0
Original Research: Disorders of the Pleura |

Improving the Predictive Accuracy of Identifying Exudative EffusionsImproving Identification of Exudative Effusions FREE TO VIEW

Carlos E. Kummerfeldt, MD; Cody C. Chiuzan, MS; John T. Huggins, MD; Matthew L. DiVietro, MD; Jennings E. Nestor, MD; Steven A. Sahn, MD, FCCP; Peter Doelken, MD
Author and Funding Information

From the Division of Pulmonary, Critical Care, Allergy, and Sleep Medicine (Drs Kummerfeldt, Huggins, DiVietro, Nestor, and Sahn), and the Department of Public Health Sciences (Ms Chiuzan), Medical University of South Carolina, Charleston, SC; and the Division of Pulmonary and Critical Care Medicine (Dr Doelken), Albany Medical College, Albany, NY.

Correspondence to: John T. Huggins, MD, Division of Pulmonary, Critical Care, Allergy, and Sleep Medicine, Medical University of South Carolina, 96 Jonathan Lucas St, Ste 812-CSBMSC 630, Charleston, SC 29425; e-mail: hugginjt@musc.edu


Part of this article was presented at the American Thoracic Society International Conference, May 17-22, 2013, Philadelphia, PA, and as a poster abstract (Kummerfeldt CE, Chiuzan CC, Huggins JT, et al. Am J Respir Crit Care Med. 2013;187:A4291).

Funding/Support: The authors have reported to CHEST that no funding was received for this study.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details.


Chest. 2014;145(3):586-592. doi:10.1378/chest.13-1142
Text Size: A A A
Published online

Background:  Application of Light’s criteria results in misclassification of some transudative effusions as exudative, particularly because of congestive heart failure (CHF). We sought to determine if the serum to pleural fluid albumin (SF-A) and serum to pleural fluid protein (SF-P) gradients increased the predictive accuracy to correctly identify exudative effusions.

Methods:  We retrospectively analyzed 1,153 consecutive patients who underwent a diagnostic thoracentesis at the Medical University South Carolina. Univariable logistic regression analyses were used to determine the statistical significance of pleural fluid tests that correctly identified exudative effusions. Tests with significant diagnostic accuracy were combined in multivariable logistic regression models, with calculation of areas under the curve (AUCs) to determine their predictive accuracy. The predictive capability of the best model was compared with Light’s criteria and other test combinations.

Results:  Pleural fluid lactate dehydrogenase (LDH), SF-A gradient, and SF-P gradient had a significant effect on the probability of identifying exudative pleural effusions. When combined together in a multivariable logistic regression, LDH (OR, 14.09 [95% CI, 2.25-85.50]), SF-A gradient (OR, 7.16 [95% CI, 1.24-41.43]), and SF-P gradient (OR, 6.83 [95% CI, 1.56-27.88]) had an AUC of 0.92 (95% CI, 0.85-0.98).

Conclusions:  Application of Light’s criteria, not uncommonly, misclassifies CHF transudative effusions as exudates. In cases where no cause for an exudative effusion can be identified or CHF is suspected, the sequential application of the fluid LDH, followed by the SF-P and then the SF-A gradients, may assist in reclassifying pleural effusions as transudates.

Figures in this Article

Pleural effusions are identified as either exudative or transudative by using Light’s criteria.1 The sensitivity of Light’s criteria for exudative effusions is 99%, but the specificity ranges from 65% to 85%.2 As a result, some transudative effusions are misclassified as exudative, particularly those due to congestive heart failure (CHF) when diuretics are given.

Other pleural fluid tests and test combinations have been described.313 When compared with Light’s criteria, the majority of these pleural fluid tests and test combinations have maintained comparable sensitivities but failed to increase specificity for correctly identifying exudative effusions. Bayesian analysis to overcome misclassification of transudative effusions as exudative was also proposed.14,15 However, another study showed that Bayesian analysis marginally improved the correct identification of exudative effusions.16

Another proposed method to reduce the misclassification of transudative effusions as exudative is the use of the serum to pleural fluid albumin (SF-A) and serum to pleural fluid protein (SF-P) gradients. In one study, an SF-A and SF-P gradient > 1.2 g/dL and 3.1 g/dL, respectively, correctly identified 86% and 91% of the transudates.17 Light’s criteria correctly identified only 75% of those transudates.17 In another study, the SF-A and SF-P gradients correctly identified 83% and 55% of misclassified transudates due to CHF, respectively.18

The use of albumin and protein gradients appears to reduce the misclassification of transudative effusions as exudative, particularly those related to CHF. Therefore, a stepwise approach combining the use of Light’s criteria with the albumin and protein gradients to correctly identify exudative effusions seems logical. We sought to determine if SF-A and SF-P gradients increased the predictive accuracy to correctly identify exudative effusions.

We retrospectively reviewed all consecutive patients who underwent a diagnostic thoracentesis at Medical University of South Carolina (Charleston, South Carolina) from 1999 until 2011. Clinical characteristics collected included age; sex; cause of the pleural effusion; serum protein, lactate dehydrogenase (LDH), and albumin concentrations; and the following pleural fluid tests: pH, protein, LDH, cholesterol, and albumin. Serum protein, LDH, albumin, and pleural fluid tests were performed in an analyzer using standard technology (UniCel DxC 800; Beckman Coulter, Inc). Pleural fluid pH was collected using heparinized syringes and transported in ice to the laboratory, where it was processed within 1 h after collection in a blood gas machine (ABL 800FLEX; Radiometer Medical ApS). The study protocol was approved by the Institutional Review Board for Human Research at Medical University of South Carolina (protocol number 00016055). Criteria used to establish the cause of an effusion is described in detail in e-Appendix 1.

Light’s criteria, which combine a pleural fluid to serum protein ratio (FP-R) > 0.5, a fluid to serum LDH ratio (FL-R) > 0.6 or the upper normal limit (UNL) serum LDH as triplets with an “or” rule, were applied to all effusions. The UNL serum LDH in our laboratory was 240 IU/L. Therefore, two-thirds UNL for serum LDH corresponded to a value of 160 IU/L. Modified Light’s criteria, defined as pleural fluid LDH > 0.45 UNL of serum (corresponding to a value of 108 IU/L) in combination with FP-R > 0.5 or FL-R > 0.6 as triplets with an “or” rule, were also applied to all effusions.6 An effusion with either a fluid LDH > 160 IU/L or a fluid cholesterol > 45 mg/dL combined in pair with an “or” rule, was considered exudative. Misclassified transudative effusions by Light’s criteria were then reclassified by applying the SF-A gradient, SF-P gradient, and SF-A ratio using previously defined cutoff values.6,17 Final consensus of the diagnosis for each pleural effusion was established by three of the authors.

Statistical Analysis

We examined the discriminative properties of eight pleural fluid tests: pH, total protein, LDH, cholesterol, FP-R, FL-R, SF-A gradient, and SF-P gradient. Cutoff points to identify exudative effusions were obtained for all eight tests from previously reported and defined cutoff values.6,17 Sensitivities, specificities, and likelihood ratios (positive and negative) with 95% CI for all eight pleural fluid tests were calculated. Because not every patient had all eight pleural fluid tests available, the size of the datasets used to generate each of the discriminative properties varied. A two-tailed P value of ≤ .05 was considered significant. Means (95% CI) and medians (interquartile range) were used to compute the descriptive statistics for transudates and exudates. A two-tailed t test or its nonparametric alternative, Wilcoxon rank-sum, were used to test the differences of pleural fluid tests between transudates and exudates. All hypotheses were tested at a Bonferroni adjusted two-sided α level of 0.00625 (0.05/8).

Univariable and multivariable logistic regression analyses were applied to individual pleural fluid tests as well as test combinations with equal or better diagnostic performance characteristics than Light’s criteria (ie, higher specificity). The area under the curve was then calculated for individual pleural fluid tests that showed statistical significance. These tests were combined and compared with Light’s criteria and other test combinations to determine if they increased the predictive accuracy to correctly identify exudative effusions. When analyzed by multivariable logistic regression, the pleural fluid tests not included by Light’s criteria (pleural fluid pH, cholesterol, SF-A gradient, and SF-P gradient) that showed statistical significance were sequentially added to those tests already included by Light’s criteria (FP-R, FL-R, and two-thirds UNL of LDH) that also showed statistical significance using an “or” rule. Internal validation of the final model was assessed using bootstrap analysis. Bootstrapping is the preferred method quantifying the uncertainty of the model, when this has already been decided.19 Compared with cross-validation, bootstrapping tends to drastically reduce the variability of the estimates by randomly selecting subsamples with replacement from the original data set. In our analysis, we used a total of 200 subsamples and averaged the results over 100 repetitions. All analyses were carried out using SAS, version 9.2 (SAS Institute Inc) software.

Approximately two-thirds of a total of 1,153 patients (67%) examined in our registry had exudates (Table 1). Exudates had significantly lower fluid pH and higher fluid protein, LDH, and cholesterol levels (Table 2). The FP-R and FL-R were significantly higher among exudates. The SF-A and SF-P gradients were significantly lower among exudates. Except for the pleural fluid pH, the sensitivities of the eight pleural fluid tests were similar (Table 3). Sensitivities were also similar between Light’s criteria, modified Light’s criteria, and two-test combination LDH and cholesterol. No single test or combination was superior.

Table Graphic Jump Location
Table 1 —Clinical Characteristics of the Study Population

Data are presented as No. (%) unless otherwise noted.

a 

Includes pancreatitis, benign asbestos pleural effusion, fibrothecoma, drug induced, radiation pleuritis, pulmonary embolus, sarcoidosis, pancreaticopleural fistula, yellow-nail syndrome, postcardiac injury syndrome, acute chest syndrome.

Table Graphic Jump Location
Table 2 —Comparison of Pleural Fluid Tests Between Transudates and Exudates

Summary statistics are presented as mean (SD) unless stated otherwise, with P values generated by the two-tailed t test. FL-R = pleural fluid to serum lactate dehydrogenase ratio; FP-R = pleural fluid to serum protein ratio; LDH = lactate dehydrogenase; SF-A = serum to pleural fluid albumin; SF-P = serum to pleural fluid protein.

a 

All hypotheses were tested using a Bonferroni corrected type 1 error (α level) of 0.00625 (0.05/8).

b 

Because of the skewness of the distribution, median (interquartile range) is presented, with P values generated by the two-tailed Wilcoxon rank-sum test.

Table Graphic Jump Location
Table 3 —Diagnostic Accuracy of Pleural Fluid Tests That Identify Exudates in Pleural Effusions

LR− = negative likelihood ratio; LR+ = positive likelihood ratio. See Table 2 legend for expansion of other abbreviations.

a 

Total number of complete observations used for calculating sensitivity, specificity, LR+, and LR−.

b 

Light’s criteria are FP-R > 0.5, or FL-R > 0.6, or pleural fluid LDH concentration greater than two-thirds of upper normal limit for serum LDH (a positive finding from any one of the three is diagnostic of an exudate).

c 

Similar to Light’s criteria but uses a pleural fluid LDH cutoff value of > 0.45 the limit of normal for serum LDH.

There were seven pleural fluid tests with similar or better specificity than Light’s criteria: (1) pH < 7.30, (2) protein, (3) LDH, (4) cholesterol, (5) FP-R, (6) FL-R, and (7) SF-P gradient. When tested in a univariable logistic regression, five tests showed a significant effect on the probability of correctly identifying exudates: (1) pleural fluid protein (OR, 10.36 [95% CI, 7.72-13.91]), (2) pleural fluid LDH (OR, 31.44 [95% CI, 20.23-48.86]), (3) cholesterol (OR, 8.25 [95% CI, 4.30-15.83]), (4) SF-A gradient (OR, 11.78 [95% CI, 6.62-20.96]), and (5) SF-P gradient (OR, 9.94 [95% CI, 7.16-13.81]) (Table 4). When combined in a multivariable logistic regression, the only tests that maintained statistical significance and were used to fit the final model were (1) LDH (OR, 14.09 [95% CI, 2.25-88.50]), (2) SF-A gradient (OR, 7.16 [95% CI, 1.24-41.43]), and (3) SF-P gradient (OR, 6.83 [95% CI, 1.56-29.88]). Two-way and three-way interactions between all these predictors were tested, but none reached statistical significance. The three-test combination using LDH, SF-A gradient, and SF-P gradient in an “or” rule generated an area under the curve of 0.92 (95% CI, 0.85-0.98) (Fig 1). Light’s criteria could not be combined with LDH, SF-A gradient, and SF-P gradient, as it produced multicollinearity between the variables and, therefore, affected the accuracy of the model estimates. The bootstrap internal validation technique was used for the final model containing LDH, SF-A gradient, and SF-P gradient. The method produced a concordance index (c-index) of 0.892, which translated into a fairly good predictive capability of approximately 89%. The c-index is measured on a scale from 0 to 1, with 0.50 representing random prediction, > 0.80 some usefulness in prediction, and 1.00 perfect prediction.20

Table Graphic Jump Location
Table 4 —Univariable and Multivariable Logistic Regression Analysis of Pleural Fluid Tests That Identify Exudates in Pleural Effusions

N/A = predictor not tested in the multivariable model because of multicollinearity or model instability; ref = reference; UNL = upper normal limit. See Table 2 legend for expansion of other abbreviations.

a 

Total values differ because of missing observations.

Figure Jump LinkFigure 1. Receiver operating characteristics curves for correct identification of exudative effusions. FL-R = pleural fluid to serum lactate dehydrogenase ratio; FP-R = pleural fluid to serum protein ratio; LDH = lactate dehydrogenase; SF-A = serum to pleural fluid albumin; SF-P = serum to pleural fluid protein.Grahic Jump Location

Light’s criteria misclassified 81 of 290 transudates (28%): 47 of 170 transudative CHF (28%), eight of 57 transudative cirrhotic (14%), and 26 of 63 non-CHF effusions (37%), respectively (e-Fig 1). Pleural fluid LDH ≤ two-thirds UNL correctly identified 37 (79%), six (75%), and 16 (62%) effusions. The SF-A gradient > 1.2 g/dL correctly identified seven (32%), three (60%), and three (33%) effusions. The SF-P gradient > 3.1 g/dL correctly identified 14 (31%), three (38%), and nine (39%) effusions, respectively. The pleural fluid to albumin ratio correctly identified nine (43%), two (40%), and two (22%) effusions. Sequential application of pleural fluid LDH, SF-A, and SF-P gradients correctly identified > 90% of misclassified CHF transudates. Transudates, misclassified by Light’s criteria, had marginal fluid test results close to the cutoff values (e-Table 1).

Our results indicate that the SF-A and SF-P gradients increased the predictive accuracy to correctly identify exudative effusions. We found that, in combination with the fluid LDH, the SF-A and SF-P gradients had a significant improvement in the diagnostic accuracy when compared with Light’s criteria. Pleural fluid LDH less than or equal to two-thirds the UNL allowed for a greater accuracy of reclassifying a transudative CHF, cirrhotic, and non-CHF effusions than SF-A and SF-P gradients > 1.2 g/dL and 3.1 g/dL, respectively.

Unfortunately, Light’s criteria and modified Light’s criteria, which both use triplet test combinations in an “or” rule, could not be combined with each of the gradients or the fluid LDH to completely resolve the dilemma of misclassifying transudates as exudates using simple, universally known, and available pleural fluid tests. Since Light’s criteria already incorporated the fluid LDH, as well as the FP-R and FL-R, multicollinearity or high correlation between variables was suspected to affect the model accuracy. The same problem occurred when we combined fluid LDH with the FL-R and SF-P gradient with the FP-R. Consequently, the fluid LDH in combination with the SF-A and SF-P gradients may be used together when encountering pleural effusions suspected of being transudative that have been misclassified as exudative by Light’s criteria.

Sequential application of the fluid LDH, followed by the SF-P gradient and then the SF-A gradient, if the fluid LDH or SF-P gradient failed to correctly identify misclassified transudates, may improve the diagnostic accuracy and maintain simplicity without increasing the number of tests performed and maintaining costs. Light’s criteria should still be used as the initial step when encountering pleural effusions, given their comparable accuracy. Our proposed triad appears to be more accurate than Light’s criteria; however, the CIs for both did overlap. In addition, Light’s criteria are universally known and simple to apply. However, if there is suspicion that the effusion is transudative, this pleural fluid triad of tests should be considered.

Although our prevalence of 28% of misclassified total transudates was similar to previous studies, the ability of SF-A and SF-P gradients to identify misclassified CHF transudates was significantly less than in previous studies.17,18 In these two studies, the SF-A gradient correctly identified 86% and 83% of misclassified CHF transudates, whereas our results showed only 32%. The SF-P gradient correctly identified 91% and 55% of misclassified CHF transudates, whereas our results showed only 31%. The differences in the ability to correctly identify misclassified CHF transudates might be due to a significantly smaller number of misclassified transudates evaluated in our study. There were only 47 misclassified CHF transudates. The study by Bielsa and colleagues18 included 107 misclassified CHF transudates.

In addition, our study was not designed to specifically evaluate the SF-A and SF-P gradients in patients with CHF who received diuretics. We could not determine the number of patients who received diuretics before thoracentesis. Romero-Candeira and colleagues17 showed that in patients who received diuretic therapy, the SF-A and SF-P gradients had higher rates of correctly identifying misclassified transudates but similar accuracy when compared with Light’s criteria.

We did not measure pleural fluid N-terminal pro-brain natriuretic peptide, a biomarker that has been used in identifying misclassified CHF transudates by Light’s criteria but does not add additional information to a serum brain natriuretic peptide determination.21,22 Given its retrospective design, missing data were also a limitation in our study, as they reduced the proportion of patients for whom both gradients could be calculated. Our study also failed to identify the pleural fluid to serum albumin ratio as a significant fluid test to correctly identify misclassified CHF and cirrhotic transudates. The pleural fluid to albumin ratio was comparable to the SF-A and SF-P gradients but not superior to the fluid LDH.

Non-CHF misclassified transudates included patients with trapped lung, nephrosis, hypoalbuminemia, and other less-common causes of transudative effusions, such as atelectasis, constrictive pericarditis, urinothorax, duropleural fistulas, and peritoneal dialysis-related effusions. The SF-A and SF-P gradients identified only 33% and 39% of these misclassified transudates, respectively. Perhaps a reason for such a low detection of misclassified transudates is the calculation of a gradient. If both the serum and the fluid albumin and protein levels are low, then the gradient will also be falsely low, such as in patients with hypoalbuminemia.

Pleural fluid LDH was able to correctly identify misclassified CHF transudates better than SF-A and SF-P gradients. The high specificity associated with the fluid LDH might have led to its high performance and accuracy when combined with SF-A and SF-P gradients. The fluid LDH specificity was 94% in our study. Two studies with large cohorts such as ours reported fluid LDH specificities of about 80%.6,16 Three smaller studies reported pleural fluid LDH specificities between 87% and 100%.5,7,23 All five studies had higher proportions of TB than in our study.

Our study does have limitations. First, its retrospective design made it vulnerable to sampling and information bias. All of the patients were collected from a single center in a university hospital setting. Therefore, its generalizability to other settings, such as a population from a community setting, is doubtful. A priori knowledge of the pleural fluid analysis results might have affected chart review interpretation. Second, data were missing, and not all patients had complete pleural fluid tests to be analyzed.

Our proposed triad should complement the patient’s history as well as pretest probability determination of the likelihood of an underlying transudative effusion. If Light’s criteria result in misclassification of the suspected transudative effusion as exudative, then sequential application of the fluid LDH, followed by the SF-P and then the SF-A gradients, should assist in reclassifying the effusion as transudative.

In conclusion, the addition of the serum to fluid albumin and protein gradients to the pleural fluid LDH increased the predictive accuracy to identify exudative effusions. Sequential application of the fluid LDH, followed by the SF-P and then the SF-A gradients, significantly reduces the misclassification of transudative effusions as exudative.

Author contributions: Drs Kummerfeldt and Huggins guarantee the integrity of the entire study.

Dr Kummerfeldt: contributed to study concept and design, drafting of the article, and data acquisition and analysis and approved the submission of this version for publication.

Ms Chiuzan: contributed to study concept and design, data analysis, and critical revision of the article and approved the submission of this version for publication.

Dr Huggins: contributed to study concept and design and critical revision of the article and approved the submission of this version for publication.

Dr DiVietro: contributed to data acquisition and critical revision of the article and approved the submission of this version for publication.

Dr Nestor: contributed to data acquisition and critical revision of the article and approved the submission of this version for publication.

Dr Sahn: contributed to study concept and design and critical revision of the article and approved the submission of this version for publication.

Dr Doelken: contributed to data analysis and critical revision of the article and approved the submission of this version for publication.

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

Additional information: The e-Appendix, e-Figure, and e-Table can be found in the “Supplemental Materials” area of the online article.

CHF

congestive heart failure

FL-R

pleural fluid to serum lactate dehydrogenase ratio

FP-R

pleural fluid to serum protein ratio

LDH

lactate dehydrogenase

SF-A

serum to pleural fluid albumin

SF-P

serum to pleural fluid protein

UNL

upper normal limit

Light RW, Macgregor MI, Luchsinger PC, Ball WC Jr. Pleural effusions: the diagnostic separation of transudates and exudates. Ann Intern Med. 1972;77(4):507-513. [CrossRef] [PubMed]
 
Heffner JE. Discriminating between transudates and exudates. Clin Chest Med. 2006;27(2):241-252. [CrossRef] [PubMed]
 
Romero S, Candela A, Martín C, Hernández L, Trigo C, Gil J. Evaluation of different criteria for the separation of pleural transudates from exudates. Chest. 1993;104(2):399-404. [CrossRef] [PubMed]
 
Costa M, Quiroga T, Cruz E. Measurement of pleural fluid cholesterol and lactate dehydrogenase. A simple and accurate set of indicators for separating exudates from transudates. Chest. 1995;108(5):1260-1263. [CrossRef] [PubMed]
 
Vives M, Porcel JM, Vicente de Vera M, Ribelles E, Rubio M. A study of Light’s criteria and possible modifications for distinguishing exudative from transudative pleural effusions. Chest. 1996;109(6):1503-1507. [CrossRef] [PubMed]
 
Heffner JE, Brown LK, Barbieri CA; Primary Study Investigators. Diagnostic value of tests that discriminate between exudative and transudative pleural effusions. Chest. 1997;111(4):970-980. [CrossRef] [PubMed]
 
Romero S, Martinez A, Hernandez L, et al. Light’s criteria revisited: consistency and comparison with new proposed alternative criteria for separating pleural transudates from exudates. Respiration. 2000;67(1):18-23. [CrossRef] [PubMed]
 
Joseph J, Badrinath P, Basran GS, Sahn SA. Is the pleural fluid transudate or exudate? A revisit of the diagnostic criteria. Thorax. 2001;56(11):867-870. [CrossRef] [PubMed]
 
Joseph J, Badrinath P, Basran GS, Sahn SA. Is albumin gradient or fluid to serum albumin ratio better than the pleural fluid lactate dehydroginase in the diagnostic of separation of pleural effusion? BMC Pulm Med. 2002;2(2):1. [CrossRef] [PubMed]
 
Atalay F, Ernam D, Hasanoglu HC, Karalezli A, Kaplan O. Pleural adenosine deaminase in the separation of transudative and exudative pleural effusions. Clin Biochem. 2005;38(12):1066-1070. [CrossRef] [PubMed]
 
Papageorgiou E, Kostikas K, Kiropoulos T, Karetsi E, Mpatavanis G, Gourgoulianis KI. Increased oxidative stress in exudative pleural effusions: a new marker for the differentiation between exudates and transudates? Chest. 2005;128(5):3291-3297. [CrossRef] [PubMed]
 
Yetkin O, Tek I, Kaya A, Ciledag A, Numanoglu N. A simple laboratory measurement for discrimination of transudative and exudative pleural effusion: pleural viscosity. Respir Med. 2006;100(7):1286-1290. [CrossRef] [PubMed]
 
Ruiz García R, Márquez de Prado Urquía MM, Borque de Larrea L. Distinguishing pleural transudates and exudates through the quantification of biochemical parameters [in Spanish]. Rev Clin Esp. 2004;204(10):511-520. [CrossRef] [PubMed]
 
Heffner JE, Sahn SA, Brown LK. Multilevel likelihood ratios for identifying exudative pleural effusions*. Chest. 2002;121(6):1916-1920. [CrossRef] [PubMed]
 
Heffner JE, Highland K, Brown LK. A meta-analysis derivation of continuous likelihood ratios for diagnosing pleural fluid exudates. Am J Respir Crit Care Med. 2003;167(12):1591-1599. [CrossRef] [PubMed]
 
Porcel JM, Peña JM, Vicente de Vera C, Esquerda A, Vives M, Light RW. Bayesian analysis using continuous likelihood ratios for identifying pleural exudates. Respir Med. 2006;100(11):1960-1965. [CrossRef] [PubMed]
 
Romero-Candeira S, Hernández L, Romero-Brufao S, Orts D, Fernández C, Martín C. Is it meaningful to use biochemical parameters to discriminate between transudative and exudative pleural effusions? Chest. 2002;122(5):1524-1529. [CrossRef] [PubMed]
 
Bielsa S, Porcel JM, Castellote J, Mas E, Esquerda A, Light RW. Solving the Light’s criteria misclassification rate of cardiac and hepatic transudates. Respirology. 2012;17(4):721-726. [CrossRef] [PubMed]
 
Efron B. Estimating the error rate of a prediction rule: improvement on cross-validation. J Am Stat Assoc. 1983;78(382):316-331. [CrossRef]
 
Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361-387. [CrossRef] [PubMed]
 
Porcel JM, Martínez-Alonso M, Cao G, Bielsa S, Sopena A, Esquerda A. Biomarkers of heart failure in pleural fluid. Chest. 2009;136(3):671-677. [CrossRef] [PubMed]
 
Porcel JM. Utilization of B-type natriuretic peptide and NT-proBNP in the diagnosis of pleural effusions due to heart failure. Curr Opin Pulm Med. 2011;17(4):215-219. [CrossRef] [PubMed]
 
Valdés L, Pose A, Suàrez J, et al. Cholesterol: a useful parameter for distinguishing between pleural exudates and transudates. Chest. 1991;99(5):1097-1102. [CrossRef] [PubMed]
 

Figures

Figure Jump LinkFigure 1. Receiver operating characteristics curves for correct identification of exudative effusions. FL-R = pleural fluid to serum lactate dehydrogenase ratio; FP-R = pleural fluid to serum protein ratio; LDH = lactate dehydrogenase; SF-A = serum to pleural fluid albumin; SF-P = serum to pleural fluid protein.Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1 —Clinical Characteristics of the Study Population

Data are presented as No. (%) unless otherwise noted.

a 

Includes pancreatitis, benign asbestos pleural effusion, fibrothecoma, drug induced, radiation pleuritis, pulmonary embolus, sarcoidosis, pancreaticopleural fistula, yellow-nail syndrome, postcardiac injury syndrome, acute chest syndrome.

Table Graphic Jump Location
Table 2 —Comparison of Pleural Fluid Tests Between Transudates and Exudates

Summary statistics are presented as mean (SD) unless stated otherwise, with P values generated by the two-tailed t test. FL-R = pleural fluid to serum lactate dehydrogenase ratio; FP-R = pleural fluid to serum protein ratio; LDH = lactate dehydrogenase; SF-A = serum to pleural fluid albumin; SF-P = serum to pleural fluid protein.

a 

All hypotheses were tested using a Bonferroni corrected type 1 error (α level) of 0.00625 (0.05/8).

b 

Because of the skewness of the distribution, median (interquartile range) is presented, with P values generated by the two-tailed Wilcoxon rank-sum test.

Table Graphic Jump Location
Table 3 —Diagnostic Accuracy of Pleural Fluid Tests That Identify Exudates in Pleural Effusions

LR− = negative likelihood ratio; LR+ = positive likelihood ratio. See Table 2 legend for expansion of other abbreviations.

a 

Total number of complete observations used for calculating sensitivity, specificity, LR+, and LR−.

b 

Light’s criteria are FP-R > 0.5, or FL-R > 0.6, or pleural fluid LDH concentration greater than two-thirds of upper normal limit for serum LDH (a positive finding from any one of the three is diagnostic of an exudate).

c 

Similar to Light’s criteria but uses a pleural fluid LDH cutoff value of > 0.45 the limit of normal for serum LDH.

Table Graphic Jump Location
Table 4 —Univariable and Multivariable Logistic Regression Analysis of Pleural Fluid Tests That Identify Exudates in Pleural Effusions

N/A = predictor not tested in the multivariable model because of multicollinearity or model instability; ref = reference; UNL = upper normal limit. See Table 2 legend for expansion of other abbreviations.

a 

Total values differ because of missing observations.

References

Light RW, Macgregor MI, Luchsinger PC, Ball WC Jr. Pleural effusions: the diagnostic separation of transudates and exudates. Ann Intern Med. 1972;77(4):507-513. [CrossRef] [PubMed]
 
Heffner JE. Discriminating between transudates and exudates. Clin Chest Med. 2006;27(2):241-252. [CrossRef] [PubMed]
 
Romero S, Candela A, Martín C, Hernández L, Trigo C, Gil J. Evaluation of different criteria for the separation of pleural transudates from exudates. Chest. 1993;104(2):399-404. [CrossRef] [PubMed]
 
Costa M, Quiroga T, Cruz E. Measurement of pleural fluid cholesterol and lactate dehydrogenase. A simple and accurate set of indicators for separating exudates from transudates. Chest. 1995;108(5):1260-1263. [CrossRef] [PubMed]
 
Vives M, Porcel JM, Vicente de Vera M, Ribelles E, Rubio M. A study of Light’s criteria and possible modifications for distinguishing exudative from transudative pleural effusions. Chest. 1996;109(6):1503-1507. [CrossRef] [PubMed]
 
Heffner JE, Brown LK, Barbieri CA; Primary Study Investigators. Diagnostic value of tests that discriminate between exudative and transudative pleural effusions. Chest. 1997;111(4):970-980. [CrossRef] [PubMed]
 
Romero S, Martinez A, Hernandez L, et al. Light’s criteria revisited: consistency and comparison with new proposed alternative criteria for separating pleural transudates from exudates. Respiration. 2000;67(1):18-23. [CrossRef] [PubMed]
 
Joseph J, Badrinath P, Basran GS, Sahn SA. Is the pleural fluid transudate or exudate? A revisit of the diagnostic criteria. Thorax. 2001;56(11):867-870. [CrossRef] [PubMed]
 
Joseph J, Badrinath P, Basran GS, Sahn SA. Is albumin gradient or fluid to serum albumin ratio better than the pleural fluid lactate dehydroginase in the diagnostic of separation of pleural effusion? BMC Pulm Med. 2002;2(2):1. [CrossRef] [PubMed]
 
Atalay F, Ernam D, Hasanoglu HC, Karalezli A, Kaplan O. Pleural adenosine deaminase in the separation of transudative and exudative pleural effusions. Clin Biochem. 2005;38(12):1066-1070. [CrossRef] [PubMed]
 
Papageorgiou E, Kostikas K, Kiropoulos T, Karetsi E, Mpatavanis G, Gourgoulianis KI. Increased oxidative stress in exudative pleural effusions: a new marker for the differentiation between exudates and transudates? Chest. 2005;128(5):3291-3297. [CrossRef] [PubMed]
 
Yetkin O, Tek I, Kaya A, Ciledag A, Numanoglu N. A simple laboratory measurement for discrimination of transudative and exudative pleural effusion: pleural viscosity. Respir Med. 2006;100(7):1286-1290. [CrossRef] [PubMed]
 
Ruiz García R, Márquez de Prado Urquía MM, Borque de Larrea L. Distinguishing pleural transudates and exudates through the quantification of biochemical parameters [in Spanish]. Rev Clin Esp. 2004;204(10):511-520. [CrossRef] [PubMed]
 
Heffner JE, Sahn SA, Brown LK. Multilevel likelihood ratios for identifying exudative pleural effusions*. Chest. 2002;121(6):1916-1920. [CrossRef] [PubMed]
 
Heffner JE, Highland K, Brown LK. A meta-analysis derivation of continuous likelihood ratios for diagnosing pleural fluid exudates. Am J Respir Crit Care Med. 2003;167(12):1591-1599. [CrossRef] [PubMed]
 
Porcel JM, Peña JM, Vicente de Vera C, Esquerda A, Vives M, Light RW. Bayesian analysis using continuous likelihood ratios for identifying pleural exudates. Respir Med. 2006;100(11):1960-1965. [CrossRef] [PubMed]
 
Romero-Candeira S, Hernández L, Romero-Brufao S, Orts D, Fernández C, Martín C. Is it meaningful to use biochemical parameters to discriminate between transudative and exudative pleural effusions? Chest. 2002;122(5):1524-1529. [CrossRef] [PubMed]
 
Bielsa S, Porcel JM, Castellote J, Mas E, Esquerda A, Light RW. Solving the Light’s criteria misclassification rate of cardiac and hepatic transudates. Respirology. 2012;17(4):721-726. [CrossRef] [PubMed]
 
Efron B. Estimating the error rate of a prediction rule: improvement on cross-validation. J Am Stat Assoc. 1983;78(382):316-331. [CrossRef]
 
Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361-387. [CrossRef] [PubMed]
 
Porcel JM, Martínez-Alonso M, Cao G, Bielsa S, Sopena A, Esquerda A. Biomarkers of heart failure in pleural fluid. Chest. 2009;136(3):671-677. [CrossRef] [PubMed]
 
Porcel JM. Utilization of B-type natriuretic peptide and NT-proBNP in the diagnosis of pleural effusions due to heart failure. Curr Opin Pulm Med. 2011;17(4):215-219. [CrossRef] [PubMed]
 
Valdés L, Pose A, Suàrez J, et al. Cholesterol: a useful parameter for distinguishing between pleural exudates and transudates. Chest. 1991;99(5):1097-1102. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Supporting Data

Online Supplement

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543