Chest
Volume 111, Issue 4, April 1997, Pages 970-980
Journal home page for Chest

Clinical Investigations: Pleural Effusions
Diagnostic Value of Tests That Discriminate Between Exudative and Transudative Pleural Effusions

https://doi.org/10.1378/chest.111.4.970Get rights and content

Study objective

To (1) determine appropriate decision thresholds and diagnostic accuracies for pleural fluid (PF) tests that discriminate between exudative and transudative pleural effusions, and (2) evaluate the quality of the primary investigations.

Design

Formal meta-analysis of studies that report the diagnostic value of pleural fluid tests.

Setting

Data collected from international academic medical centers.

Patients

Hospitalized patients undergoing thoracentesis for pleural effusions.

Interventions

Primary investigators were requested to transmit original data from patients described in their studies.

Measurements and results

Eight primary studies described 1,448 patients with one or more of the following tests: protein (P)-PF, P-PF/serum ratio (R), bilirubin (BILI)-R, lactate dehydrogenase (LDH)-PF, LDH-R, cholesterol (C)-PF, C-R, and albumin gradient. We found that all eight tests had similar diagnostic accuracies when evaluated by receiver operating characteristic (ROC) analysis except for BILI-R, which was less diagnostically accurate. Decision thresholds determined by ROC analysis differed from previously reported values for LDH-PF (>0.45 upper limits of normal) and C-PF (>45 mg/dL). Paired and triplet test combinations tended to have higher diagnostic accuracies compared with individual tests, but examination of the odds ratios with 95% confidence intervals did not identify a clearly superior test combination. Limitations of the primary studies presented a high likelihood of bias affecting their results.

Conclusions

Several strategies exist for clinicians in utilizing PF tests to classify effusions as exudates or transudates but accurate interpretations of these test results will require better designed studies.

Section snippets

Materials and Methods

The study was a meta-analysis of the published English-language literature designed to examine the diagnostic accuracy of the eight biochemical tests reported to be of value in separating exudative from transudative pleural effusions. We used statistical techniques to generate receiver operating characteristic (ROC) curves and odds ratios with their 95% confidence intervals (CIs) to compare, respectively, the discriminative properties of individual tests and tests combined in parallel (test

Results

Eleven studies were identified by the literature search.16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26 Four studies with a total of 408 patients17, 20, 21, 24 were excluded because only patients with treated heart failure were examined17 or original data were no longer available from contacted investigators and test results could not be extracted from the published report.20, 21, 24 The remaining seven studies16, 18, 19, 22, 23, 25, 26 fulfilled the entrance criteria and were included in the

Discussion

Determining the transudative or exudative nature of PF is the initial step in evaluating pleural effusions of uncertain etiology. This meta-analysis formally and critically examined the primary investigations that report the discriminative properties of various PF tests used for establishing the presence of an exudate. Our analysis resulted in two major categories of findings that pertained to (1) the diagnostic accuracies of the tests and (2) the quality of study design of the primary

References (29)

  • LeuallenEC et al.

    Pleural effusion, a statistical study of 436 patients

    N Engl J Med

    (1955)
  • LightRW et al.

    Pleural effusion: the diagnostic separation of transudates and exudates

    Ann Intern Med

    (1972)
  • LightRW.

    Pleural disease

    (1995)
  • HulleySB et al.

    Designing clinical research

    (1988)
  • Cited by (0)

    Supported by the Mercy Health Services Research Fund.

    A complete list of participants is located in the appendix.

    View full text