DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

This study reports the results of clinical, PF, and radiological characteristics as predictors of outcome with medical therapy in 85 consecutive patients with pleural infection. This is the first series describing a prospectively identified cohort of consecutive patients with pleural infection requiring chest drainage, who have all received similar investigation and management in one center. All patients were aggressively and consistently managed by antibiotics, radiologically guided chest catheter drainage, and intrapleural fibrinolytics. Of the endpoints considered in this study, the absence of frank PF purulence was the only clinically useful predictor of successful outcome with medical therapy. However, the presence of purulent PF did not predict the failure of medical treatment and the need for surgery and/or death. Negative PF microbiological culture was associated with increased hospital stay, but none of the other endpoints was associated with length of hospital stay. There is good long-term survival from pleural infection in this series of patients.

The management strategy was designed to be aggressive and “optimal.” Radiological placement of small-bore catheters was chosen as this has been used successfully both as primary drainage and as a successful salvage technique when larger drains are blocked or fail to resolve pleural infection (14-18). Similarly, there are observational data (19) and two small randomized controlled trials (20, 21) to suggest intrapleural fibrinolytics may be a useful adjunct to aid PF drainage. Both these strategies were therefore included in this management protocol, enabling a reliable comparison between the clinical characteristics of those succeeding and those failing aggressive medical treatment. It should be noted that the efficacy of intrapleural SK is still controversial and is currently being properly assessed in a large U.K. Medical Research Council/British Thoracic Society sponsored multicenter study.

There are currently no clear data to define the point at which a patient with empyema should proceed to surgical intervention. Patients should be considered for surgery if they have a residual sepsis syndrome in association with a persistent pleural collection, despite aggressive medical management. The use of video-assisted thoracoscopic surgery (VATS) to drain an infected pleural space has been a major recent innovation in the management of pleural empyema. VATS intervention is reducing the frequency with which open surgery is needed (22) and may encourage earlier surgical intervention in the fibrinopurulent phase (22-24). The exact role of VATS in the management of empyema is currently unclear, and there is only one small controlled study directly comparing surgical versus medical therapy. Wait and coworkers (25) studied 20 patients with pleural infection, who were suitable for general anesthesia, and randomized them to receive immediate VATS or intrapleural streptokinase (IPSK) for 3 d. Chest drains were not inserted under radiological guidance in the medical group. The surgical group had higher primary treatment success and all SK medical failures were salvaged by VATS, without requiring thoracotomy. VATS was not widely available for this patient group when we began this prospective study; consequently none of the patients was referred or received VATS procedures.

None of the clinical or pleural fluid characteristics studied could reliably predict the need for surgery or death. Surprisingly, delay in the administration of antibiotics before admission to hospital and the time to drainage from recognition of pleural effusion, did not affect the failure rate of medical therapy. Previous studies have suggested that delay to drainage of an infected pleural space is associated with increased morbidity and hospital stay (3, 6-9). This may be related to sampling bias in the various series. Three of the previous reports are from surgical centers and so report a selected group of patients (3, 6, 9), and the others are not a prospective, unselected series (7, 8).

Another recent study has also tried to identify predicting factors for outcome of tube thoracostomy in patients with complicated parapneumonic effusion or empyema (26). This study is not directly comparable with ours because it reports a selected series of patients managed in other centers before referral. This case selection is emphasized by some patients receiving thoracotomy and decortication as a “first-line” procedure, the most frequent organism isolated from pleural fluid being Klebsiella pneumoniae, usually an infrequent pathogen. Not all patients received radiological drain placement and fibrinolytics were only used in 18 of 121 subjects, with only 50% success. In contrast, our study includes only patients managed in a similar and aggressive manner, and consists predominantly of patients with community-acquired infection. It is therefore likely to be more generalizable in its conclusions.

In the same series by Huang and coworkers (26), failure of tube thoracostomy was associated with the presence of loculations, larger effusions, and a low PF white cell count. The finding of a low leukocyte count was surprising to the authors of the study, and seems counterintuitive. This observation is also inconsistent with the results of our study. We found that macroscopically purulent PF (indicating many white cells) was associated with a poorer outcome. We hypothesize that the differing results may be related to the case selection in the Taiwan series.

PF pH predicts the likelihood of spontaneous resolution of parapneumonic effusion and when PF pH is below 7.2 it is unusual for effusions to resolve without formal drainage (13, 27– 30). Consistent with the established evidence (13, 29), we use PF pH as a bedside test to identify patients requiring pleural drainage, but these results suggest this index is not also helpful at predicting which patients will go on to fail medical treatment after drainage. This is consistent with the pathophysiology of pleural infection where the invasion of bacteria into the pleural space is associated with the development of PF acidosis, but it is primarily the reduction in fibrinolytic activity with associated fibrin deposition and consequent fibroblast activation that will reduce the likelihood of successful PF drainage and therefore predict the success or failure of medical therapy (5).

In our series, some patients without frank purulence or acidosis demonstrated loculation of their effusion and a sepsis syndrome, and therefore had pleural drainage and fibrinolytics. This included two patients with an initial pleural pH > 7.2 who failed to resolve their sepsis syndrome and required surgical debridement despite aggressive medical management. These occasional cases confirm that although pleural pH is specific in predicting the need for pleural drainage (13), it is less than 100% sensitive; an observation consistent with that of Poe and coworkers (31), who found that PF analysis including pH did not accurately predict eventual surgery in a selected series of patients. Clinicians should be aware of this occasional trap.

Delay to antibiotic administration is thought to be of significance in the development of pleural infection (5, 32), and in this series there was a high incidence of antibiotic delay which we have defined as greater than 4 d from seeking medical advice for the current illness. Despite a delay in receiving antibiotics prior to admission in 39% patients, the difference in antibiotic use was not associated with a varying outcome in this series. There was no association with the type of antibiotic given by the general practitioner and patient outcome.

There was an increase in the total duration of hospital stay and stay after drain insertion, in patients with negative PF microbiological culture. This is difficult to explain, but may represent the presence of anaerobic bacteria not identified by our laboratory. A previous series has suggested that pleural infection caused by nonpneumococcal organisms, including anaerobes, is associated with a longer clinical recovery (30). There were, however, no clinically or statistically significant differences in the total or post–drain insertion duration of hospital stay with any of the other variables.

Our series does not support the radiological findings of the British Thoracic Society (BTS) study of empyema, where larger pleural collections (> 40% of the hemithorax) were more likely to require surgery (11). Theoretically, pleural collection size should not be the main determinant of the need for surgery—it is the failure of adequate drainage rather than initial size of pleural collection that will lead to the failure of treatment. Inadequate drainage may result from a number of possible factors including poor drain placement, viscous pus, and/or loculations and drain blockage. It is possible that our use of radiological catheters and fibrinolytics (neither widely used in the BTS series) minimized the effects of these factors and hence reduced the effect of pleural collection size.

A subgroup of 46 patients had CT scanning as part of initial assessment for placement of drainage catheter. Pleural thickening in patients with parapneumonic exudates is usually smooth with enhancement of the parietal pleura and extrapleural tissues (33-35). It has been suggested that pleural thickness is greater in patients subsequently requiring decortication for empyema (34), but this is not confirmed by our series. We found the thickness of the pleura in the group failing medical treatment was no different from that in those treated successfully, and some of those responding to medical treatment had substantially thicker pleural peels than some requiring surgery (Figure 2). Thus, pleural thickness on CT scan does not predict outcome either, with pleural peels resolving over several weeks in patients spared surgery (36). Our data do not, therefore support the common opinion that a thick pleural peel makes surgical intervention mandatory.

Perhaps the most interesting feature of this report is the low in-hospital mortality rate (4.7%), compared with some other reported series (2, 4, 9, 11, 37, 38). This shows that early and aggressive medical management for patients with pleural infection is associated with a good outcome and emphasizes the need for rapid and effective intervention in this disease. As well as aggressive medical therapy, this proactive approach should also include an early recognition of failing medical therapy and prompt radiological imaging and surgical involvement.

This is the first study reporting the long-term prognosis in a cohort of consecutive patients with pleural infection. Late mortalities (occurring after hospital discharge) all occurred in the first 400 d after drain insertion and in patients with serious comorbid conditions such as malignancy and ischemic heart disease. The postdischarge prognosis in patients without comorbidity was excellent, with no deaths in this group after discharge.

In conclusion, this study shows that an aggressive medical strategy is associated with a low failure rate in in-hospital medical therapy. It also shows that there are no reliable clinical, PF, or radiological characteristics that will predict failure of medical therapy at the time the patient is first admitted. In particular, outcome was unrelated to pleural thickness on CT scan. The absence of frank pleural purulence does predict a high success rate with medical therapy, but in this series three of 11 (27%) patients requiring surgical drainage had nonpurulent fluid. The long-term survival in these patients is 86% at 4 yr with all late deaths being related to comorbidity and not the primary pleural infection.