Published ahead of print on February 10, 2006, doi:10.1164/rccm.200510-1556OC
© 2006 American Thoracic Society doi: 10.1164/rccm.200510-1556OC
Postoperative Pneumonia after Major Lung Resection Departments of Thoracic Surgery, Intensive Care Medicine, Anesthesia, Pneumology, and Microbiology, Hotel-Dieu Hospital, Public Assistance-Paris Hospitals, Paris V University, Paris, France Correspondence and requests for reprints should be addressed to M. Alifano, M.D., Unité de Chirurgie Thoracique, Hôpital Hôtel-Dieu, 1 Place Parvis de Nôtre Dame, 75004 Paris, France. E-mail: marcoalifano{at}yahoo.com
Background: Postoperative pneumonia (POP) is a life-threatening complication of lung resection. The incidence, causative bacteria, predisposing factors, and outcome are poorly understood. Design: Prospective observational study. Methods: A prospective study of all patients undergoing major lung resections for noninfectious disease was performed over a 6-mo period. Culture of intraoperative bronchial aspirates was systematically performed. All patients with suspicion of pneumonia underwent bronchoscopic sampling and culture before antibiotherapy. Results: One hundred and sixty-eight patients were included in the study. Bronchial colonization was identified in 31 of 136 patients (22.8%) on analysis of intraoperative samples. The incidence of POP was 25% (42 of 168). Microbiologically documented and nondocumented pneumonias were recorded in 24 and 18 cases, respectively. Haemophilus species, Streptococcus species, and, to a much lesser extent, Pseudomonas and Serratia species were the most frequently identified pathogens. Among colonized and noncolonized patients, POP occurred in 15 of 31 and 20 of 105 cases, respectively (p = 0.0010; relative risk, 2.54). Death occurred in 8 of 42 patients who developed POP and in 3 of 126 of patients who did not (p = 0.0012). Patients with POP required noninvasive ventilation or reintubation more frequently than patients who did not develop POP (p < 0.0000001 and p = 0.00075, respectively). POP was associated with longer intensive care unit and hospital stay (p < 0.0000001 and p = 0.0000005, respectively). Multivariate analysis showed that chronic obstructive pulmonary disease, extent of resection, presence of intraoperative bronchial colonization, and male sex were independent risk factors for POP. Conclusions: Pneumonia acquired in-hospital represents a relatively frequent complication of lung resections, associated with an important percentage of postoperative morbidity and mortality.
Key Words: colonization • lung resection • postoperative pneumonia • risk factors • thoracic surgery Major lung resection carries high morbidity and significant mortality (1–3). It has been generally believed that bronchopleural fistulas and respiratory failure are the most important determinants of postoperative morbidity and mortality. It has been shown that, among all types of surgery, noncardiac thoracic surgical procedures are frequently complicated by postoperative pneumonia (POP) (4). POP is especially severe after thoracic surgery and is associated with high mortality (5). The exact incidence of POP after lung resection is, paradoxically, poorly known; the risk factors and causative pathogens have been rarely investigated and little is known about the pathophysiology of POP (6–11). Furthermore, POP may be difficult to diagnose because of the frequent occurrence of fever, hypoxemia, or abnormal chest X-ray after lung resection. As a consequence, strict criteria (including clinical aspect, radiographic abnormalities, laboratory results, and culture of representative specimens) have been used only rarely to define this entity. Part of our current knowledge is derived from retrospective series (1, 12–30) (Table 1). Few prospective studies (2, 7, 10, 11, 31–52) (Table 2) have been performed, most with important limitations.
Patients undergoing major lung resections carry many risk factors for nosocomial pneumonia, including a frequent smoking history (4, 17, 53), coexistent chronic bronchitis or chronic obstructive pulmonary disease (COPD) (1, 12, 54), neoplastic disease, frequent extrathoracic comorbidities (6), and poor nutritional status (1, 33). Last, but not least, surgery involves the respiratory system itself with anesthesiologic and surgical implications. Postoperative administration of opioids in patients with a recent amputation of lung tissue may further induce hypoventilation with possibly related complications (6). We performed a 6-mo prospective study including all consecutive patients undergoing major lung resection for noninfectious diseases to investigate the incidence and the characteristics of POP, the risk factors (including the presence of intraoperative airway bacterial colonization), and the outcome.
Patients All patients undergoing major lung resection between June 15, 2001, and December 15, 2001, for noninfectious disease and without signs of acute respiratory infections were eligible for this study. Patients treated with antibiotics (because of respiratory or extrarespiratory infections) in the week preceding surgery were excluded from the study. Patients presenting at the date of admission for planned surgery with clinical and radiologic signs of pulmonary infection (fever greater than 38°C, purulent sputum) were excluded from this study in case surgery was urgent. In other cases, they received antibiotic treatment as outpatients and secondarily were readmitted for the planned surgery after at least 1 wk of discontinued antibiotherapy. These last patients were included in the study. Patients receiving antibioprophylaxis other than cefamandol were analyzed separately. Patients eligible for entry into the study entry but receiving explorative thoracotomy or sublobar resections because of intraoperative findings were excluded from the study. Informed consent was obtained from all patients. The research was conducted according to recommendations outlined in the Helsinki Declaration.
Study Design Information about age, sex, weight, lung function, indication for lung resection, Karnofsky index, and C-reactive protein level was collected. White blood cell count (WBC), chest X-ray, and clinical examination were systematically performed to eliminate the possibility of underlying pneumonia or bronchitis. Nutritional status was assessed by determination of body mass index and evaluation of possible weight loss in the previous 6 mo. Lung function was evaluated by spirometry and, in almost all the cases, by calculation of predictive postoperative function with perfusion lung scanning. We studied the following risk factors for POP:
All patients were intubated with a double-lumen endobronchial tube to perform single-lung ventilation. Immediately after insertion of the tube, bilateral bronchial aspirations were performed. Quantitative endobronchial aspirate samples were obtained with a sterile suction catheter equipped with a mucus collection tube (design 534-16; Vygon, Ecouen, France). The catheter was introduced through each channel of the double-lumen tube and was blindly advanced at least 30 cm before suctioning. No liquid was instilled. Intraoperative bronchial aspirates were sent to the microbiology laboratory within 15 min of collection. A patient was considered colonized if quantitative endobronchial aspirate culture at 48 h was positive with a predominant germ (i.e., pathogenic bacterial species) exceeding a cutoff value of 104 cfu/ml in at least one side. Patients received antibiotic prophylaxis with a second-generation cephalosporin (cefamandol, 1.5 g at the induction of anesthesia and postoperatively, 3 g/24 h for 48 h) except in the case of known or suspected allergy, or if a different type of prophylaxis was indicated (e.g., in the case of valve disease). Lung resections were performed according to standard techniques. Side, type of resection, possible associated sleeve bronchial or chest wall resection, previous thoracotomy, and total procedure time were recorded. A policy of early extubation was systematically employed. Decisions concerning intensive care unit (ICU) hospitalization after resection were established on the basis of type and extent of resection, predicted postoperative lung function, and associated comorbidities. Duration of ICU stay, as well as the therapeutic intensity score in ICU (OMEGA) (57, 58), were recorded prospectively. The OMEGA score takes into account several variables related to ICU stay, including duration of invasive or noninvasive mechanical ventilation and presence of hemodynamic impairment. It has been shown that it strongly correlates with total ICU costs, medical costs, and nursing requirements (58). Postoperative analgesia was achieved by one of the following methods: patient-controlled analgesia (morphine), or spinal or thoracic epidural analgesia. Patients were kept in the semirecumbent position. A regular program of physiotherapy was started on the day of operation. Oral alimentation was started on Postoperative Day 1 after lobectomy and on the second day after pneumonectomy. In cases of previous head and neck surgery or if recurrent nerve paralysis was observed, a special program for realimentation was started.
Outcome Assessment
Statistical Analysis
Between June 2001, and December 2001, 168 of 186 patients undergoing major lung resections in our department were included in the study. Eleven patients were excluded from the study because of a preexisting infection at the time of operation (chronically suppurating bronchiectasis, n = 4; tuberculosis, n = 2; other mycobacterial infection, n = 1; infected tumor under antibiotherapy, n = 4). Patients with chronically suppurating bronchiectasis and infected tumors were hospitalized for surgery while already undergoing treatment by targeted antibiotherapy, which was further continued postoperatively. Patients with mycobacterial infections were excluded from the study because of possible interference with the aim of the study by systematic pre- and postoperative antimycobacterial therapy. Antibioprophylaxis with drugs different from cefamandol was employed in seven patients because of known allergy to penicillin/cephalosporins or because coexistent cardiac valve disease was indicated. POP occurred in three of them (microbiologically proven in one case) and was responsible for one death. Demographic data of the remaining 168 patients, principal known risk factors of POP, and surgical procedures are shown in Table 3. The global score for risk of POP established by Arozullah and coworkers (4) was 32.3 ± 7.6. The main indication for major lung resection was non–small cell lung cancer. Thirty-six of 168 patients had received neoadjuvant chemotherapy.
One hundred and thirty-six (80.9%) patients underwent bilateral intraoperative bronchial aspiration. No growth was observed in 30 of 136 (22.1%) patients, whereas in 75 of 136 (55.1%) cases germ growth not reaching the defined threshold of colonization was observed. Colonization with predominant bacteria (at least 104 cfu/ml) was identified in 31 of 136 cases (22.8%). Haemophilus and Streptococcus species represented the most frequently involved microorganism (Table 4, column 1). Pseudomonas species or gram-negative enteric bacilli were cultured in six cases. Polymicrobial colonization was proven in nine patients. Five of 19 (26.3%) cultured Haemophilus strains were found positive for  -lactamase, whereas two of nine Streptococcus pneumoniae (22.2%) had a decreased sensitivity to penicillin G (minimal inhibitory concentration [MIC] > 0.06 mg/L).
In 24 of 31 (77.4%) cases, a concordance was observed for at least one germ between the two sides. In 20 of 31 (64.5%) patients, the concordance between the two sides was strict (the same germs were found on both sides), even in the three patients in whom two germs were found on each side. Among these 20 patients, the level of colonization (in terms of colony-forming units) was identical in 14 of 20 cases (70%). In 4 of 31 patients, the concordance was not strict (an identical germ on both sides plus another germ in one of the sides). In 7 of 31 cases, germs were found in only one side (without predominance with respect to the operated side). In the first 30 cases of the series, a protected distal specimen was obtained from the resected lung. In all these cases, no germ could be cultured (no growth observed), in spite of the isolation of predominant bacteria at bronchial aspiration in seven cases. Thus, the practice of systematic protected distal sampling was discontinued. The incidence of POP was 25% (42 of 168). Documented and nondocumented POP were recorded in 24 (14.3%) and 18 (10.7%) cases, respectively. Among patients with nondocumented POP, culture of plugged telescopic catheter or protected specimen brush sample was completely negative in 16 of 18 cases and bacterial growth below the defined threshold was observed in two cases. Other thoracic infections and requirement for postoperative antibiotherapy are detailed in Table 5.
POP (both documented and nondocumented) occurred primarily during the first postoperative week (35 of 42, 83.3%; Figure 1A), with half of cases occurring by Postoperative Day 4.
Haemophilus, Streptococcus sp, Pseudomonas, and Serratia species were the predominant pathogens responsible for POP (respectively: 41.7, 37.5, 25, and 12.5%). In 33.3% of documented POP (8 of 24), more than one pathogen was documented (Table 4, column 2). All the S. pneumoniae strains (6 of 6) had decreased sensitivity to penicillin G (MIC  0.06 mg/L) and four of 10 Haemophilus isolates were -lactamase positive. Seven of 10 gram-negative bacteria other than Haemophilus species had decreased sensitivity to standard targeted antibiotics (for definition, see Table 4). A comparison between bacteria isolated in documented POP and those cultured on intraoperative bronchial aspiration is shown in Table 4. Among the 31 colonized patients, 15 (48.4%) had POP (documented in 9 cases), whereas among noncolonized patients, POP occurred in 20 of 105 cases (19.0%, p = 0.004; relative risk, 3.84). The remaining seven cases of POP occurred in patients who did not receive intraoperative bronchial aspiration (n = 32). Among the nine patients with positive intraoperative aspirates (colonized patients) who developed documented POP, a concordance between the pathogen responsible for colonization and POP could be proven in six cases (85%; in two of them POP was due to two pathogens, the germ responsible for colonization and another one). The only POP related to a pathogen different from the one isolated by operative bronchial sampling occurred late in the postoperative period (Postoperative Day 11). The different germs recovered from POP are plotted against time in Figure 1B. Streptococcus species and S. pneumoniae were recovered in all cases by Postoperative Day 6. POP occurred earlier among colonized patients than among noncolonized patients (mean days, 3.5 ± 2.7 vs. 6.7 ± 4.2; p = 0.023). Figure 1C shows the time when pathogens were isolated from fiberoptic samples in either colonized and noncolonized patients developing POP. Among the 12 pathogens responsible for POP occurring by Postoperative Day 4, 10 (83.3%) were isolated from colonized patients, with a peak on Postoperative Day 1 (5 of 6). No relationship was observed between the level of colonization (104, 105, and 106 cfu/ml) and the occurrence of POP. Overall operative mortality observed during the study period was 6.5% (11 of 168). POP represented the cause of death for 8 of 11 patients. On the other hand, among patients who developed POP, the mortality rate was 19% (8 of 42), whereas the mortality rate was 2.4% among patients who did not develop POP (p = 0.00123). Mortality rates among colonized and noncolonized patients were 11.1% (3 of 27) and 4.6% (5 of 109). This difference did not reach statistical significance (p = 0.29; Table 6).
Patients with POP required noninvasive ventilation or reintubation more frequently than did patients who did not develop POP (p < 0.0000001 and p = 0.00075, respectively). POP was associated with longer ICU and total hospital stay (p < 0.0000001 and p = 0.0000005, respectively). The OMEGA score was increased in patients with POP (p < 0.0000001; Table 6). The following risk factors for the development of POP were identified on univariate analysis (Table 7): smoking history, underlying COPD (stage II, defined according to Pauwels and coworkers [59]), FEV1 less than 80% of the predicted value, male sex, alcohol consumption, intraoperative bronchial colonization, type of resection (lobectomy), previous thoracotomy, and preoperative chemotherapy.
Multivariate analysis showed that the presence of underlying COPD (stage II), the type of resection (lobectomy or bilobectomy vs. pneumonectomy), intraoperative bronchial colonization, and male sex (Table 7) were independent risk factors for the development of POP.
In this prospective study including 168 patients undergoing major lung resection for noninfectious diseases, POP occurred in 42 of them (25%). Mortality from POP was as high as 19% and POP represented the leading cause of mortality (8 of 11 deaths). Comparison with previously published data on the topic is difficult as the subject has been relatively poorly studied. Few surgical series have specifically targeted the problem of POP after lung resection, as in most instances the relative contribution of POP to the overall mortality is not stated, and data about the incidence and characteristics of POP are poor. Few studies performed so far provide sufficient information about the incidence, the clinical characteristics, the microbiological point of view, and the outcome of POP in thoracic surgery (Tables 1 and 2). In two prospective studies, both dealing with more than 100,000 patients undergoing different types of surgery, Arozullah and coworkers showed that POP was particularly frequent in general thoracic surgery (4) and was associated with a particularly high risk of death (5). Unfortunately, all types of thoracic surgical procedure were included in these studies. Extreme variability in terms of incidence of POP after lung resection is reported in both retrospective and prospective studies, with values ranging from 2 to 40%. Such variability probably depends on the characteristics of studied populations, the type of surgical resection, antibioprophylaxis, and postoperative management. Furthermore, as stated above, the criteria for defining POP and the methods for their identification are extremely variable. In our experience, a high incidence of POP (25%) was observed. Twenty-four of 42 cases of POP (57%) could be microbiologically proven with cultures of samples obtained at fiberoptic bronchoscopy. This percentage of documentation is similar to that reported in series dealing with community-acquired pneumonias (50–70%) (60) or with pneumonia in patients with lung cancer under medical treatment (61%) (61). In our series, most POP occurred in the early postoperative week. This early incidence of POP after thoracic surgery is in agreement with a previously published study (62). The documented kinetics of POP are probably related to antibioprophylaxis: on Postoperative Day 2 (immediately after the end of antibioprophylaxis), no POP could be microbiologically documented, whereas during subsequent days the documented percentage progressively increased to reach a plateau at 80%. On Postoperative Day 1, the documented percentage was relatively elevated (56%), while antibiotic prophylaxis was still ongoing: this phenomenon can be probably explained by the fact that these POP occurred in almost all instances among colonized patients, and the inoculum was probably particularly high. Furthermore, the bacteria responsible often had a reduced sensitivity to cefamandol. In our series, pathogenic bacteria were in most instances those classically reported for early hospital-acquired pneumonia: H. influenzae (41.7%), S. pneumoniae (25%), and other streptococci (12.5%) (62). Enterobacter and Pseudomonas species were responsible for 8.7 and 25% of cases, respectively. A polymicrobial etiology was recognized in 33.3% of patients. In the few previous studies performed so far to assess microbiological characteristics of POP, results were somewhat similar. Bernard and coworkers (1) found that Streptococcus species and H. influenzae were responsible for 50% of all POP, whereas gram-negative pathogens (other than Haemophilus species) accounted for 31% of pneumonias. In the experience of Sok and coworkers (7), gram-negative pathogens (other than Haemophilus species) were responsible for 71% of POP and Streptococcus species were found in only 10% of cases. Unfortunately, in this last study, single-dose cefuroxime was used as preoperative antibioprophylaxis, but 44% of patients had been treated with the same drug for 1–6 d preoperatively, with a possible impact on their bacterial flora.
In our study, culture of intraoperative bronchial aspiration showed that 22.2% of S. pneumoniae strains had decreased sensitivity to penicillin G and 5 of 19 (26.3%) Haemophilus strains were
On the other hand, all S. pneumoniae strains isolated from patients with POP (6 of 6) had decreased sensitivity to penicillin G, whereas four of nine (44.4%) Haemophilus isolates were
The possibility that colonization should be considered as a predisposing factor for POP is further strengthened by other arguments. POP developed more frequently in colonized patients than in noncolonized patients (44.4 vs. 21%, p = 0.004). Furthermore, POP in colonized patients occurred earlier in the postoperative period. Among patients colonized with S. pneumoniae, none of those with a fully sensitive strain developed POP, whereas this last occurred frequently among patients with strains characterized by reduced sensitivity to penicillin G. Our hypothesis is that cefamandol prophylaxis may prevent POP sustained by S. pneumoniae with normal sensitivity to Our results show that POP after major lung resection is frequent, severe, and precocious. Germs typically responsible for community-acquired pneumonia (64) and classically isolated from patients with COPD (27, 58, 65) are often responsible for these instances of POP. Pathogens cultured from intraoperative bronchial samples are likely to be responsible for a significant percentage of POP, but the antibioprophylaxis based on a second-generation cephalosporin is in several instances not optimal. Although antibioprophylaxis with a second-generation cephalosporin is usually effective in the prevention of wound infection and empyema, it does not specifically target the respiratory pathogens found in these patients. A more adapted prophylaxis might be able to decrease the rate of in-hospital acquired pneumonia after thoracic surgery and should be evaluated.
 Deceased  Originally Published in Press as DOI: 10.1164/rccm.200510-1556OC on February 10, 2006 Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form October 3, 2005; accepted in final form February 10, 2006
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 or Fisher's exact test as appropriate. Data processing and analysis were performed with the statistical software system SEM (Silex Development, Mirefleurs, France). A p value less than 0.05 was considered significant. The risk factors found to be predictive of POP at univariate analysis were entered into a multivariate regression analysis, to identify independent variables. 
