Review of the Evidence for Pulmonary Rehabilitation in COPD: Clinical Benefits and Cost-Effectiveness

Abstract

COPD is a common and lethal chronic condition, recognized as a leading cause of death worldwide. COPD is associated with significant morbidity and disability, particularly among older adults. The disease course is marked by periods of stability and disease exacerbations defined by worsening respiratory status resulting in a high burden of health care utilization and an increased risk of mortality. Treatment is focused on pharmacologic therapies, but these are not completely effective. Pulmonary rehabilitation (PR) represents a key medical intervention for patients with chronic respiratory diseases, including COPD. PR provides individualized and progressive exercise training, education, and self-management strategies through a comprehensive and multidisciplinary program. PR has been associated with improvement in exercise capacity, health-related quality of life, and dyspnea in patients living with COPD. Moreover, PR has been associated with improvements in hospital readmission and 1-y survival. In addition to the clinical benefits, PR is estimated to be a cost-effective medical intervention. Despite these benefits, participation in PR remains low. We will review the evidence for PR in each of these benefit domains among patients with stable COPD and in those recovering from a COPD exacerbation.

Introduction

COPD is estimated to affect 24 million people in the United States¹ and 10% of the population 40 years and older globally² according to the Global Initiative for Chronic Obstructive Lung Disease 2021 Report (https://goldcopd.org/wp-content/uploads/2020/11/GOLD-REPORT-2021-v1.0-11Nov20_WMV.pdf. Accessed August 4, 2023). COPD is a leading cause of death around the world and associated with significant morbidity and disability (https://www.who.int/data/gho/data/themes/mortality-and-global-health-estimates. Accessed August 4, 2023). Pharmacologic treatments such as inhalers and anti-inflammatory medications have been shown to improve symptoms and reduce the frequency of COPD hospitalizations. However, none of these therapies are completely effective.

Pulmonary rehabilitation (PR) has been shown to be one of the most effective treatments in COPD. PR is a multidisciplinary program providing supervised exercise training and self-management education targeting long-term behavior modification designed to increase physical activity. PR has been shown to improve exercise capacity, dyspnea, and quality of life in participants with COPD.^3,4 While these outcomes are important, it may come as no surprise considering PR is grounded in exercise training. Maybe more surprising is the recognition that timely participation in PR after COPD hospitalization has been associated with improvement in rates of hospital readmission^4-6 and 1-y survival.^4,5,7 Despite these established benefits, only 3% of eligible Medicare beneficiaries in the United States participate in PR, with uptake similarly low globally.⁸ Beyond the clinical benefits, the rationale for increasing access to PR is supported by growing evidence that PR is not simply a cost-effective intervention but is estimated to result in cost savings.⁹ This review will discuss the benefits of PR in patients with COPD in the following domains: exercise capacity, dyspnea, quality of life, hospitalization, mortality, and cost-effectiveness.

Benefits of Pulmonary Rehabilitation on Exercise Tolerance

Exercise training is a key component of PR in patients with chronic respiratory disease. Limitations in exercise capacity arise among patients with COPD due to peripheral muscle dysfunction, dynamic hyperinflation, increased respiratory load, gas exchange abnormalities, age-related functional decline, deconditioning, and frequent concurrent comorbid conditions.¹⁰ Some of these variables may be modifiable with a comprehensive PR program. Exercise training in patients with COPD has been shown to result in physiologic benefits in skeletal muscle oxidative capacity and ventilatory requirements.^11,12 Exercise performance, a key outcome measure in PR, is assessed via field walking tests in individuals with chronic respiratory disease (eg, 6-min walk test [6MWT] or incremental shuttle walk test [ISWT]), along with laboratory-based treadmill or cycle ergometer tests.^10,13

A 2015 Cochrane review evaluated the short-term effects of PR versus usual care on exercise capacity in subjects with stable COPD or those without a recent exacerbation (65 randomized controlled trials [RCTs]; total of 3,822 participants). In subjects who participated in at least 4 weeks of PR that could encompass in-patient, out-patient, community-based, or home-based programs, functional exercise capacity as evaluated by the 6MWT was reported in 38 of the trials, encompassing 1,879 participants. A pooled analysis demonstrated a statistically, and importantly, clinically important increase in the mean meters walked associated with PR compared to usual care with a mean difference (95% CI) between groups of 43.93 (32.64–55.21) m during a median follow-up period of 12 weeks.³ For clarity, all data described in the current review are reported as mean difference (95% CI) unless otherwise stated. The mean difference and lower limit of its CI in the 6-min walk distance (6MWD) were greater than the minimal clinically important difference (MCID) of 30 m among subjects with chronic respiratory disease as reported by the European Respiratory Society (ERS) and American Thoracic Society (ATS).^14,15 The MCID is defined as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate, in the absence of troublesome side effects and excessive cost, a change in the patient’s management.”¹⁶

The ISWT, an externally paced maximal exercise test, was assessed in 8 trials encompassing 694 participants. There was a significant increase in mean meters walked among PR participants compared to usual care, 39.77 (22.38–57.15) m.³ The average improvement, but not the lower limit of its CI, exceeded the MCID of 35 m as determined using both distribution and anchor-based methods.¹⁷ Maximal exercise capacity was evaluated in 16 studies including a total of 779 participants using the incremental cycle ergometer test and revealed an mean difference in peak work load or maximal cycle exercise capacity of 6.77 (1.89–11.65) W among PR participants compared to usual care,³ which exceeded the MCID of 4 W.¹⁸ Although the magnitude of benefit varied, which impacted the precision of the CI estimate, results collectively demonstrated a beneficial effect of PR on exercise capacity.³

Exercise performance–related benefits of PR have not only been demonstrated among participants with stable COPD, but evidence also exists for the role of PR after treatment for exacerbations of COPD. Acute disease flares, referred to as COPD exacerbation, have deleterious effects on disease progression and health status including a marked reduction in physical activity.^10,19,20 A 2016 Cochrane review evaluated the effects of in-patient, out-patient, or home-based PR programs beginning within 3 weeks following treatment for an COPD exacerbation compared to usual care (20 RCTs; total of 1,477 participants). Functional exercise capacity was assessed using the 6MWT in 13 studies encompassing 819 participants. The authors found a statistically and clinically important improvement in 6MWD by an average of 62.38 (95% CI 38.45–86.31) m among those who participated in PR compared to usual care over a median follow-up of 3 months.⁴ Substantial heterogeneity of results across trials was evident (I² = 87%), but the evidence was deemed to be of high quality with heterogeneity largely driven by some studies showing very large and others showing smaller, albeit still statistically significant, effects and an overall mean difference in 6MWD along with the lower limit of the CI greater than the MCID of 30 m.^14,15 The ISWT demonstrated an mean difference of 48.14 m, greater than the MCID of 35 m¹⁷ but not statistically significant (−1.03 to 97.32 m) (4 studies, 448 participants).⁴ Overall, strong evidence supports the benefits of PR on functional and maximal exercise capacity among patients with COPD with improvement in exercise tolerance not only among those patients with stable disease but also following COPD exacerbation (Table 1).

Table 1.

Exercise Benefits for Pulmonary Rehabilitation in Subjects With Stable COPD and Post Exacerbation

Benefits of Pulmonary Rehabilitation on Health-Related Quality of Life and Dyspnea

Subjects with COPD report experiencing reduced health-related quality of life (HRQOL), or satisfaction with health,¹⁰ compared with healthy subjects.^21,22 Primary goals of PR include improvement in quality of life, reduction in symptom burden, and increased participation in everyday activities.¹⁰ The St George Respiratory Questionnaire and Chronic Respiratory Disease Questionnaire are well-established disease-specific questionnaires validated to assess health status in persons with obstructive lung disease.¹⁰ The St George Respiratory Questionnaire includes 76 weighted questions, generating an overall score ranging from 0–100 with a higher score indicating worse HRQOL, along with 3 subscores for the separate domains of symptoms, activity limitation, and impact.²³ In a meta-analysis of 19 trials encompassing 1,146 participants with stable COPD, moderate-quality evidence demonstrated a significant improvement in the aggregate St George Respiratory Questionnaire score for those persons allocated to at least 4 weeks of PR compared to usual care, −6.89 (−9.26 to −4.52) units,³ an effect larger than the MCID of 4 units.^24,25 All domains of the St George Respiratory Questionnaire demonstrated significant improvement as well (St George Respiratory Questionnaire symptoms: −5.09 [−7.69 to −2.49]; St George Respiratory Questionnaire impact: −7.23 [−9.91 to −4.55]; St George Respiratory Questionnaire activity: −6.08 [−9.28 to −2.88].³ The Chronic Respiratory Disease Questionnaire includes 4 domains: dyspnea, fatigue, emotional function and mastery, or a feeling of control over the disease. Each domain includes 5 items, each graded on a 7-point Likert scale with higher scores indicating better HRQOL.²⁶ Among results pooled from 19 studies including 1,291 participants with stable COPD, there were significant improvements in each of the Chronic Respiratory Disease Questionnaire domains for participants in PR (fatigue: 0.68 [0.45–0.92]; emotional function: 0.56 [0.34–0.78]; mastery: 0.71 [0.47–0.95]; dyspnea: 0.79 [0.56–1.03]).³ The pooled treatment effects for each of the Chronic Respiratory Disease Questionnaire domains exceeded the MCID of 0.5 points,¹⁶ and the lower limit of the CI exceeded the MCID for the dyspnea domain, providing unambiguous clinical and statistical significance.³

Similar findings have been demonstrated among participants enrolled in PR following COPD exacerbation with improvement in HRQOL. A meta-analysis of 8 studies including 846 participants revealed a significant improvement in St George Respiratory Questionnaire score among participants enrolled in PR within 3 weeks of COPD exacerbation compared to usual care (−7.80 [−12.12 to −3.47]),⁴ a mean effect size that exceeded the MCID of 4 units.^24,25 Statistically significant effects were also noted in the individual St George Respiratory Questionnaire domains of impact (−10.44 [−16.11 to −4.76]) and activities (−8.23 [−12.88 to −3.57]) but not symptoms (−2.45 [−7.33 to 2.42]). Pooled results from 5 studies including 259 participants reported significant improvement in Chronic Respiratory Disease Questionnaire on fatigue (0.81 [0.16–1.45]), emotional function (0.94 [0.46–1.42]), and dyspnea (0.97 [0.35–1.58]) domains but not the mastery domain (0.93 [−0.13 to 1.99]).⁴ The pooled effect sizes exceeded the MCID of 0.5 points, but the lower limits of the CIs did not.¹⁶ Although the magnitude of benefit varied, which impacted the precision of the CI estimate, results collectively demonstrated a beneficial effect of PR on HRQOL⁴ (Table 2).

Table 2.

Effects of Pulmonary Rehabilitation on Health-Related Quality of Life and Dyspnea in Subjects With Stable COPD and Post Exacerbation

Benefits of Pulmonary Rehabilitation on COPD Hospitalization

For many patients, COPD is associated with a disease course marked by COPD exacerbation illnesses leading to 1.5 million emergency department visits and 700,000 hospitalizations each year.⁶ Among patients hospitalized for COPD, one in 5 will require re-hospitalization within 30 d, and 64% of those adults 65 y and older are readmitted within 1 y of discharge.^27-30 The economic burden of hospitalizations for COPD exacerbation is substantial, responsible for $13.2 billion of the $50 billion in annual direct costs for COPD nationally.²⁷ Strategies to address this considerable health care burden and cost are a priority for patients and health care systems.²⁷

PR is an important intervention that addresses multiple risk factors for hospital readmission by combating physical inactivity and promoting self-management including medication adherence and education to support behavior change.⁴ A 2016 Cochrane review included 8 RCTs involving 810 participants with a median follow-up of 9 months. Pooled analysis demonstrated that participants enrolled in PR beginning within 3 weeks of hospitalization for COPD exacerbation had a 56% lower odds of hospital readmission (odds ratio 0.44 [95% CI 0.21–0.91]) compared to those not enrolled in PR. The evidence was deemed to be of moderate quality given significant heterogeneity of results (I² = 77%), which the authors speculated may be related to the methodological quality of included studies and variability in the extensiveness of rehabilitation programs.⁴ A 2018 meta-analysis included only RCTs comparing supervised PR programs initiated during or within 4 weeks of COPD hospitalization versus usual care. Among the 6 studies encompassing 365 participants with 3–12-month follow-up, participation in PR was significantly associated with a 53% lower risk of re-hospitalization for COPD (relative risk 0.47 [95% CI 0.29–0.75]) compared to subjects treated with usual care without PR. Although heterogeneity of the results was low (I² = 38%), the quality of evidence was deemed to be moderate given methodological concerns.⁵

Beyond efficacy data from highly selected subject populations enrolled in RCTs, the effectiveness of PR on hospital readmission reduction has been demonstrated in broader, more generalizable populations following less protocolized PR regimens, reflective of real-world clinical practice. A retrospective cohort study examined the effects of PR among a population of fee-for-service Medicare beneficiaries 66 y of age or older hospitalized with a COPD exacerbation in 2014. Among a cohort of nearly 200,000 subjects, only 1.5% initiated PR within 90 d of discharge. The researchers evaluated the risk of all-cause readmission and COPD-specific readmission within 1 y of discharge among subjects who participated in PR after discharge versus no PR participation or PR participation > 90 d after discharge. Among the full cohort, initiation of PR within 90 d of discharge was associated with a lower all-cause readmission rate at 1 y of 56.4% compared to 64.6% in those in the no/late PR group. The number-needed-to-treat to prevent re-hospitalization within 1 y of discharge was 12.⁶

A propensity-matched analysis adjusted for unbalanced covariates and community characteristics using a multistate model to account for the time-varying exposure to PR, recurrent hospitalizations, and the competing risk of death. The cohorts were well balanced in all measured characteristics except for travel distance to PR center. Subjects who participated in early PR were found to be 17% less likely to be re-hospitalized (hazard ratio [HR] 0.83 [95% CI 0.77–0.90]) compared to the no/late PR group. Among those who participated in early PR, there was a significantly lower risk of both COPD-specific readmissions (HR 0.86 [95% CI 0.76–0.97]) and non–COPD-specific re-hospitalizations (HR 0.79 [95% CI 0.71–0.87]). The authors postulated this effect may be related to general improvements in health because of the exercise and social components of PR.⁶ Therefore, similar, albeit attenuated, strong evidence supports the benefits of PR on reducing hospital readmissions among a broad, generalizable population of older adults.⁶

Beyond a reduction in hospital readmissions, researchers found that home time was increased among subjects who initiated PR. Home time, or the number of days alive and spent outside of a health care institution, is considered a meaningful outcome.³¹ Subjects who initiated PR within 90 d of a COPD hospitalization spent fewer days in the hospital (number of days spent in hospital per person-year of 7.9 d vs 11.7 d) and fewer days in a nursing home (number of days spent in nursing home per person-year 1.8 d vs 2.9 d).⁶

Benefits of Pulmonary Rehabilitation on Mortality

COPD represents a leading cause of death worldwide. Patients hospitalized for a COPD exacerbation are at increased risk of death⁵ with an estimated mortality of 26% within 1 y after hospital discharge.³⁰ A 2016 Cochrane meta-analysis analyzed studies investigating the impact of PR on mortality (6 RCTs; total of 670 participants). PR beginning within 3 weeks of a COPD exacerbation had a suggested 32% decreased odds of mortality compared to usual care over a median follow-up of 12 months, although this result was not statistically significant (odds ratio 0.68 [95% CI 0.28–1.67]). The evidence was reported as low quality considering none of the studies used mortality as a primary outcome, nor were they powered to detect a meaningful impact on mortality. Moreover, results were heterogenous (I² = 59%) and included one study that revealed increased mortality with PR initiation in the post-exacerbation period.^4,32 This study by Greening et al³² randomized 389 subjects to early rehabilitation versus usual care within 48 h of an admission for an exacerbation of chronic respiratory disease, 82% of whom had a primary diagnosis of COPD. Their 6-week intervention consisted of daily, supervised strength and aerobic training along with neuromuscular electrical stimulation techniques while participants were in-patient, followed by continuation of an unsupervised home program supported by telephone consultations. The investigators identified a 74% increased risk of mortality at 1 y (odds ratio 1.74 [95% CI 1.05–2.88], P = .03) among those randomized to early rehabilitation compared to usual care. The authors acknowledged that the finding of increased mortality, which appeared to begin > 5 months after the intervention in the PR group, was not fully understood and speculated it may have been unrelated to PR. The study design has been critiqued for an inadequate intervention lasting only a median of 5 d while hospitalized and their exercise-based intervention was not consistent with PR.³³ Nevertheless, these results have informed ERS and ATS guidelines that recommend initiation of PR within 3 weeks of COPD hospital discharge but not during COPD hospitalization.³⁴ In a subgroup analysis performed by the authors of the 2016 Cochrane review omitting this study by Greening et al³² and including only those trials that involved extensive rehabilitation programs, there was a statistically significant odds ratio for mortality in favor of PR (odds ratio 0.50 [95% CI 0.26–0.99]),⁴ suggesting that when PR is delivered properly it is associated with an improvement in mortality. A more recent meta-analysis⁵ that included only RCTs evaluating the effects of supervised PR programs initiated during or within 4 weeks of a hospitalization for a COPD exacerbation demonstrated a reduced risk of mortality at the end of treatment in PR participants (relative risk 0.58 [95% CI 0.35–0.98]; 4 trials, 319 participants). There was low heterogeneity, but the evidence was deemed to be of moderate quality given some methodological concerns and risks of bias within the included trials.

Considering meta-analyses of clinical trials have suggested PR was associated with an improvement in mortality, investigators evaluated the treatment effect among a broad, generalizable population in routine clinical practice. Lindenauer et al⁷ evaluated the impact of PR after COPD hospitalization on 1-y mortality among a cohort of fee-for-service Medicare beneficiaries > 65 y of age. Among a cohort of 197,376 subjects in 4,446 hospitals in 2014, only 2,721 (1.5%) enrolled in PR within 90 d of hospital discharge. An additional 3,161 subjects (1.6%) enrolled after 90 d but within a year of discharge. In the full study cohort following adjustment, participation in early PR was significantly associated with a 37% lower risk of mortality within 1 y of hospital discharge (HR 0.63 [95% CI 0.57–0.69], P < .001) and an absolute risk difference of −6.7% (95% CI −7.9 to −5.6) compared to PR after 90 d or not at all. The number-needed-to-treat to prevent one death within 12 months of hospital discharge with timely participation in PR was 8. By comparison, the number needed to screen with low-dose computed tomography to prevent one death from lung cancer is 320.³⁵

In a propensity-matched analysis, subjects who initiated PR within 90 d of hospital discharge were 50% less likely (HR 0.50 [95% CI 0.42–0.59], P < .001) to die within 1 y compared to those who initiated PR after 90 d or not at all. Cohorts in the propensity-matched analysis were well-matched except for distance to PR facility, hospital size, and urban location. The investigators utilized multiple analytic methodological approaches that yielded similar significant relationships strengthening confidence in their conclusions.⁷ Recognizing previous concerns related to appropriate timing and potential safety of PR in the immediate post-exacerbation period,^32,34 the authors conducted an exploratory analysis to evaluate the relationship between PR and mortality based on the start dates from hospital discharge. They found that PR initiation within 30 d (HR 0.74 [95% CI 0.67–0.82]), 31–60 d (HR 0.43 [95% CI 0.34–0.54]), and 61–90 d (HR 0.40 [95% CI 0.30–0.54]) were each associated with a significantly lower mortality compared to no or late (> 90 d) initiation of PR.⁷ These results alleviated prior safety concerns surrounding the need to wait to begin PR post exacerbation and provided reassurance that delayed initiation was still effective. The role of PR as a potential disease-modifying intervention³⁶ is supported by the dose-response relationship that demonstrated that every 3 additional sessions of PR beyond 9 in the first 90 d from discharge were significantly associated with a 9% lower risk of mortality (HR 0.91 [95% CI 0.85–0.98]).⁷

Cost-Effectiveness of Pulmonary Rehabilitation in COPD

PR as a worthwhile medical intervention for patients with COPD is supported by the 5 previously summarized benefits, each established through strong empirical evidence. However, the cost-effectiveness of PR, particularly in the United States, was previously not well understood. Establishment of PR as a cost-effective intervention could promote health care stakeholders (eg, health system executives and insurers) to develop novel strategies aimed at increasing access to PR. We will review the evidence on cost-effectiveness of PR in the United States and globally.

A cost-effectiveness analysis of PR in the United States health care system estimated that participation in PR after COPD hospitalization was not simply cost-effective but resulted in cost savings.⁹ Using a Markov microsimulation model, authors compared a strategy of PR versus no PR after COPD hospitalization in the United States health care system with a lifetime horizon and a discounted rate of 3% per year for both costs and outcomes. Data sources included published literature from 2001–2021, with the primary data sources stemming from analyses of Medicare beneficiaries hospitalized for COPD between January 1, 2014–December 31, 2015.^6,7 The study found that participation in PR after COPD hospitalization resulted in cost savings of $5,721 (95% prediction interval, $3,307–8,388) and quality-adjusted life expectancy gain of 0.53 y (95% prediction interval, 0.43–0.63) per patient over their lifetime.⁹

The cost savings ($5,721) associated with PR were due entirely to a reduction in hospital days (eg, 7.9 with PR vs 11.7 no PR) and skilled nursing facility days (eg, 1.8 with PR vs 2.9 no PR). The cost savings were greatest 1 y after PR ($8,226). However, considering PR is associated with improvement in 1-y survival,⁷ PR participants lived longer and as a result incurred more COPD-related costs after year one. The findings of cost savings and improved quality-adjusted life expectancy with PR did not change in univariate analyses of subjects age, Global Initiative for Chronic Obstructive Lung Disease stage, or number of PR sessions. Assuming 36 sessions in a PR program, a single PR session was estimated to remain cost saving up to $171 per session. By comparison, the cost per session of PR in the analysis was $111.32, which included billing code G0424 for 2 h and patient co-payment per session. PR was no longer cost saving beyond a cost of $171 per PR session. However, at a willingness-to-pay of $50,000/quality-adjusted life year (QALY), a standard for high-value interventions, PR remained cost-effective until $884 per session and $1,597 per session for $100,000/QALY⁹ (Fig. 1).

Fig. 1.

Threshold value estimates: total cost per pulmonary rehabilitation session. *Assumes 36 sessions. ICER = incremental cost-effectiveness ratio; QALY = quality-adjusted life year; PR = pulmonary rehabilitation. From Reference 9, with permission.

Non–United States studies performed in Canada, Ireland, United Kingdom, and the Netherlands consistently support PR as a cost-effective medical intervention.^37-40 The strengths of these studies include direct measurement of health care outcomes and associated costs of the enrolled study populations. As a result, these findings are reflective of the local study cost environment. However, their findings may be limited and less generalizable to broader populations not represented in the clinical trials. By comparison, Mosher et al⁹ used a Markov modeling approach that relied upon inputs from published literature, primarily 2 large observational studies of United States Medicare beneficiaries, reflective of a broad population of older adults in the United States. Furthermore, Liu et al⁴¹ found that PR was cost-effective across a variety of settings, including out-patient, home-based rehabilitation, and telerehabilitation. Numerous studies support PR as a cost-effective intervention in COPD and even estimated to result in cost savings in the United States. These potential cost savings could be used to address patient reported barriers to participation by subsidizing patient co-payments and transportation. PR advocates have highlighted several approaches that could be financially sustainable strategies toward increasing PR access, particularly among older adults in the United States health care system.⁴²

Summary

COPD is associated with significant morbidity and mortality resulting in substantial health care costs. Current pharmacologic therapies are beneficial but are not completely effective. In this review, we summarize the evidence demonstrating the strong merits of PR among patients living with COPD. Participation in PR has demonstrated improvement in 5 key domains including exercise capacity, HRQOL, and dyspnea in both stable disease and in those recovering from exacerbations. Moreover, studies have demonstrated that timely enrollment in PR following an exacerbation is not only safe but reduces hospital readmissions and improves 1-y survival, further strengthening the support for PR and suggesting that PR may have a role in disease modification. In recognition of the strong evidence for these 5 benefits, the latest ATS clinical practice guidelines strongly recommend PR participation for patients with stable COPD and following hospitalizations for exacerbations of COPD.⁴³ Further support for PR stems from an economic argument in favor of PR with evidence that enrollment in PR after COPD hospitalization may result in cost savings. Despite this, uptake and program completion remain low. Novel strategies to overcome barriers to enrollment and continued participation in PR should be prioritized, leveraging its potential cost savings to promote systems-level changes to support increased access to and participation in PR.

Summary Points

1. PR in COPD is associated with improvements in exercise tolerance, HRQOL, dyspnea, hospital readmission, and mortality (Figure 2).

2. Participation in PR after COPD hospitalization is associated with fewer days spent in the hospital or a nursing home over the next 12 months.

3. Participation in PR after COPD hospitalization is associated with a number-needed-to-treat of 12 to prevent one re-hospitalization and 8 to prevent one death within 12 months.

4. PR is a key medical intervention that is underutilized in patients with COPD.

5. Increasing access to PR could be a cost-effective strategy to reduce hospitalizations and improve quality of life in patients living with COPD.

Fig. 2.

Summary of benefits of pulmonary rehabilitation in COPD.

Panel Discussion

MacIntyre: Chris, that was terrific. I love your effectiveness analysis. One thing that bothers me, however, is the use of propensity scoring. Is there any way that propensity scoring can match what we heard about earlier today, motivation? You know, a person who doesn’t want to go to rehab might be a totally different person. In fact, probably is a totally different person than the person who wants to go to pulmonary rehab. And I don’t know how you can match for that in propensity matching using usual clinical criteria. Any way to get around that quandary?

Mosher: I think that’s a great point, Neil. I think you’re right. Unless you’re going to assume wholeheartedly that all of the characteristics they match for is what leads to motivation or not motivation, which I’m not sure we can say is true, there probably is a likelihood that there are just patients who are not motivated but look like those patients and they didn’t go to rehab. The ones who did, although they may have more comorbidities and look like that group, they’re highly motivated or they have transportation. They have a way to overcome a barrier that their counterpart who looks just like them on paper cannot overcome. And maybe that leads to differences in their readmission rate or death. That’s a good point.

Criner: Chris, very nice talk. When you look at the price point in the financial benefit to Medicare, what’s the financial penalty for the patient if you factored in parking, transportation when gas is $4 a gallon, loss of work of their family member bringing them, plus the co-pay? And the reason why I bring that up is those are things that patients say. Even smoking cessation programs, the most successful ones are when the patient is paid to stop smoking. There’s a financial incentive. So, if you flip the model around, could there be cost sharing that the patient gets paid to do rehab rather than being financially penalized? There’d still be a joint savings between Medicare and the participant, and the outcomes are better. Is there any credence to that kind of scheme?

Mosher: I certainly think you’re right, Jerry. In the societal perspective, which I did not include in my presentation, we do include the cost incurred by the patient; but we only included co-pays, gas prices, and the distance traveled. We thought about including time off work, but a number of these patients, particularly those in the Medicare population, largely do not work either because of age or disability. So, we didn’t feel like that was hugely important and would not have the data sources to adequately inform that modeling. But you’re right, from a patient perspective I’m sure that’s important. Or the lost work of their spouse or younger loved one who is taking them. I would love to get to a point where we as a society think that is important enough that we’re not only going to cover the cost but incentivize you. I personally—and maybe this is getting too far into politics—I feel like we’re so far away from that I would be afraid CMS would just slam the door if we proposed that. Instead, we could incrementally work toward that. If we could get to a point where the transportation is subsidized, even a portion of it, that would be a starting place to eventually get to a point where you’re actually financially incentivizing somebody to do it. I think there’s probably some credence to that.

Casaburi: Really stellar talk. I love the work you’ve done. This is a question for the group. Do we think that the pulmonary rehabilitation’s survival benefit is settled science? Do we believe that it truly exists based on Lindenauer’s work? Let me read you the last sentence of Lindenauer’s paper.¹ It says, “although the potential for residual confounding exists and further research is needed.” So, what is that further research and is it needed?

Bhatt: That’s the JAMA editor adding that.

Casaburi: Yes, I’m sure the JAMA editor encouraged them to add that. But there it is. What do you think?

Mosher: I will take the first stab at it. I don’t rule out the possibility that could be true. The part of me that, let’s say, not just wants to believe it but does believe it is using a pretty robust well-validated data set first of all, and maybe more importantly that the treatment effect size is quite large. If you don’t believe there is a mortality benefit associated with PR, I guess you’d be arguing there’s enough unaccounted for confounding to get all the way to 50% over the one. That seems like a high bar to clear for just some comorbidities or location or motivation that’s gone unaccounted for, particularly when you think about pulmonary rehab is exercise, using your medications, and taking care of yourself better. It’s plausible to think that the patient would live longer. I wouldn’t say it’s settled. The scientist in me wouldn’t say that, but I would tend to err on the side that I think it’s a true finding.

Moy: Rich [Casaburi], are you saying that the rehabilitation survival benefit is not settled science because the populations in the Lindenauer paper¹ were based on COPD subjects who had an index hospitalization and that we should be showing rehabilitation survival benefit for all comers with COPD?

Casaburi: Oh, that’s a different question. And I agree that a survival benefit hasn’t been shown in patients in a stable phase of their disease. I would bet it would be present as well, but it’s a much more difficult study. The reason why studying COPD patients who had just experienced a hospitalization for an exacerbation is easier is because these people are at high risk of mortality after hospitalization. If you take a lower-risk population, you need to study much larger numbers of people. I’m going to say this because when the statement is made that further research is needed; the further research would be a randomized trial. And I bear the scars from trying to organize the randomized trial to demonstrate the survival benefit of pulmonary rehabilitation. Barry Make was the prime mover, but we tried in 2008 and couldn’t get it funded. And then we tried it in 2018 and couldn’t get it funded. In 2008 it was a very difficult study to plan, but in 2018 it was almost impossible because of equipoise. Nobody wanted to agree that we would randomize people to not get rehabilitation. And I think in 2023 it would be absolutely impossible. So, I’m afraid that it’s as settled as it’s going to get. And I agree that the data we have are quite robust and convincing.

Evans: And to support that, although as you say we can’t account for all the unmeasured confounding, but you also have dose response data in there. That is again another element I appreciate not being the RCT that we would love as well.

Mosher: That’s a good point. I should have mentioned that as well.

Evans: You did, I’m just highlighting.

Bhatt: Chris, one comment you made I thought was very strong. You said this could be a potentially disease-modifying therapy. What do you speculate are the reasons? It doesn’t change lung function; it does change air trapping maybe to some extent; but other than that, how would you say this would be disease modifying? The outcomes are great.

Mosher: That’s a great question. My particular interest is looking at this from an immunological perspective. Particularly in those patients who are hospitalized, we understand that exacerbations are often triggered by viral infections or bacterial infections. We know that in non-COPD patients exercise augments the immune system or optimizes it. What I’m interested in understanding, is the exercise they’re doing modifying their immune system, priming them in a way that when they encounter that grandchild who blows RSV in their face, maybe they have an ability to weather that storm in a way that their counterpart who didn’t go to pulmonary rehab does not. And we’re hoping to begin work on that. I have a grant through the NIH that’s currently under review and hope to get a decision on that in a couple of months.

Moy: I love the analogy that you gave about anticoagulation, atrial fibrillation, and the comparable reduced risks in pulmonary rehab. Do you have an example like that for the cost savings of the $5,700? Is the cost saving of pulmonary rehab similar to those for breast cancer screening or lung cancer screening? Is there some other common medical intervention that we can hang our hats on that gives us the same type of cost savings that would be more understandable to the general population?

Mosher: That’s a really good point. I don’t have one offhand; that’s a good homework assignment for me. I think it will be hard, and I don’t claim to know this whole landscape, but what’s so unique is that this is cost savings. All those interventions you mentioned I’m going to bet are cost-effective. It may be hard to find an intervention. For example, vaccines which may be cost-effective but less often are they cost saving. But we as a society have decided they are important. Screening interventions to prevent cancer are usually cost-effective but may not necessarily be cost saving. So, Doctor Google will have to help me out with what’s a medical intervention that actually saves costs. But I think that’s a good point and one that we should really include in future visual graphics to help drive home the point.

Moy: For cost savings, we know that pulmonary rehab reduces hospital readmissions and hospitalizations. Why can’t we just multiply the cost of a COPD hospitalization by the average number of days of one of those hospitalizations, and say that’s what we’re saving? Why do we actually need a cost effectiveness analysis? If we have already shown that we reduce hospitalizations, can we just say the typical hospitalization is 3 d, and the typical cost of hospitalization is $1,500 a day? So here we are, we’re saving $4,500 rather than this $5,700 from a cost effectiveness analysis. Does the cost-effectiveness analysis define different thresholds or give different scenarios? What is the added benefit of a formal cost-effectiveness analysis?

Mosher: That’s a good point. I’m personally like you; if you want to just be practical, I think that would be fine. We wanted to take it all this way to try to just really silence any doubters, to say that we were going to apply the most rigid, rigorous methods to solving this question. And of course, when you’re doing that, then you’re accounting for the cost of pulmonary rehab itself, the traveling, the co-pays, the future cost of caring for COPD, all stratified by disease stage and accounting for mortality; all those things which would be a little more complicated than just adding up the days saved. We also included the utility benefit of performing rehab. We included that to impact the quality-of-life benefits. Our model includes patients who are stratified by different GOLD levels based upon a study that looked at basically what’s the demographics of patients who go to pulmonary rehab because we didn’t want people to say, “well maybe this isn’t for all patients, maybe it’s just those that are GOLD III or something.” We’ve really tried to dig down and be very detailed in our approach. And then whenever we came to a branching point, we always tried to be conservative. For example, it’s been shown, not surprisingly, that patients who go to the hospital less their quality of life is better. But we didn’t include that effect just to again err on the side of being conservative in our quality-of-life estimate.

Moy: My last question is why did you start from the year 2001 in your analyses? Is there something specific about that year?

Mosher: That’s just the time span of the data sources used. All the data sources that we used were from 2001 up to 2021. The model itself is being informed by those inputs, but the model’s just running for 1 year, and then it runs for what’s called a lifetime time horizon, essentially terminating after a patient would be 100 years old. We didn’t use calendar year, you know, 2010 to 2015. I was just summarizing that the data we used to inform the model is from the last 20 years as opposed to data from the 1980s that may be less relevant. And all the costs were based on costs converted to 2020 United States dollars.

Nici: I wanted to get back to the disease-modifying question because I think that’s a very interesting question. Chris, I think your point about immune regulation is important. But I also wonder, so many of our patients die of cardiovascular disease, much more than die of their lung disease. And so, at some point it would be interesting to see if pulmonary rehab actually impacts that comorbidity in particular because we know exercise is really good for cardiovascular health. And perhaps what we’re doing is improving cardiovascular fitness and dropping the number of patients who are dying of cardiovascular death. My other point is I think we can capitalize on the potential cost savings similar to other interventions like cardiac bypass. This could be an effective strategy for payers and receive a lot of attention from the public.

Mosher: Totally agree with your second point. And your first point really speaks to me. I’m very interested in the interplay between cardiovascular disease and COPD. I thought about looking at that using a similar approach that Lindenauer used. Unfortunately, because so few patients participate, 2,700, I think you would need a much larger sample size. Now that would just mean that probably you would look at it over many years, so there could be something to that. I’ve often wondered if it does have improvement across other diseases.

Nici: So what? You’re so young you could look at it again in 40 years.

Casaburi: Surya [Bhatt] asked the question about the mechanism of rehabilitation’s mortality benefit, and I thought it was interesting the way you asked it. So, rehabilitation doesn’t improve lung function. Well, if you improve lung function with bronchodilators are there mortality benefits? Not that we’ve been able to detect. Rehabilitation does, I believe, because of its improvement in physical fitness. It may be through associated immune effects; it may be through reduced cardiovascular risk, but I believe it’s through exercise training that the mortality benefit is achieved. And, again, that means we should be more and more focused on improving physical fitness outcomes that would maximize this benefit.

Bhatt: Agree.

Mosher: Just to echo that point, studies have shown maybe the best predictor of someone’s mortality is their V̇_O2 max. If you can measure the V̇_O2 max, that’s what will best inform their risk of mortality.