One hundred thirty-two patients (mean age, 52±19 years; mean simplified acute physiologic score,14±4; 85 men and 47 women) were included. All were suspected of presenting nosocomial bacterial pneumonia: 163 BALs and PSBs were performed. The mean duration of mechanical ventilation was 15±4 days.
The following were primary indications for ventilatory support: postoperative respiratory failure (n=40), exacerbation of COPD (n=19), other pulmonary diseases (n=10), neurologic emergencies (n=8), head injuries (n=8), multiple organ failure (n=13), severe sepsis (n=17), multisystem trauma patients (n=45), and miscellaneous (n=3).The diagnosis of bacterial pneumonia was established in 56 cases; polymicrobial growth was seen in 48.2% of the cases (27/56). There was no bacterial pneumonia in 107 cases.
Tolerance of Bronchoscopy
All the bronchoscopies were accomplished without complications (no hypoxemia, no pneumothorax, no hemorrhage); we have not observed any major hemodynamic difficulties during or following the procedure in any of the patients.
Squanwus Epithelial Cells: None of the samples had more than 1% of squamous epithelial cells (Table 1).
Count of Total Cells: The count of total cells was significantly higher in control patients: 6,684±27,826 vs 5,113±4,983xl03/mL (p<0.001).
Count of PMNs: The count of PMNs in BAL fluid was significantly higher in the group with nosocomial pneumonia: 85±13% versus 73±24% (p<0.01).
Infected Cells: The count of ICs was made on 100 cells (Figs 1 and 2). The percentage of ICs was significantly higher in nosocomial pneumonia: 12.6±12% (95% confidence interval [Cl], 9.3 to 16) vs 1.1±3.4 (95% Cl, 0.49 to 1.8) (p<0.0001). An ROC curve was plotted: the AUC was 0.888; the curve was homogeneous without break.
The presence of ICs in 2% or more on the Giemsa stain corresponded to a sensitivity of 84%, a specificity of 80%, a positive predictive value of 69%, and a negative predictive value ot 90%. There was no correlation (Fig 3) between the percentage of ICs and quantitative cultures of PSB (r=0.25; p=0.06). Concordance between final diagnosis and IC count was 0.82 (Cohen kappa=0.063). Among the patients receiving antibiotics, sensitivity and specificity were, respectively, 82% and 77%; among the patients without antibiotics, 85% and 83%.
Microscopic Examination: A Gram’s stain was performed on all BAL samples. The sensitivity and the specificity of the presence of bacteria in BAL fluid for the diagnosis were, respectively, 92% and 76.5%; the positive and negative predictive values were, respectively, 69% and 91%. However, in terms of qualitative agreement, the concordance was poor: in 44% of the cases, Gram’s stain was partially right (for example, gram-negative bacilli on Gram’s stain and Staphylococcus aureus plus Pseudomonas aeruginosa on PSB cultures) or totally false (gram-negative bacilli on Gram’s stain and gram-positive on PSB cultures).
Final Diagnosis: In 163 procedures, we noted 107 cases without pneumonia and 56 cases with pneumonia (Table 2): 18 cases with gram-negative bacilli, 11 cases with gram-positive cocci (principally S aureus), and 27 were polymicrobial infections (gram-negative bacilli and gram-positive cocci).
Table 1—Analysis of Cellularity
|Group Without Pneumonia (n=107)||Group With Pneumonia (n=56)||p Value|
|Total cells, xl03/mL||6,684 ±27,826||5,113±4,983||<0.001|
|PMNs, %||73±24||85 ±13||<0.01|
Table 2—Main Bacteriologic Results in Infected Patients (continued)
|Bacteria||PSB Cultures, cfu/mL||IC, %||PMN, %||Cells, 103/mL|
|9||E coli||5 x10s||2||90||13,320|
|10||S aureus||5 x10s||2||90||1,568|
|C itrobacter freundii||500|
Figure 1. ROC curve for the ICs count.
Figure 2. Comparison of the percentage of ICs between patients with and without YAP (mean and 95% Cl).
Figure 3. Correlation between the percentage of ICs (% of IC) and PSB cultures (log10 cfu/mL) in case of pneumonia.