Diagnosis of Pleural Effusions

Diagnosis of Pleural Effusions

Diagnosis of Pleural Effusions* Experience With Clinical Studies, 1986 to 1990 Miloslav MareZ, MD, FCCP; Boris Stastny, MD, FCCP; Libuse Melinova, MD;...

3MB Sizes 0 Downloads 64 Views

Diagnosis of Pleural Effusions* Experience With Clinical Studies, 1986 to 1990 Miloslav MareZ, MD, FCCP; Boris Stastny, MD, FCCP; Libuse Melinova, MD; Eva Svandova , PhD; and Richard W . Light, MD, FCCP

Objectives: To identify in patients with pleural effusion which procedures are most useful in separating malignant from nonmalignant pleural effusions and to identify which procedures most commonly lead to a definitive diagnosis. Design: Prospective consecutive case series. Setting: Pulmonary referral hospital in Prague, Czech Republic. Patients: One hundred seventy-one adults between ages 18 and 70 years with a pleural effusion and a Karnofsky score of 70 or above. Interventions: All patients underwent history, physical, pleural fluid cytologic study, laboratory evaluation of serum and pleural fluid, pleural biopsy, bronchoscopy, and lung scan and/ or pulmonary arteriogram. Results: In this series in which 45% of the patients had malignant effusions, 19% had paramalignant effusions, and 36% had benign diseases, the pleural fluid cytologic study was the best for establishing a diagnosis. The pleural fluid carcinoembryonic antigen (CEA) levels above 10 had a high specificity (90%) for malignancy but

Most patients who are admitted to a hospital with a pleural effusion undergo extensive diagnostic procedures in an attempt to identify the cause of the effusion. 1•2 These procedures include chest radiograph, CT study of the thorax, thoracentesis with multiple chemical determinations, microbiologic studies on the pleural fluid, pleural fluid cytologic study, needle biopsy of the pleura, pleuroscopy, bronchoscopy, and open pleural biopsy. This present study was undertaken with the following three goals: (1) to identify the relative frequency for which different diseases are responsible for pleural effusion in a large pulmonary referral hospital in Prague, Czech Republic; (2) to identify which procedures were useful in separating malignant from nonmalignant pleural effusions; and (3) to identify which procedures most commonly led to a

*From the Pneumological Clinic, Charles University, Prague, Czech Republic, and the Pulmonary Section, Department of Veterans Affairs Medical Center, Long Beach, Calif. Manuscript received December 27, 1993; revision accepted November 16, 1994. Reprint requests: Dr. Light; 5901 East 7th Street , Long Beach, CA , 90822


had low sensitivity (37%). The pleural fluid CEA level was increased only in 19% of patients with paramalignant effusions. Although there were statistically significant differences in the mean results on several biochemical tests of pleural fluid, none were very accurate in separating malignant from benign disease. Conclusion: From this study, we conclude that patients with an undiagnosed pleural effusion should be evaluated in an individualized stepwise manner. If malignancy is strongly considered, the initial three steps should be relatively noninvasive and include clinical evaluation and cytologic study. (CHEST 1995; 107:1598-1603)

CEA=carcinoembryonic antigen; LDH=lactate dehydrogenase

Key words: CEA; pleural effusion; pleural fluid ; pleural malignancy

definitive diagnosis and to identify which procedures rarely yielded significant diagnostic information . MATERIALS AND METHODS

For the 4-year period between October 1, 1986 and September 30, 1990, all patients seen at the Pneumological Clinic at Charles University with a pleural effusion evident on the chest radiograph underwent a workup as dictated by a single protocol. The Pneumological Clinic is a referral center for patients with respiratory diseases from Prague and the surrounding area. Approximately 70% of the patients have lung cancer. To be included in the study, the patients had to meet the following criteria: (1) be hospitalized during the study period with a pleural effusion; the pleural effusion could develop during the hospitalization; (2)to be between age 18 and 70 years; and (3) to have a Karnofsky score 70 or above. The patients were asked a series of questions regarding their clinical history that included the duration of their illness and whether they had weight loss, dull chest pain, pleuritic chest pain, temperature elevation, etc. The breathlessness of the patient was semiquantitated by the scheme of the World Health Organization; grade 3 is breathlessness with minimal effort while grade 4 is breathlessness at rest. Patients were said to have an occupational asbestos exposure if they worked for more than 5 years at factories producing asbestos products. Patients were said to have work exposure to carcinogens if they worked for more than 2 years with uranium, chromium, or ammonia. Patients were divided by their smoking histories into nonsmokers, light smokers (up to 200,000 cigarettes lifetime), and heavy smokers (more than 200,000 cigarettes lifetime). Clinical Investigations

The pleural fluid from all patients was examined macroscopically, ie, the color and the odor. The pleural fluid was also evaluated microscopically after it was stained with Wright's stain for the differential cell count. A pleural effusion was considered eosinophilic if it contained more than 5% eosinophils. A Gram 's stain and an acid-fast stain of the pleural fluid were also examined. The following laboratory examinations were made on the pleural fluid and serum samples: total protein, glucose, lactate dehydrogenase (LDH), cholesterol, alkaline phosphatase, amylase, and protein electrophoresis. Additional studies obtained included immunoglobulin levels (IgG, IgA, IgM ), complement levels (C3 and C4 fractions), antinuclear antibodies, and carcinoembryonic antigen (CEA). We also obtained pH measurements on pleural fluid. The 95% confidence intervals for the different measurements are as follows: total protein, 6.8 to 8.4 g/ dL; glucose, 70 to 100 mg/ dL; amylase, 0.37 to 5.1 JLkatal/L; alkaline phosphatase, 0.8 to 2.5 JLkatal/ L; cholesterol, 3.8 to 6.20 JLmol/L; LDH, 3.9 to 6.1 JLkatal/L; CEA, 0 to 10 ng/ mL; IgG, 110 to 240 IU / mL; IgA, 100 to 230 IU/ mL; IgM, 120 to 260 IU / mL; C3 fraction of complement, 0.6 to 0.9 g/ L; and C4 fraction of complement, 0.1 to 0.3 g/ L. The chemical determinations were performed with an automated biochemical analyzer (Hitashi 717). Tissue polypeptide antigen determinations were obtained from the last 68 patients. The upper limit for tissue polypeptide antigen in our laboratory is 100 IU/ L. Pleural fluid was sent for cytologic examination and for bacterial and mycobacterial cultures. The cytologic sample was graded by the Papanicolaou's grading system with class 4 and class 5 considered diagnostic of cancer. All patients in the study also underwent bronchoscopy, biopsy of the parietal pleura (unless gross pus was present in the pleural space), and perfusion lung scan. Thoracoscopy with direct visualization was performed in seven patients in whom malignancy was suspected but not proved after the other diagnostic procedures.

Pleural Effusion Criteria The criteria we used for the more common causes of pleural effusion were the following. Malignant Effusion: An effusion associated with malignancy in the pleural space as demonstrated by cytologic study, pleural biopsy specimen, or autopsy. Paramalignant Effusion: A pleural effusion occurring in a patient with known malignancy but in whom the pleural fluid cytologic study and pleural biopsy specimen did not demonstrate malignancy and for whom there was no obvious alternative diagnosis. At autopsy, there could not be any direct involvement of the pleural space by malignancy. Parapneumonic Effusion: An effusion associated with pneumonia, pulmonary abscess, or bronchiectasis. The pleural fluid was an exudate with a predominance of polymorphonuclear leukocytes. Empyema: The accumulation of pus in the pleural space on which the Gram's stain was positive or the bacterial cultures were positive. With cytologic study, the polymorphonuclear leukocytes were partially or completely disintegrated. Tuberculous Effusion: An effusion that met at least one of the following four criteria: (1) positive culture for Mycobacterium tuberculosis from pleural fluid or pleural biopsy specimen; (2) positive smear for acid-fast bacilli from pleural fluid; (3) caseating granuloma on pleural biopsy specimen; or (4) positive sputum culture forM tuberculo!is with either an exudative pleural effusion or noncaseating granuloma in pleural biopsy specimen. Heart Failure Effusion: A pleural effusion in the setting of clinical congestive heart failure with an enlarged heart, distended neck veins, and cardiac gallop that improved with therapy for the congestive heart failure. Hemothorax: The hematocrit in the pleural fluid was required to be more than 20%. For the hemothorax to be ascribed to

trauma, a history of trauma was required. For the nontraumatic hemothorax, other reasons for bleeding, eg, hemophilia or thoracic endometriosis, were required. Pancreatic Pleural Effusion: A pleural effusion with a high amylase level in a patient with acute or chronic disease of the pancreas in whom rupture of the esophagus and bronchogenic carcinoma had been excluded. Effusion due to pulmonary embolism: A pleural effusion in a patient with a lung scan showing a high probability of a pulmonary embolus or a pulmonary arteriogram demonstrating pulmonary emboli. Asbestos Pleural Effusion : A pleural effusion in a patient with a history of asbestos exposure in whom there was no other obvious cause of pleural effusion. In addition, these patients were required to have no other disease process with a follow-up of at least 2 years. Rheumatoid Effusion: An effusion that occurred in a patient with rheumatoid arthritis and that was characterized by a low glucose level, low pH, a high LDH level, and negative bacterial cultures. Uremic Pleural Effusion: An exudative effusion that occurred in a patient with uremia for which there was no other explanation. Intra-abdominal Process: An effusion that accompanied an intra-abdominal inflammatory process such as pericholecystitis or hepatic abscess. Myxedematous Effusion: An effusion that occurred in conjunction with myxedema and pericardia! effusion. At the end of the study, we reviewed each patient's chart to see what procedures were most valuable in establishing the cause of the effusion.

Statistical Methods The data were entered into a computer (Hewlett-Packard 2100 S). For data analysis, the patients were divided into those with malignancy (malignant group) and those with benign disease (benign group). The results in the two different groups were then compared via the nonparametric Mann-Whitney test for continuous variables and the x2 test for noncontinuous variables. RESULTS

During the 4-year period, a total of 171 patients with pleural effusion were evaluated and the etiologies of their pleural effusions are shown in Table 1. Table !-Causes of Pleural Effusions

in the 171 Patients

Type of Effusion



Malignant Lung cancer Other Paramalignant Lung cancer Other Para pneumonic Empyema thoracis Tuberculous pleuritis Congestive heart failure Traumatic hemothorax Pancreatic Pulmonary embolism Miscellaneous (unknown 2, benign asbestos 1, lupus 1) Total

76 44 32 32 23 9 20


11 ll

6 5 3 3 4

18.7 11.7 6.4 6.4 3.5 2.9 1.7 1.7 2.4



CHEST / 107 / 6 / JUNE, 1995


Table 2-Comparison of Medical Histories in Patients With Malignant and Benign Disease Malignant Benign p Value No. of cases Age, yr Work exposure to carcinogens, % Elevated temperature, % Chest pain dull, % Pleuritic chest pain, % Symptoms of disease less than 7 d, % Breathlessness grade 3, % Effusion developed during hospital admission, % Personal history of cancer, %


108 57 8 37 34 24 30 45 18

63 47 2 73 ll 51 50 17 0



Pleural effusions associated with malignancy accounted for 63.3% of all the effusions; 44.6% of the effusions were due to pleural malignancy while 18.7% were paramalignant effusions. Although paramalignant effusions tended to be more common with lung cancer than with other malignancies, this difference did not achieve statistical significance (X 2 =1.32; p>O.lO) . The only other types ofeffusions that accounted for more than 5% of the effusions were parapneumonic (11.7%), empyema thoracic (6.4%), and tuberculous pleuritis (6.4%) . The mean age of the 114 men with pleural effusion was 54 years while that of the 57 women was 52 years. Since one of the primary questions to answer when evaluating a patient with an undiagnosed pleural effusion is whether the patient has a malignancy, we divided our population into two groups-those with malignancies (malignant or paramalignant pleural effusion) and those with benign disease. The two groups did have statistically significant differences in their medical histories (Table 2). The patients with malignancy were more likely to have a history of working with carcinogens or to have a personal history of malignancy. The incidence of dull chest pain was higher in the group with malignancy while the incidence of pleuritic chest pain was higher in the group with benign disease. The patients with malignancy tended to have more shortness of breath. There was no significant differences in the two groups with respect to their smoking history, the prevalence of chronic bronchitis, cough, or hemotypsis. There were also statistically significant differences when we looked at the results of the laboratory tests in the two different groups (Table 3). The patients with malignancy had a significantly higher serum LDH, CEA and tissue polypeptide level than did those with benign disease, while those with benign disease had a significantly higher serum protein level. The patients with malignancy had a significantly higher pleural fluid level of alkaline phosphatase and CEA than did those with benign disease. The IgM levels in the serum and pleural fluid were signifi1600

Table 3-Comparison of Results of Biochemical Measurements in Patients With Pleural Effusions Due to Malignant Disease and Benign Disease

Protein, g/ dL 6.64±0.83 Serum Fluid 4.59±0.90 Ratio 0.7±0.13 Glucose, mg/ dL 112±41 Serum Fluid 101 ±45 Ratio 0.96±0.49 Alkaline phosphatase Serum 1.9±1.7 2.2±2.2 Fluid Ratio 0.8± 1.5 LDH Serum 6.8±4.8 Fluid 16.7±37.3 Ratio 3.2±9.6 CEA, ng/ mL Serum 25.9±53.7 Fluid 35.6±56.7 Ratio 2.9±6.0 Tissue polypeptide antigen, ng/ mL Serum 476±605 Fluid 2,080 ± 1,068 11.2 ± 11.7 Ratio IgM Serum 225±120 Fluid 77±57 Ratio 0.93±77


p Value

6.88± .76 4.68±1.1 0.7±0.14


103±36 94±49 1.0±0.7

1.6± 1.0 0.85± 1.2 0.55±0.51


4.8 ± 1.8 12.6±28.3 2.9±7.1


4.7±3.8 4.1±3.0 1.0 ± 1.0

<0.01 <0.001 <0.01

99±64 1,726± 1,154 26.9±28


270±121 119± 101 0.87±3.1

<0.01 <0.01

cantly lower in the malignant group. There were no significant differences in the serum or pleural fluid glucose level, IgG level, IgA level, amylase level, cholesterol, C3 fraction of complement, or C4 fraction of complement. The frequency of pleural fluid eosinophilia was significantly higher in those with benign disease (11 .2%) than in those with malignant disease (0.9%). The pleural fluid pH did not differ significantly between those with malignant and those with benign disease. Even though there were significant differences in the mean levels of some of the biochemical measurements in the two groups, such measurements were of limited value in definitively separating the two groups (Table 4). The three measures involving the CEA (serum CEA , pleural fluid CEA, and ratio of pleural fluid / serum CEA) all had specificities that were close to 90%, but the sensitivities of none exceeded 40%. When the biochemical parameters were compared in the patients with malignant pleural effusion and paramalignant pleural effusion, the only measure that differed significantly (p<0.05) was the CEA level. For the patients with malignant pleural effusions, the CEA level exceeded lO ng/ mL in 35 of 70 Clinical Investigations

(50%) and exceeded 20 ng/ mL in 26 of 70 (37%), while in the paramalignant effusions, the CEA level exceeded 10 ng/ mL in only 5 of 26 (19%) and exceeded 20 ng/ mL in only 2 of 26 (12 %). After the diagnostic process was completed, we evaluated which procedure was most important in delineating the causes of the effusion (Table 5). Cytologic study, which included the Wright's stain (for polymorphonuclear leukocytes, lymphocytes, mesothelial cells, and eosinophils) and the Papanicolaou stain for malignant cells, was the most definitive procedure in 60.8% of the patients, including 60 with malignancy and 18 with inflammatory-type effusions. Needle biopsy specimen of the pleura was diagnostic by itself in four patients. The combination of cytologic and histologic study was most helpful in 26 cases, including 10 malignant and 16 inflammatory effusions. Bronchoscopy was the definitive study in 12 cases, while radiographs, biochemistry , and bacteriologic cultures were the definitive study in 7, 6, and 5 patients, respectively. Based on the above analysis we could break down the diagnostic workup of patients with pleural effusions into four steps. The diagnostic yields with each of the steps are shown in Table 6. Note that the first step involves only the clinical evaluation plus the cytologic study of the pleural effusion and this gives a diagnosis in more than one third of the patients. By the time the first two steps are completed, more than 60% of the patients have a diagnosis and no invasive procedure has been necessary. With this four-step workup, thoracoscopy or an open pleural biopsy was needed in less than 5% of patients. DISCUSSION

During the 4-year study period, we evaluated 171 patients with pleural effusions who were seen at the Pneumological Clinic. The most common underlying disease in this series was malignancy (63%). The other leading underlying diseases were pneumonia (13.4%) and tuberculosis (6.4%). The high incidence of malignant effusion in the present series is influenced by the character of our clinic, which acts as a differential diagnostic center for pulmonary tumors in the Central Bohemia Region. Indeed in an epidemiologic study performed in parallel with the present study , only 22% of the newly discovered pleural effusions were due to malignancy while 46% were due to congestive heart failure.3 The results of the present study are markedly different than those reported in a similar study carried out by Tousek, 4 in the 1955 to 1960 period, in a district hospital in eastern Bohemia. He reported that the most common cause for the effusion was tuberculosis (36.5%), followed by malignancy (17%) and congestive heart failure (15%). The explanation for

the changing distribution of diagnosis is the decreasing prevalence of tuberculosis and the increasing prevalence of lung cancer. Indeed, in a previous article, Strapko and Olosova 5 demonstrated that the percentage of effusions that were due to malignancy was increasing. It should be emphasized that the distribution of diagnoses responsible for pleural effusions is heavily dependent on the locality and the type of hospital (referral vs primary care). For example, in Rwanda 6 more than 80% of pleural effusions at a referral hospital are due to tuberculosis and most of the patients with tuberculous pleuritis are HIV positive. In a recent study from a referral hospital in northern Spain, 7 nonmalignant disease was responsible for slightly more than 50% of pleural effusions, and tuberculosis was responsible for 50% of the benign effusions. One of the primary reasons for the higher percentage of malignancy in the present series is the small number of patients with tuberculosis. The explanation for the low incidence of congestive heart failure in this series is due to the fact that the Pneumological Clinic is a referral center for patients with lung disease. Indeed, in an epidemiologic report from the Czech Republic, congestive heart failure was the most common cause of pleural effusion. 3 In the present series, lung cancer accounted for approximately 60% of all malignant pleural effusions. Previous authors 7-9 have reported that only 30 to 40% of patients with malignant pleural effusions have lung cancers. We believe that the true figure in the Czech Republic is probably closer to 40%. The present study is biased toward a higher percentage of lung cancer since the Pneumological Clinic at the Charles University serves as a diagnostic center for lung cancer in the area of Prague with a catchment area of approximately 1 million. The incidence of lung cancer in the Czech Republic is 99.7 / 100,000 in men and 15/ 100,000 in women, rates that are slightly Table 4-Cutoff Values and Associated Specificities and Sensitivities for Biochemical Measurements That Differed Significantly in the Two Groups

Serum LDH CEA IgM Tissue polypeptide Fluid Alkaline phosphatase CEA IgM Ratio CEA Tissue polypeptide

Specificity, %


C utoff Value 5.8 _ukati/ L 8.0 mg/ mL 230 U/ mL 100 U/ L

72 93 72 83

41 34 54 84

l.l _ukatai/L 10 ng/ mL 80 U/ mL

70 91 73

30 37 55

2.5 5.0

88 82

23 51

CHEST / 107 / 6 / JUNE, 1995



Table 5-Procedure That Was Most Important in Delineating Diagnosis in 171 Patients With Pleural Effusion Procedure Pleural fluid cytologic study and differential Histologic study of needle biopsy specimen of pleura Histologic study of open pleural biopsy specimen Cytologic study of fluid and needle biopsy specimen of pleura Clinical findings Bronchoscopy Chest radiograph Biochemistry Bacteriologic cultures Thoracoscopy Lung scan Autopsy Type of effusion unknown


Table 6-Four Steps in the Diagnostic Workup of the Patient With a Pleural Effusion Along With the Diagnoses Established at Each Step %

78 4

45.6 2.3





22 12 7 6 5 2 3

12.9 6.6 4 .1 3 .5 2.9 2.3 1.8 0.6 1.2


lower than those in the United States.l 0 One of the important questions to answer about a patient with an undiagnosed pleural effusion is whether the patient has a benign or a malignant pleural effusion. As shown in Table 2, there are some statistically significant differences in the symptoms. The patients with malignancy are more likely to have dull chest pain, breathlessness, and a work exposure to carcinogens, while the patients with benign effusions are more likely to have an elevated temperature , pleuritic chest pain, and a more rapid development of the effusion. However, no historical finding absolutely indicates that a given patient has benign or malignant disease. In like manner, the means of several biochemical measures in the serum or pleural fluid were statistically significantly different in the two groups (Table 3). However , there was much overlap and when the sensitivities and specificities of the different measures were evaluated (Table 4) , none , with the possible exception of the CEA level, was useful in making the differentiation . The present study demonstrates that the most useful single test in establishing the diagnosis of a pleural effusion is the pleural fluid cytologic study. The first pleural fluid cytologic study in conjunction with the history , physical examination , and chest radiograph established the diagnosis in 34.5% of all patients. Other tests that were useful included the subsequent cytologic studies of the effusions, the clinical findings, the pleural biopsy, bacteriologic cultures, and the lung scan. Since most patients at the Pneumonological Clinic have lung cancer, it might seem surprising that bronchoscopy was not more useful in establishing the cause for the pleural effusion. In our clinic, the 1602

Step No. 1 History of illness, physical examination, chest radiograph, macroscopic examination, and first cytologic study 2 Second cytologic study of effusion , microscopic examination of the fluid , biochemistry, tumor markers, and bacteriology 3 Needle biopsy of parietal pleura, lung scan or pulmonary angiography , bronchoscopy 4 Thoracoscopy, open pleural biopsy

% Diagnosis This step

% Diagnosis Cumulative









diagnosis of lung cancer is most commonly established by bronchoscopy, but bronchoscopy does not delineate the cause of a pleural effusion with the exception of some paramalignant effusions. The yield from needle biopsy in the present series (25% with pleural malignancy and 36% with pleural tuberculosis) is lower than that reported in most series. 1 We attribute this low diagnostic yield to the fact that these were the first patients subjected to needle biopsy in our institution . Our diagnostic yields with this procedure have tended to improve with time. Based on our experience in the present study, we recommend that clinicians give consideration to the four-step workup outlined in Table 6 in the evaluation of a patient in a referral center where the incidence of malignancy is high. We hasten to add that the workup of a patient should be individualized. In step 1, when the patient is first evaluated, a careful history and physical examination should be obtained and a diagnostic thoracentesis should be performed. We would recommend measurement of the levels of protein and LDH to ascertain if the fluid is a transudate or an exudate 11 and pleural fluid cytologic study. If the patient has an acute febrile illness or if the pleural fluid smells putrid or is turbid, the fluid should be analyzed with Gram's stain and aerobic and anaerobic cultures should be obtained. In step 2, a second cytologic study is obtained and the pleural fluid and serum are sent for tumor markers (CEA and tissue polypeptide antigen) and the pleural fluid is sent for glucose, amylase levels, differential cell count, and bacteriologic studies if they had not been obtained previously. In step 3, the patient is subjected to one or more invasive procedures. The initial procedure will depend on the clinical situation . If pleural tuberculosis Clinical Investigations

or malignancy is strongly suspected, the procedure should probably be needle biopsy of the parietal pleura, while if pulmonary embolism is thought to be most likely, a lung scan should be obtained. Bronchoscopy is recommended at this point, although it is frequently not diagnostic if the patient does not have a parenchymal infiltrate or hemoptysis. 12 In step 4, the patient is subjected to thoracoscopy or open pleural biopsy. Thoracoscopy can establish the diagnosis of pleural malignancy, including mesothelioma in approximately 90% of patients, 13J 4 but is rarely diagnostic of benign disease. Since in most series, a sizable percentage of patients recover from an episode of a pleural effusion without any cause ever being obtained and we think that these patients have viral pleuritis, we recommend proceeding to step 5 only if the patient's condition is not improving or if the patient demands assurance that he does not have a cancer.

ACKNOWLEDGMENT: We would like to acknowledge the support of the former Chief of Pneumological Clinic, Charles University, Dr. F. Polansky for the present study. We would like to thank the many collaborating specialists from different wards of the University Hospital in Prague, including Dr. T. Fucikova, Dr. J. Musilova, Dr. J. Petrasek, Dr. Viznerova, Dr. Melinova, Dr. Svandova, Dr. Cermak, and Dr. K. Bakos among others. We would like to acknowledge the wife of the primary author, Ms. Renata Marelova, for data collection, analysis, and encouragement.


l Light RW. Pleural disease. Philadelphia: Lea & Febiger, 1990;39-83 2 Sahn SA . The pleura. Am Rev Respir Dis 1988; 138:184-234 3 Mare[ M, Zrustova M, Stastny B, et al. The incidence of pleural effusion in a well-defined region: an epidemiological study in Central Bohemia. Chest 1993; 104:1486-89 4 Tousek M. Differential diagnosis of the pleural effusion. Stat Zdravo Nak (Czech) 1962; 165 Prague 5 Strapko P, Olosova M. Pleuralne vypotky v materiali kliniky. TaPCh Stud Pneumol Phtiseol Cechoslov (Czech) 1990; 50:365-68 6 Batungwanayo J, Taelman H, Allen S, et al. Pleural effusion, tuberculosis and HIV -1 infection in Kigali, Rwanda. AIDS 1993; 7:73-9 7 Bueno CE, Clemente MG, Castro BC, et al. Cytologic and bacteriologic analysis of fluid and pleural biopsy specimens with Cope's needle. Arch Intern Med 1990; 150:1190-94 8 Spriggs AI, Boddington MM. The cytology of effusions. 2nd ed. New York: Grune & Stratton, 1968 9 Anderson CB, Philpott GW, Ferguson TB. The treatment of malignant pleural effusions. Cancer 1974; 33:916-22 10 Beckett WS. Epidemiology and etiology of lung cancer. Clin Chest Med 1993; 14:1-15 11 Light RW , MacGregor MI, Luchsinger PC, et al. Pleural effusions: the diagnostic separation of transudates and exudates. Ann Intern Med 1972; 77:507-13 12 Chang S-C, Perng RP. The role of fiberoptic bronchoscopy in evaluating the causes of pleural effusions. Arch Intern Med 1989; 149:855-57 13 Menzies R, Charbonneau M. Thoracoscopy for the diagnosis of · pleural disease. Ann Intern Med 1991; 114:271-76 14 Hucker J, Bhatnagar NK, Al-Jilaihawi AN, et al. Thoracoscopy in the diagnosis and management or recurrent pleural effusions. Ann Thorac Surg 1991; 52:1145-47

CHEST /107 / 6 / JUNE, 1995