Summary
Rationale:
While exercise capacity, expressed as maximal oxygen consumption (VO2max), has been proposed to be the best predictor of postoperative cardiopulmonary complications after surgical resection in lung cancer patients, the literature remains controversial.
The purpose of this study was to use the meta-analytic approach to determine if VO2max, expressed as either ml/kg/min or as a percentage of predicted, differed between patients who develop postoperative cardiopulmonary complications versus those that do not.
Methods:
Studies were retrieved via (1) computerized literature searches, (2) cross referencing from retrieved articles, and (3) expert review of our reference list. Trials were included if they reported preoperative VO2max values (ml/kg/min or percentage of predicted) and had patients in which postoperative cardiopulmonary complications occurred.
Results:
Fourteen studies representing a total of 955 men and women met our criteria for inclusion. Across all designs and categories, random-effects modeling demonstrated that patients without post operative pulmonary complications had significantly higher levels of VO2max in ml.kg−1.min−1 ( mean difference = 3.0, 95% CI, 1.9 to 4.0) as well as VO2max as a percentage of predicted ( mean difference = 8, 95% CI, 3.3 to 12.8).
Conclusion:
After a systematic review of the literature we found that exercise capacity expressed as VO2max, is lower in patients that develop clinically relevant complications after curative lung resection. These results are important for the practicing clinician because answers the literature controversy on the usefulness of measuring pre-operative exercise capacity and reinforce the current guidelines on decision making for lung resection.
INTRODUCTION
Lung resection is the only curative treatment for lung cancer. In almost every case, an effort is made to provide a curative surgical resection. While outcomes from lung resection have improved over the years, there is still a high rate of morbidity and mortality as a result of postoperative cardiopulmonary complications (PPC)(1-3).
Postoperative cardiopulmonary complications include (1) respiratory failure (Acute respiratory distress syndrome (ARDS), prolonged postoperative mechanical ventilation or reintubation), (2) pneumonia, (3) atelectasis requiring bronchoscopy, (4) myocardial infarction, and (5) arrhythmias requiring intravenous treatment.
Several studies which have found that exercise capacity is the best predictor of PPC have also proposed specific cut points for maximal oxygen consumption (VO2max) that discriminate whether patients will develop PPC(1, 3-7). This is usually displayed using Receiver Operating Characteristic (ROC) curves from large studies. In addition, exercise capacity is included in an algorithm for deciding whether patients with borderline lung function should undergo thoracic surgery(8, 9).
While the previously mentioned studies have found that exercise capacity is a strong predictor of PPC, others have not(10-12) given the controversy between the aforementioned studies, a need exists to examine whether exercise capacity, as measured by VO2max from an incremental test, differs between those patients that develop PPC and those patients that do not develop PPC when systematically combining all published literature. We performed the following meta-analysis to address this clinically relevant controversy.
METHODS
Data Sources
We searched MEDLINE (March 1, 1966 to February 28, 2005) and EMBASE (March, 1974 to February, 2005) using the terms lung/surgery, lung resection/cancer, bronchogenic carcinoma, thoracotomy, pneumonectomy, exercise test and postoperative pulmonary complications. We augmented our search by reviewing the reference lists of retrieved articles, including review articles, as well as the reference lists of related articles in our files. A medical librarian performed an independent search to ensure completeness. The search was not limited to English language studies but was limited to published reports.
Study Identification and Eligibility
We attempted to identify all published studies that used VO2max to determine exercise capacity in the immediate preoperative period prior to lung resection and documented cardiopulmonary complications in the first 30 days after surgery. To be included in the analysis, studies had to have (1) a maximal exercise test during the preoperative period prior to lung resection for lung cancer; (2) reported VO2max in the form of % predicted, ml.kg−1.min−1 or liters per minute; (3) defined PPC as one or more of the following: atelectasis requiring bronchoscopy, respiratory failure, prolonged mechanical ventilation, pneumonia documented by objective criteria that include chest x-ray and laboratory values and symptoms; (4) provided data on the physiologic characteristics of the two groups, i.e., those with and without PPC.
The reason for choosing specific postoperative pulmonary complications in this study is because of their clinical meaningfulness and the way (almost uniformly) those complications were presented in the publications reviewed. The manuscripts that entered the meta-analysis presented the two groups: complicated patients and uncomplicated patients (control) with their respective measurements and outcomes which made the analysis possible.
Data extraction
Two people independently coded the studies. Disagreements were resolved by consensus.
Data synthesis
The two primary outcomes in this study were VO2max in ml.kg−1.min−1 and VO2max as a percentage of predicted. These were calculated by taking the difference in preoperative exercise capacity between the group that had PPC and the group that did not have PPC (mean value differences of the group without PPC minus the group with PPC). Pooled differences using the original metric were calculated by assigning weights equal to the inverse of the variance for the net changes in VO2max. Ninety-five percent confidence intervals (CI) were generated around the mean treatment effect to establish statistical significance. We used the Dersimonian and Laird random effects model for all analyses(13). Heterogeneity (variation in study outcomes between studies), based on a fixed-effects model, was assessed using Cochran's Q statistic(14). In addition, we used the recently developed I2 statistic to determine the percentage of total variation across the studies due to heterogeneity(15). Generally, I2 values of 25%, 50%, and 75% are considered to be indicative of small, moderate, and large amounts of heterogeneity(15). Publication bias was examined using the regression approach of Egger et al. (15). In addition, sensitivity analysis was performed with each study deleted from the model once in order to assess the impact of each study on our overall results. We explored possible reasons for heterogeneity by conducting subgroup analysis with the following types of studies deleted from the model: (1) studies that did not use cycle ergometers (2) studies more than 15 years old, and (3) studies that had-fewer than 40 subjects. All data were analyzed using SPSS (version 13.0)(16) and Stata/SE 8.2(17). Given the design of the study there was no patient consenting or Review Board approval.
RESULTS
Study Selection Results and Study Characteristics
A flow diagram for the selection of studies is shown in Figure 1. As can be seen, of the 329 studies initially identified, 14 (4%), representing a total of 955 subjects, met our criteria for inclusion. A general description of the included studies is shown in Table 1. All studies were published in refereed journals. Of the 14 studies included in the analysis, all 14 provided adequate data for VO2max in ml.kg−1.min−1 while 11 provided data for VO2max as a % of predicted.
Figure 1.
Flow diagram of study selection.
Table 1.
Characteristics of included studies.
| Study(ref#) | Subjects | Criteria for Complications |
|---|---|---|
| Bechard et al. (6) |
50 men, (12WR , 28L , 10 P) for diagnosed or suspected lung carcinoma |
Acute CO2 retention, prolonged mechanical ventilation, arrhythmia, MI, pneumonia, purulent sputum, PE, atelectasis, death |
| Bolliger et al. (1) |
57 men and 23 women (14WR, 45 L, 21 P) ; 62 had clinically resectable malignancies, 12 benign disorders, 6 carcinoid tumors |
PCO2 > 45, mechanical ventilation > 48 hours, cardiac arrhythmias, myocardial infarction, pneumonia, PE, atelectasis, death |
| Boysen et al. (12) |
15 men and 2 women (7WR, 8L, 2P); known or suspected lung cancer |
PaCO2>45 mmHg, arrhythmia, need for mechanical ventilation > 48 hrs postop, MI, pneumonia, PE, atelectasis, death w/in 30 days |
| Brunelli et al. (19) |
128 men, 32 women (21WR, 111L, 28P); non-small-cell lung carcinoma |
Respiratory failure requiring mechanical ventilation > 48 h, pneumonia, atelectasis, pulmonary edema, PE, MI, arrhythmia, cardiac failure, death |
| Brutsche et al. (4) |
101 men and 24 women (36WR, 91 L, 18 P); lung cancer |
PaCO2>45 mmHg, prolonged mech. ventilation, arrhythmia, MI, pneumonia, PE, atelectasis, death |
| Epstein et al (20) |
41 men, 1 woman with lung cancer |
MI, unstable angina, CHF, arrhythmia, reintubation prolonged ventilation, pneumonia, atelectasis, high PaCO2, PE, death |
| Larsen et al (7) |
97 patients with bronchogenic carcinoma (18 T, 52 L, 27 P) |
Prolonged mechanical ventilation, PO2<55 or PCO2>65 at room air, MI, pneumonia, arrhythmia , heart failure, death |
| Markos et al (10) |
53 men and women; with suspected lung malignancy (6 T, 29 L, 18 P) |
Death, respiratory failure, pneumonia, atelectasis, PE, MI or ischemia, arrhythmia, admission to ICU or CCU |
| Morice et al.(21) |
7 men and 1 woman; pulmonary lesion consistent with clinical stage 1 lung cancer (4L, 4WR) |
Mechanical ventilation > 48 h, MI, arrhythmia, pneumonia, atelectasis, PE, death |
| Smith et al. (5) |
19 men and 3 women; suspected or diagnosed lung cancer (5 T, 1 WR, 12 L, 4 P) |
Respiratory failure (PaCO2>45), mechanical ventilation >48 h, MI, arrhythmia, pneumonia, atelectasis, PE, death |
| Torchio et al. (18) |
48 men, 8 women (3 T, 28 L, 23 P) |
Death, respiratory insufficiency with intubation for 48 h, PE |
| Villani et al. (22) |
141 men 9 women, lung carcinoma (procedure not reported) |
Respiratory failure requiring O2, atelectasis, arrhythmia, pneumonia, ARDS, PE, death. |
| Wang et al. (11) |
29 men, 11 women; 35 had non- small cell lung cancer, 5 had other reasons for thoracotomy (9 WR, 31 L) |
Ventilation >24 h, reintubation for respiratory failure, PCO2>45, pneumonia, atelectasis, need for O2 at discharge, death. |
| Wang et al. (3) |
57 men and women; non-small cell lung cancer (3 T, 10 WR, 34 L, 10 P) |
Ventilatory support >48 h, reintubation, PE, pneumonia, atelectasis, respiratory insufficiency (PCO2>45), death. |
Notes: FEV1, forced expiratory volume (liters); W, watts; mph, miles per hour; kpm, kilometers per hour; CPET, cardiopulmonary exercise testing; PE, pulmonary embolism; CO2, carbon dioxide; PCO2, partial pressure of carbon dioxide; mmHg, millimeters of mercury; GI, gastrointestinal; ICU, intensive care unit; CCU, coronary care unit; O2, oxygen; ARDS, adult respiratory distress syndrome; HR, heart rate; MI, myocardial infarction; WR, wedge resections; L, Lobectomy; P, Pneumonectomy.
Exercise assessment characteristics
Two studies reported the assessment of exercise using treadmills(12, 18) one study used stair climbing(19), while all others used a cycle ergometer(1, 3-7, 10, 11, 20-22).
Differences in primary outcomes
VO2max in ml.kg−1.min−1
Differences in VO2max in ml.kg−1.min−1 for each study are shown in Table 2 and Figure 2 while overall differences are shown in Table 3. As can be seen, subjects with no PPC had significantly higher levels of VO2max in ml.kg−1.min−1 when compared to subjects with PPC. This was equivalent to an absolute difference of approximately 3 ml.kg−1.min−1.
Table 2.
Primary outcomes for each study.
| VO2max (ml.kg−1min−1) | VO2max (% predicted) | |||||
|---|---|---|---|---|---|---|
| Study (ref#) | N | PPC (%) | No PPC | PPC | No PPC | PPC |
| Bechard et al. (6) | 50 | 14 | 17(2) | 9.9(9) | NA | NA |
| Bolliger et al. (1) | 80 | 20 | 19(5) | 14(3) | 84(19) | 61(11) |
| Boysen wt al. (12) | 17 | 12 | 20(5) | 16(3) | 62(12) | 60(3) |
| Brunelli et al. (19) | 160 | 14 | 25(4) | 23(4) | 111(22) | 113(20) |
| Brustche et al (4) | 125 | 25 | 22(5) | 17(5) | 78(20) | 66(20) |
| Epstein et al. (20) | 42 | 33 | 16(4) | 16(5) | NA | NA |
| Larsen et al. (7) | 97 | 32 | 19(4) | 18(4) | 84(17) | 77(20) |
| Markos et al. (10) | 53 | 30 | 17(6) | 16(6) | 69(20) | 75(28) |
| Morice et al. (21) | 8 | 25 | 17(2) | 15(0.7) | 67(10) | 62(9) |
| Smith et al. (5) | 22 | 50 | 22(4) | 14(2) | 73(17) | 55(9) |
| Torchio et al. (18) | 54 | 51 | 23(4) | 19(1) | 95(19) | 89(7) |
| Villani et al. (22) | 150 | 29 | 21(4) | 19(3) | 75(13) | 71(11) |
| Wang et al. (11) | 40 | 33 | 17(4) | 16(4) | NA | NA |
| Wang et al. (3) | 57 | 33 | 19(4) | 15(2) | 70(13) | 57(14) |
| Overall X̄(SD) | 955 | 28(11) | 20(2) | 16(3) | 80(13) | 72(13) |
Note: N, number of patients in the study; X̄(SD), mean ± standard deviation; %PPC, percentage of patients in the study that developed postoperative pulmonary complications; PPC, postoperative pulmonary complications; NA, not available.
Figure 2.
Forest Plot for changes in VO2max in ml.kg−1.min−1 using a random effects model. The black boxes, sized relative to random-effects weighting, represent the mean change in VO2max in ml.kg−1.min−1 for each study while the lines represent the 95% confidence intervals. The diamond and dashed lines represent the overall mean change in VO2max in ml.kg−1.min−1 across all listed studies while the left and right ends of the diamond represent the 95% confidence interval for all studies combined.
Table 3.
Weighted differences in primary outcomes and secondary outcomes.
| NPPC | PPC | RE Model | ||||
|---|---|---|---|---|---|---|
| Variable | N | X̄ (SD) | X̄ (SD) | X̄(95% CI) | Q(p value) | I2 |
| Primary outcomes | ||||||
| VO2max (ml.kg−1.min−1) | 14 | 20.0(2.7) | 16.8(3.1) | 3.0(2.0 to 4.1)* | 34.9(<0.001)* | 62% |
| VO2max (% predicted) | 11 | 80.9(13.8) | 72.9(17.6) | 8.1(3.3 to 12.8)* | 35.1(<0.001)* | 72% |
| Secondary outcomes | ||||||
| - Age (years) | 12 | 61.0(3.9) | 65.4(4.9) | −3.9(−6.3 to −1.5)* | 25.0(0.007)* | 56% |
| - Peak watts (Wp) | 9 | 110(22.0) | 97(16.0) | 13.4(7.1 to 19.7)* | 12.0(0.15) | 33% |
| - FEV1 (% predicted) | 11 | 78(13.0) | 72(10.0) | 4.8(1.6 to 8.1)* | 13.8(0.18) | 27% |
| - DLCO (% predicted) | 10 | 77(11.0) | 69(11.0) | 9.4(2.6 to 16.1)* | 24.2(0.003)* | 62% |
Notes: N, number of studies; NPPC, no postoperative pulmonary complications; PPC, postoperative pulmonary complications; RE, random-effects model; Q, heterogeneity statistic based on the fixed-effects model; p, alpha value for Q; I2, percentage of heterogeneity, calculated as (Q−df)/Q; X̄ (SD), mean ± standard deviation; X̄ (95% CI), mean and 95% confidence interval for the mean; VO2max, maximum oxygen consumption; FEV1, functional expiratory volume; DLCO, diffusing lung capacity;
statistically significant; Differences between NPPC and PPC for RE model weighted.
Statistically significant and moderate heterogeneity was observed. No statistically significant publication bias was observed (p > 0.05). With each study deleted from the model once, results remained statistically significant. In addition, we also explored potential sources of heterogeneity with the following types of studies deleted from the model: (1) studies that did not use cycle ergometers, (2) studies more than 15 years old, and (3) studies that had fewer than 40 subjects. However, no statistically significant differences were observed for any of these analyses (p > 0.05 for all).
VO2max (% predicted)
Differences in VO2max as a percentage of predicted for each study is shown in Table 2 and Figure 3 while overall differences are shown in Table 3. As can be seen, subjects with no PPC had significantly higher levels of VO2max as a percentage of predicted when compared to subjects with PPC. Statistically significant and moderate heterogeneity was observed for our overall outcomes. However, no statistically significant publication bias was observed (p > 0.05). With each study deleted from the model once, results remained statistically significant. However, when we performed our analysis with the study by Bolliger et al, deleted from the model(1), no statistically significant heterogeneity was observed (p > 0.05). No statistically significant differences were observed when the following types of studies were deleted from the model: (1) studies that did not use cycle ergometers, (2) studies more than 15 years old, and (3) studies that had fewer than 40 subjects (p > 0.05 for all).
Figure 3.
Forest Plot for changes in VO2max (% predicted) using a random effects model. The black boxes, sized relative to random-effects weighting, represent the mean change in VO2max (% predicted) for each study while the lines represent the 95% confidence intervals. The diamond and dashed lines represent the overall mean change in VO2max (% predicted) across all listed studies while the left and right ends of the diamond represent the 95% confidence interval for all studies combined.
Secondary Outcomes
Age
Differences in age are shown in Table 3. As can be seen, subjects without PPC were approximately 4 years younger than subjects with PPC. Moderate but statistically significant heterogeneity was found.
Peak watts (Wp)
Differences in Wp are shown in Table 3. Those subjects without PPC had Wp values that were significantly higher than those with PPC. This was equivalent to difference of 12 watts in favor of subjects without PPC. No statistically significant heterogeneity was found nor was any statistically significant publication bias observed (p > 0.05). With each study deleted from the model once results remained statistically significant.
FEV1 (% predicted)
Differences in FEV1 as a percentage of predicted are shown in Table 3. When compared to subjects with PPC, statistically significant and higher FEV1 values were found in those subjects that did not have PPC. No statistically significant heterogeneity was observed. With each study deleted from the model once, results remained statistically significant. No statistically significant publication bias was found (p > 0.05).
DLCO (% predicted)
Differences in DLCO as a percentage of predicted are shown in Table 3. As can be seen, DLCO values were higher in patients without PPC versus those with PPC. Statistically significant and moderate heterogeneity was observed for differences in DLCO between the two groups. With each study deleted from the model once, results remained statistically significant. No statistically significant publication bias was found (p > 0.05).
DISCUSSION
The overall results of this meta-analysis show that exercise capacity is lower in patients who developed PPC compared to those that did not. From a clinical perspective, we believe our results are clinically important because they clarify the controversy in the existing literature.
Although exercise capacity is a significant predictor of different outcomes for COPD(23, 24) its ability to predict cardiopulmonary complications in lung cancer resection patients (often carrying the diagnosis of COPD) has been controversial. It is in this kind of situation where pooled estimates from combinable studies, i.e., meta-analysis, can help identify whether any differences exist. In our study, we found that higher exercise capacity was associated with fewer PPC. Given the intent of our meta-analysis, it was not surprising that no randomized controlled trials (RCT) on this topic could be located since a randomized design is not ethically feasible. Consequently, we included observational studies which tend to have more heterogeneity (variation in study outcomes between studies) than randomized controlled trials(25).
While the random-effects model controls for statistically significant heterogeneity, we also examined for heterogeneity using the fixed-effects model (Table 3). As a result, we found statistically significant heterogeneity for both VO2max in ml.kg−1min−1 as well as VO2max as a percentage of predicted. While we could not identify any potential sources of heterogeneity for VO2max in ml.kg−1min−1 we did find that the results became homogeneous for VO2max as a percentage of predicted when the study by Bolliger et al(1) was deleted from the model. That study found a higher difference in exercise capacity between patients with and without PPCs. We believe that our results are valid despite the fact that we could not explain all heterogeneity. In addition, it is important to realize that both our random and fixed-effects analyses yielded similar and statistically significant differences for all of our reported outcomes (indicating the stability of the finding). These results support the notion that exercise capacity is greater in patients without PPC.
The mean VO2max in ml.kg−1.min−1 of 20 ml/kg/min across all studies (table 2) for non-complicated patients is consistent with the threshold proposed for patients with no risk of complications (5, 6, 18). The mean value for complicated patients (table 2) is also consistent with the currently proposed threshold for increased risk of 15 VO2max in ml.kg−1.min−1 (3, 5, 9). Our findings support the currently proposed thresholds that are based on individual studies with smaller number of patients.
The analysis of FEV1 and DLCO also showed significant diferences between the two groups. However this finding should not be viewed as those two variables being good predictors of complications. After the extensive search performed, we found that FEV1 is not consistently reported as an independent predictor of PPC. In addition, the difference found in FEV1 between the complicated and uncomplicated groups (4%) (table 3) is unlikely to be clinically meaningful. DLCO has been shown to be an independent predictor of PPC but in a minority of the publications(3, 11). FEV1 and DLCO are currently used as screening methods to detect patients with increased risk of complications but once the patients with increased risk are defined, an exercise capacity measurement is needed to further improve the risk assessment. (8, 9)
We believe that our results best reflect the evidence that currently exists regarding differences in exercise capacity between patients with PPC versus those without PPC. The former notwithstanding, we acknowledge that a meta-analysis of individual patient data (IPD) might have been more appropriate since permits the generation of ROC analysis that can provide cut points in the desired outcome that discriminates the population that developed complications from that one that not. However, given the difficulty in retrieving IPD from investigators (26) we chose to conduct a summary means meta-analysis. In addition, some studies had to be excluded because of the different criteria they used for PPC. Thus, this may be viewed as potential bias. However, even when those studies were included, they had no effect on any of our findings (results not shown). Our results using the original metric for our principal outcomes and also for the other outcomes variables (VO2max in ml.kg−1.min−1, VO2max as a percentage of predicted, Wp, FEV1, DLCO) facilitate interpretability(27) and suggest clinically important differences since those differences are anchored to clinically meaningful events (PPC). The “anchor method” is used to define the minimal clinically important difference in an outcome, by establishing the difference in that outcome is associated with a clinically meaningful event(28). For example, the 13 Wp difference we found approximates the 10 Wp cut-point considered by experts as well as reports from clinical trials as the minimally clinically important difference for that parameter(29). Finally we recognize that while cardiopulmonary complications are clinically important, surgery still remains the best opportunity for cure. Most would accept even severe complications for a chance at long term survival. With knowledge of differences that separate groups with different prognoses, interventions can be directed to overcome that difference and improve patients' prognoses. Since low exercise capacity is a potentially modifiable risk factor, we posit that an intervention that improves exercise capacity, like pulmonary rehabilitation, could improve patient outcome (fewer or-no PPC). The National Cancer Institute is currently funding a study to address that hypothesis(30). Also the National Emphysema Treatment Trial clearly showed that in some patients, exercise capacity improved sufficiently after rehabilitation to change the risk category of the patients(29).
In summary, our results suggest, after reviewing all available literature, that exercise capacity, measured as VO2max, is lower in patients that develop clinically relevant PPC after curative lung resection. This study is useful for the practicing clinician as it reinforces the current guidelines addressing the importance of assessing exercise capacity as part of the decision making process for the surgical treatment of lung cancer.
Acknowledgments
Dr Benzo is supported by grant #1K23CA106544 from the National Cancer Institute.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Bolliger CT, Jordan P, Soler M, Stulz P, Gradel E, Skarvan K, et al. Exercise capacity as a predictor of postoperative complications in lung resection candidates. Am J Respir Crit Care Med. 1995;151(5):1472–80. doi: 10.1164/ajrccm.151.5.7735602. [DOI] [PubMed] [Google Scholar]
- 2.Wang JS. Pulmonary function tests in preoperative pulmonary evaluation. Respir Med. 2004;98(7):598–605. doi: 10.1016/j.rmed.2004.01.007. [DOI] [PubMed] [Google Scholar]
- 3.Wang JS, Abboud RT, Evans KG, Finley RJ, Graham BL. Role of CO diffusing capacity during exercise in the preoperative evaluation for lung resection. Am J Respir Crit Care Med. 2000;162(4 Pt 1):1435–44. doi: 10.1164/ajrccm.162.4.2001117. [DOI] [PubMed] [Google Scholar]
- 4.Brutsche MH, Spiliopoulos A, Bolliger CT, Licker M, Frey JG, Tschopp JM. Exercise capacity and extent of resection as predictors of surgical risk in lung cancer. Eur Respir J. 2000;15(5):828–32. doi: 10.1034/j.1399-3003.2000.15e03.x. [DOI] [PubMed] [Google Scholar]
- 5.Smith TP, Kinasewitz GT, Tucker WY, Spillers WP, George RB. Exercise capacity as a predictor of post-thoracotomy morbidity. Am Rev Respir Dis. 1984;129(5):730–4. doi: 10.1164/arrd.1984.129.5.730. [DOI] [PubMed] [Google Scholar]
- 6.Bechard D, Wetstein L. Assessment of exercise oxygen consumption as preoperative criterion for lung resection. Ann Thorac Surg. 1987;44(4):344–9. doi: 10.1016/s0003-4975(10)63787-3. [DOI] [PubMed] [Google Scholar]
- 7.Richter Larsen K, Svendsen UG, Milman N, Brenoe J, Petersen BN. Exercise testing in the preoperative evaluation of patients with bronchogenic carcinoma. Eur Respir J. 1997;10(7):1559–65. doi: 10.1183/09031936.97.10071559. [DOI] [PubMed] [Google Scholar]
- 8.Diagnosis and management of lung cancer: ACCP evidence-based guidelines. American College of Chest Physicians. Chest. 2003;123(1 Suppl):D-G, 1S–337S. [PubMed] [Google Scholar]
- 9.BTS guidelines: guidelines on the selection of patients with lung cancer for surgery. Thorax. 2001;56(2):89–108. doi: 10.1136/thorax.56.2.89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Markos J, Mullan BP, Hillman DR, Musk AW, Antico VF, Lovegrove FT, et al. Preoperative assessment as a predictor of mortality and morbidity after lung resection. Am Rev Respir Dis. 1989;139(4):902–10. doi: 10.1164/ajrccm/139.4.902. [DOI] [PubMed] [Google Scholar]
- 11.Wang J, Olak J, Ultmann RE, Ferguson MK. Assessment of pulmonary complications after lung resection. Ann Thorac Surg. 1999;67(5):1444–7. doi: 10.1016/s0003-4975(99)00255-6. [DOI] [PubMed] [Google Scholar]
- 12.Boysen PG, Clark CA, Block AJ. Graded exercise testing and postthoracotomy complications. J Cardiothorac Anesth. 1990;4(1):68–72. doi: 10.1016/0888-6296(90)90449-p. [DOI] [PubMed] [Google Scholar]
- 13.DerSimonian R. Meta-analysis in the design and monitoring of clinical trials. Stat Med. 1996;15(12):1237–48. doi: 10.1002/(SICI)1097-0258(19960630)15:12<1237::AID-SIM301>3.0.CO;2-N. discussion 49-52. [DOI] [PubMed] [Google Scholar]
- 14.Cochran WG. Research techniques in the study of human beings. Milbank Mem Fund Q. 1955;33(2):121–36. [PubMed] [Google Scholar]
- 15.Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–60. doi: 10.1136/bmj.327.7414.557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.SPSS for Windows. 13.0 ed. SPSS, Inc.; Chicago, IL: 2004. [Google Scholar]
- 17.Stata/SE for windows. 8.2 ed. Sata Corporation LP; College Station, TX: 2005. [Google Scholar]
- 18.Torchio R, Gulotta C, Parvis M, Pozzi R, Giardino R, Borasio P, et al. Gas exchange threshold as a predictor of severe postoperative complications after lung resection in mild-to-moderate chronic obstructive pulmonary disease. Monaldi Arch Chest Dis. 1998;53(2):127–33. [PubMed] [Google Scholar]
- 19.Brunelli A, Al Refai M, Monteverde M, Borri A, Salati M, Fianchini A. Stair climbing test predicts cardiopulmonary complications after lung resection. Chest. 2002;121(4):1106–10. doi: 10.1378/chest.121.4.1106. [DOI] [PubMed] [Google Scholar]
- 20.Epstein SK, Faling LJ, Daly BD, Celli BR. Predicting complications after pulmonary resection. Preoperative exercise testing vs a multifactorial cardiopulmonary risk index. Chest. 1993;104(3):694–700. doi: 10.1378/chest.104.3.694. [DOI] [PubMed] [Google Scholar]
- 21.Morice RC, Peters EJ, Ryan MB, Putnam JB, Ali MK, Roth JA. Exercise testing in the evaluation of patients at high risk for complications from lung resection. Chest. 1992;101(2):356–61. doi: 10.1378/chest.101.2.356. [DOI] [PubMed] [Google Scholar]
- 22.Villani F, Busia A. Preoperative evaluation of patients submitted to pneumonectomy for lung carcinoma: role of exercise testing. Tumori. 2004;90(4):405–9. doi: 10.1177/030089160409000408. [DOI] [PubMed] [Google Scholar]
- 23.Celli BR, Cote CG, Marin JM, Casanova C, Montes de Oca M, Mendez RA, et al. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med. 2004;350(10):1005–12. doi: 10.1056/NEJMoa021322. [DOI] [PubMed] [Google Scholar]
- 24.Oga T, Nishimura K, Tsukino M, Sato S, Hajiro T. Analysis of the factors related to mortality in chronic obstructive pulmonary disease: role of exercise capacity and health status. Am J Respir Crit Care Med. 2003;167(4):544–9. doi: 10.1164/rccm.200206-583OC. [DOI] [PubMed] [Google Scholar]
- 25.Greenland S. Quantitative methods in the review of epidemiologic literature. Epidemiol Rev. 1987;9:1–30. doi: 10.1093/oxfordjournals.epirev.a036298. [DOI] [PubMed] [Google Scholar]
- 26.Simmonds MC, Higgins JP, Stewart LA, Tierney JF, Clarke MJ, Thompson SG. Meta-analysis of individual patient data from randomized trials: a review of methods used in practice. Clin Trials. 2005;2(3):209–17. doi: 10.1191/1740774505cn087oa. [DOI] [PubMed] [Google Scholar]
- 27.Kelley GA, Tran ZV. Original metric vs. standardized effect sizes for meta-analysis of clinical data. Prev Cardiol. 2001;4(1):40–5. doi: 10.1111/j.1520-037x.2001.00812.x. [DOI] [PubMed] [Google Scholar]
- 28.Wyrwich KW, Bullinger M, Aaronson N, Hays RD, Patrick DL, Symonds T. Estimating clinically significant differences in quality of life outcomes. 2005;14(2):285–95. doi: 10.1007/s11136-004-0705-2. [DOI] [PubMed] [Google Scholar]
- 29.Ries AL, Make BJ, Lee SM, Krasna MJ, Bartels M, Crouch R, et al. The effects of pulmonary rehabilitation in the national emphysema treatment trial. Chest. 2005;128(6):3799–809. doi: 10.1378/chest.128.6.3799. [DOI] [PubMed] [Google Scholar]
- 30.Benzo R. Pulmonary rehabilitation before lung cancer resection. [cited 2006]; Available from: www.cancer.gov.