limited sensitivity, brush biopsy has a limited role in this setting.

Evidence statements

• BAL has excellent yield in P jiroveci pneumonia and is now considered the gold standard. (Evidence level 3)

• BAL has a good diagnostic yield for the majority of infections and may be sufficient as the only bronchoscopic investigation in countries with a low prevalence of TB. (Evidence level 3)

• BAL can show characteristic hyphae and has moderate sensitivity for invasive aspergillosis based on cultures.(Evidence level 3)

• BAL galactomannan test has excellent sensitivity and specificity for invasive aspergillosis and should be considered if available. (Evidence level 3)

• Post-bronchoscopy sputum is a complementary method of collecting material for staining for AFB and should be collected if the patient is able to provide a sample. (Evidence level 3)

Recommendations

• In patients with pulmonary infiltrates who are immunocompromised and in whom TB is considered unlikely, BAL alone is usually sufficient to achieve a diagnosis. In areas or populations with high prevalence of TB, TBLB may be considered in addition. (Grade C)

• BAL or bronchial washings should be sent for microscopy for AFB and for mycobacterial culture in patients with pneumonia who are immunocompromised. (Grade C)

• Post-bronchoscopy sputum could be collected in patients who are immunocompromised and suspected to have TB. (Grade D)

• TBLB and EBB for invasive aspergillosis may be avoided if BAL galactomannan test is available due to the high sensitivity and specificity of the latter and inherent risks with the biopsies. (Grade C)

• In patients suspected to have invasive aspergillosis, BAL should be sent for microscopy for hyphae and fungal culture; a BAL galactomannan test should be considered to further improve diagnostic yield. (Grade C)

Bronchoscopes used in the ICU

The specifications of a bronchoscope suitable for use in the ICU require a degree of compromise. The internal diameter of the ET may restrict the size of the bronchoscope, while efficient suctioning requires a larger bronchoscope with a wide suction channel. The internal diameter of the tracheal tube relative to the external diameter of the bronchoscope is an important consideration. Bronchoscopes in the non-intubated patient occupy only 10–15% of the cross-sectional area of the trachea. In contrast, a 5.7 mm bronchoscope occupies 40% of a 9 mm ET and 66% of a 7 mm tracheal tube. Failure to recognise this may lead to inadequate ventilation of the patient and impaction of or damage to the bronchoscope.

Lubrication is essential to facilitate passage of the bronchoscope. Tracheostomy tubes are prone to damage the bronchoscope, particularly during withdrawal when the rigid edge of the end of the tracheostomy tube can abrade the covering of the bronchoscope.

A single model (non-human) study addresses the issue of airway support device and size of bronchoscope.225 In this study, the laryngeal mask airway (LMA), reinforced LMA (RLMA) and ET were assessed in a simulated model of ventilation, measuring flow resistance during simulated inspiration and the maximum size of flexible bronchoscope (using oesophageal dilators to mimic six different sizes of bronchoscope (2.7–8.9 mm external diameter)). This study demonstrated that the LMA can accommodate a larger bronchoscope than the corresponding sizes of RLMA and ET. Mean flow resistance was 2.3 (1.7–3.5) times higher with the RLMA and 2.1 (1.2–4.2) times higher with the ET compared with the LMA, and 1.2 (0.7–1.8) times lower with the ET compared with the RLMA. Removal of the LMA aperture bars resulted in a mean decrease in flow resistance of 3.6%. The study provides useful information on the maximum external diameter of the bronchoscope to be used with different airway devices. This information is provided in appendix 11 of this guideline.

Recommendation

• The external diameter of a bronchoscope used in the ICU setting should be carefully selected according to the external diameter of the bronchoscope, the size of the airway support device (ET or laryngeal mask) and the type of airway device used. (Grade D)

Indications for FB in the ICU

FB may be performed in the ICU for a wide range of diagnostic or therapeutic indications. These are considered separately in the sections below.

Precautions and contraindications to bronchoscopy on the ICU

Critically ill patients represent a high risk group for most invasive procedures. They will often present with hypoxia, electrolyte disturbances, clotting abnormalities and arrhythmias. It is therefore important that the potential benefits of bronchoscopy outweigh the risks. Elevated prothrombin time, increased activated partial thromboplastin time, reduced fibrinogen titre or thrombocytopenia indicate clotting dysfunction which makes biopsy procedures hazardous. Critically ill patients may be more susceptible to the toxic effects of local anaesthetics. However, in the ventilated patient intravenous sedation or anaesthesia is probably the most appropriate alternative.

A case series of 37 ventilated patients with severe thrombocytopenia (platelets<50) undergoing bronchoscopy (33 BALs, 5 TBLBs)263 demonstrated only two deaths associated with the procedure (5.4%), one of which was secondary to a major airway bleed. Minor tracheobronchial bleeds requiring no treatment were seen in 32%. In this study, it is not clear which patients required platelet transfusions. However, overall, it would appear that bronchoscopy in patients with low platelets is feasible (if corrected) but associated with significant potential harm.

In a case series of 110 mechanically ventilated, highly selected patients with ARDS on standard criteria (excluded if partial pressure of O₂ <80 mm Hg on FiO₂ of 1.0, hypotension, ischaemic heart disease, raised ICP or ET <7 mm diameter), bronchoscopy and BAL appeared to be relatively safe in the short term (1 h) with no clinically significant changes in oxygenation.264 One episode of pneumothorax (without transbronchial biopsy) occurred. Thus bronchoscopy and BAL appear to be safe in the short term in highly selected patients with ARDS in terms of oxygenation.

Recommendations

• Patients in the ICU should be considered at high risk from complications when undergoing bronchoscopy. (Grade D)

• All potential risk factors (ventilator parameters, clotting dysfunction) should be corrected as far as possible before undertaking bronchoscopy. (Grade D)

Good practice point

• The risks and benefits of bronchoscopy should be carefully considered in mechanically ventilated patients. (√)

Monitoring

A full range of physiological monitoring is available in the ICU, including ECG (for heart rate and rhythm), continuous intra-arterial blood pressure or intermittent cuff blood pressure measurement, pulse oximetry (SpO₂) and end-tidal CO₂ monitoring. Setting appropriate alarm limits for heart rate, blood pressure and SpO₂ and requesting other attendant staff to monitor physiological variables during the bronchoscopy allows close monitoring for deterioration during the procedure. Adverse events require immediate withdrawal of the bronchoscope and resuscitation of the patient. The clinician must then weigh the benefits against the risks of proceeding further.

A total of 13 studies have addressed potential physiological parameters which may be altered by bronchoscopy in ventilated/ICU patients.239 ,243 ,246 ,263^–271 A randomised trial of different BAL volumes using bronchoscopy in mechanically ventilated patients demonstrated the procedure to be associated with a worsening of PaO₂/FiO₂ ratio.265 A number of case series suggest bronchoscopy and BAL are associated with variable and sometimes non-significant increases in mean arterial pressure, mean pulmonary artery pressure and CO₂ in ventilated patients.264 ,266^–271 Bronchoscopy and BAL are associated with worsened hypoxia and tachycardia in several studies in mechanically ventilated patients,239 ,243 ,263^–269 ,271 although these changes were clinically insignificant in most patients. In a specific study with a specific subgroup of patients with ARDS, a more than 30% drop in PaO₂ was observed in 35% of patients.271 A more detailed physiological study in 18 patients undergoing bronchoscopy and BAL has demonstrated adverse change in lung compliance and resistance.267 Post-bronchoscopy pneumothorax has been reported in ventilated patients in whom transbronchial biopsy was not undertaken.246 ,264

Evidence statements

• Bronchoscopy with BAL in mechanically ventilated patients has been shown to be associated with a worsened oxygenation, with an 80–86% reduction in PaO₂/FiO₂ ratio from baseline regardless of the volume of BAL used. (Evidence level 1+)

• BAL has been associated with a change in lung compliance, an increase in mean arterial pressure, mean pulmonary artery pressure and rise in CO₂ in ventilated patients. Hypoxia and tachycardia have been shown to be a result of BAL in several studies in mechanically ventilated patients, although these were clinically insignificant in most patients. (Evidence level 3)

• Post-bronchoscopy pneumothorax can occur in mechanically ventilated patients in the absence of TBLB. (Evidence level 3)

Recommendations

• Continuous multimodal physiological monitoring should be undertaken during and after bronchoscopy in the ICU setting. (Grade C)

• Patients should be monitored post procedure for complications, including pneumothorax even when a biopsy has not been taken. (Grade D)

Ventilator settings

Preoxygenation should be achieved by increasing the inspired oxygen concentration to 100%. One hundred percent oxygen should be given during bronchoscopy and in the immediate recovery period. The ventilator should be adjusted to a mandatory setting. Triggered modes such as pressure support or assist control will not reliably maintain ventilation during bronchoscopy. The ventilator pressure limit should be increased to ensure that adequate tidal volumes are delivered during each respiratory cycle and the ventilator rate increased if necessary. Most modern microprocessor-controlled ventilators will monitor tidal volume and minute ventilation.

A special swivel connector (Portex, Hythe, UK) with a perforated diaphragm, through which the bronchoscope can be inserted, allows continued ventilation and maintenance of (positive end-expiratory pressure PEEP)/CPAP. This is particularly important when performing a bronchoscopy in patients with hypoxia (eg, ARDS) when maintenance of PEEP is important.

There is a small case series in normal human volunteers272 and a single study conducted in a dog model and six healthy human volunteers273 in support of the use of jet ventilation via the bronchoscope in ventilated patients. An 11-patient case series in healthy volunteers is in support of the use of high-frequency positive pressure ventilation during bronchoscopy.274 These techniques may be useful during bronchoscopy in the ICU, but more robust comparative studies are needed to quantify any potential benefit.

A total of three case series (totalling 46 patients)24 ,275 ,276 suggest non-invasive ventilation via mask or helmet may be helpful in preventing progression to mechanical ventilation in non-ventilated patients with hypoxia undergoing bronchoscopy, but further studies are needed in this area. A small (n=30) but well conducted randomised trial suggests CPAP plus oxygen is superior to oxygen alone during bronchoscopy of patients with hypoxia, preventing reduction in saturations and the need for ventilator support post procedure.268 A single case series of 40 patients assessed those requiring non-invasive ventilation for respiratory failure prior to a decision to conduct bronchoscopy.277 In these patients, 4/40 (10%) required mechanical ventilation within 8 h. Within 48 h, 45% were intubated, and in total 55% required intubation (median time from bronchoscopy to intubation=27 h). This implies that patients requiring non-invasive ventilator support prior to bronchoscopy should be considered high risk for requiring intubation post procedure, and that the procedure should therefore be conducted by an experienced operator in a setting where intubation and ventilation can occur.

Monitoring intracranial pressure (ICP) in patients with a head injury is essential if sudden rises in ICP are to be avoided due to CO₂ retention or other causes. Monitoring endotracheal CO₂ in such patients may also help to detect falls in minute ventilation caused by the presence of the bronchoscope within the ET. Profound anaesthesia, including effective neuromuscular blockade, is often used in patients with head injury while undergoing bronchoscopy. Four studies (one randomised controlled trial270 and three others278^–280) addressing this issue have consistently demonstrated that bronchoscopy raises ICP. A small but well conducted randomised study assessing hyperventilation via the ventilator during bronchoscopy demonstrated no difference in peak ICP,270 but suggested hyperventilation may accelerate return to baseline ICP after bronchoscopy in patients with closed head injury. The significance of this treatment effect on clinically important recovery is unknown. Three small non-randomised studies suggest that while ICP is significantly raised during bronchoscopy, cerebral perfusion pressure is probably maintained.278^–280

Evidence statements

• A small but well conducted randomised study suggests hyperventilation via the ventilator is useful in reducing the expected rise in ICP seen during bronchoscopy in ventilated patients with closed head injury. (Evidence level 1−)

• A single but well conducted randomised trial suggests CPAP plus oxygen is superior to oxygen alone during bronchoscopy of patients with hypoxia, preventing reduction in saturations and the need for ventilator support post procedure. (Evidence level 1+)

• A total of three case series suggest non-invasive ventilation may be helpful in preventing mechanical ventilation in patients with hypoxia undergoing bronchoscopy, but further studies are needed in this area. (Evidence level 3)

Recommendations

• CPAP plus oxygen support may be considered in patients with hypoxia undergoing bronchoscopy to prevent desaturation and post-procedure requirement for mechanical ventilation. (Grade B)

• When patients require non-invasive ventilation prior to bronchoscopy, the procedure should be conducted in an environment where intubation and ventilatory support are readily accessible. (Grade D)

• Bronchoscopy should be undertaken cautiously in patients with documented or suspected raised ICP. (Grade D)

Good practice point

• Care must be exercised to ensure adequate ventilation and oxygenation is maintained during bronchoscopy in intubated patients. (√)

Sedation and analgesia

The clinical status of the patient will often determine the type and level of sedation required for bronchoscopy. Synthetic narcotics such as alfentanil or fentanyl will suppress cough and provide profound analgesia. Sedation can be induced using incremental doses of a benzodiazepine or propofol. Some patients may only require light sedation for comfort during mechanical ventilation. In such cases bronchoscopy can be undertaken with supplemental topical anaesthesia with lignocaine injected through the bronchoscope. However, care must be exercised in patients with combinations of renal failure, liver dysfunction and congestive heart failure when accumulation of lignocaine may be associated with side effects (please see section on ‘Tpoical anaesthesia’).

One case series of 104 patients undergoing bronchoscopy during mechanical ventilation indirectly addressed the issue of sedation level during bronchoscopy.271 The primary purpose of the study was to assess gas exchange differences before and after the procedure, however the study demonstrated significant falls in oxygenation (PaO₂ < 60 mm Hg with FiO₂ = 0.8) in 14 patients, and this fall was independently associated in multivariate analysis with the presence of ARDS and a clinician's observation of the patient ‘fighting the ventilator’. This study therefore provides weak indirect evidence that adequate sedation, to the point of the patient not fighting the ventilator, may be helpful in preventing hypoxia during bronchoscopy in ventilated patients.

Evidence statement

• Adequate sedation, to the point of the patient not fighting the ventilator, may be helpful in preventing hypoxia during FB in ventilated patients. (Evidence level 3)

Recommendations

• Adequate sedation and analgesia should be provided for patients undergoing bronchoscopy in an intensive care setting. The risks of these procedures should be carefully balanced with their potential benefit in ventilated patients. (Grade D)

• Clinicians administering sedation/anaesthesia/analgesia should be acquainted with the use of these agents, and the anaesthetist/intensivist is usually best placed to fulfil this role. (Grade D)

Nursing skill mix and staffing levels

There is limited evidence available regarding skill mix and nurse staffing levels required during bronchoscopy procedures (www.nrls.npsa.nhs.uk/resources/?entryid45=59896NPSA/2008/RRR011). National Patient Safety Alert RRR011 (2008), concerning administration of sedatives during procedures such as bronchoscopy, must be considered when assessing nurse skill mix and staffing levels. The BSG guidelines (2001)289 have made recommendations for two service models for endoscopy procedures with the recommendation that levels increase when more than one procedure room is available.

Patients require very careful monitoring of sedation levels during the procedure and throughout the recovery phase. This requires knowledge and experience of the effects and adverse effects of the sedatives and the ability to administer drugs when required.

Recommendations

• A minimum of two qualified nurses are required during bronchoscopy procedures: one assistant nurse and another dedicated to monitoring patients response to the medication and procedure. (Grade D)

• A qualified nurse is required to recover a patient after bronchoscopy. (Grade D)

• Advanced procedures may require additional staff. (Grade D)

Numbers of ancillary staff, for support and cleaning of equipment, may vary according to location and cleaning methods of bronchoscopes.