Mortality in Advanced Chronic Obstructive Pulmonary Disease and Heart Failure Following Cardiopulmonary Rehabilitation

Abstract

Cardiopulmonary rehabilitation (CR) improves physical function and quality of life (QoL) in chronic obstructive pulmonary disease (COPD) and heart failure (HF), but it is unknown if CR improves outcomes in very severe disease. This study’s purpose was to describe functional capacity (6-min walk distance [6MWD], steps/day), symptoms (dyspnea, depression), QoL (Short-Form Health Survey–Veterans [SF-36 V]) and cardiopulmonary function (N-terminal pro-brain natriuretic peptide [NT-proBNP], forced expiratory volume in 1 s [FEV₁]), and derive predictors of mortality among patients with severe COPD and HF who participated in CR.

Methods and Results:

In this secondary analysis of a randomized controlled trial comparing two CR methods in severe COPD and HF, 90 (COPD = 63, HF = 27) male veterans, mean age 66 ± 9.24 years, 79% Caucasian, and body mass index 31 kg/m², were followed for 12 months after CR. The COPD group had greater functional decline than the HF group (6MWD, p = .006). Dyspnea was lower (p = .001) and QoL higher (p = .006) in the HF group. Mean NT-proBNP was higher in the HF group at all time points. FEV₁ improved over 12 months in both groups (p = .01). Mortality was 8.9%, 16.7%, and 37.8% at 12, 24, and 60 months, respectively. One-year predictors of mortality were baseline total steps (<3,000/day), 6MWD (<229 meters), and NT-proBNP level (>2,000 mg/pg).

Conclusions:

In very severe COPD and HF, risks of mortality over 12 months can predict patients unlikely to benefit from CR and should be considered at initial referral.

In the United States, chronic obstructive pulmonary disease (COPD) and heart failure (HF) are among the top four leading causes of death (Li, Caughey, & Johnston, 2014; Osthoff, Jenkins, & Leuppi, 2013). The incidence of these diseases, both of which are incurable and chronic, is increasing (American Heart Association, 2016; American Lung Association, 2016). Although exacerbations are preventable and treatable with pharmacologic and nonpharmacologic interventions, both diseases are associated with significant morbidity and mortality (American Lung Association, 2016; Carson et al., 2015; Duncan, 2016; Singh & Yu, 2016). Although the trajectories of disease progression differ, symptoms and disability characteristics of HF and COPD are similar at the end of life (Gavazzi et al., 2015). Patients with COPD or HF frequently suffer from the incapacitating symptoms of shortness of breath, fatigue, and weakness. Consequently, both groups have profoundly impaired functional capacity in walking, activities of daily living, and quality of life (QoL; Apostolovic et al., 2011; Li et al., 2014; Padeletti, Jelic, & LeJemtel, 2008; Troosters et al., 2016). These bothersome symptoms cause significant physical and psychological distress that is related to mortality (Janssen, Spruit, Uszko-Lencer, Schols, & Wouters, 2011).

In the past two decades, research has identified similarities in the pathophysiological and functional concomitants of COPD and HF (Louvaris & Vogiatzis, 2015). For this reason, investigators are increasingly aware of the value of applying the traditional cardiopulmonary rehabilitation (CR) model to the management of both HF and COPD (Troosters & VanRemoortel, 2009). Exercise- and activity-based interventions to improve physical functioning and disease self-management in COPD and HF populations improve exercise performance, dyspnea, daily activity, illness exacerbation, QoL, and health-care costs (McCarthy et al., 2015; Tzanis et al., 2016). Recently, exercise-based rehabilitation programs have shown promise in the reduction of mortality in HF (Sagar et al., 2015), though similar benefits for COPD are less evident (McCarthy et al., 2015). The effect of such programs on symptoms and mortality in advanced disease, however, remains to be determined. This issue is particularly relevant to CR management because persons with very severe, end-stage HF and/or COPD comprise a high proportion of those referred to CR.

In order to impact survival and determine the efficacy of interventions that may reduce mortality, it is important to identify risk factors for mortality among patients with COPD and HF (Khazanie et al., 2015). Selected indicators, including the 6-min walk distance (6MWD), forced expiratory volume in 1 s (FEV₁), and natriuretic peptides (N-terminal pro-brain natriuretic peptide [NT-proBNP]), show promise for predicting mortality in both groups (Fonarow, 2012; Gaggin & Januzzi, 2013; Nunez et al., 2016; Richter et al., 2013; Singh & Yu, 2016). Similarly, reduced physical activity, defined by the World Health Organization as “any force exerted by skeletal muscles that results in energy expenditure above resting level” (Cavill, Kahlmeier, & Racioppi, 2006), is associated with increased risk of mortality in the elderly and those with cardiopulmonary illness (Waschki et al., 2015).

The present analysis was prompted by the findings of our previous randomized controlled trial comparing two forms of outpatient CR in patients in very severe HF or COPD (Steele, Dougherty, Burr, Gylys-Colwell, & Hunziker, 2017). In that study, we found no benefit in functional capacity (6MWD), daily activity (accelerometer), symptoms (dyspnea), or QoL (Short-Form Health Survey–Veterans [SF-36 V]) following outpatient CR. These results suggested that CR interventions in severe end-stage COPD or HF may have little ability to impact outcomes. Furthermore, it is likely that the lack of benefit in very ill subjects is also associated with mortality, rendering CR a less optimal choice. The purpose of the present secondary analysis was to (1) describe functional capacity, symptoms, and QoL by diagnostic group (severe COPD vs. HF) and by mortality group over time following participation in CR and (2) derive predictors of mortality among patients with very severe COPD and HF following participation in CR.

Method

Design of Parent Study

The purpose of the parent study was to compare functional capacity and daily activity outcomes between a standard 8-week outpatient CR program and a 4-week outpatient/home-based self-efficacy-based rehabilitation program. The results at 6 months (Steele et al., 2017) demonstrated that neither form of CR was successful in improving any of the outcomes, and in some instances, functional capacity and daily activity declined during the year after completion of CR in both COPD and HF. Because neither form of CR resulted in significant benefits in either diagnostic group, we combined all patients into one sample for the current analysis.

Setting, Participants, and Interventions

We recruited participants between 2009 and 2012 from the VA Puget Sound Health Care System, Seattle, WA. We randomly assigned outpatients (HF: n = 27 or COPD: n = 63) to one of the two intervention groups when they met the following criteria: optimally managed, severe COPD (FEV₁ predicted < 50%) or HF with reduced ejection fraction (HFrEF; EF < 40%), one disease-related hospitalization in the past 2 years or at least two unscheduled outpatient visits over the past year related to cardiopulmonary disease, a working phone, and willingness to participate in an outpatient exercise and self-management program. Exclusion criteria were unstable disease or recent surgery that precluded exercise, a supplemental oxygen requirement at rest of >4 L/min, already exercising 3 times a week, inability to ambulate, uncontrolled mental illness, alcohol or drug abuse, or a life expectancy of less than 1 year. Participants provided written informed consent, and the VA Research and Development and Human Subjects Committees Institutional Review Board at the VA Puget Sound Health Care System approved the trial. An independent data and safety monitoring board provided oversight for procedures, protocol, and events. We carried out randomization and preprogram (Time 1) measures prior to beginning the CR programs. A team member blinded to group assignment completed postprogram measures.

CR Interventions

The two CR programs we tested were based on social cognitive theory, incorporating concepts of self-efficacy and self-regulation. Briefly, the 4-week outpatient CR adherence program consisted of an outpatient/home-based program of 4 hr of exercise and integrated self-management education weekly, followed by 5 months of weekly phone calls encouraging home exercise for 30 min 3×/week (n = 46). Standard outpatient CR consisted of 8 weeks of 2 hr of exercise and self-management instruction weekly (n = 44). Outpatient exercise sessions in both groups began with 5 min of upper- and lower extremity stretching, followed by walking on a treadmill for a goal of 20 min and exercise on the NuStep for 20 min at a pace that reached a Borg rating of 3–5 (moderate to strenuous). Following the aerobic exercise, there were 5 min of free arm weights and 5 min of arm ergometry. A 5-min cooldown consisting of upper- and lower extremity stretching concluded each session. All outpatient exercise sessions were monitored using blood pressure, telemetry (heart rate and arrhythmias), and O₂-saturation monitoring. The number of minutes and type of aerobic activity was individually prescribed and was based on the baseline 6MWD.

Following the 8 weeks of standard CR, we gave patients a pedometer to track steps/day and encouraged them to walk 30 min 3×/week. Following the 4-week outpatient phase for the adherence intervention, we completed a home safety evaluation. We asked patients to walk for 30 min 3×/week, record walking sessions in an exercise log, and self-monitor symptoms and were followed them with weekly coaching phone calls for 3 months to help them maintain exercise after the outpatient program. For the final 2 months of this program, we made biweekly coaching phone calls and asked participants to record home walking and self-monitor symptoms.

Outcome Measures

Primary outcomes in the parent study included functional capacity measured by (1) 6MWD in meters (best of two walks; Cavanaugh, Coleman, Gaines, Laing, & Morey, 2007) and (2) daily physical activity (total steps/24 hr) measured using the StepWatch® Activity Monitor (SAM; Orthocare Innovations, LLC), a pager-sized device worn above the dominant ankle that records stride cycles. Participants wore the SAM for a minimum of 10 hr for a least 5 consecutive days. The measure used from the SAM was total steps taken during 10 hr worn (stride counts doubled; Haeuber, Shaughnessy, Forrester, Coleman, & Macko, 2004; Nguyen, Burr, Gill, & Coleman, 2011). Secondary outcomes were symptoms of dyspnea (Clinical COPD Questionnaire; van der Molen et al., 2003) and depressed mood (Geriatric Depression Scale; Watson & Pignone, 2003), general health-related QoL-Veterans version (SF-36 V; Boueri, Bucher-Bartelson, Glenn, & Make, 2001; Kazis et al., 1998; Sprenkle, Niewoehner, Nelson, & Nichol, 2004), and cardiopulmonary function [NT-pro-BNP; Jelic & Le Jemtel, 2006] and FEV₁ [Miller et al., 2005]). Measures were taken at (1) baseline (study entry), (2) 6 months, and (3) 12 months.

We determined mortality by reports taken from the electronic health record (EHR) from study entry through 60 months following randomization or when we discovered that a patient had died. We noted causes of death from the EHR.

Analysis (Present Study)

To describe the longitudinal trajectories of functional capacity, symptoms, and QoL in the two diagnostic groups (COPD or HF) and the “mortality groups” (dead vs. alive), longitudinal generalized estimating equations (GEEs) controlling for baseline age, body mass index (BMI), and Charlson comorbidity index were used (Charlson, Pompei, Ales, & MacKenzie, 1987). Adjusted standardized odds ratios for potential risk factors for mortality were estimated using univariate logistic regression. Standardized variables were employed to enhance comparability because the scales of the logistic regression coefficients, and hence the estimated odds ratios, depended on the scales of the independent variables in the model. In addition, because the variable NT-proBNP typically has a non-normal right-skewed statistical distribution, it was log-transformed prior to standardization using a Tukey-bias natural logarithm transform (LnNT − proBNP = Ln(1 + NT-proBNP)).

Empirical multivariate models predicting mortality at 12, 24, and 60 months were developed based on the previously described univariate analyses. Potential risk factors for mortality were entered into a multivariate logistic regression model using a forward stepwise procedure. The robust Eicker–Huber–White “sandwich” estimator and the autoregressive lag 1 form for the initial working covariance matrix were specified for the GEE model. Because this was a secondary analysis, we interpret statistical significance at the p ≤ .05 level conservatively.

Results

Patient Characteristics

There were a total of 90 participants, two thirds of whom had a primary diagnosis of COPD. The mean age for the total group was 66 ± 9.24 years, 79% of participants were Caucasian, the average BMI was 31 kg/m², more than half had a college degree or higher, and all were male veterans (Table 1). There were statistically significant differences between the COPD and HF groups in age (COPD > HF), oxygen use (COPD > HF), NT-proBNP level (HF > COPD), and FEV₁ (COPD < HF). The percentage of COPD patients with severe and very severe symptoms (GOLD stages III and IV) was significantly higher than that of patients with HF. There were 12 patients with a diagnosis of HFrEF who also had an initial baseline FEV₁ < 50% predicted. Mortality across both groups was 8.9% at 12 months, 16.7% at 24 months, and 37.8% at 60 months. The COPD group had higher mortality at all time points compared to the HF group, with the exception of 24 months.

Table 1.

Baseline Demographic and Clinical Characteristics (Mean ± SD or n [%]).

Characteristic	Total (N = 90)	COPD (n = 63)	HF (n = 27)	p (D × Gp)
Age (years)	66.0 ± 9.24	67.3 ± 9.43	63.0 ± 8.17	.03
Race				.53
Caucasian	71 (78.9)	49 (77.8)	22 (81.5)
African American	14 (15.6)	10 (15.9)	4 (14.8)
Asian/Pacific Islander	3 (3.3)	2 (2.2)	1 (1.1)
Hispanic	1 (1.1)	1 (1.1)	0
American Indian	1 (1.1)	1 (1.1)	0
BMI (kg/m2)	31.2 ± 7.46	30.7 ± 7.06	32.4 ± 8.35	.37
Education				.36
Elementary school	5 (5.6)	5 (7.9)	0 (0)
High school	32 (35.6)	24 (38.1)	8 (29.6)
College/tech school	47 (52.2)	30 (47.6)	17 (63.0)
Graduate school	6 (6.7)	4 (6.3)	2 (7.4)
Marital status				.31
Never married	9 (10.0)	6 (9.5)	3 (11.1)
Married/partnered	27 (30.0)	22 (34.9)	5 (18.5)
Separated, divorced, widowed	54 (60.0)	35 (55.6)	19 (70.4)
Employment status				.42
Paid employment	12 (13.3)	10 (15.9)	2 (7.4)
Unemployed	7 (7.8)	4 (6.3)	3 (11.1)
Disabled	33 (36.7)	25 (39.7)	8 (29.6)
Retired	38 (42.2)	24 (38.1)	14 (51.9)
Oxygen use	43 (47.8)	39 (61.9)	4 (14.8)	<.001
Current smoker	24 (26.7)	16 (25.4)	8 (29.6)	.79
Charlson score	2.17 ± 1.20	2.05 ± 1.11	2.44 ± 1.37	.19
EF% (HF group only)	30.9 ± 6.68		30.9 ± 6.68
NT-proBNP	1,086.3 ± 2,013.6	468.6 ± 900.9	2,527.5 ± 2,977.0	.001
FEV1	39.8 ± 11.3	32.2 ± 12.3	57.7 ± 19.2	<.001
GOLD staging				<.001
Stage I: FEV1 > 80%	4 (4.4)	0 (0.0)	4 (14.8)
Stage II: FEV1 > 50% and < 80%	17 (18.9)	6 (9.5)	11 (40.7)
Stage III: FEV1 > 30% and < 50%	40 (44.4)	31 (49.2)	9 (33.3)
Stage IV: FEV1 < 30%	29 (32.2)	26 (41.3)	3 (11.1)

Note. The p values reported for the continuous variables are based on independent groups. Student’s t-test (unequal variance formula) was used to contrast the COPD and HF groups. The p values reported for the categorical variables are based on a general (R×C) Fisher’s exact test. BMI = body mass index; COPD = chronic obstructive pulmonary disease; EF = ejection fraction; FEV₁ = forced expiratory volume in 1 s; HF = heart failure; NT-proBNP = N-terminal pro-brain natriuretic peptide; SD = standard deviation.

Outcomes by Diagnostic Group

Participants with COPD had significantly greater functional decline compared to those with HF as indicated by the 6MWD (p = .006; Table 2). Those with HF had higher 6MWD at baseline study entry and experienced an increase in 6MWD at 6 months following CR. Those with HF reported lower levels of dyspnea symptoms and higher physical composite scores on the SF-36 V, but these were not significantly different between the groups. Self-reported SF-36 V mental composite scores were comparable to other COPD and HF samples and were low in both groups. Although the score on the physical function subscale of the SF-36 V was lower in the COPD group than in the HF group across all time points, physical function improved at 6 months in both COPD and HF groups (p = .04). The mean level of NT-proBNP was higher in the HF group than the COPD group at all time points and declined in the HF group over time (p = .02). FEV₁ showed a significant improvement over 12 months in both groups (p = .02).

Table 2.

Functional Capacity, Symptoms, and Quality of Life: Baseline and 6 and 12 Months, by Diagnostic Group.

	Time Point			Wald χ2		p Value
Variable	Baseline (Mean ± SD)	6 Months (Mean ± SD)	12 Months (Mean ± SD)	Groupa	Group × Timeb	Groupa	Group × Timeb
Functional capacity
6MWD (m)				3.68	10.37	.05	.006
Total	354.69 ± 117.39	343.17 ± 113.89	314.29 ± 131.29
COPD	339.21 ± 115.82	317.06 ± 123.52	302.77 ± 129.66
HF	390.82 ± 115.12	407.82 ± 139.54	343.08 ± 134.24
Total steps/24 hr				0.22	2.33	.64	.31
Total	5,337.27 ± 3,053.33	5,244.92 ± 3,120.11	5,327.01 ± 3,324.02
COPD	5,065.39 ± 2,668.28	5,145.79 ± 3,174.35	5,185.51 ± 3,237.43
HF	5,971.66 ± 3,786.54	5,522.48 ± 3,024.71	5,680.72 ± 3,593.27
Symptoms
Dyspnea				15.92	3.51	<.001	.17
Total	3.02 ± 1.40	2.69 ± 1.36	2.85 ± 1.58
COPD	3.29 ± 1.22	3.13 ± 1.12	3.16 ± 1.43
HF	2.39 ± 1.60	1.54 ± 1.25	2.13 ± 1.70
Quality of life (SF-36 V)
PCS				7.42	4.10	.006	.12
Total	31.96 ± 8.18	32.83 ± 10.26	32.02 ± 10.99
COPD	30.17 ± 7.47	30.87 ± 8.92	31.14 ± 10.64
HF	36.13 ± 8.39	37.81 ± 11.91	34.04 ± 11.75
MCS				0.15	0.76	.70	.68
Total	48.84 ± 11.82	51.47 ± 11.66	50.04 ± 11.53
COPD	49.35 ± 12.24	51.72 ± 11.29	49.45 ± 12.27
HF	47.66 ± 10.89	50.87 ± 12.83	51.37 ± 9.73
Physical Function				7.09	6.19	.008	.04
Total	38.63 ± 20.68	43.53 ± 23.55	41.76 ± 25.00
COPD	34.76 ± 18.01	37.60 ± 21.43	39.34 ± 24.55
HF	50.00 ± 22.36	58.36 ± 22.31	47.39 ± 25.66
General Health				3.07	1.35	.08	.51
Total	46.69 ± 18.41	48.67 ± 20.83	48.85 ± 20.58
COPD	45.07 ± 18.97	46.29 ± 21.41	48.17 ± 20.65
HF	50.48 ± 16.76	54.72 ± 18.55	50.56 ± 20.80
Cardiopulmonary function
NT-proBNP				14.20	7.33	<.001	.02
Total	1,086.28 ± 2,013.60	727.10 ± 958.32	708.86 ± 1,124.69
COPD	468.62 ± 900.88	370.93 ± 598.79	452.60 ± 787.30
HF	2,527.48 ± 2,977.00	1,374.25 ± 1,149.25	1,319.00 ± 1,534.14
FEV1				37.32	8.23	<.001	.02
Total	39.85 ± 19.20	41.82 ± 22.35	44.62 ± 20.87
COPD	32.18 ± 12.30	34.18 ± 14.86	35.81 ± 15.32
HF	57.74 ± 20.65	61.30 ± 26.51	66.42 ± 16.37
Self-efficacy
Self-efficacy exercise				2.09	1.16	.14	.56
Total	6.44 ± 2.33	5.62 ± 2.44	5.28 ± 2.88
COPD	6.36 ± 2.31	5.58 ± 2.54	4.99 ± 2.81
HF	6.63 ± 2.40	5.74 ± 2.23	5.94 ± 2.99
Self-efficacy walking				3.93	4.81	.05	.09
Total	2.88 ± 2.42	3.82 ± 3.07	3.19 ± 2.92
COPD	2.25 ± 2.04	3.38 ± 2.96	3.11 ± 2.89
HF	4.14 ± 2.74	4.95 ± 3.24	3.39 ± 3.02

Note. Total N = 90 (COPD: n = 63, HF: n = 27). 6MWD = 6-min walk distance; COPD = chronic obstructive pulmonary disease; HF = heart failure; FEV₁ = forced expiratory volume in 1 s; MCS = mental component score; NT-proBNP = N-terminal pro-brain natriuretic peptide; PCS = physical component score; SF-36 V = Short-Form 36–Veterans version.

^aGroup differences between COPD and HF. ^bCOPD and HF Trajectory × Time.

Predictors of Mortality

Risk factors for mortality at 12, 24, and 60 months for the entire sample (Table 3) showed that baseline total steps/24 hr, 6MWD, and log NT-proBNP level were significantly associated with mortality at all time periods in both COPD and HF groups. At 12 months, the odds of death were higher for those with a higher dyspnea score and lower self-efficacy for exercise, SF-36 V physical function, and SF-36 V general health. At 24 months, additional predictors of mortality were FEV₁, SF-36 V physical function, and the dyspnea score. At 60 months, the same factors that were strong predictors of mortality at both 12 and 24 months continued to be statistically significant, with the addition of self-efficacy for walking.

Table 3.

Risk Factors for Mortality for Veterans With HF and COPD, Combined, at 12, 24, and 60 Months After Cardiopulmonary Rehabilitation.^a

Variable	Odds Ratio (95% CI)	p Value
At 12 months	Alive: 82; Dead: 8
Total steps/24 hr	0.35 [0.13, 0.90]	.03
6MWD	0.15 [0.04, 0.55]	<.01
Log NT-proBNP	4.16 [1.43, 12.1]	.01
COPD Dyspnea Scale	3.22 [1.22, 8.53]	.02
Self-efficacy exercise	0.33 [0.12, 0.90]	.03
SF-36 V, physical function	0.28 [0.10, 0.78]	.02
SF-36 V, general health	0.19 [0.05, 0.69]	.01
At 24 months	Alive: 75; Dead: 15
Total steps/24 hr	0.34 [0.17, 0.70]	.01
6MWD	0.28 [0.13, 0.60]	.01
FEV1	0.37 [0.17, 0.77]	.01
Log NT-proBNP	2.98 [1.43, 6.19]	.01
COPD Dyspnea Scale	2.45 [1.26, 4.78]	.01
SF-36 V, physical function	.50 [0.26, 0.94]	.03
At 60 months	Alive: 56; Dead: 34
Total steps/24 hr	0.38 [0.22, 0.66]	<.01
6MWD	0.37 [0.21, 0.66]	<.01
Log NT-proBNP	2.81 [1.58, 4.99]	<.01
Self-efficacy walking	0.50 [0.30, 0.83]	.01

Note. N = 90. 6MWD = 6-min walk distance; CI = confidence interval; COPD = chronic obstructive pulmonary disease; FEV₁ = forced expiratory volume in 1 s; HF = heart failure; Log NT-proBNP = N-terminal pro-brain natriuretic peptide; SF-36 V = Short-Form 36–Veterans version.

^aControlling for age, body mass index, and Charlson score.

In the multivariate logistic regression model for the entire sample (Table 4), COPD Dyspnea Scale scores and self-efficacy for walking were predictive of death at 12 months. Similarly, FEV₁ and log NT-proBNP were predictive of death at 24 months, and total steps/24 hr and log NT-proBNP were predictive of death at 60 months. The Nagelkerke R ² was 0.65 at 12 months, 0.40 at 24 months, and 0.47 at 60 months. The percentage correctly classified was 94.4% at 12 months, 85.6% at 24 months, and 82.2% at 60 months.

Table 4.

Multivariate Predictors of Mortality for Veterans With HF and COPD, Combined, at 12, 24, and 60 Months After Cardiopulmonary Rehabilitation.

Variable	β	SE	Wald Z 2	p Value
At 12 months
COPD Dyspnea Scale	1.56	.75	4.33	.04
Self-efficacy walking	−2.61	.99	6.83	<.001
Overall	χ2 = 21.7, df = 6. Nagelkerke R 2 = .65 Correct classification = 94.4%
At 24 months
FEV1	−1.42	.47	9.06	.003
Log NT-proBNP	1.58	.50	9.76	.002
Overall	χ2 = 24.5, df = 5. Nagelkerke R 2 = .40 Correct classification = 85.6%
At 60 months
Total steps/24 hr	−0.81	.37	4.89	.027
Log NT-proBNP	1.02	.30	7.59	.006
Overall	χ2 = 38.4, df = 6. Nagelkerke R 2 = .47 Correct classification = 82.2%

Note. N = 90. COPD = chronic obstructive pulmonary disease; FEV₁ = forced expiratory volume in 1 s; HF = heart failure; NT-proBNP = N-terminal pro-brain natriuretic peptide; SE = standard error.

Outcomes by Mortality Groups

The trajectories in functional capacity, QoL, symptoms, and self-efficacy by mortality groups (Table 5) demonstrate that four variables differed significantly between the survivors and nonsurvivors over 12 months. These included baseline levels of dyspnea symptoms (p = .04), mental health (p = .02), self-efficacy for walking (p = .05), and FEV₁ (p = .03).

Table 5.

Functional Capacity, Symptoms, and Quality of Life at Baseline and 6 and 12 Months After Cardiopulmonary Rehabilitation in Veterans With HF and COPD, Combined, by Mortality Group.

Variable	Time Point			Wald χ2		p Value
	Baseline (Mean ± SD)	6 Months (Mean ± SD)	12 Months (Mean ± SD)	Groupa	Group × Timeb	Groupa	Group × Timeb
Functional capacity
6MWD (m)				2.36	1.13	.12	.29
Total	354.69 ± 117.39	343.17 ± 113.89	314.29 ± 131.29
Alive at 12 months	367.28 ± 113.48	349.62 ± 131.62	314.29 ± 131.29
Dead at 12 months	225.70 ± 73.13	192.63 ± 110.74	—
Total steps/24 hr				1.37	0.89	.24	.34
Total	5,337.27 ± 3,053.33	5,244.92 ± 3,120.10	5,327.00 ± 3,324.02
Alive at 12 months	5,551.47 ± 3,077.57	5,327.79 ± 3,154.98	5,327.00 ± 3,324.02
Dead at 12 months	3,141.74 ± 1,699.58	3,558.53 ± 611.42	—
Symptoms
COPD Dyspnea Scale				3.99	0.76	.04	.38
Total	3.02 ± 1.40	2.69 ± 1.36	2.85 ± 1.58
Alive at 12 months	2.90 ± 1.38	2.67 ± 1.32	2.85 ± 1.58
Dead at 12 months	4.25 ± 0.93	3.17 ± 2.27	—
Quality of life
SF-36 V, PCS				0.29	0.54	.59	.46
Total	31.96 ± 8.19	32.83 ± 10.27	32.02 ± 10.99
Alive at 12 months	32.63 ± 7.95	33.35 ± 10.06	32.02 ± 10.99
Dead at 12 months	25.12 ± 7.84	19.96 ± 7.56	—
SF-36 V, MCS				5.92	11.84	.02	.001
Total	48.84 ± 11.81	51.47 ± 11.67	50.04 ± 11.53
Alive at 12 months	49.11 ± 12.06	51.20 ± 11.81	50.04 ± 11.53
Dead at 12 months	46.06 ± 9.08	58.05 ± 3.46	—
SF-36 V, physical function				5.04	1.08	.025	.29
Total	38.63 ± 20.68	43.53 ± 23.55	41.76 ± 25.00
Alive at 12 months	40.27 ± 20.70	44.67 ± 22.90	41.76 ± 25.00
Dead at 12 months	21.85 ± 11.31	15.00 ± 25.98	—
SF-36 V, general health				0.33	2.33	.57	.12
Total	46.69 ± 18.41	48.67 ± 20.88	48.89 ± 20.58
Alive at 12 months	48.44 ± 17.85	50.16 ± 19.72	48.89 ± 20.58
Dead at 12 months	28.50 ± 14.82	11.67 ± 16.07	—
Cardiopulmonary function
FEV1				4.65	2.23	.03	.14
Total	39.85 ± 19.20	41.82 ± 22.35	44.62 ± 20.87
Alive at 12 months	40.83 ± 19.58	42.49 ± 22.49	44.62 ± 20.87
Dead at 12 months	29.75 ± 11.18	25.50 ± 3.53	—
NT-proBNP				1.15	2.79	.28	.09
Total	1,086.28 ± 2,013.60	727.10 ± 2,013.60	708.86 ± 1,124.69
Alive at 12 months	985.93 ± 1,913.75	722.55 ± 968.84	708.86 ± 1,124.69
Dead at 12 months	2,114.88 ± 2,800.81	863.50 ± 775.70	—
Self-efficacy
Self-efficacy exercise				0.30	0.14	.59	.71
Total	6.44 ± 2.33	5.63 ± 2.45	5.28 ± 2.89
Alive at 12 months	6.61 ± 2.24	5.72 ± 2.42	5.28 ± 2.89
Dead at 12 months	4.67 ± 2.62	3.22 ± 2.14	—
Self-efficacy walking				4.02	0.45	.05	.83
Total	2.82 ± 2.42	3.82 ± 3.08	3.20 ± 2.92
Alive at 12 months	2.96 ± 2.48	3.92 ± 3.09	3.20 ± 2.92
Dead at 12 months	1.38 ± .92	1.33 ± 1.53	—

Note. Total N = 90; alive at 12 months = 82; dead at 12 months = 8. 6MWD = 6-min walk distance; COPD = chronic obstructive pulmonary disease; HF = heart failure; FEV₁ = forced expiratory volume in 1 s; MCS = mental component score; NT-proBNP = N-terminal pro-brain natriuretic peptide; PCS = physical component score; SF-36 V = Short-Form 36–Veterans version.

^aGroup differences between those who were alive at 12 months compared with those who had died by 12 months. ^bDead and alive Trajectory × Time.

Discussion

The parent intervention study compared two forms of CR and demonstrated that those with very severe COPD or HF experienced few benefits in functional capability, symptoms, and QoL following participation in CR. These findings are in stark contrast with the CR literature that demonstrates clear benefit in these outcomes in less severe disease (Fleg et al., 2015; McCarthy et al., 2015). In general, those with COPD in the present study had worse outcomes than those with HF. In both groups, however, there were three consistent risk factors for mortality across the 60-month study period: lower baseline 6MWD and total steps/day and higher NT-proBNP levels. Our findings suggest that those with baseline 6MWD < 229 m, total steps < 3,000/day, or NT-proBNP > 2,000 mg/pg are not likely to benefit from CR or survive beyond 1 year. There were a number of other predictors of mortality that were not the same at 12, 24, and 60 months, but that were related to lower physical capacity at baseline entry into CR. Some of these risk factors are interrelated; for example, 6MWD may be limited by dyspnea in both COPD and HF groups (Troosters et al., 2016).

In the present study, mortality rates in both the COPD and HF groups increased from 8% at 12 months to 34% at 60 months. Mortality rates in these two patient groups have varied across studies using different interventions, with COPD patients having higher mortality rates than those with HF. Other CR programs in HF have demonstrated reductions in hospitalizations but not mortality rate (Fleg et al., 2015; Jolly et al., 2009). In COPD, patients with moderate/severe disease did not experience improved mortality over 3 years (Guell et al., 2017). However, a meta-analysis suggested that patients with COPD who participated in a pulmonary rehabilitation program after an exacerbation had improved mortality (Puhan, Scharplatz, Troosters, & Steurer, 2005). None of these prior studies included participants with very severe disease, as in the current investigation.

Functional Capacity and Physical Activity

Patients with COPD or HF have impaired skeletal muscle function that decreases functional capacity, resulting in poor physical activity and lower self-reported ratings on the SF-36 physical component scale (Janssen et al., 2011; Jelic & Le Jemtel, 2006). Researchers have frequently identified the 6MWD as a predictor of mortality as well as a prognostic tool in both COPD and HF (Dajczman et al., 2015; O’Connor et al., 2012; Waatevik et al., 2016). Studies have shown the 6MWD to be a better predictor of mortality than FEV₁ (Martinez et al., 2006; Waatevik et al., 2016). It has also been associated with mortality specifically in COPD patients with severe disease (Durheim et al., 2015; Polkey et al., 2013). Dajczman et al. (2015) recently noted that preprogram 6MWD predicted survival at 3 years in severe COPD, where those who were able to walk <150 m had reduced likelihood of survival. These findings are similar to those of Steele et al. (2008), in which improvements in 6MWD, SF36 V physical functioning, and self-efficacy for walking after pulmonary rehabilitation were related to 2-year survival.

Past data have shown that physical activity (total steps/24 hr or self-reported physical function) is associated with mortality in COPD and HF (Blumenthal et al., 2016; Chamberlain et al., 2013; Dwyer et al., 2015). Our results demonstrated that total steps/24 hr and 6MWD were associated with mortality over 60 months. Survivors after 12 months had higher mean total steps/24 hr at baseline than nonsurvivors. Regardless of the disease state, there was an association between higher total steps/24 hr at baseline and reduced risk of mortality in adults (Dwyer et al., 2015).

Symptoms

A high COPD Dyspnea Scale score was a risk factor for mortality at 12 months and 24 months in the entire sample in the present study, though we found higher mean scores in patients with COPD compared to patients with HF. In general, more patients with COPD experience dyspnea than patients with HF (Janssen et al., 2011; Man et al., 2016). Nearly 60% of patients with COPD complain of moderate or severe shortness of breath related to walking distance (Padeletti et al., 2008; Theander et al., 2014). Shortness of breath is a predictor of both short- and long-term mortality in patients with COPD (Mentz et al., 2012; Yohannes et al., 2016). Depression is another condition prevalent in both COPD and HF (Janssen et al., 2011; Pelle, Gidron, Szabo, & Denollet, 2008). In the present study, we did not find depression to be related to mortality at any time point. Others, however, have found depression to be associated with increased mortality in advanced COPD (Yohannes et al., 2016).

Cardiopulmonary Function

Researchers have commonly used NT-proBNP and BNP levels to predict negative outcomes among patients with HF, to evaluate HF management, and to differentiate COPD from HF in those with shortness of breath (Gaggin & Januzzi, 2013; Jelic & Le Jemtel, 2006; O’Connor et al., 2012; Richter et al., 2013; Troosters et al., 2016). Authors have recommended measurement of NT-proBNP among patients experiencing acute dyspnea, and researchers have used it to predict mortality in patients with acute HF as well as in patients hospitalized with HF (Baggish, van Kimmenade, & Januzzi, 2008; Fonarow, 2012; Richter et al., 2013). In the present analysis, NT-proBNP levels were higher in patients with HF than in those with COPD at every time point. Acute hypoxemia or left ventricular systolic dysfunction in COPD can cause an increase in natriuretic peptide levels, even though the values in COPD are not typically as high as in HF (Rutten et al., 2007). Thus, NT-proBNP may be used to diagnose left ventricular systolic dysfunction in the presence of dyspnea and to predict mortality among patients with both COPD and HF.

Limitations

The parent study was not originally powered to find differences in mortality, so readers are cautioned to interpret statistically significant results conservatively. The sample size in each diagnostic group was small and not normally distributed. A larger proportion of participants had COPD, and most of these had severe disease, with at least 84% classified as GOLD III and IV. Mortality results may be biased because of unbalanced numbers in each diagnostic group. We conducted the study in a male VA population, so generalizability to other groups is limited.

Conclusions

Risk factors for mortality at 12, 24, and 60 months for the entire sample of male veterans with very severe COPD or HF showed that fewer total steps/24 hr, lower 6MWD, and higher NT-proBNP levels were significantly associated with mortality at all time periods in both diagnostic groups. More rigorous selection criteria for CR will result in the participation of fewer patients who are unlikely to receive benefits from the program and who may experience greater discomfort and risk from an exercise-based program. Such patients may attain greater benefit from a less demanding course of more limited exercise, rest, and socialization. More effective selection of CR participants will thus result in more efficient and efficacious use of staff and resources in this highly valued therapeutic field.