Meta-analysis of respiratory rehabilitation in chronic obstructive pulmonary disease

Transcription

1 Articles Meta-analysis of respiratory rehabilitation in chronic obstructive pulmonary disease Yves Lacasse, Eric Wong, Gordon H Guyatt, Derek King, Deborah J Cook, Roger S Goldstein Summary Background Respiratory rehabilitation is increasingly recognised as an important part of the management of patients with chronic obstructive pulmonary disease (COPD). The widespread application of such programmes should be preceded by evidence of directly attributable improvements in function. We assessed the effect of respiratory rehabilitation on exercise capacity and healthrelated quality of life (HRQL) in patients with COPD. Methods We carried out a meta-analysis of randomised controlled trials of respiratory rehabilitation in patients with COPD that assessed functional or maximal exercise capacity, HRQL, or both. Respiratory rehabilitation was defined as exercise training (for at least 4 weeks) with or without education,, or both. The most commonly used measure for HRQL was the chronic respiratory questionnaire, in which responses were presented on a 7-point scale. The control groups received no rehabilitation. Within each trial and for each outcome an effect size was calculated; the effect sizes were then pooled by a random-effects model. The overall effect of treatment was compared with its minimum clinically important difference (MCID) defined as the smallest difference perceived as important by the average patient. Findings We included 14 trials. Significant improvements were found for all the outcomes. For two important features of HRQL, dyspnoea and mastery, the overall treatment effect was larger than the MCID: 1 0 (95% CI ) and 0 8 ( ), respectively, compared with an MCID of 0 5. For functional exercise capacity (6-min walk test), the overall effect was 55 7 m ( ), and for maximum exercise capacity (incremental cycle ergometer test), 8 3 W ( ). Functional exercise capacity showed heterogeneity that could not be explained by the sensitivity analyses. Interpretation Respiratory rehabilitation relieves dyspnoea and improves control over COPD. These improvements are clinically important. The value of the improvement in exercise capacity is not clear. Respiratory rehabilitation is an effective part of care in patients with COPD. Lancet 1996; 348: See Commentary page 1111 Departments of Clinical Epidemiology and Biostatistics (Y Lacasse MD, E Wong MD, G H Guyatt MD, D King BMath, D J Cook MD) and Medicine (G H Guyatt, D J Cook), McMaster University, Hamilton, Canada; and Department of Medicine, Division of Respiratory Medicine, University of Toronto, West Park Hospital, Toronto, Ontario M6M 25J, Canada (R S Goldstein MB) Correspondence to: Dr Roger S Goldstein Introduction Chronic obstructive pulmonary disease (COPD) is the fifth leading cause of mortality in north America and its prevalence continues to increase. 1 The associated loss of physical capacity and the adverse psychological effects of the disorder contribute greatly to morbidity. Official organisations in north America and Europe endorse respiratory rehabilitation as integral to the long-term management of COPD, 2,3 but reports of the benefits of this approach are mostly from uncontrolled trials and unsupervised therapy. Controlled trials have been limited by the lack of standard measurements of exercise tolerance and quality of life. Given the commitment asked of patients, their families, and health-care professionals, the interventions required must be justified by proof of improvement in exercise tolerance and quality of life. Moreover, if rehabilitation does benefit patients with COPD then it is important, before any widespread application, to know of the size of the treatment effect. To establish the influence and effect size of respiratory rehabilitation on functional exercise capacity, maximum exercise capacity, and health-related quality of life (HRQL) in patients with COPD, we undertook a meta-analysis of all randomised controlled trials in which rehabilitation, including systemic exercise for at least 4 weeks, was offered to patients with COPD and in which treated patients were compared with control patients whose care in the community did not include rehabilitation. Methods We searched Medline (1966 to October, 1995) and CINAHL (Cumulated Index to Nursing and Allied Health, 1982 to October, 1995) for original articles published in any language. For this search we used the following items: (exp, lung diseases, obstructive), (exp, rehabilitation or exp, exercise therapy), and (research design or longitudinal studies or evaluation study or randomised controlled trial). The reference lists of relevant articles were reviewed. In addition, abstracts presented at international meetings (American Thoracic Society, , European Respiratory Society, ) were searched. We contacted investigators of studies included in the meta-analysis and experts in respiratory rehabilitation to locate any unpublished material. We used the following criteria to select randomised controlled trials for inclusion in the meta-analysis. Target population We accepted trials in which more than 90% of patients had a clinical diagnosis of COPD and either a best recorded ratio of forced expiratory volume in 1 s (FEV 1 ) to forced vital capacity (FVC) of less than 0 7, or a best recorded FEV 1 of less than 70% of the predicted value. Respiratory rehabilitation programmes We included any inpatient, outpatient, or home-based rehabilitation programme of at least 4 weeks duration that included exercise therapy with or without any form of education, for patients with exercise limitation attributable to COPD, or both. Vol 348 October 26,

2 Trial Respiratory rehabilitation programme Outcome measures Quality of methods Setting Components Duration Exercise HRQL Concealed Blinded assessment capacity randomisation of outcomes McGavin, Home-based LLE Continuous 12 min walk test, Interviews Yes No Cockcroft, ,14 Inpatient LLE, ULE 6 weeks 12 min walk test, Interviews, POMS, Yes Yes ITT EPS Booker, Home-based LLE, BE, PD, education, 9 weeks 6 min walk test Anxiety/depression scale Not available Yes Jones, Home-based LLE, ULE 10 weeks 12 min walk test, Daily diary, Lubin, Yes 12 min walk test,, SSCET affectometer SSCET: yes; : no Busch, Home-based LLE, BE 18 weeks, multistep CRQ (dyspnoea only) Yes No stage test Lake, Outpatient LLE, ULE 8 weeks 6 min walk test, Bandura scale of Yes : yes, 6 min, IAET well-being walk test: no Simpson, Outpatient LLE, ULE 8 weeks 6 min walk test, CRQ Yes CRQ: yes; others: no, SSCET Weiner, Outpatient LLE, ULE, IMT, BE 6 months 12 min walk test, Not measured Yes Yes, SSCET Goldstein, Inpatient LLE, ULE, BE, education, 8 weeks 6 min walk test, CRQ, BDI/TDI Yes Yes, SSCET Reardon, Outpatient LLE, ULE, BE, education, 6 weeks ITT BDI/TDI Yes Yes Vallet, Inpatient LLE, BE 8 weeks Not measured Yes No Wijkstra, Home-based LLE, ULE, IMT, BE, 12 weeks 6 min walk test, CRQ Yes No education, psychological support Güell, Outpatient LLE, BE, PD 6 months 6 min walk test, CRQ Yes Yes Strijbos, Outpatient LLE, BE, PD, education, 12 weeks 4 min walk test, Interview No Yes LLE=lower-limb exercise, ULE=upper-limb exercise, IMT=inspiratory-muscle training, BE=breathing exercises, PD=postural drainage. =incremental cycle ergometer test, ITT=incremental treadmill test, SSCET=steady-state cycle ergometer test, IAET=incremental arm ergometer test, POMS=profile of mood state (Lorr-NcNair Mood questionnaire), EPS=Eysenck personality questionnaire, Lubin=Lubin depression adjective checklist, CRQ=chronic respiratory disease questionnaire, BDI/TDI=baseline/transitional dyspnoea index. Table 1: Characteristics of randomised trials included in meta-analysis Primary outcome measures were maximum or functional exercise capacity, HRQL, or both. Methodological criteria We required that randomised controlled trials compared respiratory rehabilitation with conventional community care or any other intervention unlikely to have any effect on exercise capacity or quality of life. We looked at two potential sources of bias that have proved to be important determinants of the magnitude of the effect size in clinical trials: unconcealed randomisation and study investigators who were aware of treatment allocation. The former has been associated with an overestimation of the treatment effect by up to 40%; 4 whereas the latter may result in differential encouragement during tests of performance, which might distort the findings (eg, up to 30 5 m in a 6-min walk test). 5 When details of the randomisation process and the outcome assessment were not specified in the original trial publication, we contacted the investigators for clarification. Two reviewers (YL, EW) abstracted the data from the papers selected for inclusion in the meta-analysis. The missing data from the primary study reports were requested from the investigators. Investigators were asked to provide a copy of the questionnaire used to measure HRQL. For multiple-group comparisons (eg, exercise therapy plus inspiratory muscle training compared with exercise therapy alone or with conventional community care), only the treatment group that received the more comprehensive therapy was included. This treatment group was compared with the group that received conventional community care. Statistical analysis The primary outcome measures, exercise capacity and HRQL, were treated as continuous outcomes. Several protocols have been recommended for exercise testing. 6 Exercise capacity is commonly assessed by tests of maximum exercise capacity measured in terms of workload, energy, or oxygen consumption (eg, incremental cycle ergometry or treadmill tests), and tests of functional exercise capacity (eg, timed-walk tests). We decided to analyse maximum and functional test results separately because in this study we found only moderate correlations (r= ) between maximum exercise capacity and functional exercise capacity. 7,8 Only instruments for which there was evidence of validity 9 (the ability of an instrument to measure what it claims to measure) Trial reference Respiratory rehabilitation group Control group Sample size* M/F Age (years) FEV 1 Sample size* M/F Age (years) FEV /0 61 (6) 0 97 L (0 33) 12 12/0 57 (8) 1 15 L (0 72) 13, /0 61 (5) 1 53 L (0 70) 16 16/0 60 (5) 1 32 L (0 44) Not available 66 (8) 0 85 L (0 29) 37 Not available 65 (7) 0 97 L (0 37) /2 64 (6) 0 78 L (0 27) 6 41/5 63 (8) 0 68 L (0 12) /2 65 (16) 26% (9) 7 46/1 66 (16) 27% (11) /1 66 (7) 0 83 L (0 25) 7 44/3 66 (4) 0 97 L (0 29) /9 73 (5) 40% (19) 14 10/4 70 (6) 39% (21) /6 67 (9) 34% (9) 12 45/7 61 (9) 39% (10) /17 66 (7) 35% (15) 40 17/23 65 (8) 35% (12) /5 66 (8) 35% (10) 10 45/5 66 (7) 33% (15) /3 60 (9) 1 80 L (0 54) 10 48/2 58 (6) 1 77 L (0 76) /5 64 (5) 44% (11) 15 14/1 62 (5) 45% (9) /0 64 (7) 31% (12) 27 27/0 66 (6) 39% (14) /1 61 (6) 40% (20) 15 12/3 63 (5) 43% (9) *Number of patients at study completion. Mean (SD). Mean (SD) in L or % predicted. Data from eight patients excluded because FEV 1 70% predicted. Table 2: Baseline characteristics of study populations of trials included in meta-analysis 1116 Vol 348 October 26, 1996

3 and responsiveness 9 (the ability to detect real change, even when it is small) were included in the meta-analysis of HRQL. In each trial and for every outcome measure, we calculated the treatment effect from the difference between the preintervention and postintervention changes in the treatment and control groups. Most related outcome measures were expressed across studies within different units; we, therefore, standardised the resulting treatment effects to obtain an effect size by division of treatment effects by the pooled SD value of the postintervention outcome measure in the treatment and control groups. The effect sizes were weighted by the inverse of the population variance and combined according to a random-effects model. This model assumes that: the studies included in the meta-analysis are a random sample from a larger population of studies (to include uncovered, uncompleted, or planned trials); and that the estimate of effect size in each study differs from the population effect size because of sampling error. 10 Homogeneity among study results was tested. 10 The pooled effect sizes and corresponding 95% CIs were reported for each outcome in SD units, and then converted back to natural units of the most commonly used measure. Whenever possible, the overall treatment effect expressed in natural units was compared with its minimum clinically important difference (MCID) defined as the smallest difference perceived by the average patient. When the magnitude of the treatment effect equals or exceeds the MCID, the management of a patient should be changed, unless there are adverse side-effects or excessive costs. 11 Subgroup analyses were indicated when significant heterogeneity was found among the primary findings of the trials, or when no heterogeneity was found among trial results but when the clinical significance of the overall treatment effect was borderline ie, when the CI for the overall effect encompassed the MCID. To explain anticipated heterogeneity among trial findings, three respiratory disease physicians (YL, EW, RSG) identified, a priori, potential sources of heterogeneity among the outcomes of exercise capacity and HRQL. Since the treatment effect might vary according to the severity of disease, population was judged a likely source of heterogeneity. In addition, we postulated that: the more comprehensive the rehabilitation programme, the larger the effect size in improving exercise capacity and HRQL; that patients in short-duration rehabilitation programmes (<8 weeks) might have smaller improvements than those in longer-duration Study McGavin, 1977 Cockcroft, 1981 Booker, 1984 Jones, 1985 Lake, 1990 Simpson, 1992 Weiner, 1992 Goldstein, 1994 Wijkstra, 1994 Güell, 1995 Strijbos, 1996 Favours control Favours treatment Overall effect 0 6 (0 3 to 1 0) Effect size (SD units) Figure: Effect of respiratory rehabilitation on functional exercise capacity programmes ( 8 weeks, >12 weeks); and that patients in supervised inpatient and outpatient programmes might have greater improvements than those in unsupervised, home-based programmes. We also postulated that the findings of trials would be affected by the quality of the methods used particularly in terms of double-blind conditions for the assessment of the primary outcome measures. Results 301 publications were retrieved from the computer searches and 54 abstracts were identified. Three studies were found through our contact with experts. The publications were reduced to 81 potentially eligible papers. 64 trials were excluded because: the population of patients assessed was inappropriate (two studies); the intervention did not meet the definition of rehabilitation (11); the control groups did not receive conventional community care (27); or trials were not randomised (24). Both reviewers (YL, EW) agreed to include 15 papers in the meta-analysis (quadratic weighted =0 79 [95% CI ]). The reviewers disagreed about two papers, but both were included after consultation with the third reviewer (RG). Of the 17 reports of the 16 trials judged acceptable, two were excluded because they did not have adequate data, even after contact with the investigators. 16,27 Table 1 shows the 14 trials that were included in the meta-analysis. We contacted the authors of these papers for any additional information required; the response rate was 100%. Of the 14 trials, one did not meet our criteria for randomisation. 28 Consequently, we defined subgroups only on the basis of whether the outcome assessment was masked. Table 2 shows important demographic and clinical characteristics of the patients at study entry in each trial. Most patients were elderly and had severe COPD. The main exclusion criteria were: ischaemic heart disease, heart failure, intermittent claudication, disabling musculoskeletal disorders, domiciliary oxygen requirement, hypercapnia, and other medical disorders that limited exercise tolerance. 12,13,15,17 20,22 26,28 Maximum exercise capacity was measured in 11 trials (309 patients). The pooled effect size achieved significance (0 3 SD units [ ]) and corresponded, in incremental cycle ergometer test natural units, to 8 3 W ( ). Homogeneity was found among study findings (p=0 85), which suggests that the effect of rehabilitation on maximum exercise capacity was constant across studies, irrespective of the duration or composition of the rehabilitation programme. The effect of respiratory rehabilitation on functional exercise capacity are shown in the figure. 11 trials (413 patients) were included in this analysis. After conversion back to natural units for the 6-min walk test (m), a difference in response between the treatment and control groups of 55 7 m ( ) was found. We estimated the MCID of the walk test at about 50 m from a study in which COPD patients rated their walking ability by subjective comparisons with one another. 29 The limits of the CI for the pooled estimate of effect size ( ) were wider than those of the estimate of the MCID of the 6-min walk test (37 71 m). Heterogeneity among study results (p=0 0008) could not be explained by the subgroup analyses. A post-hoc analysis based on the observed results showed a significant difference between the programmes of 6 months duration and the other Vol 348 October 26,

4 Outcome Number Total number Effect size Treatment effect p measures of trials of patients (SD units) (natural units)* (treatment/ Overall effect Overall effect control) (95% CI) (95% CI) Maximum / ( ) 8 3 W 0 85 exercise ( ) capacity Functional / ( ) 55 7 m exercise ( ) capacity Dyspnoea 6 126/ ( ) 1 0 ( ) 0 12 Fatigue 4 111/ ( ) 0 8 ( ) 0 36 Emotional 4 111/ ( ) 0 6 ( ) 0 68 function Mastery 4 111/ ( ) 0 8 ( ) 0 77 *Natural units are from individual items (7-point scale) of the chronic respiratory questionnaire (dyspnoea, fatigue, emotional function, mastery), an incremental cycle ergometer test (maximum exercise capacity), and a 6-min walk test (functional exercise capacity). Table 3: Primary results of meta-analysis programmes (93 8 m [homogeneity, p=0 31] vs 39 2 m [homogeneity, p=0 10]; difference, p=0 0004). Since this analysis was data-driven the results should be interpreted as hypothesis-generating. 12 of the 14 trials measured HRQL and ten different instruments were used for this assessment (table 1). Evidence of validity and responsiveness for only two of these instruments has been published (the transitional dyspnoea index, 30 and the chronic respiratory questionnaire 31 ). We, therefore, confined the analysis of HRQL to the six trials in which one of these questionnaires had been used. The most frequently used measure was the chronic respiratory questionnaire, in which responses were presented on a 7-point scale. For each outcome, the overall effect size exceeded the MCID (0 5 points). CI values suggest that the smallest treatment effect exceeded the MCID for the dimensions of dyspnoea and mastery (the extent to which patients feel they can cope with the disease and its manifestations); whereas for fatigue and emotional function the CI values encompassed the MCID. We analysed fatigue and emotional function for sources of heterogeneity, but the small number of trials limited the power of the subgroup analyses. No a-priori hypotheses accounted for differences in individual trial results. Discussion The development of objective HRQL outcome measures and better understanding of the physiological rationale for exercise training in patients with COPD 32 have facilitated the widespread acceptance of respiratory rehabilitation. We would like to comment on three features of our meta-analysis. First, we assessed the acute effect of respiratory rehabilitation on COPD ie, the benefits of rehabilitation at the completion of a programme. Further studies are needed to assess the long-term benefits of rehabilitation and strategies to maintain the early benefits. Second, the criterion we used to define a clear benefit of respiratory rehabilitation was conservative: the lower limit of the CI had to be greater than the MCID. Third, several well-conducted studies were excluded from the meta-analysis because we focused on the effect of rehabilitation when added to conventional care. We, therefore, excluded one well-conducted trial, because the control group received an educational intervention. 33 Similarly, we excluded studies that compared interventions, such as inspiratory muscle training, psychosocial support, or breathing exercises, with exercise training. The care of patients with COPD largely aims to address symptoms and, therefore, we believe that quality of life and functional exercise capacity should be the primary outcome measures for respiratory rehabilitation. Our meta-analysis shows that respiratory rehabilitation relieved dyspnoea and improved mastery, because the magnitude of the improvement was greater than the MCID. By contrast, the clinical importance of functional exercise capacity is not clear because the CI of the effect size was wider than that of the MCID for the 6-min walk test. Although rehabilitation programmes included in the metaanalysis differed in several ways, for example, in clinical settings, duration, and composition, the homogeneity among study results suggests that less sophisticated rehabilitation programmes may be as effective as the more comprehensive programmes in improving HRQL. The importance of measures of maximum exercise capacity is not clear. We believe that an initial test may be useful to establish the appropriate level of training. Retesting may provide physiological evidence that a training response has occurred and may, therefore, be used to adjust intensity levels during rehabilitation. 6 Since the results of maximum exercise tests correlate poorly with HRQL measures, 7 such tests should not be substituted for measures of HRQL in the assessment of a rehabilitation programme. Our findings strongly support respiratory rehabilitation that includes at least 4 weeks exercise training as part of management for patients with COPD. We found clinically and statistically significant improvements in dyspnoea and mastery. When compared with the treatment effect of other important approaches for patients with COPD, such as bronchodilators or oral theophylline, 34,35 rehabilitation led to greater improvements in HRQL and functional exercise capacity. We thank the authors of the primary studies included in the meta-analysis who provided additional data and information about their work. References 1 Feinlieb M, Rosenberg HM, Collins JG, Delozier JE, Pokras R, Chevarley FM. Trends in COPD morbidity and mortality in the United States. Am Rev Respir Dis 1989; 140: S American Thoracic Society. Pulmonary rehabilitation. Am Rev Respir Dis 1981; 124: Donner CF, Howard P. Pulmonary rehabilitation in chronic obstructive pulmonary disease (COPD) with recommendations for its use: report of the European Respiratory Society Rehabilitation and Chronic Care Scientific Group (SEPCR Rehabilitation Working Group). Eur Respir J 1992; 5: Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995; 273: Guyatt GH, Pugsley SO, Sullivan MJ, et al. Effect of encouragement on walking test performance. Thorax 1984; 39: Jones NL. Clinical exercise testing. Philadelphia, Pennsylvania: W B Saunders, 1988: Guyatt GH, Thompson PJ, Berman LB, et al. How should we measure function in patients with chronic heart and lung disease? J Chronic Dis 1985; 38: Wijkstra PJ, TenVergert EM, van der Mark ThW, et al. Relation of lung function, maximal inspiratory pressure, dyspnoea, and quality of life with exercise capacity in patients with chronic obstructive pulmonary disease. Thorax 1994; 49: Kirshner B, Guyatt G. A methodological framework for assessing health indices. J Chronic Dis 1985; 38: Fleiss JL. The statistical basis of meta-analysis. Stat Methods Med Res 1993; 2: Vol 348 October 26, 1996

5 11 Jaeschke R, Singer J, Guyatt GH. Measurement of health status: ascertaining the minimal clinically important difference. Control Clin Trials 1989; 10: McGavin CR, Gupta SP, Lloyd EL, McHardy GJR. Physical rehabilitation for the chronic bronchitis: results of a controlled trial of exercises in the home. Thorax 1977; 32: Cockcroft AE, Saunders MJ, Berry G. Randomised controlled trial of rehabilitation in chronic respiratory disability. Thorax 1981; 36: Cockcroft A, Berry G, Brown EB, Exall C. Physiological changes during a controlled trial of rehabilitation in chronic respiratory disability. Thorax 1982; 37: Booker HA. Exercise training and breathing control in patients with chronic airflow limitation. Physiotherapy 1984; 70: Reid WD, Warren CPW. Ventilatory muscle strength and endurance training in elderly subjects and patients with chronic airflow limitation: a pilot study. Physiother Can 1984; 36: Jones DT, Thomson RJ, Sears MR. Physical exercise and resistive breathing in severe chronic airways obstruction are they effective? Eur J Respir Dis 1985; 67: Busch AJ, McClements JD. Effects of a supervised home exercise program on patients with severe chronic obstructive pulmonary disease. Phys Ther 1988; 68: Lake FR, Henderson K, Briffa T, Openshaw J, Musk AW. Upper-limb and lower-limb exercise training in patients with chronic airflow obstruction. Chest 1990; 97: Simpson K, Killian K, McCartney N, Stubbing DG, Jones NL. Randomised controlled trial of weightlifting exercise in patients with chronic airflow limitation. Thorax 1992; 47: Weiner P, Azgad Y, Ganam R. Inspiratory muscle training combined with general exercise reconditioning in patients with COPD. Chest 1992; 102: Goldstein RS, Gort EH, Stubbing D, Avendano MA, Guyatt GH. Randomised controlled trial of respiratory rehabilitation. Lancet 1994; 344: Reardon J, Awad E, Normandin E, Vale F, Clark B, ZuWallack RL. The effect of comprehensive outpatient pulmonary rehabilitation on dyspnea. Chest 1994; 105: Vallet G, Varray A, Fontaine JL, Préfaut C. Intérêt du réentraînment à l effort individualisé, au niveau du seuil ventilatoire, au cours de la bronchopneumopathie chronique obstructive de sévérité modérée. Rev Mal Respir 1994; 11: Wijkstra PJ, van Altena R, Kraan J, Otten V, Postma DS, Koëter GH. Quality of life in patients with chronic obstructive pulmonary disease improves after rehabilitation at home. Eur Resp J 1994; 7: Güell R, Morante F, Sangenis M, Casan P. Effects of respiratory rehabilitation on the effort capacity and on the health-related quality of life of patients with chronic obstructive pulmonary disease. Eur Respir J 1995; 8 (suppl): 356 (abstr). 27 Stanton M, Beauchamp C, Weinberger M. A randomized controlled trial of pulmonary rehabilitation in chronic obstructive airways disease. J Gen Intern Med 1995; 10 (suppl): Strijbos JH, Postma DS, van Altena R, Gimeno F, Koëter GH. A comparison between an outpatient hospital-based pulmonary rehabilitation program and a home-care pulmonary rehabilitation program in patients with COPD. Chest 1996; 109: Goldstein RS, Redelmeier DA, Baksh L, Guyatt GH. Subjective comparison ratings of walking ability in patients with COPD. Proceedings of the 5th International Conference on Pulmonary Rehabilitation and Home Ventilation. Denver, Colorado, 1995: Mahler DA, Weinberg DH, Wells CK, Feinstein AR. The measurement of dyspnea: contents, interobserver agreement, and physiologic correlates of two new clinical indexes. Chest 1984; 85: Guyatt GH, Berman LB, Townsend M, Pugsley SO, Chambers LW. A measure of quality of life for clinical trials in chronic lung disease. Thorax 1987; 42: Casaburi R, Patessio A, Ioli F, Zanaboni S, Donner CF, Wasserman K. Reductions in exercise lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease. Am Rev Respir Dis 1991; 143: Ries AL, Kaplan RM, Limberg TM, Prewitt LM. Effects of pulmonary rehabilitation on physiologic and psychological outcomes in patients with chronic obstructive pulmonary disease. Ann Intern Med 1995; 122: Jaeschke R, Guyatt GH, Willan A, et al. Effect of increasing dose of beta-agonists on spirometric parameters, exercise capacity, and quality of life in patients with chronic airflow limitation. Thorax 1994; 49: McKay SE, Howie CA, Thomson AH, Whiting B, Addis GJ. Value of theophylline treatment in patients handicapped by chronic obstructive lung disease. Thorax 1993; 48: Vol 348 October 26,