Technology Report. Issue 48 December Comparison of Lung Volume Reduction Surgery with Medical Management for Emphysema

Transcription

1 Technology Report Issue 48 December 2004 Comparison of Lung Volume Reduction Surgery with Medical Management for Emphysema

2 Publications can be requested from: CCOHTA Carling Avenue Ottawa ON Canada K1S 5S8 Tel. (613) Fax. (613) or download from CCOHTA s web site: Cite as: Banerjee S, Babidge W, Miller J, Smith J, Noorani HZ, Cuncins-Hearn A, Mensinkai S. Comparison of lung volume reduction surgery with medical management for emphysema [Technology report no 48]. Ottawa: Canadian Coordinating Office for Health Technology Assessment; Reproduction of this document for non-commercial purposes is permitted provided appropriate credit is given to CCOHTA. CCOHTA is a non-profit organization funded by the federal, provincial and territorial governments. Legal Deposit National Library of Canada ISBN: (print) ISBN: (online) PUBLICATIONS MAIL AGREEMENT NO: RETURN UNDELIVERABLE CANADIAN ADDRESSES TO CANADIAN COORDINATING OFFICE FOR HEALTH TECHNOLOGY ASSESSMENT CARLING AVENUE OTTAWA ON K1S 5S8

3 Canadian Coordinating Office for Health Technology Assessment Comparison of Lung Volume Reduction Surgery with Medical Management for Emphysema Srabani Banerjee MSc PhD 1 Wendy Babidge PhD 2 John Miller MD FRCSC 3 Julian Smith MS FRACS 4 Hussein Z. Noorani MSc 1 Astrid Cuncins-Hearn MSc 2 Shaila Mensinkai MA MLIS 1 December Canadian Coordinating Office for Health Technology Assessment (CCOHTA), Ottawa ON Canada 2 ASERNIP-S, Royal Australasian College of Surgeons, Adelaide, Australia 3 McMaster University, Hamilton ON Canada 4 Monash University, Clayton, Australia

4 Reviewers These individuals kindly provided comments on this report. External Reviewers Donna Maziak, MDCM FRCSC FACS Associate Professor Ottawa Hospital General Campus University of Ottawa Ottawa ON Robert Dales, MSc MD FRCPC Professor of Medicine University of Ottawa Ottawa ON Jeremy Road, MD FRCPC Professor of Medicine Vancouver Hospital and Health Sciences Centre Faculty of Medicine University of British Columbia Vancouver BC CCOHTA Scientific Advisory Panel Reviewers David Hailey, MSc PhD Department of Public Health Sciences University of Alberta Edmonton AB Andrew B Hill, MSc MDCM Associate Professor Chair, University of Ottawa Division of Vascular Surgery The Ottawa Hospital Ottawa ON This report is a review of existing public literature, studies, materials and other information and documentation (collectively the source documentation ) which are available to CCOHTA. The accuracy of the contents of the source documentation on which this report is based is not warranted, assured or represented in any way by CCOHTA and CCOHTA does not assume responsibility for the quality, propriety, inaccuracies or reasonableness of any statements, information or conclusions contained in the source documentation. CCOHTA takes sole responsibility for the final form and content of this report. The statements and conclusions in this report are those of CCOHTA and not of its Panel members or reviewers. Authorship Srabani Banerjee coordinated the project and was involved in the protocol development; article selection; data extraction and analysis; and preparation of the report. Wendy Babidge was involved in the protocol development, article selection, data extraction and preparation of the report. John Miller was involved in the protocol development and preparation of the report. Julian Smith was involved in the protocol development and preparation of the report. i

5 Hussein Noorani was involved in the article selection, data extraction and preparation of the report. Astrid Cuncins-Hearn was involved in the article selection, data extraction and preparation of the report. Shaila Mensinkai was involved in developing the protocol, in conducting the literature search and in writing the literature search method and associated appendix for the report. Acknowledgements The authors thank Catherine Allison for her input and for writing the Report in Brief, Lynn Brodsky and Chuong Ho for their input and help in checking data, Christine Murray and Isabella Steffensen for their assistance with the data extraction and Clara Pham for her assistance with the data extraction and checking. Conflicts of Interest For all authors: no competing interests. ii

6 REPORT IN BRIEF December 2004 Lung Volume Reduction Surgery Compared with Medical Management for Emphysema Technology Name Lung volume reduction surgery (LVRS) Disease or Condition Emphysema is a chronic obstructive pulmonary disease that mainly affects adults over the age of 65. The disease causes shortness of breath and reduces quality of life. Technology Description LVRS involves the partial removal of severely affected lung tissue. This allows healthy tissue to expand and increases oxygen uptake. The Issue Mortality from COPD continues to rise in Canada. LVRS is an expensive procedure with high risks for postoperative death in older patients who are severely compromised by lung disease. There is a need to assess the efficacy of surgery compared to medical management. Assessment Objectives To assess the evidence on whether LVRS improves quality of life; defers death; or affects lung function in patients with emphysema, compared to medical management To identify the risks associated with LVRS and to evaluate what level of risk is appropriate in patients with a compromised quality of life. Methods We performed a systematic literature review and identified seven randomized controlled trials (involving 1,412 patients) that compared LVRS to medical management. The outcome measures included quality of life, complications arising from treatment, death, shortness of breath, lung function and exercise function. Where possible, metaanalyses were done to derive a statistical summary. Conclusions LVRS is a palliative treatment for patients with severe emphysema. It improves quality of life, lung function and exercise tolerance, but the trade-off compared with medical management alone is an increase in short-term death. There is no reduction in the overall death rate at approximately two years of follow-up. LVRS offers a survival advantage compared with medical management for patients whose emphysema mainly affects the upper lobes of the lung and whose baseline exercise capacity is low. Patients who have a forced expiry volume in one second that is no more than 20% of the predicted value and who have either a homogenous distribution of emphysema in the lung or a carbon monoxide diffusing capacity that is no more than 20% of the predicted value are at high risk of death after LVRS. As with most surgical procedures, LVRS is associated with significant risk. Based on the compromised quality of life experienced by patients with severe emphysema, it is difficult to define an acceptable level of surgical risk. More trials are needed to evaluate the safety issues involved. This summary is based on a comprehensive health technology assessment available from CCOHTA s web site ( Banerjee S, Babidge W, Miller J, Smith J, Noorani HZ, Cuncins-Hearn A, Mensinkai S. Comparison of lung volume reduction surgery with medical management for emphysema. Canadian Coordinating Office for Health Technology Assessment (CCOHTA) Carling Avenue, Ottawa, ON, Canada K1S 5S8 Tel: Fax: CCOHTA is an independent, not-for-profit organization that supports informed health care decision-making by providing unbiased, reliable information about health technologies. iii

7 EXECUTIVE SUMMARY The Issue There is no cure for emphysema, which is a debilitating disease that is common among seniors. A subtype of chronic obstructive pulmonary disease (COPD), emphysema is characterized by the destruction of alveolar walls and hyperinflation of the lungs, which result in shortness of breath, decreased exercise capacity and reduced quality of life (QoL). Mortality from COPD continues to escalate in Canada and the health care costs are enormous. Lung volume reduction surgery (LVRS) is used to remove damaged lung tissue and seems to be a promising treatment for patients with severe emphysema. It is expensive, however, and the risks of early postoperative mortality are high in older patients who are compromised by lung disease. Consequently, there is a need to assess the efficacy of LVRS compared to medical management (M) with drugs, oxygen and lung rehabilitation. Objective This systematic review and meta-analysis evaluates the clinical efficacy of LVRS compared to M (with or without rehabilitation) in the treatment of emphysema. Four questions are addressed. Does LVRS improve QoL? What are the risks associated with LVRS and what level of risk is appropriate to consider for patients with compromised QoL? Does LVRS defer mortality? Does LVRS affect morbidity with respect to parameters such as pulmonary function and exercise tolerance? Methods Relevant trials and studies were identified by searching electronic databases and web sites. Experts in emphysema treatment were consulted. Randomized controlled trials (RCTs) were selected for inclusion if they compared LVRS with M (with or without rehabilitation) for the treatment of emphysema. The outcomes analyzed included QoL, complications associated with treatment, mortality, shortness of breath (dyspnea), level of blood gases, exercise function and pulmonary function. Case-series studies were included to obtain more information on the complications and mortality associated with LVRS. Results Seven RCTs involving 1,412 patients met the inclusion criteria for the systematic review. Because of the lack of complete reporting of data, not all trials could be included in the metaanalyses for all outcomes. Overall QoL improved after LVRS, compared with M. Three trials used the chronic respiratory disease questionnaire (CRQ). At six months follow-up, pooled QoL data from three RCTs indicated more improvement with LVRS compared with M. The weighted mean differences (WMD) and 95% confidence intervals (CI) were 1.23 (0.80, 1.65) for emotion, 1.48 (1.07, 1.90) for fatigue and 1.42 (0.98, 1.87) for mastery (feeling of control over the disease). The remaining trials had different follow-up times (three months to two years) and used different assessment tools, but also showed an improved QoL with LVRS compared with M. iv

8 Hospitalization and complications associated with LVRS or M were sparsely documented in the RCTs. The duration of hospital stay ranged from 12 to 22 days in three RCTs. Prolonged air leaks from the lung into the chest cavity ranged from 13% to 37% in three RCTs. Other complications documented in some RCTs included infection, bleeding, pneumonia, transient ischemic attack, transient confusion, transient Horner syndrome, small bowel ileus, benign dysrhythmia and atrial fibrillation. The 59 case-series studies included did not provide significant additional insight into the risks of surgery. A number of RCTs showed that the relative risk (RR) of death at three, six and 12 months were higher with LVRS compared to M. This value was statistically significant at three months [RR (95% CI)=5.71 (2.89, 11.26)] but not at six and 12 months [RR (95% CI)=1.56 (0.68, 3.56) and 2.90 (0.83, 10.10) respectively]. The 17-centre National Emphysema Trial (NETT) conducted in the US, showed that at a mean follow-up of 29 months, the mortality with LVRS was similar to that with M [RR (95% CI)=0.98 (0.81, 1.19)]. Patients who had a low baseline exercise capacity and whose emphysema was mainly in the upper lobes of the lung were more likely to survive longer after LVRS compared with similar patients who received M only [RR (95% CI)=0.55 (0.37, 0.84)]. Patients with a forced expiratory volume in one second (FEV 1 ) that was 20% of the predicted value and who had homogenous emphysema or a carbon monoxide diffusing capacity (DLCO) 20% of the predicted value were at high risk of death when undergoing LVRS [RR (95% CI) for 90-day mortality was (2.53, )]. Pulmonary function data (FEV 1 ) was reported in some trials in litres (L) and in other trials as % of predicted (% pr). LVRS produced a statistically significantly greater improvement in FEV 1 (L) at three and six months of follow-up [WMD (95% CI)=0.19 (0.04, 0.34) and 0.26 (0.16, 0.37) respectively]. But at nine and 12 months, the improvement in FEV 1 (L) obtained with LVRS over M was not statistically significant [mean difference (95% CI) and WMD (95% CI)=0.30 ([ 0.14, 0.74) and 0.16 ([ 0.02, 0.34) respectively]. Those trials that reported FEV 1 (% pr) showed that between three and 24 months, there was a statistically significant greater improvement with LVRS. Compared to M, LVRS provided greater improvement in exercise function (measured by the six-minute walk test). The difference was statistically significant at six, nine, 12 and 24 months [WMD (95% CI)=45.45 (29.58, 61.33), (25.32, 94.68), (30.43, 68.87) and (51.39, ) respectively], but not at three months [WMD (95% CI)=17.18 ( 14.22, 48.58)]. Conclusions LVRS is mostly a palliative treatment for patients with severe emphysema. Compared to M, LVRS does improve QoL. As with any surgical procedure, LVRS is associated with a number of postoperative complications. Data on the risks associated with LVRS are poorly documented. Compared to M, LVRS is associated with increased short-term mortality. There is no reduction in overall mortality. A recent large trial showed that in a certain subgroup of patients with emphysema, LVRS is associated with a reduction in mortality. Compared to M, LVRS produces greater improvement in exercise capacity. LVRS results in improved pulmonary function. RCTs are needed to confirm which subgroup of patients is likely to benefit from LVRS and to compare safety issues and the occurrence and extent of adverse events between LVRS and M. v

9 GLOSSARY 6MWT CRQ DLCO Dyspnea FEV 1 FVC Six-minute walk test (distance walked by patient in six minutes) Chronic respiratory disease questionnaire (has components for assessing dyspnea, emotion, fatigue and mastery) Diffusing capacity of carbon monoxide (CO) Shortness of breath, difficult or laboured breathing Forced expiratory volume in one second [maximal quantity of air patient can exhale in one second, expressed in litres (L) or as percentage of predicted value based on age, size, sex and race; most frequently used parameter to measure pulmonary function in emphysema patients] Forced vital capacity (maximal volume of air forcefully expired, expressed in L or as percentage of predicted) Hypercapnia Increased levels of carbon dioxide (CO 2 ) in blood Hypoxemia PaCO 2 PaO 2 RV SF-36 Shuttle walking distance SIP SGRQ Decreased levels of oxygen (O 2 ) in blood Partial pressure of CO 2 in arterial blood Partial pressure of O 2 in arterial blood Residual volume (amount of air that remains in lung after maximal expiration, expressed in L or as percentage of predicted) 36-item Short Form Questionnaire to determine QoL (higher scores indicate better QoL) Distance walked by patient on treadmill where speed is increased every 10 minutes Sickness Impact Profile (136-item questionnaire to determine QoL; higher scores indicate poorer QoL) St. George s respiratory questionnaire (51-item questionnaire to determine QoL; higher scores indicate poorer QoL) vi

10 ABBREVIATIONS % pr percentage of predicted value 6MWT six-minute walk test CI confidence interval CLVR Canadian Lung Volume Reduction trial COPD chronic obstructive pulmonary disease DLCO diffusing capacity of carbon monoxide ICU intensive care unit IQR interquartile range L litre LVRS lung volume reduction surgery M medical management (with or without rehabilitation) MD mean difference MS median sternotomy NC not calculated NETT National Emphysema Trial NR not reported NS not significant OBEST Overholt-Blue Cross Emphysema Surgery Trial QoL quality of life QWB Quality of Well Being questionnaire RCT randomized controlled trial RR relative risk SD standard deviation SE standard error SMD standardized mean difference TLC total lung capacity VATS video-assisted thoracoscopic surgery WMD weighted mean difference vii

11 TABLE OF CONTENTS EXECUTIVE SUMMARY... iv GLOSSARY... vi ABBREVIATIONS... vii 1 INTRODUCTION Background Technology Overview THE ISSUE OBJECTIVE METHOD Literature Search Strategy Selection Criteria and Method Data Extraction Strategy Strategy for Quality Assessment Data Analysis Methods RESULTS Quantity of Research Available Trial Characteristics RCTs Case series Data Analysis and Synthesis QoL Complications and mortality Dyspnea Pulmonary function Exercise function Blood gases DISCUSSION CONCLUSIONS REFERENCES...33 Appendix 1: Literature Search Strategies Appendix 2: Data Extraction Form Appendix 3: Trial Quality Assessment Form Appendix 4: Excluded Reports of Trials and Studies viii

12 Appendix 5: Characteristics and Complications of Case Series Appendix 6: QoL Appendix 7: QoL Measurement Instruments Appendix 8: Mortality Data from RCTs Appendix 9a: Dyspnea Appendix 9b: Dyspnea (using UCSD, Shortness of breadth questionnaire) Appendix 10a: Pulmonary Function Measured as FEV Appendix 10b: Pulmonary Function Measured as FEV 1 (% predicted) Appendix 11a: Pulmonary Function Measured as FVC (L) Appendix 11b: Pulmonary Function Measured as FVC (% predicted) Appendix 12a: Pulmonary Function Measured as RV (L) Appendix 12b: Pulmonary Function Measured as RV (% predicted) Appendix 13a: Pulmonary Function Measured as TLC (L) Appendix 13b: Pulmonary Function Measured as TLC (% predicted) Appendix 13b: Pulmonary Function Measured as TLC (% predicted) Appendix 14: Exercise Capacity Using 6MWT (m) or Shuttle Walking Distance (m) Appendix 15: Blood Gases Measured as PaO 2 (mm Hg) Appendix 16: Blood Gases Measured as PaCO 2 (mm Hg) Appendix 17a: Blood Gases Measured as DLCO (mm/min/mm Hg) Appendix 17b: Blood Gases DLCO Measured as (% predicted) ix

13 1 INTRODUCTION 1.1 Background Emphysema is one of several conditions collectively known as chronic obstructive pulmonary disease (COPD). 1 It is a debilitating disease characterized by a progressive destruction of the alveolar wall and hyperinflation of the lung. 2 As the disease develops, exercise capacity and quality of life (QoL) are reduced; and airflow limitation, hypoxemia and hypercapnia can worsen and lead to early death. 3 There is no known cure and the aim of treatment is to relieve symptoms and improve the QoL. 4 As COPD mainly affects adults over the age of 65, the rate of occurrence increases as the population ages. In Canada, it is the fourth leading cause of death in men and the seventh for women, with 9,618 deaths in A Canadian National Population Health Survey (1994 to 1995) found that COPD was present in women at the rate of 2.1% in non-smokers, 2.7% in exsmokers and 8.2% in smokers. In men, the corresponding percentages were 0.8%, 2.9% and 3.5%. 6 The 1998 National Population survey reported that 3.2% of adult Canadians had been diagnosed with chronic bronchitis or emphysema. 5 The total cost of COPD in Canada was estimated to be $1.67 billion in Preventive measures for COPD include smoking cessation strategies. Traditional first-line management of the disease is symptomatic treatment with bronchodilators, anti-inflammatory medications, oxygen supplementation and pulmonary rehabilitation. When medical management (M) fails, surgical procedures such as lung transplantation are considered. Surgical options to treat emphysema include lung volume reduction surgery (LVRS) and lung transplantation. LVRS may be used as a bridge to lung transplantation or when transplantation is inappropriate. In practice, patients are evaluated for both procedures. 8 Suitable organs for transplantation are scarce, leading to the search for surgical alternatives such as LVRS. 1.2 Technology Overview LVRS was originally developed in 1957 by Otto Brantigan et al., who published a small case series of 33 patients. 9 Brantigan et al. postulated that a loss of elastic recoil was mainly responsible for the expiratory airway collapse observed in emphysema. Their LVRS technique consisted of a unilateral thoracotomy with resection of 20% to 30% of the lung, coupled with lung denervation using radical hilar stripping. The procedure was repeated on the contralateral lung, as a staged procedure, once the patient recovered from the first operation. 3 It is postulated that LVRS provides benefit because the excision and plication of emphysematous lung tissue restores circumferential traction in small airways. 10 Others postulate that an outward circumferential pull on the airways diminishes expiratory airway collapse 11 and that the reduced thoracic volume facilitates better movement of the rib cage and diaphragm. 3 1

14 Brantigan et al. 9 documented a clinical improvement in 75% of patients but provided no objective physiologic measurements. Because of the high rates of severe morbidity and mortality (16%) that were reported, the procedure never gained widespread acceptance. In 1994, interest in LVRS was renewed after results were published by Joel Cooper et al. 11 They observed that after lung transplantation, the hyper-expanded chests and flattened diaphragms of COPD patients quickly reverted to a more normal configuration. They refined the original LVRS technique by buttressing the staple lines with bovine pericardial strips and including bilateral parenchymal reduction via median sternotomy. Two LVRS approaches have been developed: one involves a midline sternotomy with stapled resection of the lung tissue 12 and the other, videoassisted thoracoscopic surgery (VATS) with stapled resection of tissue. 13 Indications for LVRS vary among centres. They generally include severe expiratory airflow limitation despite maximal M, air trapping, marked hyperinflation and ability to complete a vigorous preoperative pulmonary rehabilitation program. Contraindications include severe pulmonary hypertension, continued smoking, severe co-morbidities that increase surgical risk, previous lung surgery and a low tolerance to exercise and pulmonary rehabilitation THE ISSUE Emphysema is a common health problem among seniors in Canada. Mortality from COPD continues to escalate and the health care costs are enormous. LVRS is one treatment option for patients with severe emphysema. It is an expensive procedure and the risks of early postoperative mortality are high in patients who are older and severely compromised by lung disease. An assessment of this procedure is warranted. 3 OBJECTIVE The aim of this systematic review is to evaluate the clinical efficacy of LVRS compared to M (with or without rehabilitation) for the treatment of emphysema. Four questions are addressed. Does LVRS improve QoL? What are the risks associated with LVRS and what level of risk is appropriate to consider for patients with compromised QoL? Does LVRS defer mortality in patients with emphysema? Does LVRS affect morbidity with respect to parameters such as pulmonary function and exercise tolerance? 2

15 4 METHOD A protocol for this systematic review was written a priori. It was later modified to expedite the review process and to better organize the large amount of available data. As a result, the systematic review was split into two. The first, which is reported here, is a comparison of LVRS with M. The second systematic review, which will be published separately, is a comparison of different LVRS procedures (e.g., median sternotomy, VATS). 4.1 Literature Search Strategy A DIALOG OneSearch was performed from 1992 onward across MEDLINE, EMBASE, BIOSIS Previews, PASCAL and Current Contents Search databases using a broad search strategy (Appendix 1). There were no language or publication restrictions. A soft study design filter was used to restrict retrieval to relevant studies. The original search was performed in 2002 with regular database alerts set up on MEDLINE, EMBASE and BIOSIS Previews to capture new studies. Parallel searches were performed and updated on PubMed and the Cochrane Library to capture additional studies (last update on PubMed was on March 2, 2004; Cochrane Library, Issue 1, 2004). Grey literature was obtained through searching the web sites of regulatory agencies, health technology assessment and related agencies and specialized databases, such as those of the University of York NHS Centre for Reviews and Dissemination (CRD) and the Latin American and Caribbean Center on Health Sciences Information (LILACS). The Internet was searched using Google and alltheweb search engines. Web sites of professional associations such as the American Thoracic Society and the Canadian Thoracic Society were also searched. 4.2 Selection Criteria and Method Trials were selected for inclusion if they were randomized controlled trials (RCTs) involving patients with emphysema, compared LVRS with M (with or without rehabilitation) and reported data on QoL, mortality, complications, dyspnea, pulmonary function, exercise tolerance or blood gas level. Case-series studies on patients with emphysema undergoing LVRS were included if the studies reported data on complications of surgery or mortality. Although case-series studies are considered to be weak, they were used to obtain additional information. Only case series studies published in English and involving 20 or more patients were considered. Trials and studies were independently selected for inclusion by two reviewers (SB and WB or AC or HN). Differences in opinion were resolved by consensus reached after discussion. Citations downloaded in Reference Manager (bibliographic software) were exported into Microsoft Excel (spreadsheet software) to document the selection process. Differences in decisions between reviewers at different geographic locations could then be easily compared and resolved. 3

16 4.3 Data Extraction Strategy After the selection of relevant trials and studies, data were extracted by two of six individuals (SB, WB, AC, CM, HN or LB) working independently and using a structured form (Appendix 2). 4.4 Strategy for Quality Assessment The quality of the included RCTs was evaluated using the Jadad five-point scale. This scale is used to assess randomization (0 to 2 points), double-blinding (0 to 2 points) and withdrawals and dropouts (0 to 1 point), with higher scores indicating superior quality. Information on the concealment of allocation was categorized as adequate, inadequate or unclear. Quality was scored using an assessment form (Appendix 3) based on the Jadad scale. 4.5 Data Analysis Methods The outcomes analyzed, which are listed in Table 1, were reported by the investigators as continuous data. In the case of outcomes where data from more than one trial were available, the data were pooled and summary estimates [weighted mean difference (WMD)] and corresponding 95% confidence intervals (CI) were calculated. If only data from one trial were available, the mean difference (MD) and corresponding 95% CI were calculated. The WMD, SMD or MD was statistically significant if the 95% CI excluded zero. QoL Outcome Dyspnea Pulmonary function Exercise function Level of blood gases Table 1: Outcomes analyzed Parameter or Scale Used to Measure Outcome Chronic respiratory questionnaire (CRQ), 36-item Short Form Questionnaire (SF-36), Quality of Well Being (QWB) Sickness Impact Profile (SIP), St. George s respiratory questionnaire (SGRQ) Dyspnea score + UCSD shortness of breadth questionnaire - Forced expiratory volume in 1 second (FEV 1 ), + Forced vital capacity (FVC) Residual volume (RV), - Total lung capacity (TLC) Six-minute walk test (6MWT), + Shuttle walking distance Partial pressure of arterial oxygen (PaO 2 ), + Diffusing capacity of carbon monoxide (DLCO) Partial pressure of arterial carbon dioxide (PaCO 2 ) - LVRS Produces a Better Result than M if WMD or MD is + LVRS=lung volume reduction surgery, M=medical management, UCSD=University of California, San Diego, QoL=quality of life. - 4

17 Complication and mortality data were also examined. For mortality reported as binary data, the relative risk (RR) and the corresponding 95% CI were calculated. RR<1 indicated a reduced mortality with LVRS. The corresponding 95% CI if it excluded the value of one indicated a significant change in mortality. The Cochrane software ReviewManager was used for the data analyses and generation of Forest plots. Computations were performed using the random effects model. For the analyses of continuous data, when standard deviation (SD) values were not reported, they were calculated from reported standard error (SE), interquartile range (IQR) or p values. Most of the trials reported post-treatment values (final values). Hence, post-treatment values were used for the analyses. When only change values from baseline values were available, the post-treatment values and corresponding SD were calculated where possible using appropriate assumptions and formulae. 5 RESULTS 5.1 Quantity of Research Available The flow chart for the trial selection is shown in Figure 1. A total of 1,643 citations were identified from the original literature search. From these, 84 potentially relevant reports were selected for scrutiny. Ten potentially relevant reports were identified from other sources. A total of 94 potentially relevant reports (16 reports of RCTs and 78 reports of case-series studies) were retrieved. Of the 16 RCT reports, eight reports describing seven unique RCTs could be used for the analyses and eight reports were excluded. Of the eight included reports, one report (JM, unpublished observations, 2004) described two unique RCTs (CLVR and OBEST), three reports described three unique RCTs, two reports (by Fishman et al. and Ramsey et al.) 18,19 were used to extract data for one unique RCT and another two reports (by Mineo et al. and Pompeo et al.) 20,21 were used to extract data for another unique RCT. The unpublished data for CLVR and OBEST were provided by Dr. John Miller of McMaster University and are referred to as JM, unpublished observations, 2004 throughout the report. Of the 78 case-series reports, 62 reports describing 59 unique studies could be included and 16 were excluded. Case-series studies were used only for documenting additional information on the risks associated with LVRS. The excluded reports are listed in Appendix 4. 5

18 Figure 1: Flow chart for selection of trials and studies 1,643 citations identified from electronic search and broad screened 1,559 citations excluded 84 potentially relevant reports retrieved for further scrutiny 10 potentially relevant reports identified from other sources 94 potentially relevant reports (16 reports on RCTs and 78 reports on case-series) 8 RCT reports excluded (3 did not contain sufficient data and 5 were multiple publications of same trial) 16 case-series reports excluded as they did not contain relevant data 8 reports describing 7 unique RCTs 62 reports describing 59 unique case-series studies 6

19 5.2 Trial Characteristics RCTs Characteristics of the RCTs comparing LVRS with M are shown in Table 2. Of the seven RCTs included, two were conducted in Canada, three in the US and one each in the UK and Italy. There was one large RCT 18,19 with 1,218 patients and follow-up time up to 24 months. The number of patients in four RCTs 15-17,20,21 and two unpublished RCTs (JM, unpublished observations, 2004) varied between 37 and 60 with follow-up times ranging between three and 12 months. LVRS included median sternotomy or VATS using a unilateral or bilateral procedure. Details of M were described in one RCT 16 and two unpublished RCTs (JM, unpublished observations, 2004). M included pulmonary rehabilitation, bronchodilators, anti-inflammatory agents, antibiotics and oxygen supplementation. Trial Enrolment Period Table 2: Characteristics of RCTs included for meta-analyses Participating Country Centres Number of Patients Intervention Aug 1997 CLVR, 2004 * to Oct 2001 Criner et NR US Single 37 Bilateral MS al., (performed by same surgeon) Geddes et al., Goldstein et al., Apr 1996 to Feb to 2001 (study period) NETT, Jan 1998 to Jul 2002 Oct 1998 to OBEST, 2004 * Jan 2002 Pompeo Jan 1996 to et al., Jan (study and period) Mineo et al Comparator Canada Multi- 60 MS M 6 UK Single 48 Bilateral MS or thoracoscopy Canada 55 VATS or MS (less often), whenever possible bilateral US Multi- 1,218 Bilateral stapled wedge resection through MS or VATS M 3 M 12 M 12 M 24 US Multi- 35 MS or VATS M 6 Italy Single 60 Unilateral or bilateral VATS M 6 Followup Time (months) MS=median sternotomy, VATS=video-assisted thoracoscopic surgery, M=medical management, * JM, unpublished observations,

20 Patient characteristics are described in Table 3. All RCTs included patients with severe emphysema. In the included trials, the mean age of the patients varied between 58.8±6.4 and 66.7±5.9 years; and the proportion of female patients ranged from 3.3% to 68.4%. The quality scores for the seven RCTs, assessed using the five-point Jadad scale, varied between 2 and 3 (mean±sd=2.6±0.5). Allocation concealment was adequate in one RCT 17 and unclear in the remaining six. Trial Arm Table 3: Characteristics of patients in RCTs Number of Patients Age in Years % Female Smoking Status CLVR, LVRS ±7.19 a * M ±7.38 a 35.7 Criner et LVRS 19 59±8.3 a 68.4 al., M ±6.4 a 55.5 Geddes et LVRS (56 to 67) c 29.2 al., M (53 to 69) c 25.0 Goldstein et LVRS 55 al., versus M eligible NETT, NR Smoking cessation 6 months NR 64.9±0.91 b 40.0 Smoking cessation 6 months Disease Stage Severe emphysema Severe diffuse emphysema Severe emphysema Severe, stable COPD; heterogeneous patients LVRS ±6.3 a 41.6 NR Heterogeneous (54.3%), homogeneous (45.7%) M ±5.9 a 35.9 NR Heterogeneous (55.1%), homogeneous (44.9%) OBEST, LVRS ±5.25 a * M ±5.98 a 36.3 Pompeo et LVRS ±7.3 a 3.3 al., M ±5 a 6.7 and Mineo et al., NR Smoking cessation 4 months a Mean±SD, b mean±se, c median (interquartile range), NR=not reported, LVRS=lung volume reduction surgery, M=medical management, * JM, unpublished observations, Case series Advanced emphysema Heterogeneous Details of the trial and patient characteristics are shown in Appendix 5. The 59 studies included over 5,000 patients. All patients had advanced or end-stage emphysema at recruitment. The range of means for age of patients recruited was from 53 to 68 years (Table 4). The proportion of female patients ranged from 0% to 70% (Table 4). 8

21 The country of origin for patient recruitment was reported in 58 studies: US (n=28), Japan (n=7), five trials each in Austria and Switzerland, Australia (n=4), Italy (n=3), one study each in Belgium, Canada, France, Germany, the United Kingdom and one study recruiting participants from both Canada and the US. There were three multi-centre studies (5%). Three studies were reported in multiple publications: Moersig (1998) 32 reported earlier by Cooper (1996), 31 and Malthaner (2000) 51 reported earlier by Miller (1999), 52 both on the same number of patients; and an update by Zollinger (1997) 82 on four additional patients (n=24) reported earlier by Bingisser (1996) (n=20). 83 Table 4: Summary of patient characteristics (case-series studies) Characteristic Range of Values Number of Studies (n=59) Sample size (number) 14 to Age (years) 53 to 68* 54 Females (%) 0 to *range of means. The type of surgical procedure used was reported in 57 studies (97%) (Table 5). Table 5: Types of surgical procedures (case-series studies) Surgical Procedure Number of Studies (n=57) MS 15 VATS and thoracoscopy 22 Thoracotomy 1 MS, VATS and thoracoscopy 10 MS and thoracotomy 6 MS, VATS and thoracotomy 3 MS=median sternotomy, VATS=video-assisted thorascopic surgery. 5.3 Data Analysis and Synthesis Data were obtained from seven reports and one unpublished report (JM, unpublished observations, 2004) describing seven RCTs that satisfied the inclusion criteria. Mean difference (MD) or relative risk (RR) were calculated where appropriate for the different parameters at the different follow-up times. For outcomes where more than one study was available, summary estimates [weighted mean difference (WMD) or relative risk (RR)] were computed using the random effects model. The complications of surgery and mortality related to LVRS were also examined using 59 case series studies that met the inclusion criteria. 9

22 5.3.1 QoL Five RCTs 15-18,20 and two unpublished RCTs (JM, unpublished observations, 2004) reported QoL data (Appendix 6) using different measurement scales (CRQ, SF-36, QWB and SGRQ). Details of the scales are given in Appendix 7. a) CRQ One RCT 17 and two unpublished RCTs (JM, unpublished observations, 2004) had CRQ (emotion, fatigue and mastery) data. The Forest plot (Figure 2) with these three RCTs shows that at six months, in comparison with M, LVRS produced a significantly better improvement in QoL [WMD (95% CI) for CRQ-emotion, CRQ-fatigue and CRQ-mastery were 1.23 (0.80, 1.65), 1.48 (1.07, 1.90) and 1.42 (0.98, 1.87) respectively], where 0.5 is considered to be a minimal clinically important difference for CRQ. CRQ data were also available at three, nine, and 12 months from one RCT 17 (Appendix 6). Figure 3 shows that the extent of improvement obtained with LVRS when compared with M, changed over time, suggesting an increasing trend followed by a constant trend or a decreasing trend. The results shown at three, nine and 12 months are from one RCT and the result at six months is from three RCTs. b) SF-36 Two RCTs (JM, unpublished observations, 2004) reported SF-36 (mental, physical) data at six months (Table 6). There was greater improvement with LVRS as compared with M. The difference was clinically important and statistically significant for SF-36-physical [WMD (95% CI) was 8.95 (5.21, 12.70)] but not significant for SF-36-mental [WMD (95% CI) was 1.89 (-3.29, 7.06]. A change of 5 is considered to be a minimal clinically important difference for SF-36. The WMDs are pooled values. One RCT 16 reported SF-36 overall values (Table 6). The extent of improvement obtained with LVRS as compared with M was significantly greater clinically and statistically. It increased over time up to six months and then leveled off. MD (95% CI) were (2.32, 19.68), (20.03, 37.97) and (15.89, 44.11) at three, six and 12 months respectively. Another RCT 20 reported SF-36 values and demonstrated greater improvement with LVRS compared with M, but insufficient data were available to determine the statistical significance. Details of the SF-36 data are included in Appendix 6. c) Quality of Well Being (QWB) and St. George s Respiratory Questionnaire (SGRQ) NETT 18 reported QoL status using the QWB and SGRQ (Table 6 and Appendix 6). The degree of improvement in QoL obtained with LVRS as compared with M was significantly greater and remained practically the same over time. WMDs (95% CI) at six, 12 and 24 months were respectively 0.05 (0.03, 0.07), 0.04 (0.02, 0.06) and 0.05 (0.03, 0.07) for QWB; and (-15.13, ), (-15.76, ) and (-13.63, -8.37) for SGRQ. The differences obtained with QWB and SGRQ were also clinically significant. Changes of 0.03 and -4 are considered as minimal clinically important differences for QWB and SGRQ respectively. d) Sickness Impact Profile (SIP) One RCT 15 presented QoL status at three months using SIP (mental, physical and overall) scores (Appendix 6). It demonstrated greater improvement with LVRS compared with M, but insufficient data were available to determine the statistical significance. 10

23 Figure 2: WMD in QoL (at six-month follow-up) between patients undergoing LVRS and those on M alone: QoL assessed with CRQ (emotion, fatigue and mastery) Study Control WMD (random) Weight WMD (random) or sub-category N Mean (SD) N Mean (SD) 95% CI % 95% CI CRQ (emotion) Goldstein, (1.17) (1.17) [0.68, 1.96] CLVR, (1.13) (1.20) [0.58, 1.88] OBEST, (1.06) (1.66) [-0.29, 2.07] Subtotal (95% CI) [0.80, 1.65] Test for heterogeneity: chi²=0.40, df=2 (p=0.82), I²=0% Test for overall effect: Z=5.67 (p< ) 11 CRQ (fatigue) Goldstein, (1.02) (1.27) [1.19, 2.45] CLVR, (1.27) (1.23) [0.53, 1.91] OBEST, (1.14) (1.13) [0.35, 2.13] Subtotal (95% CI) [1.07, 1.90] Test for heterogeneity: chi² = 1.95, df=2 (p=0.38), I²=0% Test for overall effect: Z=7.06 (p< ) CRQ (mastery) Goldstein, (1.17) (1.22) [0.95, 2.25] CLVR, (1.46) (1.25) [0.69, 2.19] OBEST, (0.97) (1.40) [-0.04, 1.98] Subtotal (95% CI) [0.98, 1.87] Test for heterogeneity: chi²=1.06, df=2 (p=0.59), I²=0% Test for overall effect: Z=6.32 (p< ) Favours M Favours LVRS WMD=weighted mean difference, QoL=quality of life, LVRS=lung volume reduction surgery, M=medical management, CRQ=chronic respiratory disease questionnaire. 11

24 Figure 3: Change with time in magnitude of improvement in QoL (measured using CRQ) of LVRS group as compared to M group (bars represent 95% CI) CRQ-emotion (LVRS versus M) Follow-up time (months) 3.0 CRQ-fatigue (LVRS versus M) Follow-up time (months) 3.0 CRQ-mastery (LVRS versus M) Follow-up time (months) QoL=quality of life, CRQ=chronic respiratory disease questionnaire, LVRS=lung volume reduction surgery, M=medical management. 12

25 Specific Measures Used Table 6: QoL results using various measurement instruments Number of Studies in Analysis Number of Patients in Analysis Time of Follow-up (months) WMD or MD (95% CI) CRQ-emotion (0.80, 1.65) CRQ-fatigue (0.07, 1.90) CRQ-mastery (0.98, 1.87) SF-36 mental (-3.29, 7.06) SF-36 physical (5.21, 12.70) SF-36 overall QWB SGRQ Complications and mortality (2.32, 19.68) (20.03, 37.97) (15.89, 44.11) (0.03, 0.07) (0.02, 0.06) (0.03, 0.07) (-15.13, ) (-15.67, ) (-13.60, -8.37) a) RCT Data for length of hospital stay and use of health care resources from the single large RCT (NETT) are shown in Table 7. Compared with the M group, the LVRS group had a significantly higher number of hospital days (p<0.001) and days of ambulatory care (p=0.005) during the first 12 months after randomization. Between 13 and 24 months, the number of hospital days was significantly less in the LVRS group than in the M group. Emergency room visits and claims for supplemental oxygen were not significantly different in the two groups. In the six-month period, 50% of the patients in the CLVR M arm were admitted for 38 hospitalizations, of which 30 were for pulmonary causes. In the OBEST M arm, 9% of the patients were hospitalized. In the CLVR surgical arm, 66.7% of patients were readmitted for 27 hospitalizations and in the OBEST surgical arm, 12.5% were readmitted for three hospitalizations. One RCT 20 showed that there was decreased dependence on supplemental oxygen in the LVRS group compared with the M group. Two other RCTs (CLVR and OBEST) did not demonstrate a significant difference in oxygen use between the two groups. Data comparing adverse events in the LVRS and M groups were unavailable. Data related to hospitalization and complications of surgery for the six smaller RCTs are presented in Table 8. Of the six, two RCTs 16,21 and two unpublished RCTs (JM, unpublished observations, 2004) reported length of hospital stay as mean or median values that ranged from 12 to 22 days. Three RCTs 16,17,21 reported the number of prolonged air leaks, which ranged from 3 to 11 (i.e., 13% to 37%). Other complications included infection, bleeding, pneumonia, transient ischemic attack, transient confusion, transient Horner syndrome, small bowel ileus, benign dysrhythmia and atrial fibrillation (Table 8). 13

26 Variable Hospital days Emergency room visits Table 7: Data on use of health care resources (from NETT) Period after Randomization (months) LVRS Group Number Mean (95% CI) of Patients M Group Number Mean (95% CI) of Patients p Value 0 to (22.3, 27.6) (4.0, 5.8) < to (2.3, 4.1) (4.5, 7.6) to (2.3, 5.8) (3.8, 6.7) to (0.5, 0.7) (0.6, 0.9) to (0.4, 0.6) (0.6, 0.8) to (0.4, 0.6) (0.5, 0.8) to (9.5, 11.2) (7.8, 9.4) to (4.4, 5.6) (4.2, 5.5) 0.49 Days of ambulatory care 25 to (3.8, 5.2) (3.5, 5.2) 0.43 Claims for 0 to (6.2, 7.3) (6.6, 7.7) 0.19 supplemental 13 to (5.2, 6.4) (5.9, 7.1) 0.09 oxygen 25 to (5.1, 6.6) (4.8, 6.3) 0.39 CI=confidence interval Interim analysis of NETT identified subgroup of 140 patients [patients with FEV 1 (% predicted) 20% and homogeneous emphysema or DLCO (% predicted) 20%] with high risk of death and little chance of improved function after LVRS. Patients in this subgroup became ineligible for enrolment in the trial as of May 2001 and are excluded here. Four RCTs 16-18,21 and two unpublished RCTs (JM, unpublished observations, 2004) reported mortality data (Appendix 8). One RCT 15 reported mortality data only after crossover and hence was excluded in Appendix 8. The Forest plot (Figure 4) shows the risk of death at different times after surgery. At three months post-surgery, the relative risk (RR) of death is significantly greater in the LVRS group compared with the M group [RR (95% CI) was 5.71 (2.89, 11.26)]. With increasing time up to 12 months, the RR seems to be greater in the LVRS group but is not significantly different. The single large trial 18 shows that the risk of death in the two groups at a mean follow-up of 29 months is practically the same [RR (95% CI) was 0.98 (0.81, 1.19)]. The risk of death, i.e., RR, in different patient categories in the NETT is shown in Table 9. Patients with FEV 1 (% pr) 20% and either homogeneous emphysema or DLCO (% pr) 20% are at very high risk of death when undergoing LVRS [RR (95% CI) for 90-day mortality was (2.53, )] when compared with patients treated with M. Patients with non-upperlobe emphysema and high exercise capacity are at high risk of death when undergoing LVRS [RR (95% CI) for 90-day mortality was (1.47, 85.29)]. Risk of death is low in patients with upper-lobe emphysema and low exercise capacity [RR (95% CI), 0.87 (0.24, 3.17) and 0.55 (0.37, 0.84) at 90-day and at mean follow-up of 29 months respectively]. 14

27 Table 8: Hospitalization and complications associated with LVRS (data from RCTs) 15 Trial Number of Patients Length of Hospital Stay (days) Number (%) of Prolonged Air Leaks Number of Readmissions Number of Reoperations Number of Infections Number of Bleedings CLVR, (4 to 161) c NR 27 NR NR NR NR 2004 * Criner et al., (13 from NR NR 2 NR NR NR Pneumonia (2), ER visit (2) crossover) Geddes et (18 to 64) d 3 NR NR 2 NR NR al., Goldstein et al., NR TIA (1), transient confusion (6), small bowel ileus (2), benign dysrhythmia (6) OBEST, (4 to 57) c NR 3 NR NR NR NR 2004 * Pompeo et al., ±7.1 a 11 NR NR NR NR Atrial fibrillation (3), pneumonia (2), TIA (1), empyema (1), transient Horner syndrome (1) a Mean±SD, b median (interquartile range), c median (range), d mean (range), TIA=transient ischemic attack, NR=not reported, * JM, unpublished observations, Other 15

28 Figure 4: RR of death for patients undergoing LVRS as compared to those on M alone, at three, six, 12 and 29 months Study Control RR (random) Weight RR (random) or sub-category n/n n/n 95% CI % 95% CI 3 months Geddes, /24 1/ [0.48, 33.22] Goldstein, /28 0/ [0.24, 96.16] NETT, /608 8/ [2.87, 12.62] Subtotal (95% CI) [2.89, 11.26] Total events: 54 (treatment), 9 (control) Test for heterogeneity: chi²=0.14, df=2 (p=0.93), I²=0% Test for overall effect: Z=5.02 (p< ) 16 6 months Geddes, /24 1/ [0.48, 33.22] Pompeo, /30 1/ [0.19, 20.90] Goldstein, /28 1/ [0.19, 20.05] CLVR, /31 4/ [0.35, 3.93] OBEST, /24 1/ [0.09, 9.07] Subtotal (95% CI) [0.68, 3.56] Total events: 15 (treatment), 8 (control) Test for heterogeneity: chi²=1.28, df=4 (p=0.87), I²=0% Test for overall effect: Z=1.05 (p=0.30) 12 months Geddes, /24 2/ [0.54, 11.65] Goldstein, /28 1/ [0.46, 32.35] Subtotal (95% CI) [0.83, 10.10] Total events: 9 (treatment), 3 (control) Test for heterogeneity: chi²=0.11, df=1 (p=0.75), I²=0% Test for overall effect: Z=1.67 (p=0.09) 29 months NETT, / / [0.81, 1.19] Subtotal (95% CI) [0.81, 1.19] Total events: 157 (treatment), 160 (control) Test for heterogeneity: not applicable Test for overall effect: Z=0.16 (p=0.87) RR=relative risk Favours LVRS Favours M 16