AAAMJDS

qpcido

wcsstrabssbp

PptcipstctcPtp

A

PS

AOb

©P

GEN

ERA

LT

HO

RA

CIC

GENERAL THORACIC SURGERY:The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org.To take the CME activity related to this article, you must have either an STS member or anindividual non-member subscription to the journal.

Scoring System to Predict the Risk of Prolongedir Leak After Lobectomy

lessandro Brunelli, MD, Gonzalo Varela, MD, PhD, Majed Refai, MD,arcelo F. Jimenez, MD, Cecilia Pompili, MD, Armando Sabbatini, MD, and

ose Luis Aranda, MDivision of Thoracic Surgery, Ospedali Riuniti, Ancona, Italy; and Division of Thoracic Surgery, Salamanca University Hospital,

alamanca, Spain

md

1wpvmgst

ptpeio

Background. Prolonged air leak (PAL) remains a fre-uent complication after lung resection. Perioperativereventative strategies have been tested, but their effi-acy is often difficult to interpret due to heterogeneousnclusion criteria. The objective of this study was toevelop and validate a practical score to stratify the riskf PAL after lobectomy.Methods. Six hundred fifty-eight consecutive patientsere submitted to pulmonary lobectomy (2000 to 2008) in

enter A and were used to develop the risk-adjustedcore predicting the incidence of PAL (> 5 days). Exclu-ion criteria were chest wall resection and postopera-ive assisted mechanical ventilation. No sealants, pleu-al tent, or buttressing material were used. To build theggregate score numeric variables were categorizedy receiver operating curve analysis. Variables werecreened by univariate analysis and then used intepwise logistic regression analysis (validated byootstrap). The scoring system was developed by pro-

ortional weighing of the significant predictor esti-

ao

P

Tpbg

slswacteote

spedali Riuniti, Via Conca 1, Ancona, IT60020, Italy; e-mail:runellialex@gmail.com.

2010 by The Society of Thoracic Surgeonsublished by Elsevier Inc

ates and was validated on patients operated on in aifferent center (center B).Results. The incidence of PAL in the derivation set was

3% (87 of 658 cases). Predictive variables and their scoresere the following: age greater than 65 years (1 point);resence of pleural adhesions (1 point); forced expiratoryolume in one second less than 80% (1.5 points); and bodyass index less than 25.5 kg/m2 (2 points). Patients were

rouped into 4 risk classes according to their aggregatecores, which were significantly associated with incremen-al risk of PAL in the validation set of 233 patients.

Conclusions. The developed scoring system reliablyredicts incremental risk of PAL after pulmonary lobec-

omy. Its use may help in identifying those high-riskatients in whom to adopt intraoperative prophylactic strat-gies; in developing inclusion criteria for future random-zed clinical trials on new technologies aimed at reducingr preventing air leak; and for patient counseling.

(Ann Thorac Surg 2010;90:204–9)

© 2010 by The Society of Thoracic Surgeons

rolonged air leak (PAL) remains a frequent compli-cation after pulmonary resection, occurring in ap-

roximately 10 to 15% of patients after pulmonary lobec-omy [1, 2]. It may prolong hospital stay, impacting onosts [3–6], and increase the risk of other complications,ncluding empyema [5–7]. For these reasons, severalreventative strategies, including surgical techniques,ealants, or buttressing materials, have been tested inhe clinical investigations. However, their reportedost efficacy has been often difficult to interpret owingo heterogeneous inclusion criteria. A commonly ac-epted system that may assist in stratifying the risk ofAL and standardize the selection of patients for these

rials or for using prophylactic measures in the clinicalractice is lacking.

ccepted for publication Feb 3, 2010.

resented at the Forty-sixth Annual Meeting of The Society of Thoracicurgeons, Fort Lauderdale, FL, Jan 25–27, 2010.

ddress correspondence to Dr Brunelli, Division of Thoracic Surgery,

The objective of this study was to develop a practicalnd user-friendly aggregate risk score to stratify the riskf PAL after pulmonary lobectomy.

atients and Methods

his is an observational multicenter analysis performed onrospective electronic databases. The study was approvedy the local Institutional Review Boards and all patientsave their consent to use their data in the dataset.All consecutive pulmonary lobectomies operated on

ince January 2000 through April 2008 in center A forung cancer were used as a derivation set to develop thecoring system predicting the risk of PAL. The risk scoreas then validated on a sample of patients operated on in

nother center (center B) from 2006 to 2008. In bothenters, major lung resections were contraindicated inhose patients with a predicted postoperative forcedxpiratory volume in one second and a predicted post-perative carbon monoxide lung diffusion capacity lesshan 30% of predicted, in addition to an insufficient

xercise tolerance (height reached at stair climbing test

0003-4975/$36.00doi:10.1016/j.athoracsur.2010.02.054

lt

watbtoswppfipptaosrImate(dnaesmpebaT(mnpfpp

es(apae

STwsps

5fahostlbtaeoptmmw(m

nd(uddosInstpbfpsoob1sdpeepa

opttooe

s

205Ann Thorac Surg BRUNELLI ET AL2010;90:204–9 SCORING SYSTEM FOR RISK OF PAL AFTER LOBECTOMY

GEN

ERA

LT

HO

RA

CIC

ess than 12 m or maximum oxygen consumption lesshan 10 mL/kg/min) [8].

Patients undergoing lung resections including chestall or diaphragm resection, or those needing postoper-

tive mechanical ventilation at any time after the opera-ion were not included in this series. Mortality rate inoth centers was 1% and these cases were not used for

he analysis. All patients in both centers were operatedn by qualified thoracic surgeons through a muscle-paring anterolateral thoracotomy. Mechanical staplersere used to develop incomplete fissures in 80% ofatients and to close the bronchus in all cases. Twentyercent of patients had completely developed or filmyssures that did not require the use of staplers. Noleural tents, sealants, buttressing material, or pneumo-eritoneum were applied in any of these patients. Sys-

ematic lymphadenectomy was performed in all casesfter the pulmonary resection [9]. At completion of theperation the presence of an air leak was tested byubmerging the lung parenchyma in sterile saline andeinflating the lung up to a pressure of 25 to 30 cm H2O.f any significant air leak was detected, any attempt was

ade to reduce it by applying sutures. One (center After July 2007 and center B in all cases) or two chestubes (center A before July 2007) were positioned at thend of the operation. Chest tubes were left on suction�20 cm H2O) until the morning of the first postoperativeay and then either alternate suction (suction duringight and no suction during day 10) or no suction waspplied according to institutional policies. Patients werextubated in the operating room and admitted to apecialized dedicated thoracic ward. Perioperative treat-ent was standardized and focused on the control of

ostthoracotomy chest pain, chest physiotherapy, andarly as possible mobilization, antibiotic and antithrom-otic prophylaxis. Postoperative chest pain was assessedt least twice a day during morning and evening rounds.reatment was titrated to achieve a pain score below 4

range 0 to 10) during the first 72 postoperative hours byeans of epidurals or continuous intravenous infusion of

onopioid analgesics. Physical rehabilitation and chesthysiotherapy were performed in all patients starting

rom postoperative day 1 by qualified and dedicatedhysiotherapists according to standardized protocols. Noositive pressure ventilation was used.Air leak was assessed twice daily (during morning and

vening rounds). Patients were instructed to performtandardized repeated forced expiratory maneuverscoughing and blowing). Chest tubes were removed if noir leak was detectable in the chest drain unit and theleural effusion was less than 400 mL in the last 24 hours,fter a chest X-ray was obtained to show satisfactory lungxpansion.

tatistical Analysishe derivation set consisted of 658 consecutive patientsho underwent pulmonary lobectomy in center A. This

ample was used to develop the risk-adjusted scoreredicting the incidence of PAL. For the purpose of this

tudy, PAL was defined as an air leak lasting longer than t

days. Initially the following series of perioperativeactors were screened by univariate analysis for possiblessociation with PAL: age, gender, serum albumin level,emoglobin level, forced expiratory volume in one sec-nd (FEV1%), ratio of forced expiratory volume in oneecond to total lung capacity, ratio of residual volume tootal lung capacity, body mass index, diffusing capacity ofung for carbon monoxide %, smoking pack-years, dia-etes, preoperative systemic steroids, induction chemo-

herapy, side and site of lobectomy, presence of pleuraldhesions (defined as dense adhesions occupying anntire lobe or at least 30% of the lung surface), and lengthf stapled parenchyma. All data were at least 95% com-lete. Sporadic missing data were imputed by averaging

he nonmissing values (numeric variables) or taking theost frequent category (categoric variables). To avoidulticollinearity, only one variable in a set of variablesith a correlation coefficient greater than 0.5 was selected

by bootstrap procedure), and used in the regressionodel.For the purpose of this analysis, significant (p � 0.05)

umeric variables were tested for a threshold effect andichotomized by using receiver operating curve analysis

for identifying the best cutoff). Significant variables atnivariate analysis were then used as independent pre-ictors in a stepwise logistic regression analysis (depen-ent variable: presence of PAL � 5 days). The reliabilityf the predictors was finally assessed by using a boot-trap resampling technique with 1,000 samples [10–12].n the bootstrap procedure, repeated samples of the sameumber of observations as the original database wereelected with replacement from the original set observa-ions. For each sample, stepwise logistic regression waserformed. The stability of the final stepwise model cane assessed by identifying the variables that enter most

requently in the repeated bootstrap models and com-aring those variables with the variables in the finaltepwise model. If the final stepwise model variablesccur in a majority (�50%) of the bootstrap models, theriginal final stepwise regression model can be judged toe stable. Only reliable (bootstrap frequency � 50% in,000 simulated samples) predictors were used to con-truct the final aggregate score. The scoring system waseveloped by proportional weighing of the significantredictors estimates, assigning a value of 1 to the small-st coefficient. An aggregate risk score was generated forach patient by summing each estimate [13]. Finally,atients were grouped in classes of incremental riskccording to their total score.The risk score was then validated on the patients

perated on in center B (external validation set of 233atients) and the risk of PAL was verified in each class in

his external population. Moreover, to further validatehe score, the population of center B was bootstrapped tobtain 1,000 simulated external samples. The proportionf patients with PAL was then verified for each class inach of these bootstrapped samples.Statistical analysis was performed on the Stata 9.0

tatistical software (StataCorp, College Station, TX). All

ests were 2-tailed with a statistical significance of 0.05.

R

TvP6

cgpcqclbtbitFktopapacv((

s

9rc(P

C

TaaildET

cotcmcohplilriimwests

tgrDs(p

TS

V

ABFP

TV

V

AMBFFDRAPDISSSP

Rs

BmFT

206 BRUNELLI ET AL Ann Thorac SurgSCORING SYSTEM FOR RISK OF PAL AFTER LOBECTOMY 2010;90:204–9

GEN

ERA

LT

HO

RA

CIC

esults

he characteristics of patients in the derivation andalidation sets are illustrated in Table 1. The incidence ofAL in the derivation and validation sets were 13% (87 of58 cases) and 14% (32 of 233 cases), respectively.After receiver operating curve analysis was used to

ategorize the numeric variables, stepwise logistic re-ression identified the following significant and reliableredictors of PAL: age greater than 65 years (p � 0.04,oefficient 0.558, standard error [SE] 0.27, bootstrap fre-uency 57%), presence of pleural adhesions (p � 0.01,oefficient 0.616, SE 0.25, bootstrap frequency 68%), FEV1

ess than 80% (p � 0.001, coefficient 0.795, SE 0.24,ootstrap frequency 88%), body mass index (BMI) less

han 25.5 kg/m2 (p � 0.0001, coefficient 1.03, SE 0.25,ootstrap frequency 98%). Based on their coefficients, the

ndividual factor scores were the following: age greaterhan 65 years, 1 point; presence of adhesions, 1 point;EV1 less than 80%, 1.5 points; and BMI less than 25.5g/m2, 2 points (Table 2). To obtain a cumulative scorehe individual points were summed in each patient tobtain a range from 0 to 5.5. For example, a 70-year-oldatient, with a FEV1 of 60% predicted, a BMI of 23 kg/m2,nd with pleural adhesions would have a score of 5.5oints. Patients were then grouped into four risk classesccording to their aggregate scores, which were signifi-antly associated with incremental risk of PAL in thealidation set of 233 patients (Table 3). Patients in class Ascore � 0) had no PAL, whereas patients in class Dscore � 3) had 26% risk of developing PAL (p � 0.003).

When the risk classes were assessed in 1,000 boot-

able 1. Characteristics of Patients in the Derivation andalidation Sets

ariablesDerivation Set

(n � 658)Validation Set

(n � 233)

ge (years) 66.9 (10.2) 63.2 (12)ale gender (n,%) 514 (78%) 176 (79%)

MI m/kg2 26.2 (4) 26.2 (4.4)EV1% 87.3% (18.5) 87.3 (24)EV1/FVC ratio 0.7 (0.1) 0.74 (0.1)

lco% 79.5% (19.5) 81.9% (18.7)V/TLC 0.43 (0.08) 0.41 (0.1)lbumin level (g/dL) 4.2 (2.2) 4.1 (2.5)ack-years 42.8 (32.4) 45.5 (35)iabetes (n,%) 70 (11%) 34 (15%)

nduction chemotherapy (n,%) 63 (9.6%) 25 (11%)ystemic steroids (n,%) 26 (3.9%) 7 (3.1%)ide of resection (right, n,%) 368 (56%) 132 (56%)ite of resection (upper, n,%) 409 (62%) 139 (60%)leural adhesions (n,%) 176 (27%) 96 (41%)

esults are expressed as means � standard deviations unless otherwisepecified.

MI � body mass index; Dlco � diffusing capacity of lung for carbononoxide; FEV1 � forced expiratory volume in one second; FEV1/

VC � forced expiratory volume to forced volume capacity ratio; RV/LC � ratio of residual volume to total lung capacity.

trapped samples from center B, we found that in class ABs

8% of samples had a PAL risk less than 5%. Class B had aisk less than 10% in 99% of samples. On the other hand,lass C had a risk greater than 10% in 91% of samplesalthough in no cases greater than 20%) and class D had aAL risk greater than 20% in 99% of samples.

omment

he objective of this study was to develop and validate anggregate score to stratify the risk of prolonged air leakfter pulmonary lobectomy. For the purpose of thisnvestigation, an air leak was defined as prolonged if itasted more than 5 days in line with recent recommen-ations [14] and with the current definition in bothuropean Society of Thoracic Surgeons and Society ofhoracic Surgeons thoracic databases.Prolonged air leak remains a frequent and bothering

omplication after lung resection. It has been reported toccur in approximately 10 to 15% of patients after lobec-omy [1, 2]. Besides obliging the patients to live with ahest tube in place causing distress, anxiety, and pain, itay increase the risk of other cardiopulmonary compli-

ations [5, 7] and empyema [6]. It has been shown to bene of the most important determinants of a prolongedospital stay and increased hospital costs [3–6]. Severalapers have tried to identify risk factors for prolonged air

eak after lung resection [14], but to our knowledge theres no user-friendly risk model or score available in theiterature that could be used in the clinical setting toapidly stratify the risk of PAL. Although previous stud-es have shown that certain factors are associated withncreased risk of PAL, there is the need to integrate the

ost reliable factors in a single scoring system. Thisould greatly simplify the selection of patients for future

fficacy studies on preventative interventions (like use ofealants or buttressed staple lines), making the interpre-ation of results across different investigations more con-istent and meaningful.

We chose to limit the analysis to lobectomies only, forhe sake of homogeneity and because lobectomies carry areater risk of PAL compared with minor resections aseported in the European Society of Thoracic Surgeonsatabase 2009 Annual Report (10% vs 4%). We found that

ignificant and reliable predictors of PAL were older age�65 years), reduced pulmonary function (FEV1 � 80%),resence of pleural adhesions, and low BMI (�25.5 kg/

able 2. Points Assigned at Individual Variables and to beummed to Derive the Aggregate PAL Risk Score

ariablesPoints Assigned at

Individual Variables

ge � 65 1MI � 25.5 m/kg2 2EV1 � 80% 1.5resence of pleural adhesions atoperation

1

MI � body mass index; FEV1 � forced expiratory volume in oneecond; PAL � prolonged air leak.

mshhrtfpcrtana

pamfimccChbstTtpenpttitpboppdmtw

upbTsw

taAhoc

vatttep

R

T

P

CCCC�

C

C � pr

207Ann Thorac Surg BRUNELLI ET AL2010;90:204–9 SCORING SYSTEM FOR RISK OF PAL AFTER LOBECTOMY

GEN

ERA

LT

HO

RA

CIC

2). Although older age has never been reported to be aignificant risk factor for PAL [14], elderly patients mayave a more fragile lung parenchyma with a reducedealing capacity, which may predispose to the occur-ence of this complication. A reduced pulmonary func-ion has been reported as one of the most consistent riskactors for PAL [2, 3, 15–17]. Presence of importantleural adhesions has been previously found to be asso-iated with this complication [2]. Tears in the lung pa-enchyma may ensue during mobilization of the lung andaking down of the adhesions. Finally, a low BMI may be

marker of a poor nutritional status, which in turn mayegatively influence the healing of the surgically dam-ged tissue.We chose to construct an aggregate risk score to

rovide a practical tool for stratifying the risk for clinicalnd scientific purposes following previously reportedethods [13]. In this way, four risk classes were identi-

ed. It readily appears that patients in class A had ainimal risk of PAL. Certainly these patients are not

andidates for any preventative interventions or for effi-acy trials testing new prophylactic devices or materials.onversely, patients in class D were the ones to have theighest risk of PAL. These are the patients who couldenefit most from the application and use of materialsuch as sealants or buttressed staples or preventativeechniques such as pleural tents or pneumoperitoneum.rials testing the cost efficacy of these measures or

reatment should be ideally performed on these high-riskatients in order to obtain the most meaningful infer-nces. If they will show that such treatments or tech-iques will be able to reduce the risk of PAL in this set ofatients then their systematic use could be warranted in

he clinical practice. A risk stratification system such ashis may therefore have a positive clinical and economicmpact, avoiding unnecessary costly treatments in pa-ients who most likely would not need them. This isarticularly true for those devices, such as sealants oruttressed staple lines, which may significantly impactn surgical costs, or for those surgical techniques, such asleural tent, which may increase the risk of other com-lications (bleeding). The risk score may further assisturing preoperative counseling identifying the patientsore at risk to develop PAL and permitting to instruct

hem in advance on the possibility to be discharged home

able 3. Distribution of Patients and Incidence of PAL in the

AL Risk Class

Derivation Set (658 Pati

No. Patients (%) PA

lass A (score 0) 68 (10%)lass B (score 1) 136 (21%)lass C (score 1.5–3) 313 (48%)lass D (score � 3) 141 (21%)2, p value �0.0001-index 0.71 (95% CL 0.65 and

-index � receiver operating curve; CL � confidence limits; PAL

ith a portable chest drainage unit.

The scoring system was validated in an external pop-lation showing a good discrimination across differentopulations. It was further validated in 1,000 externalootstrapped samples drawn from the validations set.his new statistical approach allowed us to assess thetability of the score across multiple external populationsarranting its generalization.Bootstrap has been shown to be superior to the tradi-

ional training and test splitting of the population tossess reliability of predictors and risk models [12].lthough the model has been developed in a sizeable,omogeneous population and further validated in an-ther external sample of patients operated on in anotherenter, this study may have potential limitations.

● The retrospective and multicenter nature of thestudy may have implied some problems of defi-nition and recording of variables and outcomes.

● Patients had partly different chest tube manage-ments at the two centers (although most of thetime tubes were left without suction in both cen-ters) that could have potentially influenced dura-tion of air leak [18, 19–22]. Nevertheless, the scoreappears to stratify the risk of PAL in both centersirrespective of the chest tube management.

● Air leak was assessed by several and differentoperators in two centers (staff surgeons on duty).Interobserver variability has been reported tooccur in air leak assessment [23], but we think thisis a general phenomenon which reflects the reallife situation in the clinical ward and minimallyaffected, if any, the results of this study.

In conclusion, we were able to develop and externallyalidate an easy-to-use risk score for prediction of PALfter pulmonary lobectomy. The primary application ofhis prediction score is for the targeted use of prophylac-ic therapy and interventions and for appropriate selec-ion criteria in efficacy trials, thereby minimizing thexpense and risk of such interventions or trials in thoseatients unlikely to derive benefit.

eferences

1. Cerfolio RJ, Bass CS, Pask AH, Katholi CR. Predictors and

vation and Validation Sets by Class of Risk

Validation Set (233 Patients)

cidence No. Patients (%) PAL Incidence

4% 23 (10%) 030 (13%) 6.7%

5% 110 (47%) 10.9%70 (30%) 25.7%

0.0030.69 (95% CL 0.59 and 0.77)

olonged air leak.

Deri

ents)

L In

1.5%

12.29%

0.77)

treatment of persistent air leaks. Ann Thorac Surg 2002;73:1727–31.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

D

Dtmp

Dn

Dby

Dpwrtteos

DuaTmayh

208 BRUNELLI ET AL Ann Thorac SurgSCORING SYSTEM FOR RISK OF PAL AFTER LOBECTOMY 2010;90:204–9

GEN

ERA

LT

HO

RA

CIC

2. Brunelli A, Monteverde M, Borri A, Salati M, Marasco R,Fianchini A. Predictors of prolonged air leak after pulmo-nary lobectomy. Ann Thorac Surg 2004;77:1205–10.

3. Bardell T, Petsikas D. What keeps postpulmonary resectionpatients in hospital? Can Respir J 2003;10:86–9.

4. Irshad K, Feldman LS, Chu VF, et al. Causes of increasedlength of hospitalization on a general thoracic surgery service:a prospective observational study. Can J Surg 2002;45:264–8.

5. Varela G, Jiménez MF, Novoa N, Aranda JL. Estimatinghospital costs attributable to prolonged air leak in pulmo-nary lobectomy. Eur J Cardiothorac Surg 2005;27:329–33.

6. Brunelli A, Xiumé F, Al Refai M, et al. Air leaks after lobectomyincrease the risk of empyema but not of cardiopulmonarycomplications: a case-matched analysis. Chest 2006;130:1150–6.

7. Okereke I, Murthy SC, Alster JM, Blackstone EH, Rice TW.Characterization and importance of air leak after lobectomy.Ann Thorac Surg 2005;79:1167–73.

8. Brunelli A, Charloux A, Bolliger CT. ERS/ESTS clinical guide-lines on fitness for radical therapy in lung cancer patients(surgery and chemo-radiotherapy). Eur Respir J 2009;34:17–41.

9. Lardinois D, De Leyn P, Van Schil P, et al. ESTS guidelinesfor intraoperative lymph node staging in non-small cell lungcancer. Eur J Cardiothorac Surg 2006;30:787–92.

0. Blackstone EH. Breaking down barriers: helpful break-through statistical methods you need to understand better.J Thorac Cardiovasc Surg 2001;122:430–9.

1. Grunkemeier GL, Wu YX. Bootstrap resampling method:something for nothing? Ann Thorac Surg 2004;1142–4.

2. Brunelli A, Rocco G. Internal validation of risk models inlung resection surgery: bootstrap versus training and testsampling. J Thorac Cardiovasc Surg 2006;131:1243–7.

3. Passman RS, Gingold DS, Amar D, et al. Prediction rule foratrial fibrillation after major noncardiac thoracic surgery.

Ann Thorac Surg 2005;79:1698–703.

ave done in this line is you always give us the derivation set

asfitsaww

DErrrn

DmmtrWmm

DBwaW

4. Singhal S, Ferraris VA, Bridges CR, et al. Management ofalveolar air leaks after pulmonary resection. Ann ThoracSurg 2010;89:1327–35.

5. Linden PA, Bueno R, Colson YL, et al. Lung resection inpatients with preoperative FEV1 � 35% predicted. Chest2005;127:1984–90.

6. Stolz AJ, Schützner J, Lischke R, Simonek J, Pafko P. Predic-tors of prolonged air leak following pulmonary lobectomy.Eur J Cardiothorac Surg 2005;27:334–6.

7. Abolhoda A, Liu D, Brooks A, Burt M. Prolonged air leakfollowing radical upper lobectomy: an analysis of incidenceand possible risk factors. Chest 1998;113:1507–10.

8. Brunelli A, Sabbatini A, Xiumé F, Refai MA, Salati M,Marasco R. Alternate suction reduces prolonged air leakafter pulmonary lobectomy: a randomized comparison ver-sus water seal. Ann Thorac Surg 2005;80:1052–5.

9. Cerfolio RJ, Bass C, Katholi CR. Prospective randomized trialcompares suction versus water seal for air leaks. Ann ThoracSurg 2001;71:1613–7.

0. Marshall MB, Deeb ME, Bleier JI, et al. Suction vs water sealafter pulmonary resection: a randomized prospective study.Chest 2002;121:831–5.

1. Alphonso N, Tan C, Utley M, et al. A prospective random-ized controlled trial of suction versus non-suction to theunder-water seal drains following lung resection. Eur J Car-diothorac Surg 2005;27:391–4.

2. Brunelli A, Monteverde M, Borri A, et al. Comparison ofwater seal and suction after pulmonary lobectomy: a pro-spective, randomized trial. Ann Thorac Surg 2004;77:1932–7.

3. Varela G, Jiménez MF, Novoa NM, Aranda JL. Postoperativechest tube management: measuring air leak using an elec-tronic device decreases variability in the clinical practice.

Eur J Cardiothorac Surg 2009;35:28–31.

ISCUSSION

R DAVID HARPOLE, JR (Durham, North Carolina): I noticedhat as a prerequisite to the trial you used sort of no provocative

easures for spaces or leaks; you didn’t do pleural tents orneumoperitoneum and so forth.

R BRUNELLI: No. Pleural tents or pneumoperitoneum wereot used in this series.

R HARPOLE: So in your center and in this trial, for ailobectomy there was clearly a space afterwards. What wasour usual mechanism for taking care of those?

R BRUNELLI: Although we previously published on theleural tent we now use it sporadically only in some patientshere a large residual pleural space is anticipated, but not

outinely anymore. In this regard the cohort of patients used forhe analysis was homogeneous and we really had the chance toest the influence of other factors on the duration of air leakliminating the confounder of prophylactic measures. Since thebjective was indeed to find a classification system to helpelecting those patients who could benefit from their use.

R BRYAN F. MEYERS (St. Louis, MO): I just rise to congrat-late you on your paper. I think that usually in papers like this,t meetings like this at the AATS (American Association forhoracic Surgery), people will analyze their data, create aultivariable model, give us the result, and then expect us to

ccept that those data will extrapolate to other datasets. But whatou have done here and what you generally do in work that you

nd then have another set of data to validate it all within theame paper, and I think that really adds to the reliability of yourndings, so I commend you on that. The one thing I see is that

he overall numbers that you used to build the models were a bitmaller than you have used in the past for other similar modelsnd I just wondered if you ran into any challenges in doing thisork with the smaller numbers of patients that you had to dealith.

R BRUNELLI: We feel quite confident in using bootstrap.specially when you have rare events or small numbers, weeally feel that bootstrap adds to the analysis. We stronglyecommend the use of bootstrap to assess model and factorseliability after logistic regression even in the case of smallumbers, and particularly if you have rare events.

R TODD L. DEMMY (Buffalo, NY): I’m curious regarding yourodel as far as the pathophysiology, particularly the low bodyass index having a greater risk of air leak. One could look at

hese risk factors and wonder if it’s a compliance problem, aelative disproportion of chest space and remaining lung to fill it.

hat do you think is the primary mechanism and is thereaybe a more precise way to measure it, like lung compliance oraximum negative intrapleural pressure, at the end of the case?

R BRUNELLI: It is even more surprising when you think thatMI was the strongest, the most influential factor associatedith PAL, and this was the first time we found this, and I’m not

ware in the literature of any study reporting on this association.

e think a low BMI may be associated with perhaps a low

nTiBtc

Dri

D

D

D

Dqfa

Dautp

209Ann Thorac Surg BRUNELLI ET AL2010;90:204–9 SCORING SYSTEM FOR RISK OF PAL AFTER LOBECTOMY

utritional status, hindering tissue healing and regeneration.his is our interpretation, but we cannot exclude that a low BMI

s a surrogate measure reflecting or influencing chest mechanics.ut I agree this should be analyzed further and using other

echnologies. With the data we have in our hands now, weannot speculate too much on this.

R SCOTT J. SWANSON (Boston, Massachusetts): It’s kind of aare finding in the U.S. to have somebody with a low body massndex. What percentage of your patients were below 25?

R BRUNELLI: We had a rather high percentage, around 40%.

R SWANSON: They were below 25? T

R BRUNELLI: Yes.

R JOHN A. ODELL (Jacksonville, FL): I think my planneduestion has briefly been answered. Italians are built differently

rom Americans and I just wonder whether your data wouldpply to people of American build, where obesity is so common.

R BRUNELLI: I don’t know. That’s a difficult question tonswer. Although the model appeared quite generalizable bysing external validation and bootstrap, it would be interesting

o cross-validate the model in the North American population,erhaps using the STS [Society of Thoracic Surgeons] database.

hat could be an interesting analysis.

GEN

ERA

LT

HO

RA

CIC