bev 3 stats assignment

Management of the Second Episode of Spontaneous Pneumothorax: A Decision Analysis Pierre-Emmanuel Falcoz, MD, Christine Binquet, MD, François Clement, MD, Djamel Kaili, MD, Catherine Quantin, MD, PhD, Sidney Chocron, MD, PhD, and Joseph-Philippe Etievent, MD Department of Thoracic and Cardiovascular Surgery, Jean-Minjoz Hospital, Besançon, Department of Biostatistics, University Hospital, Dijon, France

Background. Optimal management for patients pre- senting a second episode of spontaneous pneumothorax remains controversial. The aim of this study was to compare two possible treatment strategies, video-assisted thoracic surgery (VATS) and conservative management, in order to assess which of the two was better adapted for the treatment of the second episode of spontaneous pneumothorax.

Methods. The authors propose a decision analytic model including a cost-effectiveness study to compare two clinical strategies: VATS (reference strategy) and conservative management (alternative strategy). Data were obtained from a Medline search for English lan- guage articles and cost estimates were derived from the financial and public health departments of our hospital. The model was analyzed to determine the baseline strat- egy leading to the highest expected effectiveness and the lowest expected cost.

Results. Conservative management offered a slight advantage in expected effectiveness value (99.99 vs 99.93 for VATS). VATS produced the lowest expected cost (€4347 vs €7536 for conservative management). The incre- mental cost-effectiveness ratio was €57,750. Within the ranges tested, the sensitivity analysis presented consis- tent results in terms of effectiveness and advocated conservative management as the best strategy. In terms of cost, with the exception of length of stay, the sensitivity analysis was insensitive in estimating the different prob- abilities, and favored VATS over conservative management.

Conclusions. In the management of the second episode of spontaneous pneumothorax, VATS offers substantial savings in cost for only a slight decrease in effectiveness, when compared with conservative management.

(Ann Thorac Surg 2003;76:1843– 8) © 2003 by The Society of Thoracic Surgeons

Spontaneous pneumothorax (SP) is a disabling disor-der that may present either in young and otherwise healthy patients (primary pneumothorax) or as a compli- cation of an underlying lung disease (secondary pneu- mothorax). Since the 1960s clinical trials have led to opposing recommendations for management of SP, an issue still under debate particularly for the management of the second episode.

Conservative management (CM) has long been the generally accepted treatment strategy for the second episode of SP (pleural drainage and a “wait and see” approach) [1, 2]. Thanks to advances in thoracoscopic instrumentation (such as optics, instruments, and endo- scopic staplers) treatment strategy has progressively changed since the early 1990s, and video-assisted tho- racic surgery (VATS) has set a new standard of excellence for treating the second episode of SP [3–5].

Nevertheless, there is still controversy as to what the best strategy is for a patient presenting a second episode of SP. The answer to this question remains unclear due to

the lack of controlled clinical trials comparing the two possible treatment strategies. We propose a decision analytic model to examine this clinically important issue.

The aim of this study was to compare two possible treatment strategies, VATS and CM, in order to assess which of the two was better adapted for the treatment of the second episode of SP.

Material and Methods

The Model This study was conducted using a decision analysis design. Decision analysis is a quantitative method for synthesizing data from numerous sources to evaluate treatment alternatives [6]. All decision analyses involve the following basic components: the alternative strategies and potential outcomes associated with each strategy are specified in the decision model; the probabilities for each of these outcomes are estimated from the most appropri- ate available data and assigned to each decision point in the model; and an analysis is performed to calculate the expected value of each treatment alternative. By calculat- ing the expected value, a favored strategy is identified. All analyses were performed with Data version 3.5 (Tree- Age Software Inc, Williamstown, MA), a decision analy-

Accepted for publication June 3, 2003.

Address reprint requests to Dr Falcoz, Department of Thoracic and Cardiovascular Surgery, Hôpital Jean-Minjoz, Boulevard Fleming, 25000 Besançon, France; e-mail: [email protected].

© 2003 by The Society of Thoracic Surgeons 0003-4975/03/$30.00 Published by Elsevier Inc doi:10.1016/S0003-4975(03)01324-9

G E

N E

R A

L T

H O

R A

C IC

sis software program. The decision tree used in this analysis is illustrated in Figure 1.

In the present study, the outcomes of each strategy were examined both in terms of effectiveness and cost. By performing a cost-effectiveness analysis, we obtained a more integrated view of the decision-making issue. Thus, the model was analyzed to determine the baseline strat- egy that leads to the highest expected effectiveness and lowest expected cost.

The financial impact of the management of the second episode of SP may be considered from different perspec- tives, for example: the physician’s time or health-care costs. Our model looks at financial impact from the national health insurance perspective. Because it was not possible to determine real costs, only direct costs were taken into account. No attempt was made to include noninstitutional costs, either direct or indirect, born by the patients and their families or by society.

Assumptions In designing the decision tree, several assumptions were made in order to simplify the analysis. Complications that followed an episode of SP were categorized into three groups, regardless of the chosen treatment strat- egy: short-term (less than 1 month), medium-term (less than 3 months), and long-term (less than 2 years). All potential complications were considered either as imper-

fect results or major complications. Imperfect results were: minor air leak, pleural effusion, incomplete reex- pansion of the lung (failing pleurodesis), and short-term ipsilateral recurrent pneumothorax. Major complications were: hemothorax or pleural infection following surgery or chest tube insertion, prolonged air leak, and middle- and long-term ipsilateral recurrent pneumothorax. For each VATS, the probability of conversion to thoracotomy was taken into account.

Utility and Probabilities of Chance Events The concept of utility is a measure of a decision maker’s relative preference for an outcome state. Each of the health states at the terminal nodes of the decision tree, which represent the final outcomes, are thus assigned a utility value that quantifies preferences for these states [7]. In the context of the present decision tree, the final outcomes under study were either definitive resolution or ensuing complications. They were estimated over a 2-year period because most recurrences occur during the first 2 years after the initial pneumothorax [8]. The utility of a patient having a definitive resolution, which is the optimal outcome, was assigned a value of 100. The utility of a patient experiencing a course with ensuing compli- cations was assigned a value of 0.

Table 1 summarizes the main data probabilities and estimates of their values used in the decision analysis.

Fig 1. Decision trees representing the choice of management strategies for the second episode of spontaneous pneumothorax. The two clinical strategies to be chosen from are represented at the square decision node of the first tree (A). The probabilities and estimates of their values are listed in Table 1. � � the choice between strategies (decision node); E � the occurrence of chance events (chance node); and � � a logic check in the simulation (terminal node). (CM � conservative management; VATS � video-assisted thoracic surgery.)

1844 FALCOZ ET AL Ann Thorac Surg MANAGEMENT OF SPONTANEOUS PNEUMOTHORAX 2003;76:1843– 8

G E

N E

R A

L T

H O

R A

C IC

These values are based on a critical review of the avail- able literature regarding SP management. We performed a systematic Medline search for all English language articles dating from 1966 and falling under the medical subject headings “spontaneous pneumothorax,” “pleural drainage,” “thoracoscopy,” and “video-assisted thoracic surgery” alone and in combination with the terms “ran- domized controlled trials,” “meta-analysis,” and “guide- lines.” This survey of the medical literature was done to obtain a baseline value and a range for all variables of interest. Decision tree values were assigned by applying critical appraisal criteria (methodologic quality, largest number of cases reported, most recent information, and relevance to our study population) in the articles re- trieved [9].

Cost Data Costs, expressed in euros (€), were provided by the financial department of our hospital. They are those of the actual procedures for pneumothorax. The total cost of both surgical and anesthetic procedures was taken into account. The anesthetic cost was €188.1 for VATS and thoracotomy, and €20.9 for CM. The surgical cost was €418 for VATS and thoracotomy, and €41.8 for CM. The cost per day of hospitalization was €584.19 (ranging from €416.93 to €724.03).

The public health department provided the mean and standard deviation of length of stay over a 5-year period (1998 to 2002) of 133 patients suffering a second episode of SP: 5.4 � 1.8 days for VATS (n � 82 patients), 10.5 � 3.5 days for thoracotomy (n � 17 patients), and 9.4 � 6.9 days for CM (n � 34 patients).

Sensitivity Analysis The stability of the results obtained from the model for mean probabilities and costs was assessed through sen- sitivity analysis. Sensitivity analysis is a method of vary- ing probabilities over a defined range to determine how the optimal choice would change if the value of a chance event changed [10]. Sensitivity analysis can be performed on individual probabilities (one-way sensitivity analysis) or by varying two probabilities at the same time (two-way sensitivity analysis). One-way sensitivity analysis was performed for each probability in the decision tree, and the value at which the optimal strategy changed (thresh- old value) was identified. Two-way sensitivity analysis was performed for all probabilities found to be relevant in one-way sensitivity analysis.

Results

Baseline Analysis The baseline analysis (Table 2) was based on the assump- tions described above and used the aforementioned probabilities (Table 1), costs estimates, and formalized utility scores.

Conservative management offered a slight increase in expected effectiveness as compared to VATS. The incre- mental effectiveness, obtained by calculating the differ- ence between the expected effectiveness values associ- ated with the two strategies, was 0.06. This value indicates that CM would have provided six additional recoveries for 10,000 patients. Similarly, when focusing on the hierarchical choice between VATS and thoracot- omy, our analysis suggests that VATS would have pro- vided 95 additional recoveries for 10,000 patients under- going surgery.

The strategy producing the lowest expected cost was VATS. The cost-effectiveness ratio for one recovery was €44 for VATS and €75 for CM. The incremental cost- effectiveness ratio was €57,750 for CM, which means that the cost of one additional recovery by CM, when com- pared to VATS, was €57,750.

Sensitivity Analysis Because the probabilities for the various outcomes used in this analysis could conceivably vary, sensitivity testing was performed to assess the validity of the conclusions over a wide range of probabilities.

In terms of effectiveness, the sensitivity analysis dem-

Table 1. Literature-Based Range of Probabilities Comparing VATS With Conservative Management

Parameter

Baseline Value

(%)

Sensitivity Analysis

Range (%) References

Definitive resolution after CM

70 48 – 84 [11a–13]

Complications after CM 30 16 –52 [11a, 14a] Imperfect results 27 5–50 [15a, 16] Major complications 3 0 –3 [15a, 17, 18]

Definitive resolution after VATS

90 86 –98 [11a, 12, 14a]

Complications after VATS

10 5–10 [11a, 19]

Imperfect results 7.5 0 –22 [20 –22] Major complications 2.5 0 –24 [20, 21, 23]

Conversion to thoracotomy

7 2–10 [12, 19]

Definitive resolution after thoracotomy

99 99 –99.5 [14a, 24]

a Review articles compiling different studies.

The baseline value is the best estimate for each probability (average of the results from the different studies to estimate the probability). The range reflects the lowest and highest estimates from the retrieved studies.

CM � conservative management; VATS � video-assisted thoracic surgery.

Table 2. Summary of Baseline Results Comparing VATS With Conservative Management

VATS CM

Effectiveness 99.93 99.99 Incremental effectiveness — 0.06 Cost (€) 4,347 7,536 Incremental cost (€) — 3,189 Cost-effectiveness ratio (€) 44 75 Incremental cost-effectiveness ratio (€) — 57,750

CM � conservative management; VATS � video-assisted thoracic surgery.

1845Ann Thorac Surg FALCOZ ET AL 2003;76:1843– 8 MANAGEMENT OF SPONTANEOUS PNEUMOTHORAX

G E

N E

R A

L T

H O

R A

C IC

onstrated consistent results and advocated CM as the better of the two strategies regardless of the probability concerned. Within the ranges tested, the different vari- ables in the model influenced the weight of VATS, but did not change the preferred strategy. Examples are given in Figure 2, which illustrates the effect of definitive resolution after VATS (Fig 2A) and the effect of definitive resolution after CM (Fig 2B) on the two management strategies.

In terms of costs, the sensitivity analysis was insensi- tive to changes in the estimate of the values, except for length of stay, and favored VATS over CM for the treatment of the second episode of SP. One-way sensi- tivity analysis for VATS did not reveal any threshold (Fig 2C). One-way sensitivity analysis for CM revealed that the threshold value at which strategies were equivalent for length of stay was 4.9 days (Fig 2D). Two-way sensi- tivity analysis indicated how the optimal strategy changed as the length of stay related to each strategy varied (Fig 3). A combination of these two lengths of stay corresponds to a point on the graph. For any combina- tion, if the point lies in the black area, VATS is the optimal choice and in the white area, it is CM. The boundary line between the two areas represents a series

of threshold values at which the two strategies yield an equivalent outcome.

Comment

In the field of SP, a consensus emerging in the literature indicates clearly what to do for an initial episode of SP (CM) or the third episode (VATS). Conversely, the need for and method of promoting adequate pleural symphy- sis at the second episode has given rise to great contro- versy. The usefulness of VATS as opposed to CM for the second episode of SP remains controversial. Although some centers recommend VATS [5, 19, 25], others, who report a high recurrence and complication rate and recommend more caution [1, 2], advocate CM. Clearly there is still much debate over which approach is more appropriate in any given patient. The British Thoracic Society [26] and the American College of Chest Physi- cians [27] have both published guidelines for the imme- diate and subsequent management of patients with SP. On one hand, the guidelines of the British Thoracic Society make no distinction between patients with first and recurrent episodes of SP. On the other hand, most of the panel members of the American College of Chest

Fig 2. One-way sensitivity analysis demonstrating the effect of different variables in terms of effectiveness and cost. The upper graphs illustrate the effect of definitive resolution after VATS (A) and after CM (B) on expected effectiveness. The lower graphs illustrate the effect of length of stay for VATS (C) and for CM (2 d) on expected cost (D). The threshold value at which strategies were equivalent was 4.9 days. ■ � CM; E � VATS. (CM � conservative management; T � threshold value; VATS � video-assisted thoracic surgery.)

1846 FALCOZ ET AL Ann Thorac Surg MANAGEMENT OF SPONTANEOUS PNEUMOTHORAX 2003;76:1843– 8

G E

N E

R A

L T

H O

R A

C IC

Physicians recommend VATS at the second episode. Because a consensus has not been clearly established from these basic recommendations, even among experts, and given the lack of controlled clinical trials, we decided to address this controversial question using formal deci- sion analysis.

The use of decision analysis in surgery is gaining interest [28, 29]. In a complex decision-making environ- ment, where medical practice is influenced by published clinical trials, consensus statements and detailed formal examination of procedures, physicians may perform in- adequately in establishing the standard of care if they rely only on their clinical judgment based uniquely on past experience. Since none of us can intuitively estimate probabilities accurately, decision analysis becomes use- ful when it is desirable to compare alternative treatment strategies quantitatively. Nevertheless, this method has its own limitations. Some of the limitations of our deci- sion analysis imposed by the assumptions and tree structure deserve mention. First, the possibility of having to perform a thoracotomy was taken into account. We decided not to exclude this option from our model in order to leave the surgeon the choice of which technique he judged the more appropriate to treat a major compli- cation (eg, hemothorax). Second, we used estimated costs of the actual procedures for SP to compare the cost of the two strategies. The ideal approach would have been to look at the real cost of each procedure and model those costs explicitly. Such a cost analysis was beyond the

scope of the current work and not really justified in a comparative study.

Despite the above limitations from the standpoint of baseline probabilities, our analysis reveals only a slight advantage in terms of expected effectiveness for CM. The difference we take into account in our analysis between CM and VATS (incremental effectiveness value of 0.06) may be somewhat disturbing to readers. This value is rather small and, consequently, the difference between the two strategies is only slight. However, from the perspective of decision analysis, it is not the quantity but the sense of the difference between two treatment strat- egies that guides the choice of the decision [6]. It is especially noteworthy that sensitivity analysis did not have a significant impact on the baseline results. In the absence of variation in the sensitivity analysis, the inter- pretation of results can be considered reliable and robust. In addition, although the primary goal of this study was to compare two possible treatment strategies, VATS and CM, in the management of a second episode of SP, the present model also provides information for a hierarchi- cal choice between VATS and thoracotomy, keeping in mind that thoracotomy is often reserved for patients with technically more complex problems. Be that as it may, VATS would have provided 95 additional recoveries for 10,000 patients undergoing surgery for a second episode of SP. Hence, it should be considered whenever surgery is necessary. This finding has already been reported in many studies dealing with VATS versus thoracotomy for SP [19, 24, 27], especially in primary SP [25].

In the present study, VATS was the strategy found to produce the lowest expected cost. This result was prin- cipally due to a shorter “cost-consuming” length of stay. Schramel and colleagues [30] highlighted this in a cost analysis study comparing VATS and CM. The length of stay for VATS was 5.4 � 1.8 days in our study, 6.9 days (ranging from 2 to 15 days) in Crisci and Coloni’s study [31], whereas it was 11 � 4 days in Schramel associates’ study [30] and 2.9 � 0.9 in Hazelrigg and coworkers’ study [32]. In considering the discordance in the length of stay between these studies (postoperative length of stay for European patients is typically much longer than for US patients) it was necessary to examine the impact on the results of these variations in duration. By varying the length of stay for CM, the one-way sensitivity analysis revealed that the threshold value at which outcomes were equivalent was 4.9 days. CM should be preferred below this threshold value, and VATS above it. The two-way sensitivity analysis indicates the preferred strat- egy in function of the mean length of stay (Fig 3). The shorter the length of stay for VATS, the smaller the area in which CM was the optimal strategy. Therefore, de- pending on the length of stay in a given institution, the clinician may well choose one strategy or the other.

Cost effectiveness is becoming more and more impor- tant in the choice of treatment. The incremental cost- effectiveness ratio found in our study was approximately €58,000 for CM. Deciding whether a strategy is cost effective is a subjective interpretation of the cost- effectiveness ratio. Although there is no absolute cutoff

Fig 3. Two-way sensitivity analysis of the relation between the length of stay for VATS and CM that illustrates how the optimal strategy changes according to the length of stay when all other prob- abilities are maintained at the baseline. The black area indicates VATS as the preferred strategy; the white area indicates CM as the preferred strategy. The base-case scenario, indicated by a star, is defined by a mean length of stay of 5.4 days for VATS and 9.4 days for CM. (CM � conservative management; VATS � video-assisted thoracic surgery.)

1847Ann Thorac Surg FALCOZ ET AL 2003;76:1843– 8 MANAGEMENT OF SPONTANEOUS PNEUMOTHORAX

G E

N E

R A

L T

H O

R A

C IC

level, there is a consensus that a cost-effectiveness ratio under the accepted limit of €50,000 corresponds to a cost-effective strategy [33]. Hence, CM is definitely not cost effective for the management of the second episode of SP in our study.

In conclusion, the results we obtained using decision analysis for the management of the second episode of SP reveal that VATS offers substantial savings in cost for only a slight decrease in effectiveness when compared with CM. Institutions whose probabilities reflect those presented in our analysis may well want to consider performing VATS in all patients undergoing a second episode of SP. Nevertheless, further clinical trials should help to validate the formal analytic results found in this study.

The authors thank Nancy Richardson-Peuteuil for her editorial assistance, and Catherine Lejeune for her helpful comments.

References

1. Weissberg D, Refaely Y. Pneumothorax: experience with 1,199 patients. Chest 2000;117:1279 –85.

2. Miller AC. Management of spontaneous pneumothorax: back to the future. Eur Respir J 1996;9:1773–4.

3. Hatz RA, Kaps MF, Meimarakis G, Loehe F, Müller C, Fürst H. Long-term results after video-assisted thoracoscopic sur- gery for first-time and recurrent spontaneous pneumotho- rax. Ann Thorac Surg 2000;70:253–7.

4. Cardillo G, Facciolo F, Giunti R, et al. Videothoracoscopic treatment of primary spontaneous pneumothorax: a 6-year experience. Ann Thorac Surg 2000;69:357–62.

5. Chan P, Clarke P, Daniel FJ, Knight SR, Seevanayagam S. Efficacy study of video-assisted thoracoscopic surgery pleu- rodesis for spontaneous pneumothorax. Ann Thorac Surg 2001;71:452–4.

6. Weinstein M, Fineberg H. Clinical decision analysis. Phila- delphia: WB Saunders, 1980.

7. Naglie G, Krahn MD, Naimark D, Redelmeier DA, Detsky AS. Primer on medical decision analysis: part 3. Estimating probabilities and utilities. Med Decis Making 1997;17:136 –41.

8. Lippert HL, Lund O, Blegvad S, Larsen HV. Independent risk factors for cumulative recurrence rate after first sponta- neous pneumothorax. Eur Respir J 1991;4:324 –31.

9. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. 2nd edition. Boston, MA: Little, Brown and Company, 1991.

10. Krahn MD, Naglie G, Naimark D, Redelmeier DA, Detsky AS. Primer on medical decision analysis: part 4. Analyzing the model and interpreting the results. Med Decis Making 1997;17:142–51.

11. Sahn SA, Heffner JE. Primary care: spontaneous pneumo- thorax. N Engl J Med 2000;342:868 –74.

12. Inderbitzi RGC, Leiser A, Furrer M, Althaus U. Three years’ experience in video-assisted thoracic surgery (VATS) for spontaneous pneumothorax. J Thorac Cardiovasc Surg 1994; 107:1410 –5.

13. Cole FH Jr, Cole FH, Khandekar A, Maxwell JM, Pate JW, Walker WA. Video-assisted thoracic surgery: primary ther- apy for spontaneous pneumothorax? Ann Thorac Surg 1995; 60:931–5.

14. Massard G, Thomas P, Wihlm JM. Minimally invasive man-

agement for first and recurrent pneumothorax. Ann Thorac Surg 1998;66:592–9.

15. Miller KS, Sahn SA. Chest tubes: indications, technique, management and complications. Chest 1987;91:258 –64.

16. Chan L, Reilly KM, Henderson C, Kahn F, Salluzzo RF. Complications rates of tube thoracostomy. Am J Emerg Med 1997;15:368 –70.

17. Milikan JS, Moore EE, Steiner E, Aragon GE, VanWay CW. Complications of tube thoracostomy for acute trauma. Am J Surg 1980;140:738 –41.

18. Etoch SW, Bar-Natan MF, Marcos F, Miller FB, Richardson JD. Tube thoracostomy: factors related to complications. Arch Surg 1995;130:521–6.

19. Naunheim KS, Mack MJ, Hazelrigg SR, et al. Safety and efficacy of video-assisted thoracic surgical techniques for the treatment of spontaneous pneumothorax. J Thorac Cardio- vasc Surg 1995;109:1198 –204.

20. Dumont P, Diemont F, Massard G, Tourmieux B, Wihlm JM, Morand G. Does a thoracoscopic approach for surgical treatment of spontaneous pneumothorax represent progress? Eur J Cardiothorac Surg 1997;11:27–31.

21. Boutin C, Viallat JR, Aelony Y. Thoracoscopy in the diagno- sis and treatment of spontaneous pneumothorax. In Boutin C, Viallat JR, Aelony Y, eds. Practical thoracoscopy. 1st edition. Berlin: Springer, 1991:73–80.

22. Bertrand PC, Regnard JF, Spaggiari L, et al. Immediate and long-term results after surgical treatment of primary spon- taneous pneumothorax by VATS. Ann Thorac Surg 1996;61: 1641–5.

23. Casadio C, Rena O, Giobbe R, Rigoni R, Maggi G, Oliaro A. Stappler blebectomy and pleural abrasion by video-assisted thoracoscopy for spontaneous pneumothorax. J Cardiovasc Surg (Torino) 2002;43:259 –62.

24. Waller DA, Forty J, Morritt GN. Video-assisted thoraco- scopic surgery versus thoracotomy for spontaneous pneu- mothorax. Ann Thorac Surg 1994;58:372–7.

25. Passlick B, Born C, Häussinger K, Thetter O. Efficiency of video-assisted thoracic surgery for primary and secondary spontaneous pneumothorax. Ann Thorac Surg 1998;65: 324 –7.

26. Miller AC, Harvey JE. Guidelines for the management of spontaneous pneumothorax: standards of care committee, British Thoracic Society. BMJ 1993;307:114 –6.

27. Baumann MH, Strange C, Heffner JE, et al. Management of spontaneous pneumothorax: an American College of Chest Physicians Delphi Consensus Statement. Chest 2001;119: 590 –602.

28. Plume SK. Choices, chances, values. Ann Thorac Surg 1992; 53:373.

29. Birkmeyer JD, Birkmeyer NJO. Decision analysis in surgery. Surgery 1996;120:7–15.

30. Schramel FMNH, Sutedja TG, Braber JCE, Van Mourik JC, Postmus PE. Cost-effectiveness of video-assisted surgery versus conservative treatment for first time or recurrent spontaneous pneumothorax. Eur Respir J 1996;9:1821–5.

31. Crisci R, Coloni GF. Video-assisted thoracoscopic surgery versus thoracotomy for recurrent spontaneous pneumotho- rax. A comparison of results and costs. Eur J Cardiothorac Surg 1996;10:556 –60.

32. Hazelrigg SR, Landreneau RJ, Mack M, et al. Thoracoscopic stapled resection for spontaneous pneumothorax. J Thorac Cardiovasc Surg 1993;105:389 –93.

33. Elixhauser A, Halpern M, Schmier J, Luce BR. Health care cost-benefit analysis and cost-effectiveness analysis from 1991 to 1996: an update bibliography. Med Care 1998;36(5 Suppl):MS18 –147.

1848 FALCOZ ET AL Ann Thorac Surg MANAGEMENT OF SPONTANEOUS PNEUMOTHORAX 2003;76:1843– 8

G E

N E

R A

L T

H O

R A

C IC