A comparison of mini-bronchoalveolar lavage and blind-protected specimen brush sampling in ventilated patients with suspected pneumonia

Mini-bronchoalveolar lavage (m-BAL) and blind-protected brush sampling (b-PSB) are minimally invasive methods of diagnosing pneumonia in mechanically ventilated patients. The aim of this study was to compare these techniques in a prospective study at a medical and surgical intensive care unit in a university-affiliated community teaching hospital. One hundred and ninety episodesof pneumonia was suspected in 175 mechanically ventilated patients. Sequential b-PSB followed by m-BAL were performed by respiratory therapists who had undergone specialized training. A b-PSB quantitative culture greater than 1,000 cfu/mL and a m-BAL greater than 10 4 cfu/mL was considered diagnostic of pneumonia. Colony counts between 500 and 1,000 efu/mL and 5,000 − 10 4 cfu/mL, respectively, were considered borderline positive. One hundred and ninety paired specimens were obtained from 175 patients. The diagnostic agreement between the two techniques was 90%. Sixty-six patients (37%) were considered to have bacterial pneumonia. In 108 episodes, patients were receiving concurrent antibiotics; pneumonia was diagnosed in 30 (27%) of these cases compared with 36 of 82 (43%) episodes off antibiotics ( P = .03). In 6 episodes, m-BAL was negative and b-PSB was positive (1 patient receiving antibiotics). In 13 episodes, b-PSB was negative and m-BAL was positive (7 patients were receiving antibiotics). The b-PSB took 30 ± 8 seconds to perform and was complicated by minor bleeding in 3 cases. The m-BAL took 5 ± 2 minutes to perform, was considered easy in 105 cases, difficult/very difficult in 63, and failed in 2 patients. Significant coughing occurred in 98 patients with other minor reversible complications occurring in a further 20 cases. In mechanically ventilated patients withsuspected pneumonia both b-PSB and m-BAL can be performed safely by respiratory therapists. The tests complement each other and likely reduce the number of false-negative results.