An integrated approach to diagnosis and management of severe haemoptysis in patients admitted to the intensive care unit: a case series from a referral centre

The study was conducted between January 1999 and December 2001 in Tenon hospital, a tertiary university hospital and referral centre for haemoptysis in Paris, France. All consecutive patients admitted to the respiratory intermediate care ward or ICU for severe haemoptysis were eligible. Exclusion criteria were iatrogenic bleeding, bleeding of gastrointestinal and oropharyngeal origin, heart failure, intra alveolar haemorrhage and incomplete data. For each patient, the following information were recorded: baseline demographics, comorbid conditions, initial clinical presentation and vital signs, laboratory tests results, chest radiography, fiberoptic bronchoscopy and CT scan findings when performed, severity of haemoptysis, and pre-ICU and in-ICU management. The persistence or recurrence of bleeding, the patients' ICU and hospital lengths of stay and their vital status at discharge were recorded, as well as the occurrence of long-term rebleeding. Patients with recurrent haemoptysis were included at the first episode only.

Our approach to initial management favoured conservative measures and BAE over surgery, whenever possible. Conservative measures included strict bed rest, nothing by mouth, and continuous monitoring of oxygen saturation, respiratory rate, heart rate and arterial blood pressure. Oxygen was delivered to obtain a pulse oxymetry value > 90%; two large-bore intravenous lines were inserted and all medications potentially increasing the risk of bleeding were stopped. Broad-spectrum antibiotics were frequently administered and no attempt was made to suppress cough. Bronchoscopic techniques were attempted to control the bleeding, using cold saline solution lavage, instillation of topical vasoconstrictive agents and/or balloon tamponade therapy. As the administration of systemic terlipressin may interfere with the success of BAE, its use was avoided whenever possible.

The selection of BAE as the first-line approach was based on the presence of severity criteria on admission. A standardized BAE procedure was used as follows: a catheter was introduced into the right femoral artery through an introducer sheath using the Seldinger technique. A 5-French pigtail catheter (Angioflex, biosphere medical, Roissy, France) with the tip located at the origin of the ascending aorta was used, and 40 ml of contrast medium was administered at 20 ml/s. Selective bronchial artery angiography was then performed, using catheters ranging from 5 to 6.5 French. Embolisation was performed when the bronchial arteries appeared to be the source of haemoptysis (tortuous hypertrophy, systemic-to-pulmonary shunting, extravasation of contrast material, or peribronchial hyper-vascularisation) or when they had a near-normal aspect but supplied the site of bleeding identified by fiberoptic bronchoscopy and/or CT scan. The material used for embolisation was 400- to 1000-μm polyvinyl alcohol particles and/or gelfoam. A visualisation of an anterior spinal artery arising from an intercostal artery deriving from the right bronchointercostal trunk was considered an absolute contraindication to embolisation. Microcatheters were not used at the time of the study. BAE was considered successful when bleeding stopped immediately after embolisation.

During the three-year study period, 230 consecutive patients were referred to our unit for severe haemoptysis. Thirty-four patients (15%) were excluded because of bleeding secondary to bronchial biopsies (n = 1), digestive tract bleeding (n = 1), pharyngeal bleeding (n = 1), heart failure (n = 3), intra alveolar haemorrhage (n = 2) and incomplete data (n = 26). Overall, 196 patients were thus included in this study. Most patients (n = 149, 76%) were referred to our unit from another hospital for consideration of BAE within 24 hours after hospital admission (1 ± 1.8 days; median 0) because haemoptysis persisted or worsened. The patients (148 males) were 51 years old. Cough, persistent bloody sputum and dyspnea were the main respiratory symptoms on admission. Physical examination revealed localized crackles in 60% of the cases (Table 1). There were mild biological consequences of bleeding regarding blood spillage and gas exchanges (Table 2).

Chronic obstructive pulmonary (n = 50, 26%) and/or cardiovascular (n = 53, 27%) disease were frequently recorded. Using our scale, the mean cumulated volume of blood loss averaged 240 ± 200 ml on admission to our unit (range, 10 to 1000 ml; median 200 ml) (Figure 1). Active tuberculosis, cancer and mycetoma were associated with a larger volume as compared with the cryptogenic group (p = 0.03; p = 0.03 and p = 0.003, respectively; Figure 2.). There were severe consequences of bleeding in 73 patients (37%) leading to the following interventions prior to the referral or during the first 24 hours of ICU admission: local (n = 23) or systemic (n = 56) terlipressin, mechanical ventilation (n = 17), blood transfusion (n = 22), vasoactive drugs support (n = 3) or cardiopulmonary resuscitation (n = 2). Patients receiving the above mentioned interventions had a higher respiratory rate on admission (24 ± 7 vs. 21 ± 6 per min; p = 0.04), a higher heart rate (91 ± 20 vs. 85 ± 20 bpm; p = 0.04), a lower room air partial pressure of oxygen in arterial blood (73 ± 15 vs. 80 ± 17 mm Hg; p = 0.03), a higher cumulated volume of blood loss (360 ± 240 ml vs. 180 ± 150 ml; p < 0.0001), and a lower haemoglobin value (11.5 ± 2.6 vs. 13.3 ± 2; p < 0.001); they also had more often active bleeding on bronchoscopy (31/68 vs. 22/114, p = 0.0003), a first-line attempt at bronchial arteriography (66/73 vs. 81/123; p < 0.0001), a need for surgery (25/73 vs. 9/123; p < 0.0001) and specific aetiologies [mycetoma (11/73 vs. 3/123; p = 0.002) and cancer (22/73 vs. 11/123; p < 0.001)], but not a higher frequency of cardiovascular and pulmonary pre-existing diseases.

Bronchiectasis (n = 78, 40%), lung cancer (n = 33, 17%), active tuberculosis (n = 27, 14%) and mycetoma (n = 14, 7%) accounted for 87% of all causes. Emphysema (n = 10, 5%), pneumonia (n = 6, 3%), pulmonary embolism (n = 2, 1%) and miscellaneous causes (n = 5, 3%) accounted for the remaining probable causes. In 21 patients (11%), no cause was evidenced. The cause of bleeding was identified in 69% (n = 111/162) of patients at bedside when combining history, comorbid conditions, physical examination, chest-X-Ray and fiberoptic bronchoscopy findings, as compared with 91% (n = 148/162) after a further CT scan (p < 0.001). The CT scan examination was especially useful for diagnosing bronchiectasis.

All patients received conservative measures. Local (n = 6) or systemic (n = 37) terlipressin, mechanical ventilation (n = 3) and blood transfusion (n = 3) were administered to the most severe patients before referral. Forty-three (22%) patients were receiving aspirin, coumadin or clopidrogel that may have worsened the bleeding, and these drugs were temporarily stopped whenever possible. Broad-spectrum antibiotics were administered to 153 patients (78%). A fiberoptic bronchoscopy was performed within 24 hours of bleeding onset in most patients (n = 184, 94%). Diffuse and bilateral (n = 13) or localized endobronchial bleeding (n = 163) was evidenced in 176 patients (96%). The bronchoscopic findings revealed a localized active endobronchial bleeding in 53 patients and a localized endobronchial clotting in 41. Otherwise, a localized endobronchial bleeding was evidenced in the upper (n = 43) or lower bronchia (n = 26) without active bleeding or clotting. In the remaining 8 patients (4%), a few signs of endobronchial blood were present. Bronchoscopic techniques were combining blood aspiration and local instillation of cold saline lavage. Vasopressors were bronchoscopically delivered in 23 patients, and a balloon was placed in one patient.

A first-line bronchial arteriography was attempted in 147 patients (75%), whereas 46 (23%) received conservative treatment. Emergency surgery was performed in 3 patients (bleeding of 700 ml revealing a cancer complicated by a cardiac arrest; bleeding of 300 ml revealing a cancer nearby the pulmonary artery; bleeding of 200 ml complicating repeated obstructive pneumonias in a patient diagnosed with a cancer) (Figure 3). The following parameters on admission were associated with the first attempt of arteriography as opposed to conservative treatment alone: a higher respiratory rate (23 ± 7 vs. 20 ± 4; p = 0.03), a greater amount of bleeding (290 ± 205 vs. 80 ± 50; p < 0.0001), a persistent bloody sputum (87/119 vs. 18/35; p = 0.02), an active bleeding on bronchoscopy (49/141 vs. 3/36; p = 0.002), the identification of a definite cause of haemoptysis (120/149 definite causes vs. 9/21 cryptogenic; p = 0.0005) and the absence of renal impairment (creatinin, μmol/l; 73 ± 22 vs. 82 ± 25; p = 0.03).

Technical failure of the attempted arteriography occurred in 15/147 (10%) patients, mostly those with mycetoma (n = 4) and cancer (n = 6). This led to either maintaining conservative measures in 9 patients or to surgery in six; 5 of these 15 patients died within the first month (4/9 patients managed conservatively and 1/6 undergoing surgery). In another patient, bleeding was related to a pulmonary artery aneurysm. Bronchial artery embolisation was eventually completed in 131/147 patients (89%) leading to an immediate control of bleeding in 106 patients (81%), 8 of whom had a secondary scheduled surgery (Figure 3).

Bleeding recurred in 7/46 patients (15%) managed conservatively, 2 of whom received BAE secondarily. Bleeding recurred in 35/131 patients (27%) receiving completed BAE. Haemoptysis recurred after 3 ± 3 days (range, 0 to 11 days) in 25 patients, who received conservative treatment (n = 4), BAE (n = 7) or surgery (n = 14). Mycetoma and cancer accounted for 50% of the early recurrences. There were 10 late recurrences (9 ± 4 months; range, 2 to 14 months) managed conservatively (n = 4) or with a second BAE (n = 3) or surgery (n = 3). Overall, surgery (pneumonectomy, n = 3; lobectomy, n = 11) was performed after 7 ± 7.5 days for early recurrences (mycetoma, n = 6; bronchiectasis, n = 3; pneumonia, n = 3; cancer, n = 2). A lobectomy was performed for late recurrences 12 ± 5 months after the initial episode.

Bronchial artery embolisation was associated with a 5% rate of complications (minor arterial dissection, n = 2; coronary ischemia, n = 2; chest pain, n = 1; transient neurological episode, n = 1; and dysphagia; n = 1). The outcome of these patients was uneventful without further intervention.

The lengths of ICU and hospital stay were respectively 5.4 ± 4.9 days (range, 0 to 47 days) and 10.7 ± 14 days (range, 0 to 142 days). Lengths of ICU (6 ± 5.4 days vs. 3.8 ± 2.5 days) and hospital stay (11 ± 14.4 days vs. 8.8 ± 12.4 days) were significantly longer for patients in whom a bronchial arteriography was first attempted, as compared with patients receiving conservative treatment alone (all p < 0.01). Therapy was withheld or withdrawn in 12 patients (6%), 7 of whom died in ICU. The ICU and hospital mortality rates were 4% (n = 8) and 8% (n = 15), respectively. At short term follow-up (one month), successful control of haemoptysis was obtained using completed BAE, conservative management or surgery in respectively 112 (57%), 44 (22%) and 22 (11%) patients. No recurrence of haemoptysis occurred in 116 (89%) of the 131 patients in whom BAE was completed, after a mean (median) follow-up duration of 20 (8) months.