Pleural effusion (PE) is a common problem in patients seen in internal medicine and pneumology departments.1 Etiology can vary according to geographical area, healthcare setting, patient age or the time period studied, among other factors. Knowledge of the main etiologies of PE (pre-test probability) enables clinicians to correctly choose and interpret different diagnostic tests following a Bayesian clinical reasoning model.

Only 3 studies on the etiology of PE (with 484,2 6423 and 10004 patients) have been published in Spain in the past 25 years; the last of these was published more than a decade ago.4 The aim of this study was to describe the causes of PE and the accuracy of some basic tests such as pleural fluid (PF) cytology and culture in over 3000 consecutive patients diagnosed in a university hospital over the past 19 years.

Our hospital created a database in which the demographic and clinical characteristics of all consecutive adult patients undergoing diagnostic thoracentesis since 1994 have been entered prospectively, irrespective of the medical or surgical service from which they were referred. Most procedures were performed on hospitalized patients, and very few on outpatients. In this study, we analyze the etiologies of PE from January 1994 to June 2013. In total, 206 patients that had either no definite diagnosis or insufficient PF testing, imaging and clinical follow-up to establish a presumptive diagnosis or to consider the PE idiopathic was excluded. Each patient was included only once if 2 or more thoracenteses were performed for the same reason, i.e., to determine the etiology of PE. The data collection protocol was approved by the local ethics committee.

This study shows that in 75% patients undergoing diagnostic thoracentesis, the etiology of PE was, in order of frequency, cancer, heart failure, pneumonia and tuberculosis. These findings are consistent with previous Spanish series2–4 (Table 4), although the relative frequency of these 4 conditions varies. Specifically, a higher percentage of heart failure patients are found in a series taken from internal medicine departments,2 since this is one of the diagnosis-related groups most frequently seen by these specialists. Pleural tuberculosis etiology predominates in Spanish regions with a greater prevalence of this disease, such as Galicia,3 although the number of cases has steadily diminished over the past 10 years,6 a trend that was also reflected in our series.

The causes of PE in patients at either end of the age scale differ significantly. While the cause of approximately half of all PE cases among young patients (<34) is tuberculosis, among octogenarians PE is secondary to heart failure, which is consistent with the findings of Villena et al.4 It is important to note that the higher prevalence of pericardial (4%) and hepatic hydrothorax (3%) etiologies in our study compared with previous series3,4 is probably due to the fact that these studies recruited, either solely or chiefly, pulmonology patients.

Distinguishing between transudative and exudative effusion is traditionally considered a fundamental first step in the diagnosis of PE. In clinical practice, these are distinguished using Light's criteria.5 Most transudates are associated with heart failure (75%–80%),2–4 although pericardial effusion meets Light's criteria for exudate, albeit narrowly, in around 25% of cases.7 Contrary to the belief that pulmonary embolism can produce transudative PE, we agree with other authors8 who claim that PE due to pulmonary embolism always meets Light's criteria for exudative effusion.

As far as radiological evidence is concerned, differential diagnosis is further refined by the presence of bilateral PE or effusion clouding almost the entire hemithorax. Cancers cause over half of massive PE,9 and heart failure causes over half of bilateral PE. Occasionally, bilateral PE can have a different causative factor for each side, a condition known as Contarini's syndrome.10

In both this and other series published,2–4,11 lung cancer is responsible for a third or more of malignant PE. Experts, with a few exceptions,4 also agree that around 10% of malignant PE are associated with an unknown primary tumor.2,3,11,12 However, the relative frequency of mesotheliomas varies according to geographical region, in relation to the risk of asbestos exposure. Therefore, while mesotheliomas in ours and other series2,3 account for only 3% of all malignant PE, this increases by four- or five-fold (13%–14.5%) in other Spanish studies.4,12 In line with the literature,12,13 overall accuracy of PF cytology in diagnosing malignancy was found to be approximately 60%. Nevertheless, it is important to note that accuracy was considerably lower in squamous cell lung cancer and mesotheliomas (25% and 75%, respectively). Mesothelioma is a prime example of how the combination of an experienced cytologist (in parallel to the prevalence of the disease) and the use of immunocytochemistry can significantly improve the accuracy of PF cytology. These factors were combined in the study conducted by the Australian group Segal et al. who recently reported a PF cytology sensitivity of 73% in 517 cases of mesothelioma.14 Finally, the diagnostic yield of PF cytology for hematological malignancies (61%) reported in our study contrasts with the findings of another Spanish group (18%).12 However, this could be due to the fact that in our hospital it is common practice to request a flow cytometric analysis of PF whenever a lymphoma is suspected.

With regard to PF microbiology, in our study only 30% of patients with infectious PE tested positive using standard PF culture, a percentage that increases to 60% in the case of purulent fluid (empyema). Two large studies have also reported positive results in 56.4% and 58% of 40615 and 43416 empyemas, respectively. In our series, streptococci accounted for 43% of all isolates, the most common being viridans (formerly milleri), which accounted for 25.5% of all positive cultures, a finding that is consistent with the results of Maskell et al. (24% of 336 isolations in community-acquired empyemas).16 Anaerobic bacteria isolation is dependent not only on correct collection, transport and processing procedures, but also on the culture medium used. These bacteria made up 7% of the isolates in our series, compared with 5.7%,15 8.5%,17 and 20%16 reported in three other large studies. Of course, inoculating PF into blood culture bottles18 or the use of nucleic acid amplification techniques19 can significantly increase the number of bacteria isolated. With regard to pleural tuberculosis, we were only able to isolate Mycobacterium tuberculosis from PF using the Lowenstein–Jensen method in 18.5% of cases, a percentage well within the 7%20 to 36%21 range described in the literature. The low yield in this case is due to the fact that tuberculous pleuritis is a paucibacillary disease. However, yield can be greatly increased by using liquid culture media (63% sensitivity in a series of 382 pleural tuberculosis cases).22

This study has certain limitations. Only patients undergoing diagnostic thoracentesis were included. This could mean that some PE etiologies are poorly represented, such as heart failure (many patients presenting with characteristic symptoms are diagnosed with pericardial effusion and do not undergo thoracentesis) or pulmonary embolism (effusion is generally insignificant and the patient is usually given anticoagulants while awaiting confirmatory lab results). In addition, the lack of cardiothoracic surgery services in our hospital could explain the paucity of PE cases associated with these procedures.

In conclusion, cancer, pneumonia and tuberculosis are the causes of most pleural exudates, while heart failure is the main cause of transudates. Pericardial and hepatic hydrothorax conditions are also relatively frequent causes of PE. Pleural tuberculosis predominates in young adults, while in octogenarians the main etiology is heart failure. The bacterial yield from infectious pleural effusion culture is low, except in the case of purulent PF. Likewise, solid medium culture identifies very few cases of tuberculous PE. PF cytology is a simple method for diagnosing malignancies, although it can yield false negatives in at least 40% of cases.

No funding was received for this study.

The authors have no conflicts of interest.