Postoperative air leak grading is useful to predict prolonged air leak after pulmonary lobectomy
- Sang Gi Oh1,
- Yochun Jung†1Email author,
- Sanghoon Jheon2,
- Yunhee Choi3,
- Ju Sik Yun4,
- Kook Joo Na4 and
- Byoung Hee Ahn†1Email author
© The Author(s). 2017
Received: 11 July 2016
Accepted: 19 January 2017
Published: 23 January 2017
Results of studies to predict prolonged air leak (PAL; air leak longer than 5 days) after pulmonary lobectomy have been inconsistent and are of limited use. We developed a new scale representing the amount of early postoperative air leak and determined its correlation with air leak duration and its potential as a predictor of PAL.
We grade postoperative air leak using a 5-grade scale. All 779 lobectomies from January 2005 to December 2009 with available medical records were reviewed retrospectively. We devised six ‘SUM’ variables using air leak grades in the initial 72 h postoperatively.
Excluding unrecorded cases and postoperative broncho-pleural fistulas, there were 720 lobectomies. PAL occurred in 135 cases (18.8%). Correlation analyses showed each SUM variable highly correlated with air leak duration, and the SUM4to9, which was the sum of six consecutive values of air leak grades for every 8 h record on postoperative days 2 and 3, was proved to be the most powerful predictor of PAL; PAL could be predicted with 75.7% and 77.7% positive and negative predictive value, respectively, when SUM4to9 ≥ 16. When 4 predictors derived from multivariable logistic regression of perioperative variables were combined with SUM4to9, there was no significant increase in predictability compared with SUM4to9 alone.
This simple new method to predict PAL using SUM4to9 showed that the amount of early postoperative air leak is the most powerful predictor of PAL, therefore, grading air leak after pulmonary lobectomy is a useful method to predict PAL.
KeywordsProlonged air leak Lobectomy Air leak grade
There is no clear consensus on the duration of prolonged air leak (PAL) , which is usually considered as lasting longer than 5 or 7 days postoperatively. Air leaking after pulmonary resection is natural, but its prolongation increases the risks of other pulmonary complications such as empyema  and unnecessarily lengthens hospital stay [3, 4]. Therefore, accurate prediction of PAL could enable early, selective postoperative management to prevent PAL possible, which in turn help to reduce the complication risks and hospital costs.
Many studies to elucidate the risk factors of PAL have been made to predict its occurrence [5–10], but the results were inconsistent and therefore, of limited use clinically. Thus, rather than identifying the risk factors, we sought to determine whether observing the pattern of postoperative air leak might be a more direct and accurate way, based on a simple assumption: ‘the larger, the longer’.
In this study, the authors developed a new quantitative scale to express the amount of early postoperative air leak to determine its correlation with air leak duration and possibility as a predictor of PAL.
The medical records of 779 lobectomies conducted consecutively at our institution from January 2005 to December 2009 were reviewed retrospectively. This study was approved by Seoul National University Bundang Hospital’s Institutional Review Board, and the need for informed consent was waived (IRB number: B-1010-113-103).
Grading of air leak and definition of time period ‘P’
Air leak grading
No air bubble on three serial volitional coughs
More than one air bubbles on three serial volitional coughs
Persistent air bubbles on volitional coughs
Persistent, small amount of air bubbles on spontaneous respiration
Persistent, large amount of air bubbles on spontaneous respiration
Definition of air leak cessation and PAL
We generally do not remove the chest tube until the drainage amount decreases to less than 150 – 200 mL within the previous 24 h. Therefore, chest tube removal time cannot reflect the time when air leak stops, so prior to performing this study, air leak cessation was necessary to be defined clearly based on the grading records. Air leak cessation was defined as air leak grade 0 or 1 being continued for 3 Ps (24 h). In cases with 2 chest tubes, air leak cessation was timed based on the late-ceased one. In this study, PAL was defined as air leak persisting more than 5 days (15 Ps).
‘SUM’ variables as predictors of PAL
Definitions of 6 ‘SUM’ variables
Air leak gradea
N1 + N2 + N3
N4 + N5 + N6
N1 + N2 + N3 + N4 + N5 + N6
N7 + N8 + N9
N4 + N5 + N6 + N7 + N8 + N9
N1 + N2 + N3 + N4 + N5 + N6 + N7 + N8 + N9
Validation of the reliability of air leak grading
For 1 week while conducting this study, we had one or two pairs of nurses from each shift assess the air leak grade simultaneously but independently, and one of the authors (Jung Y) compared the two grades to monitor the agreement on air leak grading.
Spearman rank correlation coefficients were calculated to investigate the correlation between each SUM variable and air leak duration. Receiver operating characteristic (ROC) curve analyses were performed and areas under the ROC curves (AUC) were compared to select the best predictive SUM variable, then its cutoff value that optimized both sensitivity and specificity was obtained.
The distribution of preoperative and intraoperative variables were compared between PAL (+) and PAL (−). Categorical variables were expressed as the frequency and percentage of cases and compared by either the χ2 test or Fisher’s exact test, as necessary. For a continuous variable with a nonnormalized distribution, data were reported as a median with range and compared by the Mann-Whitney U-test. The significant factors determined by a two-tailed nominal P value of <0.2 in univariable analysis were entered into a multivariable logistic regression to identify predictors of PAL. Stepwise logistic regression was performed to control multicollinearity among predictors and for each element remaining in the multivariable model, a P value, odds ratio (OR), and 95% confidence interval (CI) were calculated. The Hosmer-Lemeshow goodness-of-fit statistic was used to evaluate the model fit. The resultant predictors and the optimal SUM variable were combined to create a multivariable logistic regression, the AUC of which was compared with that of the optimal SUM variable regarding predictive ability . A two-sided P value of <0.05 was considered to indicate a statistically significant difference. In order to assess the agreement on air leak grading, the weighted kappa statistic was performed. The analysis was carried out using PASW Statistics, version 17.0 (SPSS Inc.; Chicago, IL) and SAS, version 9.1 (SAS Institute; Cary, NC).
Baseline characteristics and results of univariable analyses
No. of cases (%)
(N = 720)
(N = 585)
(N = 135)
Aspirin or NSAIDs
Steroid (including inhalator)
FEV1/FVC < 70%
FVC% < 80
FEV1% < 80
DLCO% < 80
Preoperative Hb < 10 g/dL
Preoperative albumin < 3.0 g/dL
BMI < 25.5 kg/m2
Right side operation
Combined chest wall resection
Pleural adhesion (+)
Incomplete fissure (+)
Median duration of postoperative air leak was 2.7 (0.7–58.0) days and the median of chest tube removal was at postoperative day (POD) 4.7 (0.7–67.0), and that of discharge was POD 7 (2–187). The numbers of cases with persistent air leak at P4, P7, P10, P13 and P16 are 577, 410, 275, 184, and 127, respectively. A total of 127 cases with persistent air leak at P16 were considered as PAL and 8 other cases were also regarded as PAL despite their air leak duration being ≤ 15 Ps, because pleurodesis had been conducted for prophylaxis of PAL, so the total occurrence of PAL in this study cohort was 135 (18.8%).
The larger, the longer?
Statistical values of SUM variables
Spearman rank correlation coefficient
Prediction of PAL by the SUM scale
PAL occurred in 134 cases (48.7%) out of 275 cases with persistent air leak at P10. ROC curves were generated for each SUM variable. The calculated AUC revealed that SUM4to9 to be the most powerful predictor of PAL (the largest AUC; 0.819. Table 4), and PAL could be predicted with 76.9% sensitivity and 76.6% specificity with 75.7% positive predictive value and 77.7% negative predictive value, when SUM4to9 ≥ 16.
Results of multivariable logistic regression
Right side operation
BMI < 25.5 kg/m2
Reliability of air leak grading
A total of 33 pairs of air leak grades were collected. Twenty two pairs of ward nurses participated in the grading. There were 7 disagreements on air leak grading, but the degrees of difference were never more than a single grade. The weighted kappa coefficient between the 2 measurements of air leak grade was 0.88 (95% CI: 0.79–0.96), indicating a very good interobserver agreement .
We set out to determine whether our newly developed scale could be useful as a predictor of PAL. Although there have been many studies to predict PAL, most of which have tried to determine predictive risk factors by comparing PAL groups with other groups and reported various risk factors such as poor pulmonary function [6, 8–10], poor nutrition , or specific operative findings [8, 10], but their results were not consistent with each other and therefore are of limited clinical use. Brunelli et al.  in 2010, unlike previous studies, reported a risk factor scoring system for prediction of PAL. He elicited 4 predictors from a study group of 658 lobectomy patients: age > 65 years, presence of pleural adhesion, FEV1% < 80, and BMI < 25.5 kg/m2, and different scores were given to the factors according to their weights. The scoring system was validated externally in 233 other hospital patients. However it also had a limitation because of its low positive predictive value of about 25%.
Various investigators have examined the concept that grading the amount of early postoperative air leak might be helpful in predicting PAL [13, 14]. In 2001, Cerfolio et al.  reported that PAL can be predicted by air leak grade on POD 1. In their study, a commercially available air-leak meter, scoring leaks from 1 to 7 with 7 being the highest, was used. However, originally his work aimed to evaluate the effectiveness of water seal for stopping air leaks. Thus he did not give much weight to air leak grading as a predictor of PAL. In addition, the air leak grading system devised by Cerfolio et al. requires special equipment, and thus the system is not widely used.
To the best of our knowledge, the present study is the first practical attempt to focus only on quantifying air leakage in the early postoperative period to predict PAL. The easy-to-use variable, SUM4to9, has the highest positive predictive value among reports until now. Our air leak grading system, to obtain SUM4to9, needs no special equipment, and yet it is very convenient to apply in the clinical field. Based on our results, we can now decide whether to wait or perform a reintervention (e.g. pleurodesis or redo surgery) for air leak cessation on POD 3.
In addition to proposing a practical and effective method to predict PAL, this study tangibly confirms our hypothesis that the amount of early postoperative air leak predicts air leak duration, by correlation analyses of SUM variables with air leak duration. Furthermore, it reveals that other preoperative or intraoperative variables do not increase the predictive power of SUM4to9. These findings can be integrated to mean that 1) grading postoperative air leak might be the only factor needed to predict PAL, and 2) the effects of various possible factors contributing to prolongation of air leak might combine to result in the grade of early postoperative air leak. Therefore, future studies aiming at providing more accurate prediction of PAL will have to focus on the evaluation of early postoperative air leak in terms of when, by what method, and how frequently it should be measured, not on other indirect factors.
However, this study has the following potential limitations. First of all, the retrospective nature of the study might have incurred some problems in defining and recording the variables. In particular, since air leak cessation was determined retrospectively from medical records, there might have been some discrepancy between the real and the defined cessations. Essentially, our air leak grading is based on subjective assessment, albeit by specially trained nurses, so some might question its reliability. However, over the past several years, we have empirically recognized that its simplicity enables strong interobserver agreement, as shown in the test to see the reliability of our air leak grade. Recently developed digital airflowmetry may be helpful in further increasing interobserver agreement in future studies.
We developed an easy-to-use method to predict PAL using a new scale, SUM4to9, derived from our own air leak grading protocol. This study proved that the amount of early postoperative air leak is the most powerful predictor of PAL, therefore, grading air leak after pulmonary lobectomy is a useful method to predict PAL.
Area under the ROC curve
Body mass index
% of predicted normal diffusing capacity of the lung for carbon monoxide
% of predicted normal forced expiratory volume in one second
% of predicted normal forced vital capacity
Non-steroidal anti-inflammatory drug
Prolonged air leak
- ROC curve:
Receiver operating characteristic curve
Video-assisted thoracoscopic surgery
The authors give their warmest thanks to Professor J. Patrick Barron of Department of International Communications Center of Tokyo Medical University for his kind pro vono review of this manuscript, and to Professor Cheong Lim of Department of Thoracic and Cardiovascular Surgery, Seoul National University Bundang Hospital, for his editing of the figure. We also appreciate the statistical analysis provided by the Medical Research Collaborating Center at Seoul National University College of Medicine / the Seoul National University Hospital, and thank the DSYU English Editing Service and Editage (www.editage.co.kr) for English language editing.
This study received no external funding and was completely supported by the budget of Department of Thoracic and Cardiovascular Surgery, Seoul National University Bundang Hospital.
Availability of data and materials
The dataset supporting the conclusions of this article is included within the article and its additional file.
YJ and BHA initially proposed the basic idea of air leak grading for prediction of PAL. SGO and YJ designed this study and were principally involved in data collection, statistical analysis, interpretation, and manuscript writing. SJ also participated in designing this study and was involved in the writing of manuscript. YC was principally involved in selecting appropriate statistical method and validating the results. JSY, KJN, and BHA made a significant contribution to this manuscript in the process of initial research design, data analysis, and revision of the draft. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
This study was approved by Seoul National University Bundang Hospital’s Institutional Review Board, with patient consent waived (IRB number: B-1010-113-103).
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Adebonojo SA. How prolonged is “prolonged air leak”? Ann Thorac Surg. 1995;59:549–50.View ArticlePubMedGoogle Scholar
- Brunelli A, Xiume F, Al Refai M, Salati M, Marasco R, Sabbatini A. Air leaks after lobectomy increase the risk of empyema but not of cardiopulmonary complications: a case-matched analysis. Chest. 2006;130:1150–6.View ArticlePubMedGoogle Scholar
- Bardell T, Petsikas D. What keeps postpulmonary resection patients in hospital? Can Respir J. 2003;10:86–9.View ArticlePubMedGoogle Scholar
- Varela G, Jimenez MF, Novoa N, Aranda JL. Estimating hospital costs attributable to prolonged air leak in pulmonary lobectomy. Eur J Cardiothorac Surg. 2005;27:329–33.View ArticlePubMedGoogle Scholar
- Singhal S, Ferraris VA, Bridges CR, Clough ER, Mitchell JD, Fernando HC, et al. Management of alveolar air leaks after pulmonary resection. Ann Thorac Surg. 2010;89:1327–35.View ArticlePubMedGoogle Scholar
- Abolhoda A, Liu D, Brooks A, Burt M. Prolonged air leak following radical upper lobectomy: an analysis of incidence and possible risk factors. Chest. 1998;113:1507–10.View ArticlePubMedGoogle Scholar
- Isowa N, Hasegawa S, Bando T, Wada H. Preoperative risk factors for prolonged air leak following lobectomy or segmentectomy for primary lung cancer. Eur J Cardiothorac Surg. 2002;21:951.View ArticlePubMedGoogle Scholar
- Brunelli A, Monteverde M, Borri A, Salati M, Marasco RD, Fianchini A. Predictors of prolonged air leak after pulmonary lobectomy. Ann Thorac Surg. 2004;77:1205–10. discussion 10.View ArticlePubMedGoogle Scholar
- Stolz AJ, Schutzner J, Lischke R, Simonek J, Pafko P. Predictors of prolonged air leak following pulmonary lobectomy. Eur J Cardiothorac Surg. 2005;27:334–6.View ArticlePubMedGoogle Scholar
- Brunelli A, Varela G, Refai M, Jimenez MF, Pompili C, Sabbatini A, et al. A scoring system to predict the risk of prolonged air leak after lobectomy. Ann Thorac Surg. 2010;90:204–9.View ArticlePubMedGoogle Scholar
- DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–45.View ArticlePubMedGoogle Scholar
- Altman DG. Practical statistics for medical research. London: Chapman and Hall; 1991. p. 404.Google Scholar
- Cerfolio RJ, Bass C, Katholi CR. Prospective randomized trial compares suction versus water seal for air leaks. Ann Thorac Surg. 2001;71:1613–7.View ArticlePubMedGoogle Scholar
- Sakamoto T, Nishio W, Okada M, Harada H, Uchino K, Tsubota N. Management of air leak after pulmonary resection. Jpn J Thorac Cardiovasc Surg. 2004;52:292–5.View ArticlePubMedGoogle Scholar