Pleural effusions in the medical ICU: prevalence, causes, and clinical implications.

OBJECTIVE: To determine the prevalence and causes of pleural effusions in patients admitted to a medical ICU (MICU).

DESIGN: Prospective.

SETTING: MICU in a tertiary care hospital.

PATIENTS: One hundred consecutive patients admitted to the MICU at the Medical University of South Carolina whose length of stay exceeded 24 h had chest radiographs reviewed daily and chest sonograms performed within 10 h of their latest chest radiograph.

RESULTS: The prevalence of pleural effusions in 100 consecutive MICU patients was 62%, with 41% of effusions detected at admission. Fifty-seven of 62 (92%) pleural effusions were small. Causes of pleural effusions were as follows: heart failure, 22 of 62 (35%); atelectasis, 14 of 62 (23%); uncomplicated parapneumonic effusions, seven of 62 (11%); hepatic hydrothorax, five of 62 (8%); hypoalbuminemia, five of 62 (8%); malignancy, two of 62 (3%); and unknown, three of 62 (5%). Pancreatitis, extravascular catheter migration, uremic pleurisy, and empyema caused an effusion in one instance each. Heart failure was the most frequent cause of bilateral effusions (13/34 [38%]). When compared with patients who never had effusions during their MICU stay, patients with pleural effusions were older (54+/-2 years, mean+/-SEM, vs 47+/-2 years [p=0.04]), had lower serum albumin concentration (2.4+/-0.1 vs 3.0+/-0.01 g/dL [p=0.002]), higher acute physiology and chronic health evaluation II scores during the initial 24 h of MICU stay (17.2+/-1.1 vs 12+/-1.2 [p=0.010]), longer MICU stays (9.8+/-1.0 vs 4.6+/-0.7 days [p=0.0002]), and longer mechanical ventilation (7.0+/-1.3 vs 1.9+/-0.7 days [p=0.004]). No patient died as a direct result of his or her pleural effusion. Chest radiograph readings had good correlation with chest sonograms (p<0.0001).

CONCLUSION: Pleural effusions in MICU patients are common, and most are detected by careful review of chest radiographs taken with the patient in erect or semierect position. When clinical suspicion for infection is low, observation of these effusions is warranted initially, because most are caused by noninfectious processes that should improve with treatment of the underlying disease.