Although air leaks continue to be one of the most

Transcription

1 ORIGINAL ARTICLES: GENERAL THORACIC Prospective Randomized Trial Compares Suction Versus Water Seal for Air Leaks Robert J. Cerfolio, MD, Cyndi Bass, MSN, CRNP, and Charles R. Katholi, PhD Department of Surgery, Division of Cardiothoracic Surgery, and Department of Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama Background. Surgeons treat air leaks differently. Our goal was to evaluate whether it is better to place chest tubes on suction or water seal for stopping air leaks after pulmonary surgery. A second goal was to evaluate a new classification system for air leaks that we developed. Methods. Patients were prospectively randomized before surgery to receive suction or water seal to their chest tubes on postoperative day (POD) #2. Air leaks were described and quantified daily by a classification system and a leak meter. The air-leak meter scored leaks from 1 (least) to 7 (greatest). The group randomized to water seal stayed on water seal unless a pneumothorax developed. Results. On POD #2, 33 of 140 patients had an air leak. Eighteen patients had been preoperatively randomized to water seal and 15 to suction. Air leaks resolved in 12 (67%) of the water seal patients by the morning of POD #3. All 6 patients whose air leak did not stop had a leak that was 4/7 or greater (p < ) on the leak meter. Of the 15 patients randomized to suction, only 1 patient s air leak (7%) resolved by the morning of POD #3. The randomization aspect of the trial was ended and statistical analysis showed water seal was superior (p 0.001). The remaining 14 patients were then placed to water seal and by the morning of POD #4, 13 patients leaks had stopped. Of the 32 total patients placed to seal, 7 (22%) developed a pneumothorax and 6 of these 7 patients had leaks that were 4/7 or greater (p 0.001). Conclusions. Placing chest tubes on water seal seems superior to wall suction for stopping air leaks after pulmonary resection. However, water seal does not stop expiratory leaks that are 4/7 or greater. Pneumothorax may occur when chest tubes are placed on seal with leaks this large. (Ann Thorac Surg 2001;71:1613 7) 2001 by The Society of Thoracic Surgeons Although air leaks continue to be one of the most common complications after elective pulmonary resection, few studies have been performed that study them. Many physicians argue about whether chest tubes should be placed to water seal or suction when a patient has an air leak [1 5]. The majority believes that suction is better. Moreover, there is not even an accepted classification system for air leaks. In 1998 [6], we described a classification system and showed that placing chest tubes on water seal was safe for patients with air leaks after 48 hours of suction. This allowed us to obtain our institutional review board s (IRB) approval to perform this prospective randomized trial to compare suction versus water seal. We measured the size of the air leak with a leak meter. This adjunct was supplemented to our previously described classification system that was qualitative only. Our new classification system is now both qualitative and quantitative. There were two goals to this trial. The first was to determine if wall suction or water seal was superior in stopping air leaks. The second goal was to see if this new classification system was useful in guiding patient care and if it was reproducible between two different observers. Accepted for publication Jan 17, Address reprint requests to Dr Cerfolio, Department of Surgery, Division of Cardiothoracic Surgery, University of Alabama at Birmingham, 1900 University Blvd, THT 712, Birmingham, AL 35294; robert.cerfolio@ccc.uab.edu. Patients and Methods Between August 1, 1997 and June 1, 1998, 140 consecutive patients underwent elective pulmonary resection by a single surgeon (R.J.C.) at The University of Alabama at Birmingham. Exclusion criteria included any patient who had pulmonary resection performed exclusively by video-assisted thoracoscopy, had lung volume reduction surgery, pneumonectomy or any type of broncho-plastic procedure (i.e., sleeve resection). No patients refused entry into this trial secondary due to its safe design. Standard operative techniques were used and lobectomy or segmentectomy was performed for patients with primary bronchogenic carcinoma and, in general, wedge resection was performed for metastasectomy. Incomplete fissures were stapled. Air leaks were pinpointed prior to chest closure by squirting warm, sterile water over the inflated lung. Leaks were sutured with 4-0 Prolene (Ethicon, Somerville, NJ) on a small half needle. The surgeon was blinded to the group to which the patient was randomized until the morning of postoperative day (POD) #2. The 2 groups were matched to ensure that preoperative risk factors for air leaks were controlled. The variables controlled for were type of pulmonary resection, pulmonary function testing, age, use of steroids, preoperative radiation, or previous ipsilateral thoracotomy. An algorithm was devised so water seal could be 2001 by The Society of Thoracic Surgeons /01/$20.00 Published by Elsevier Science Inc PII S (01)

2 1614 CERFOLIO ET AL Ann Thorac Surg SUCTION VERSUS WATER SEAL 2001;71: Fig 1. Prospective randomized algorithm used for all patients in trial. (CT chest tube; CXR chest roentgenogram; POD postoperative day; pts patients.) compared to suction (see Figs 1, 2) and so we could obtain IRB approval. The complex design of this algorithm is explained by three facts: (1) In our previous report, patients were on suction until POD #2. Because we did not show that water seal applied sooner was safe, our IRB required patients to be on suction until the morning of POD #2. Then they were randomized into their 2 groups, based on their preoperative randomization. (2) The members of the IRB had a preconceived notion that a pneumothorax with an air leak was bad. Therefore, we had to check a chest roentgenogram (CXR) when patients were placed to water seal. (3) If a patient who was randomized to water seal developed a new or enlarging pneumothorax, his chest tubes were placed to 10 cm of suction and another CXR was performed. If the new or enlarging pneumothorax was still present, the tubes were then placed to 20 cm of suction. By POD #3, the randomized portion of the trial was ended and statistical analysis was done. However, because our first trial on air leaks implied that water seal might be better at stopping air leaks, we informed patients that if they were randomized to suction and still had an air leak after POD #3, we would place their chest tubes to water seal. These facts are all incorporated in the algorithm. It allowed us to obtain IRB approval, provide safe and expedient care for all patients without potentially prolonging their hospitalization, recruit a consecutive series of patients and yet perform a sound prospective randomized trial comparing suction versus seal. Importantly, statistical analysis comparing water seal to suction was done on POD #3, prior to any patient crossing over to the other group. Air leaks were classified daily by two observers. One was the cardiothoracic fellow who spent 6 months on the general thoracic service and the other was the surgical attending (R.J.C.). Air leaks were qualitatively described by when they occurred in the respiratory cycle. They were labeled as expiratory, inspiratory, continuous, or forced expiratory. If the leak occurred during expiration, it was called expiratory. If it occurred during inspiration, it was called inspiratory. If it occurred during both inspiration and expiration, it was called continuous. If there was no leak during these maneuvers, the patient was asked to cough forcefully several times. If an air leak was present only with cough, it was classified as forced expiratory. It is important to note that any patient with an

3 Ann Thorac Surg CERFOLIO ET AL 2001;71: SUCTION VERSUS WATER SEAL 1615 Fig 2. Number of patients in algorithm. (CXR chest roentgenogram; POD postoperative day; pts patients.) expiratory leak will also have a forced expiratory leak but, in our classification system, it is labeled as an expiratory leak. Air leaks were also quantitatively labeled. A commercially available air-leak meter that comes as part of a pleura vac system was used in all patients (Sahara Pleura-vac, Deknatel, Boston, MA). This meter scores leaks from 1 to 7, with 7 being the highest. The patient takes several deep breaths in and out (or coughs several times) and the bubbles from the leak go into the chambers. The larger the leak, the more chambers it goes into. The largest size leak consistently seen is recorded. In this manner, an air leak is fully scored using our qualitative and quantitative classification system. An example of a fully scored air leak is an expiratory 2 (E-2) or a forced expiratory 3 (FE-3). Patients who still had an air leak on POD #6 received a Heimlich valve and were discharged. The tubes were then removed in the clinic once the leak resolved. If the patient still had an air leak after 2-week follow-up, he was readmitted and provocative chest tube clamping was performed [7]. Prolonged air leaks were defined as an air leak present on POD #6. All data are reported with median and ranges. Bi-variable comparisons were made using a Kruskal-Wallis or chi-square test with a Fisher s exact test. Multivariable analysis was performed using logistic regressions. Results Suction Versus Water Seal The median age at the time of pulmonary resection in these 140 patients (96 men, 44 women) was 54 years (range 3 years to 83 years). Forty-one patients had a previous history of cardiac disease (either coronary artery bypass surgery, balloon angioplasty, or exertional angina), 28 had insulin-dependent diabetes mellitus, 13 had chronic renal failure (on dialysis), 7 had a previous

4 1616 CERFOLIO ET AL Ann Thorac Surg SUCTION VERSUS WATER SEAL 2001;71: ipsilateral thoracotomy, and 6 were on chronic steroids (at least 20 mg of prednisone per day). Twelve patients had previous radiation to the chest and 11 had a history of chemotherapy. The indications for operation were a nodule(s) or mass in 120 patients, destroyed lung from radiation or tuberculosis in 7, empyema in 3, persistent infiltrate with fever in 3, bronchiectasis in 3, and others in 4. Types of pulmonary resections were bilobectomy in 4 patients, lobectomy and wedge resection in 13, lobectomy in 61, segmentectomy in 3, multiple wedge resections in 25, and wedge resection in 34. Figure 2 shows the breakdown of patients as they went through each postoperative day in the planned algorithm. Eighteen patients with air leaks were randomized to receive water seal, and 15 were randomized to receive 20 cm of wall suction. There were no statistically significant differences in variables previously shown to predict air leaks between these 2 groups. As shown in Figure 2, by the morning of POD #3, 12 of the 18 patients (67%) who had been randomized to water seal no longer had an air leak. All 6 patients whose air leak did not stop had a leak that was 4/7 or greater on the leak meter. Both patients with continuous leaks were also in this group. Of the 18 patients initially randomized to water seal, 5 developed a pneumothorax and all had air leaks that were 4/7 or greater on the leak meter. They then went to 10 cm of wall suction and the pneumothorax resolved in 3 of the 5. The other 2 remaining patients needed 20 cm of wall suction to resolve the pneumothorax. Both of these patients had air leaks that were 5/7 on the leak meter. Review of Figure 2 also shows that of the 15 patients randomized to stay on 20 cm of wall suction, only 1 patient s (7%) air leak stopped by the morning of POD #3. Statistical analysis performed shows water seal to be superior to suction for stopping air leaks ( p 0.001). The randomized aspect of the trial was completed. The remaining 14 patients with air leaks who had been initially randomized to suction were then placed on water seal. By the morning of POD #4, 13 of the 14 patients air leaks had stopped. Further analysis of the data shows that a total of 32 patients who had air leaks were eventually placed on seal. In 25 patients (78%), the air leak stopped and all had air leaks that were 3/7 or less on the leak meter. Moreover, 23 of the 25 patients leaks stopped within 24 hours of being placed on seal. The seven patients whose air leaks did not stop on water seal all had air leaks on the leak meter of 4/7 or greater ( p ). Seven of these 32 patients placed to water seal developed a pneumothorax. Six of the 7 patients had an air leak that was 4/7 or greater on the leak meter. Of the 25 patients who did not develop a pneumothorax on seal, all except 1 had an air leak that was 3/7 or less on the leak meter ( p 0.001). Six of the 7 patients with an air leak on POD #6 had a large leak on POD #1 (5/7 or greater on the leak meter). All patients whose leaks stopped prior to POD #6 had air leaks on POD #1 that were less than 5/7. Therefore, having an air leak of 5/7 or greater on POD #1 predicted a prolonged air leak ( p ). Statistical analysis was performed in an attempt to identify predictors of air leaks. The only variable found to be a statistically significant predictor of having an air leak on POD #1 was having had a lobectomy ( p 0.006). Classification System for Air Leaks Two different observers who measured and graded the air leaks disagreed on the size of the air leak six times out of the 35 patients (17%). However, they never disagreed by more than one on the air-leak meter. There was only one disagreement between the two observers on the type of the leak. This occurred when one observer called a leak an expiratory 1 and the other called it a forced expiratory 1. Median day of discharge was POD #4 range (3 to 20 days). Postoperative complications occurred in 21 patients (15%). These included prolonged air leaks in 7, atrial fibrillation in 5, pneumonia in 3, reintubation in 2, retained secretions requiring bronchoscopy in 2, right ventricular infarct in 1, neuropraxia to a left recurrent laryngeal nerve in 1, stroke in 1, and a chylothorax in 1. Operative mortality was 1.4% (2 patients). One patient aspirated on POD #5 and eventually died on POD #20 from pneumonia and multisystem failure. Another patient suffered a pulmonary embolism on POD #4 and eventually died on POD #12. Comment The decision to place chest tubes on suction versus water seal for many surgeons is based more on when, where, and how he or she trained as opposed to any scientific data. Because of this fact, we have performed several studies on air leaks and developed a classification system. This is an attempt to bring science to this common problem. Our previous trial [1] showed that water seal was safe for patients with air leaks. It enabled us to perform this prospective randomized trial with both IRB and patient approval. This study compared 2 matched groups in a prospective randomized fashion and found a statistically significant advantage by POD #3 favoring the group randomized to water seal. The study was actually stopped early because of the clinical and statistical benefit shown for those patients randomized to water seal. Water seal seems superior to wall suction for stopping expiratory air leaks after pulmonary resection. Lending even more credence to this conclusion is the fact that by POD #4, 13 of the 14 patients who had been on suction for 3 days who still had an air leak, had resolution of their leak after just 1 day of water seal. However, water seal may not be safe for everyone. It did cause a pneumothorax in patients with large air leaks (greater than 4/7 on the leak meter). Whether a pneumothorax and an air leak together truly represent a clinical problem was not answered by this trial (because of its design). However, 4 of these 7 patients who developed a pneumothorax on water seal did start to develop subcutaneous emphysema.

5 Ann Thorac Surg CERFOLIO ET AL 2001;71: SUCTION VERSUS WATER SEAL 1617 The second goal of this trial was to evaluate our new classification system for air leaks and to see if it was clinically useful in guiding patient care and was reproducible. The qualitative aspect of the classification system is highly reproducible between different observers. However, the quantitative aspect of our system that relies on a leak meter still has some minor interpretation. The observers disagreed on the size of the leak 17% of the time, but the disagreements were never more than 1 or 1 on the leak meter. The real questions are, did this small level of disagreement make a difference in patient care, and is the system useful to the surgeon at the bedside? We believe the classification system is extremely useful. Because of it, we can now predict which leaks on POD #1 are going to continue until POD #6 regardless of what the physician does with the chest tubes. Air leaks that are 5/7 or greater on the leak meter on POD #1 are likely to continue. Armed with this information, the physician can prepare the patient with the news that he or she will probably need to go home with a Heimlich valve. The patient can be discharged from the hospital on the typical POD #3 or 4 regimen, rather than waiting 3 to 4 extra days in the hospital to see if the air leak will seal. This has the potential for significant cost savings. The leak meter also predicted what size leaks would not stop on seal and which patients would get a pneumothorax or subcutaneous emphysema on seal. This may allow the surgeon to know when to avoid water seal. The qualitative aspect of our classification system was also clinically important. Only 2 patients had a continuous air leak, and both had prolonged air leaks. Both leaks were large as well. Air leaks that are forced expiratory are less severe than those that are expiratory, regardless of their size. Oftentimes, a leak will change from an expiratory leak to a forced expiratory leak, and this change suggests that the leak is sealing. These facts help direct management and save hospital days. In conclusion, we have shown in this study of 33 patients with air leaks after elective pulmonary resection that water seal seems statistically superior to suction for stopping air leaks. However, large leaks (4/7 or greater on the leak meter) or inspiratory leaks will not stop with water seal, and pneumothorax may occur. We also found that most leaks are expiratory or forced expiratory. Air leaks that are continuous, or 5/7 on the day after surgery, are unlikely to stop by POD #6 despite chest tube settings. Our newly designed classification system for air leaks is both qualitative and quantitative. It is easy to use, reproducible, and can help guide the clinical management of patients with air leaks. These results have allowed us to embark on our third prospective trial on air leaks. References 1. LoCicero J III, Hartz RS, Frederikson JW, et al. New applications of laser in pulmonary surgery: hemostasis and sealing of air leaks. Ann Thorac Surg 1985;40: Matar AF, Hill JG, Duncan W, et al. Use of biological glue to control pulmonary air leaks. Thorax 1990;45: Turk R, Weidringer JW, Hartel W, et al. Closure of lung leaks by fibrin gluing. Experimental investigations and clinical experience. Thorac Cardiovasc Surg 1983;31: Hausmann M, Keller R. Thoracoscopic pleurodesis in spontaneous pneumothorax. J Suisse Med 1994;124: Hatta T, Tsubota N, Yoshimura M, et al. Intrapleural minocycline for postoperative air leakage and control of malignant pleural effusion. Jpn J Thorac Surg 1990;43: Cerfolio RJ, Tummala RP, Holman WL, et al. A prospective algorithm for the management of air leaks after pulmonary resection. Ann Thorac Surg 1998;66: Kirschner PA. Provocative clamping and removal of chest tubes despite persistent air leak [Letter]. Ann Thorac Surg 1992;53:740 1.