Pleural Effusions Following Cardiac Surgery: Prevalence, Risk Factors, and Clinical Features FREE TO VIEW

It is estimated that >600,000 persons undergo coronary artery bypass graft (CABG) surgery annually in the United States,^1,2 which is one of the most frequently performed major surgical interventions in North America. During the period immediately following myocardial revascularization, 41% to 87% of patients have chest radiograph findings consistent with pleural effusion.^2–5 Most are small-volume, left-sided effusions associated with the harvesting of the left internal thoracic artery that resolve without intervention. However, in approximately 10% of patients, the effusion occupies >25% of the hemithorax on chest radiograph.⁶ In addition to the morbidity that ensues, pleural effusion frequently causes hospital readmission following CABG surgery. Despite this, the prevalence of symptomatic effusion requiring therapeutic intervention is still largely unexplored.

Post-CABG surgery pleural effusions can be categorized into early and late effusions.⁷ Early effusions are those occurring during the first postoperative month, whereas late effusions develop after this period. According to Sadikot et al,⁷ early effusions are generally hemorrhagic and associated with a high pleural fluid lactate dehydrogenase (LDH) level and an elevated eosinophil count. Late effusions are usually nonhemorrhagic inflammatory exudates that are characterized by lower LDH level and a predominance of lymphocytes.

The present study was designed to determine the prevalence of early-onset pleural effusions requiring thoracentesis or drainage following cardiac surgery in our institution and to identify preoperative, perioperative, and postoperative predisposing factors. Another aim was to assess the cellular and biochemical characteristics of the pleural fluid.

Approximately 1,200 CABG surgeries and 500 valve replacement interventions are performed annually at Laval University Heart and Lung Institute (Québec City, QC, Canada), a tertiary care hospital specializing in respiratory medicine and cardiology. Baseline characteristics, preoperative and perioperative parameters, and follow-up data for all patients undergoing heart surgery are kept in a specific database in our institution. For the purpose of this retrospective cohort study, information was examined from this database on all consecutive patients who underwent CABG surgery, valve replacement, or both between January 1, 2004, and December 31, 2005. Patients with previous CABG, pulmonary or aortic artery surgery, or heart or lung transplants were excluded. The local institutional ethics committee approved the use of anonymized clinical information contained in the database in a scientific report.

Demographic information, anthropometric parameters, cardiac and renal function assessments, medications, and comorbid conditions were retrieved prospectively for all patients and surgery characteristics. We also collected data on the type of procedure, whether it was elective or urgent; the number and origin of the grafts; the duration of the surgery and the cardiopulmonary bypass; and the postoperative complications, including pleural effusions. Patients were considered to have a clinically significant pleural effusion when they met at least one of the following criteria: need for thoracentesis, chest tube drainage, or hospital readmission due to a pleural effusion that occurred within 30 days of surgery. Patients whose pleural effusions were already present at the time of the surgery (n = 16) were excluded from this analysis. Figure 1 shows details on patients included or excluded from this study. The timing of effusion development was determined according to the date of pleural drainage or the initiation of treatment. The medical files of patients with a clinically significant effusion were reviewed to obtain information on the clinical presentation and outcome. We also reviewed all available prethoracentesis chest radiographs in these patients to evaluate the volume of these effusions. We quantified the volume of the effusions on a scale from 0 to 4 (Table 1), according to the estimated proportion of the hemithorax occupied by the effusion.

The results for continuous and categorical variables are presented as the mean±SD and percentages, respectively. Categorical variables were compared between patients with and without pleural effusion using contingency tables and χ² tests, and t tests were used to compare continuous variables between the two groups.

Univariate analyses using preoperative and perioperative variables were performed to identify predictors of pleural effusion. Variables that were associated with the occurrence of pleural effusion with a p value <0.20 then were entered into a stepwise multivariate logistic regression model. Postoperative outcome and pleural fluid findings were examined separately using only univariate analysis. Data were analyzed with statistical software (SAS, version 9.1.3; SAS Institute Inc; Cary, NC).

A total of 2,892 patients who underwent heart surgery during the study period met our inclusion criteria. Of these patients, 192 (6.6%) had a clinically significant pleural effusion in the immediate postoperative period (30 days). One patient experienced a chylothorax that was attributed to thoracic duct injury, and a second patient had a pleural effusion that was associated with a malignancy; these patients were not included in the pleural effusion group.

The pleural effusions were generally symptomatic. Seventy percent of patients complained of shortness of breath, 56% of cough, and 48% of bronchial secretions; 34% were tachypneic, and 23% complained of chest pain. Pleural effusions were drained a mean duration of 6.6±5.9 days following surgery (median duration, 5.0 days). The majority of the pleural drainages (73.8%) were conducted within the first 7 postoperative days. In 21% of patients, a recurrence of the effusion was diagnosed during the first postoperative month, despite initial thoracentesis.

The patients' preoperative characteristics are presented in Table 2. Those with pleural effusion were older than those without (p<0.05). The proportion of female patients was significantly greater in the pleural effusion group (p<0.001). Patients with a clinically significant pleural effusion had a smaller body surface than those without such an effusion (p<0.05). Left ventricular ejection fraction, assessed in the immediate period before the surgery, was lower in patients with pleural effusion (p<0.05).

The comorbidities are presented in Table 3. A greater proportion of patients in the pleural effusion group had a history of congestive heart failure (p<0.01), atrial fibrillation (p<0.05), or peripheral vascular disease (p<0.001).

Table 4 presents selected medications that were used before the surgery. The development of a postoperative pleural effusion was associated with preoperative anticoagulation (heparin, dalteparin, or enoxaparin; p<0.01), warfarin (p<0.01), and clopidogrel (p<0.05). Diuretic and antiarrythmic agents (other than β-blockers and calcium channel blockers) also were more frequently used among patients in whom a pleural effusion developed after the surgery (p<0.05).

The surgical characteristics are reported in Table 5. A total of 2,242 patients (77.5%) underwent CABG surgery, 330 patients (11.4%) underwent valve replacement surgery, and 320 patients (11.1%) underwent combined procedures. The proportion of patients who had valve replacement surgery either alone or in combination with CABG was significantly higher in the patients with pleural effusion (p<0.001). Elective surgeries were significantly less associated with pleural effusions (p<0.01). The number and type of grafts were similar in both groups, except for the use of saphenous vein grafts, which was lower in the pleural effusion group (p<0.05). With regard to other technical considerations, all patients underwent postoperative mediastinal drainage and often pleural drainage, the pleura was opened during mammary harvest most of the time, and the criteria for tube removal on postoperative day 1 were the absence of significant blood loss from the mediastinum or the reaccumulation of fluid in the pleural space (all were standard Argyle chest tubes).

A multivariate analysis was conducted using the variables that yielded a p value of <0.20 in the intergroup analyses. Male gender was associated with a reduced risk of the development of a pleural effusion (odds ratio [OR], 0.57; 95% CI, 0.41 to 0.79). Significant predictors of pleural effusion were a history of peripheral vascular disease (OR, 2.17; 95% CI, 1.45 to 3.24); use of antiarrythmic agents (OR, 2.03; 95% CI, 1.18 to 3.5); and surgery length, with an additional 30 min of surgery time being associated with an OR for developing pleural effusion of 1.18 (95% CI, 1.08 to 1.28) and an additional 60 min of surgery time being associated with an OR of 1.38 (95% CI, 1.17 to 1.64). We found no difference between off-pump and on-pump CABG with regard to the frequency of pleural effusion, although the percentage of patients undergoing surgery off-pump was small in our practice.

The development of a clinically significant pleural effusion was associated with a greater number of cardiac, infectious, hematologic, and renal postoperative complications, as shown in Table 6. Only one case among the 192 drained pleural effusions was diagnosed as empyema. Three other patients with mediastinitis and one patient with pneumonia and sepsis had pleural fluid changes compatible with an infectious process (pleural fluid with elevated total leukocyte and neutrophil counts). Length of stay was higher in the presence of a pleural effusion vs no pleural effusion in the hospital ICU (142±643 h vs 35±565 h, respectively; p<0.05) and in the hospital (16.2±20.9 vs 7.5±8.6 days, respectively; p<0.001).

For 116 of the 192 patients, the attending physician submitted their pleural fluid to biochemical and cellular analyses for diagnostic purposes; the results are presented in Table 7. On average, 672 mL of pleural fluid was withdrawn. The criteria of Light et al⁸ for an exudate were fulfilled in 100% of these samples (116 samples), and 50% of samples (58 samples) were hemorrhagic. Pleural fluid from effusions occurring >15 days postoperatively had lower erythrocyte counts and LDH levels than did the earlier effusions. Moreover, these later effusions were predominantly lymphocytic, whereas the earlier effusions primarily were neutrophilic. To explain why pleural fluid analysis was ordered in only some cases, patient characteristics were compared between those whose pleural fluid samples had undergone fluid analysis and those whose samples had not. The results of this comparison are reported in Table 8. The variable most strongly associated with pleural fluid analysis was the involvement of a pulmonology consultant (95.7% vs 49.2%, respectively; p = <0.0001). Moreover, a higher percentage of patients were reported as having dyspnea in those with pleural fluid analysis than in those without (78.3% vs 57.8%, respectively; p = 0.002).

We were able to review 176 prethoracentesis chest radiographs of the total 192 patients who underwent pleural drainage (Table 1). Ninety patients (51.1%) had a grade 1 pleural effusion, whereas 58 patients (38.6%) had a grade 2 effusion and 15 patients (8.5%) had a grade 3 effusion. Only three patients (1.7%) had a grade 4 effusion. The pleural effusion drained was on the left for 140 patients (76.1%).

This study demonstrates that the prevalence of pleural effusions requiring drainage after CABG, valve replacement, or both was approximately 6.6%, an observation confirming the few data available from previous studies.⁶ However, this study provided original and clinically useful information on the significant impact of pleural effusions on postoperative outcomes. Patients with pleural effusion had significantly longer hospital ICU and hospital stays, and experienced higher rates of complications than those without pleural effusion.

This study also demonstrates that a history of heart failure, more advanced and diffuse arteriosclerosis (ie, peripheral vascular disease), and atrial fibrillation were significantly more prevalent among patients with pleural effusion. Anticoagulants (including heparin, low-molecular-weight heparins, and warfarin), clopidogrel, antiarrythmic agents, and diuretic agents were associated with an increased risk of the development of a pleural effusion, which may partly be explained by the underlying disease for which such medications are generally prescribed. However, in the multivariate analysis, antiarrythmic agents were independently associated with the development of an effusion, whereas among comorbidities, only peripheral vascular disease was also significantly associated. Antiarrythmic medication may have been prescribed more often in patients with a history of heart failure and associated ventricular dysfunction.

Valve replacement was more strongly associated with postoperative pleural effusions than CABG, contrary to previous observations of Light et al,⁶ who reported that chest radiographs obtained 28 days after surgery showed a significantly higher rate of effusions among patients who had undergone either CABG surgery (63%) or combined CABG and valve surgery (62%) than among those who had undergone valve replacement alone (45%). This discrepancy is probably due to our inclusion of patients who required a thoracentesis or pleural drainage.

Female gender and small body surface area, although associated with each other, are known to be independent risk factors for acute morbidity and mortality after CABG.⁹ It is also recognized that women present with higher risk profiles for elevated morbidity and mortality following CABG,¹⁰ valve replacement surgery, or both¹¹ than men. This finding may be related to the older age of female surgical candidates, technical issues related to anatomy, the increased prevalence of comorbidities, and more severe heart disease than their male counterparts.¹² In this study, female gender was clearly associated with increased risk of developing clinically significant pleural effusion. Although the reason for this association is unclear, it may be related to the smaller body surface and vascular volume, although the between-group difference with regard to this parameter was very small in our analysis. The perioperative hemodilution associated with the priming volume of the cardiopulmonary bypass, through its effects on oncotic pressure and fluid hemostasis, also may have contributed to effusion in smaller patients. In addition, patients who eventually developed an effusion were shown to have a lower left ventricular ejection fraction and a higher proportion of history of heart failure and diuretics intake, suggesting that hypervolemia may be critical in the development of this complication. We believe that fluid control during cardiopulmonary bypass, with perioperative ultrafiltration in patients in whom fluid overload is suspected, may possibly reduce the incidence of pleural effusion.

Another finding of this study is that more complex surgeries, such as urgent or combined CABG and valve replacement surgery, and prolonged surgeries were associated with a higher prevalence of pleural effusion. Because the trauma caused to the pleura during these procedures has been linked to the development of pleural effusions,¹³ it makes sense that more invasive and longer surgeries would be associated with a greater likelihood of the development of pleural effusion. Furthermore, other phenomena associated with prolonged surgeries, such as blood transfusion, endothelial injury, and inflammatory response secondary to the use of an extracorporeal bypass circuit may contribute to the development of this complication. Anticoagulants and clopidogrel also may contribute to this phenomenon by facilitating hemothorax formation. Besides, it has been reported^14,15 that preservation of the pleural membrane during the surgery was associated with a significantly reduced rate of postoperative pleural effusions.

In our series, the use of the internal mammary artery in CABG was not associated with an increased risk of pleural effusion. Some investigators⁹ have reached the same conclusion, whereas others^2,5,16 have found the opposite. Topical hypothermia, through phrenic nerve injury and diaphragmatic elevation, has been reported¹⁷ to be associated with an increased rate of pleural effusion.

Sadikot et al⁷ reported distinct pleural fluid characteristics, depending on the time of onset of the effusion in the postoperative period. Although we used a different time of onset cutoff than Sadikot et al,⁷ the present study corroborates these observations. Indeed, patients with very early pleural effusions (ie, those occurring ≤15 days postoperatively) were hemorrhagic, had higher neutrophil counts, and had an elevated LDH level. In turn, patients with later effusions (ie, between 15 days and 30 days following surgery) had lower RBC counts, were predominantly lymphocytic, and had lower LDH levels than those with early effusions. Thus, we think that early effusions may be due to trauma to the pleura, whereas late effusions are more likely due to an immune-inflammatory process related to postcardiac injury syndrome. In this regard, our subanalysis of the first 15 days and 15 subsequent days shows that even in patients with the type of pleural effusion observed 2 weeks postsurgery, significant changes occur compared to those with early-onset pleural effusions, therefore expanding observations made on effusions occurring after a longer time period following surgery.

In this study, pleural effusions requiring drainage were significantly associated with some postoperative complications that may have contributed to the development of the pleural effusion. Our study only provides associations between those events, with sometimes uncertain causality relationships, although the mechanisms by which this may occur are more obvious for complications such as myocardial infarction, atrial fibrillation, and arrhythmias.

Some methodological considerations should be taken into account for a proper interpretation of our results. We elected to consider only patients for whom pleural drainage was deemed necessary by the treating physician. As such, effusions are more likely to be clinically relevant. Furthermore, we wanted to report on the characteristics of the pleural fluid. So, the decision to perform a thoracentesis was made by the treating physician or consultant who decided on the need for such a procedure, which probably reflects the clinical significance of such a complication of the surgery better than using predetermined criteria. With regard to sample size calculation, we initially had insufficient data to provide an adequate estimate but were able to include a large number of patients in whom the variables were statistically different between the groups. We studied consecutive and unselected patients who had undergone heart surgery over a >2-year period in a tertiary care hospital. Thus, our results should apply to this specific clinical setting and may not be generalizable to community hospitals. Finally, we recognize the limitation related to the 1-month follow-up, which excluded from our analysis patients in whom a late pleural effusion developed. Because of the retrospective nature of the study, occasional difficulties in interpreting information from medical records were encountered. Finally, pleural fluid analysis was requested by the treating physician in only 59% of the cases.

This study is among the largest reports of pleural fluid analyses in postcardiac surgery patients. It suggests that patients with the observed risk factors for pleural effusion should be closely monitored to detect this complication. Avoiding fluid overload, reducing pleural trauma during the procedure, or optimizing cardiac function and anticoagulant therapy may likely prevent some of these events.

In conclusion, we found that pleural effusions significant enough to require drainage developed in 6.6% of patients who had undergone CABG with or without valve surgery, mostly because of troublesome respiratory symptoms. We also determined that pleural effusion was more frequent in women and in patients with heart failure, peripheral vascular disease, or atrial fibrillation, and in the presence of some medications, including anticoagulants. Moreover, the level of complexity and the length of the surgery also seemed to influence the development of pleural effusions. Finally, pleural effusions were associated with numerous postoperative complications and added postoperative morbidity, resulting in longer hospital and hospital ICU stays.

Author contributions: Dr. Labidi performed the analysis. Dr. Baillot reviewed the data. Ms. Dionne participated to data gathering. Drs. Lacasse and Maltais reviewed the analysis. Dr. Boulet planned the study. All authors contributed to the writing of the manuscript.

Financial/nonfinancial disclosures: The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Other contributions: We thank Serge Simard for the statistical analyses and Mylène Bertrand for her significant contribution to this project.