Surgical Sealant for the Prevention of Prolonged Air Leak After Lung Resection: Meta-Analysis

Transcription

1 Surgical Sealant for the Prevention of Prolonged Air Leak After Lung Resection: Meta-Analysis Guislain Malapert, MD, Halim Abou Hanna, MD, Pierre Benoit Pages, MD, and Alain Bernard, MD Department of General Thoracic Surgery, Centre Hospitalier Universitaire Dijon, Hospital du Bocage, Dijon, France Background. We performed a systematic and metaanalysis of randomized controlled trials comparing a surgical with buttressed staple lines using standard methods. The aim of our meta-analysis was to determine the effectiveness and safety of different techniques to reduce the proportion of patients with prolonged air leakage after pulmonary resection. Methods. We searched the Medline, Embase, Science Direct, Food and Drug Administration, Cochrane controlled trials register, and clinical trial databases for publications between January 1995 and May 2009 that included terms related to prolonged air leak after lung resection. We included randomized controlled trials comparing glue or patch or buttressed staple line with suture or staple in patients undergoing lung resection (wedge resection or lobectomy). The prespecified primary outcome of our meta-analysis was prolonged air leak more than 7 days. Secondary outcomes were the occurrence of adverse effects. Results. Thirteen trials were included in the metaanalysis. Overall, the trials had allocated 1,335 patients to glue or patch (1,064 patients) or buttress (271 patients) for the prevention of prolonged air leak after lung resection. The type of buttress used to reinforce the staple line was bovine pericardial strips (271 patients). In the control group of all trials for air-leakage management, single or continuous running sutures or staples were used according to the routine of the center. The use of glue or a patch or buttressing compared with control groups (1,335 patients) decreased prolonged air leak more than 7 days. Indeed, the pooled effect size odds ratio was 0.55 (95% confidence interval: to 0.79). An I 2 of 0% indicated low between-trial heterogeneity. The funnel-plot asymmetry coefficient was significantly different from zero (asymmetry coefficient 1.23 (95% confidence interval: 2.38 to 0.086; p < 0.04), indicating the presence of publication bias. Neither glue nor a patch nor buttressing influenced the occurrence of postoperative complications such as atelectasis, hemothorax, pneumonia, pneumothorax, and mortality. Eight trials (1,020 patients) showed that, compared with control groups, the use of glue or a patch or buttressing decreased postoperative arrhythmia, which yielded a pooled odds ratio of 0.44 (95% confidence interval: to 0.72). Conclusions. The use of surgical s and buttressing decreased the risk of prolonged air leakage and postoperative arrhythmia after pulmonary resection. However, given the possibility of publication bias, the conclusions should be interpreted with caution. (Ann Thorac Surg 2010;90: ) 2010 by The Society of Thoracic Surgeons Prolonged air leak is defined as air leakage lasting more than 7 days after pulmonary resection. Prolonged air leak is the most common complication after lung resection and the reported incidence ranges from 15 to 18% [1, 2]. Prolonged air leak has a detrimental effect on the postoperative course as it results in a longer need for a chest tube with the associated pain, reduced mobility, and increased risk of further complications [2]. Prolonged air leak for more than 7 days is a complication that penalizes the patient by requiring a chest tube. One may question the clinical relevance of endpoints such as duration of air leak. For patients, what is the clinical impact of reducing the duration of air leakage by 1 or 2 days, as reported by some trials. Classical methods for intraoperative control of air leaks include suture and Accepted for publication July 9, Address correspondence to Dr Bernard, Division of Thoracic Surgery, CHU Hospital du Bocage, Blvd de Lattre de Tassigny, Dijon Cedex, 21034, France; alain.bernard@chu-dijon.fr. stapling, but have the disadvantage of causing further trauma to lung tissue. Various additional techniques, including the use of a variety of s, such as fibrin glue, synthetic materials, and collagen patches coated with fibrinogen and thrombin, have been employed to minimize the intensity and duration of air leaks. Other techniques using staples and strips of reinforcing material to buttress the staple lines have also been tried to reduce air leaks after lung resection. Several trials have compared these different techniques to reduce air leakage with classical methods. We performed a systematic and meta-analysis of randomized controlled trials that compared the use of surgical s or buttressed staple lines with the use standard methods. The aim of our meta-analysis was to determine the effectiveness and safety of different techniques to reduce the percentage of patients with a prolonged air leak after pulmonary resection by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 1780 MALAPERT ET AL Ann Thorac Surg PROLONGED AIR LEAK AFTER LUNG RESECTION 2010;90: Material and Methods Literature Search We searched the Medline, Embase, Science Direct, Food and Drug Administration, Cochrane controlled trials register, and clinical trial databases for publications between January 1995 and May 2009 that included terms related to prolonged air leak after lung resection. We entered the relevant articles into a science citation index to retrieve reports that cited these articles, manually searched for conference proceedings and thoracic congresses, screened reference lists of all the articles obtained, and checked the proceedings of the US Food and Drug Administration advisory and health technology assessment agencies. Trial Selection We included randomized controlled trials comparing glue or patch or buttressed staple line with suture or staple in patients undergoing a lung resection (wedge resection or lobectomy). We included every randomized controlled trial that reported on the primary or secondary outcome of prolonged air leak for more than 7 days. Two reviewers (G.M. and H.A.) independently evaluated the reports for their eligibility. Disagreements were resolved by consensus. Quality Assessment Two reviewers (G.M. and H.A.) independently assessed concealment of treatment allocation, blinding (masking) and analyses. Concealment of allocation was considered adequate if the investigators responsible for patient selection were unable to determine before allocation which treatment was next in line (central randomization, sealed, opaque, sequentially numbered assignment envelopes). Patient blinding was considered adequate if it was stated that patients were blind to the assigned treatment. Therapist blinding was considered adequate if participants assessing the outcomes were blinded to group assignment. The number of patients randomized per group and the number of patients analyzed per group were extracted to distinguish between trials that had included all randomized patients in the analysis (considered as an intent-to-treat analysis and therefore adequate) and trials that had not. Disagreements were resolved by consensus. OUTCOME MEASURES. The prespecified primary outcome of our meta-analysis was prolonged air leak longer than 7 days. Secondary outcomes were the occurrence of adverse effects. DATA COLLECTION. Data on publication status, trial design, patient characteristics (age, pulmonary pathology, type of procedure) treatment regimens, prolonged air leak more than 7 days, adverse effects, quality assessment and funding were extracted in duplicate (by G.M. and H.A.) using a standardized form. Any disagreements were resolved by discussion. Statistical Analysis For prolonged air leak more than 7 days considered as dichotomous outcomes, effect sizes were estimated by odds ratio (OR). An effect size less than 1 indicates the superiority of treatment compared with suture or staple. We used a random-effects meta-analysis to pool effect sizes and calculated the I 2 statistic for each metaanalysis. This statistic describes the percentage of total variation across a trial that is attributable to statistical heterogeneity rather than chance [3]. The I 2 values of 25%, 50%, and 75% correspond to low, moderate, and high between-trial heterogeneity, respectively. We performed stratified meta-analyses to investigate potential sources of statistical heterogeneity. The following trial characteristics were considered for stratification: adequacy of concealment of allocation, blinding of patients, adequacy of the analysis, trial size, type of pathology, type of procedure, and type of. We used prespecified cutoffs of 100 randomized patients to distinguish between small- and large-scale trials. Univariable random-effects meta-regression analysis was used to examine whether effect sizes were affected by these factors. We performed post hoc sensitivity analyses excluding outlier studies from the main meta-analysis. Studies were considered outliers if the confidence interval of the estimated effect size from these studies did not overlap with pooled overall effect size. Sensitivity analyses also focused on the buttressed staple line and surgery for emphysema when studies were excluded from the main meta-analysis. Asymmetry of the funnel-plot was assessed by the asymmetry coefficient to search for likely effects of publication bias [4]. The Egger test detects funnel plot asymmetry by determining whether the intercept deviates significantly from zero in a regression of standardized effect estimates against their precision [4]. For these trials, we determined the contribution to Cochran s heterogeneity Q statistic in the overall analysis [5]. An approach developed by the DerSimonian and Laird methods was used to perform random-effects meta-analysis for overall effect measures [6]. All confidence intervals (CI) relate to the 95% limit, and p values were two-sided. Analyses were performed using STATA 11 statistical software (StataCorp, College Station, TX). In addition, we conducted a Bayesian approach to random-effects meta-analysis [7]. We performed a sensitivity analysis to prior distributions. The three Bayesian models differ only with respect to the prior placed on the between-study variance 2: (1) a noninformative inverse gamma (IG [0.0001, ]); (2) a half-normal (0, 1,000); and (3) a Dumouchel prior [8]. Bayesian analysis was performed using WinBugs software [9]. Results From 1995 to 2009, we identified 811 references and considered 60 to be potentially eligible for the metaanalysis (Fig 1). Thirteen trials [10 22] were included in the meta-analysis (Fig 1). Forty-seven articles were ex-

3 Ann Thorac Surg MALAPERT ET AL 2010;90: PROLONGED AIR LEAK AFTER LUNG RESECTION 1781 Fig 1. Flow diagram of articles evaluated for inclusion or exclusion. cluded for the following reasons: no randomized trials (n 37), no surgical (n 4), pleural tenting (n 2), outcome after prolonged air leak more than 7 days not reported (n 4; Fig 1). In the group, glue or patch was used in nine trials [11, 14 18, 20 22] and buttress in four trials [10, 12, 13, 19] (Table 1). Overall, the trials had allocated 1,335 patients to glue or patch (1,064 patients) or buttress (271 patients) for the prevention of prolonged air leak after lung resection. Fibrin glue was used in three trials (210 patients) [11, 16, 22], and four trials used synthetic s (500 patients) [14, 15, 17, 20]. Two trials used a collagen patch (354) coated with fibrinogen and Table 1. Characteristics of Identified Randomized Controlled Trials Author, Year [Ref #] Sealant Group No. Patients Sealant Group No. Patients Control Group Age (Years) Emphysema Procedure Concealed Allocation Therapist Blinding Intent to Treat Hazzelrig, 1997 [10] Wong, 1997 [11] Venuta, 1998 [12] Stammberger, 2000 [13] Wain, 2001 [14] Porte, 2001 [15] Fabian, 2003 [16] Allen, 2004 [17] Lang, 2004 [18] Gomes-Caro, 2006 [19] Tansley, 2006 [20] Anegg, 2007 [21] Moser, 2008 [22] Buttress Yes Wedge Unclear No Yes Fibrin glue No Lobectomy Adequate No No Buttress No Lobectomy Unclear No Yes Buttress Yes Wedge Unclear Yes No No Lobectomy Adequate Yes Yes No Lobectomy Unclear No No Fibrin glue No Lobectomy No Lobectomy Adequate Yes Yes Unclear No No Collagen patch No Lobectomy Unclear No No Buttress No Lobectomy Adequate No Yes No Lobectomy Adequate No Yes Collagen patch No Lobectomy Adequate Yes Yes Fibrin glue Yes Wedge Adequate Yes No

4 1782 MALAPERT ET AL Ann Thorac Surg PROLONGED AIR LEAK AFTER LUNG RESECTION 2010;90: thrombin [18, 21]. The type of buttress used to reinforce the staple line was bovine pericardial strips (271 patients) [10, 12, 13, 19]. In the control group of all trials [10 22], single or continuous running sutures or staples were used for air leakage management according to the routine of the centre. The average age of patients ranged from 55 to 68 years, and in two trials [10, 13], the lung resection was used to treat emphysema (Table 1). The type of lung resection was lobectomy in four trials [12, 15, 18, 19], wedge resection in three trials [10, 13, 22], and lobectomy in six trials [11, 13, 16, 17, 20, 21] (Table 1). Assessment of Quality Concealment of allocation was judged to be adequate in seven trials [11, 14, 16, 19 22] (Table 1). For all other trials, concealment of allocation remained unclear. The therapist who assessed the outcomes was blinded to the assigned intervention in five trials [13, 14, 16, 21, 22]. Seven trials [10, 12, 14, 16, 19 21] were considered to have performed an intent-to-treat analysis. Pooled Effect Sizes A total of 13 trials contributed to the meta-analysis of the prevention of prolonged air leak more than 7 days. The glue or patch or buttress compared with control groups (1,335 patients) decreased prolonged air leak of more than 7 days. Indeed, the pooled effect size was 0.55 (95% CI: to 0.79; Fig 2). An I 2 of 0% indicated low between-trial heterogeneity (p 0.5 for heterogeneity). The results from stratified analyses are presented in Table 2. Concealment of allocation, blinding, intent-totreat, trial size, emphysema, procedure, and type of had no influence on between-trial heterogeneity and p values for interactions between trial characteristics (Table 2). Sensitivity Analysis One trial [16[ was found to be an outlier. When the study of Fabien and coworkers [16] was excluded from the analysis, the pooled effect size decreased to 0.58 (95% CI: to 0.833) with an I 2 of 0%. When four trials [10, 12, 13, 19] with a buttressed staple line were excluded from the analysis, the pooled effect size was 0.58 (95% CI: 0.37 to 0.89) with an I 2 of 9.6%. When three trials [10, 13, 22] including surgery for emphysema were excluded from the analysis, the pooled effect size was 0.58 (95% CI: 0.39 to 0.86) with I 2 of 1.6%. Bayesian approaches with a sensitivity analysis to prior distributions demonstrated that glue,, patch, or buttress decreased prolonged air leak of more than 7 days compared with control groups (Table 3). The effect size was comparable to that estimated by the frequentist method. Publication Bias The funnel-plot asymmetry coefficient was significantly different from zero (asymmetry coefficient 1.23 (95% CI: 2.38 to 0.086; p 0.04), indicating the presence of publication bias (Fig 3). Figure 3 suggests that publication bias may be present, as the funnel appears to be skewed with a scarcity of small studies (a high standard error) in the bottom left-hand corner of the plot. Probable, missing studies were in the bottom right-hand corner of the plot. Adverse Events The principal adverse events reported in the trials were atelectasis, arrhythmia, hemothorax, pneumonia, pneumothorax, empyema, and postoperative death (Table 4). Neither glue nor patch nor buttress influenced the occurrence of postoperative complications such as atelectasis, hemothorax, pneumonia, pneumothorax, and death (Table 4). Seven trials (817 patients) with glue or patch or buttress showed a decreased risk of postoperative arrhythmia compared with control groups (Table 4) [12, 14, 17 21], which yielded a pooled odds ratio of 0.48 (95% CI: 0.28 to 0.81) and no between-trial heterogeneity (I 2 0%). Five trials with glue or patch or buttress compared with control groups contributed to the meta-analysis of empyema [11, 14, 15, 18, 19], which showed a pooled odds ratio of 2.2 (95% CI: 0.57 to 8.5; p 0.25) and no between-trial heterogeneity (I 2 0%; Table 4). Comment The various methods such as glue or patch or buttress can reduce the occurrence of prolonged air leak for more than 7 days compared with the conventional methods of suturing or stapling alone. The Cochrane review for the Fig 2. Forest plot of the metaanalysis of outcomes of air leak more than 7 days. (CI confidence interval; OR odds ratio.)

5 Ann Thorac Surg MALAPERT ET AL 2010;90: PROLONGED AIR LEAK AFTER LUNG RESECTION Table 2. Results of Stratified Meta-Analyses Related to Methodologic Characteristics of Trials With Glue or Patch or Buttress Compared With Control Groups a No. of Trials No. of Patients Effect Size OR (95% CI) I 2,% p Value for Interaction All 13 1, (0.386, 0.79) 0 Concealed allocation Adequate (0.21, 0.7) Unclear (0.4, 1.1) 0 Therapist blinding Yes (0.209, 0.841) No (0.38, 0.99) 0 Intent to treat Yes (0.26, 0.752) No (0.395, 1.122) 7 Trial size (0.155, 0.785) , (0.405, 0.933) 5 Emphysema No 10 1, (0.387, 0.855) Yes (0.156, 1.108) 10 Procedure Wedge (0.156, 1.108) Wedge or lobectomy (0.369, 1.023) 2 Lobectomy (0.213, 1.16) 22 Type of Fibrin glue (0.032, 0.511) 0 glue (0.368, 1.059) Collagen patch (0.392, 1.426) 0 Buttress (0.167, 0.974) a Outcome: prolonged air leak more than 7 days. CI confidence interval; OR odds ratio. same endpoint selected eight studies [23]. The authors [23] concluded that using a surgical could reduce the incidence of prolonged air leak after pulmonary resection, but the duration of hospitalization was not influenced. They did not recommend this practice routinely, but advised that other trials be conducted [23]. The prolonged air leak endpoint is nevertheless relevant because it indicates that patients are obliged to keep their chest tube, which causes pain and sometimes infectious complications. Secondly, the duration of the chest tube is not the only factor that explains the duration of hospitalization. Presumably, a publication bias was identified as shown in the funnel plot (Fig 3). The distribution of studies is asymmetrical around zero because smaller studies, which showed no statistically significant effects, remain unpublished [24]. Small studies are often more signifi- Table 3. Results of Pooling Trials With Glue or Patch and Buttress Compared With Control Groups Using Bayesian Model With Sensitivity Prior a Gamma ( ) Half-Normal Prior (0, 1000) Dumouchel Prior Between-study variance Pooled/posterior mean (odds ratio) 95% confidence interval to to to 0.72 a Outcome: prolonged air leak more than 7 days. Fig 3. Funnel plot to detect publication bias. Odds ratios (OR) are plotted on a logarithmic scale.

6 1784 MALAPERT ET AL Ann Thorac Surg PROLONGED AIR LEAK AFTER LUNG RESECTION 2010;90: Table 4. Meta-Analysis of Adverse Effects of Glue or Patch or Buttress Compared With Control Groups No. of Trials No. of Patients Effect Size Odds Ratio (95% Confidence Interval) I 2,% p Value Atelectasis (0.37, 1.99) Arrhythmias (0.29, 0.8) Hemothorax (0.6, 9.6) Pneumonia (0.44, 1.3) Pneumothorax (0.38, 1.52) Empyema (0.57,8.47) Mortality (0.31, 1.97) cant than studies involving larger cohorts. Because of the risk of publication bias, the findings of our meta-analysis must be interpreted with caution. Our meta-analysis did not reveal heterogeneity between studies for this endpoint. All authors had probably applied the same definition for this endpoint, thus reinforcing its clinical relevance. However, the quality of the studies included in this meta-analysis is intermediate. Indeed, for six of the 13 trials [10, 12, 13, 15, 17, 18], no information on the randomization method was provided. These studies generate a potential selection bias. In only five of the 13 studies [13, 14, 16, 21, 22] were outcomes evaluated without knowledge of the assigned treatment. However, it is unlikely that there is a measurement bias. Indeed, patients with a prolonged air leak for more than 7 days are obliged to keep their chest tube. Consequently, this outcome is objective enough to prevent a systematic error in measurement. The homogeneity between studies argues in favor of this hypothesis. The sensitivity analysis excluding trials with a buttressed staple line, and the surgical treatment of emphysema did not change the pooled effect of the meta-analysis. For most adverse events, no excess risk was reported. Surprisingly, the surgical s and buttress significantly reduced the occurrence of postoperative arrhythmia. The prevalence of arrhythmia after lung surgery [25] is between 8% and 42%. Factors influencing postoperative arrhythmia were age, a history of cardiac disease and the type of pulmonary resection including pneumonectomy [25]. In a randomized controlled trial, these factors are distributed equally in each group and are therefore unlikely to influence the occurrence of arrhythmia. The only factor that may differ in the two treatment groups is a persistent chest tube in patients with prolonged air leak. The chest tube could favor the occurrence of arrhythmia. The effect size of the four methods is in the same direction, as indicated by the absence of significant interaction. This naive indirect comparison does not show whether one method is superior to the others in its ability to reduce prolonged air leak after lung resection. The systematic search of the literature did not identify trials that made a direct comparison between different products. To answer this question, indirect comparisons may help to classify these different methods, depending on the effect sizes. The use of a surgical to prevent prolonged air leak increases costs. But only medicoeconomic studies may show that the cost of these methods are offset by the reduction in the length of hospital stay [26]. In conclusion, this meta-analysis showed that the use of glues, patches, or buttressing reduces the occurrence of prolonged air leak after pulmonary resection. However, publication bias should temper the conclusions of this meta-analysis. Further trials will confirm the effectiveness of these methods on the prevention of postoperative prolonged air leak. References 1. Rice TW, Kirby TJ. Prolonged air leak. Chest Surg Clin North Am 1992;2: 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: Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327: Steichen TJ, Egger M, Sterne J. Test for publication bias in meta-analysis. Stata Tech Bull 1998;44: Baujat B, Mahe C, Pignon JP, Hill C. A graphical method for exploring heterogeneity in meta-analyses: application to a meta-analysis of 65 trials. Stat Med 2002;21: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trial 1986;7: Smith TC, Spiegelhalter DJ, Thomas A. Bayesian approaches to random-effects meta-analysis: a comparative study. Stat Med 1995;14: Sutton AJ, Abrams K. Bayesian methods in meta-analysis and evidence synthesis. Stat Meth Med Res 2001;10: Spiegelhalter DJ, Thomas A, Best NG. WinBUGS version 1.2. User manual. Cambridge, UK: MRC Biostatistics Unit, Hazelrigg S, Boley TM, Naunheim KS, et al. Effect of bovine pericardial strips on air leak after stapled pulmonary resection. Ann Thorac Surg 1997;63: Wong K, Goldstraw P. Effect of fibrin glue in the reduction of postthoracotomy alveolar air leak. Ann Thorac Surg 1997;64: Venuta F, Rendina EA, De Giacomo T, et al. Technique to reduce air leaks after pulmonary lobectomy. Eur J Cardiothorac Surg 1998;13: Stammberger UZ, Klepetko W, Stamatis G, et al. Buttressing the staple line in lung volume reduction surgery: a randomized three-center study. Ann Thorac Surg 2000;70: Wain JC, Kaiser LR, Johnstone JW, et al. Trial of a novel synthetic in preventing air leaks after lung resection. Ann Thorac Surg 2001;71: Porte HL, Jany T, Akkad R, et al. Randomized controlled trial of a synthetic for preventing alveolar air leaks after lobectomy. Ann Thorac Surg 2001;71:

7 Ann Thorac Surg MALAPERT ET AL 2010;90: PROLONGED AIR LEAK AFTER LUNG RESECTION 16. Fabian T, Federico JA, Ponn RB. Fibrin glue in pulmonary resection: a prospective, randomized, blinded study. Ann Thorac Surg 2003;75: Allen MS, Wood DE, Hawkinson RW, et al. Prospective randomized study evaluating a biodegradable polymeric for sealing intraoperative air leaks that occur during pulmonary resection. Ann Thorac Surg 2004;77: Lang G, Csekeo A, Stamatis G, et al. Efficacy and safety of tropical application of human fibrinogen/thrombin-coated collagen patch (TachoComb) for treatment of air leakage after standard lobectomy. Eur J Cardiothorac Surg 2004;25: Gomez-Caro A, Roca Calvo MJ, Lanzas JT, et al. The approach of fused fissures with fissureless technique decreases the incidence of persistent air leak after lobectomy. Eur J Cardiothorac Surg 2007;31: Tansley P, Al-Mulhm F, Lim E, Ladas G, Goldstraw P. A prospective, randomized, controlled trial of the effectiveness of Bioglue in treating alveolar air leaks. J Thorac Cardiovasc Surg 2006;132: Anegg U, Lindenmann J, Matzi V, et al. Efficiency of fleecebound sealing (Tachosil) of air leaks in lung surgery: a prospective randomized trial. Eur J Cardiothorac Surg 2007; 31: Moser C, Opitz I, Zhai W, et al. Autologous fibrin reduces the incidence of prolonged air leak and duration of chest tube drainage after lung volume reduction: a prospective randomized blinded study. J Thorac Cardiovasc Surg 2008;136: Serra-Mitjans M, Belda-Sanchis J. Surgical for preventing air leaks after pulmonary resections in patients with lung cancer. Cochrane Database System Rev 2005;3:CDOO Sterne JAC, Egger M. Funnel plot for detecting bias in meta-analysis: guidelines on choice of axis. J Clin Epidemiol 2001;54: Roselli EE, Murphy SC, Rice TW, et al. Atrial fibrillation complicating lung cancer resection. J Thorac Cardiovasc Surg 2005;130: Varela G, Marcelo FJ, Nuria N, Aranda JL. Estimating hospital costs attributable to prolonged air leak in pulmonary lobectomy. Eur J Cardiothorac Surg 2005;27: INVITED COMMENTARY This meta-analysis by Malapert and colleagues [1] evaluates the impact of s and buttresses on the incidence of prolonged air leak (AL) after pulmonary resection. It represents a novel and substantial contribution to the literature surrounding the controversial use of intraoperative adjuncts to reduce postoperative AL. I believe the authors have reasoned correctly that perhaps it is only the prolonged ALs ( 7 days by their definition) that have a significant economic impact and are closely associated with the development of difficult complications, such as empyema and pneumonia. Despite this reasoning, most of the randomized studies evaluating s and buttresses for AL have used mean duration of air leak as their primary outcome measure. These studies have generally not shown clinically significant differences, although a few have shown statistically significant differences in favor of s on this outcome measure. As a result, many physicians (including myself) have challenged whether costly s should be widely used, particularly outside of patients who are at high risk for the development of problematic ALs. By focusing on prolonged AL, the authors bring attention to the outcome that may really matter, and they find that on this outcome measure the performance of s and buttresses seems to be more impressive. In the 13 trials analyzed that had reported data on the incidence of prolonged AL, and buttress studies evaluated together had a pooled effect size of 0.55 (95% confidence interval, 0.39 to 0.79) on this outcome. Results were similar when only the seven studies evaluating s (glues) were evaluated, as well as when the three studies that focused on severe emphysema patients were excluded. Another novel finding is that the s and buttresses seemed to reduce the incidence of atrial fibrillation (pooled effect size, 0.48), although not the incidences of other complications. The authors are appropriately circumspect regarding their conclusions. They point out that there are a variety of flaws in many of the studies that nevertheless they decided to include in their analysis; only 7 of 13 studies had adequate concealment of allocation, only 5 of 13 had blinded outcomes assessors, and only 7 of 13 used intentto-treat. They demonstrate elegantly by funnel analysis that publication bias is likely present. I would agree with their assessment that the conclusions should be interpreted with caution. Future study should focus on the use of s in subgroups of patients at high risk for problematic AL (eg, those with emphysema and a forced expiratory volume in 1 second below some cut-off, perhaps 60% predicted). In addition, clearly, prolonged AL should be an important outcome measure. However, in the end, only the analysis of cost-effectiveness will allow the determination of whether a reduction in the incidence of prolonged AL by approximately 50% will be sufficient to justify the substantial cost of using commercial buttress and products that are usually quite expensive. I suspect we will ultimately find that it is appropriate to use these products in those patients at the highest risk for prolonged or complicated AL, but the routine use of these products in all comers undergoing pulmonary resection is not indicated. Joseph B. Shrager, MD Division of Thoracic Surgery Stanford University School of Medicine Falk Building, 2nd Floor 300 Pasteur Dr Stanford, CA shrager@stanford.edu Reference 1. Malapert G, Hanna HA, Pages PB, Bernard A. Surgical for the prevention of prolonged air leak after lung resection: meta-analysis. Ann Thorac Surg 2010;90: by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur