Nejmoa 2026141
Nejmoa 2026141
Nejmoa 2026141
The
journal of medicine
established in 1812 July 15, 2021 vol. 385 no. 3
a bs t r ac t
BACKGROUND
Older patients who are hospitalized for acute decompensated heart failure have From the Department of Internal Medi-
high rates of physical frailty, poor quality of life, delayed recovery, and frequent cine, Sections of Cardiovascular Medi-
cine (D.W.K., M.B.N., B.U.) and Geron-
rehospitalizations. Interventions to address physical frailty in this population are tology and Geriatric Medicine (D.W.K.,
not well established. M.A.E.), and the Departments of Neurol-
METHODS ogy (P.D.) and Biostatistics and Data Sci-
ence (H.C., M.A.E.), Wake Forest School
We conducted a multicenter, randomized, controlled trial to evaluate a transi- of Medicine, Winston-Salem, the Depart-
tional, tailored, progressive rehabilitation intervention that included four physical- ment of Orthopedic Surgery, Doctor of
function domains (strength, balance, mobility, and endurance). The intervention Physical Therapy Division (A.M.P.), the
Department of Medicine, Division of Car-
was initiated during, or early after, hospitalization for heart failure and was con- diology (R.J.M.), and the Department of
tinued after discharge for 36 outpatient sessions. The primary outcome was the Population Health Sciences (S.D.R.),
score on the Short Physical Performance Battery (total scores range from 0 to 12, Duke University School of Medicine,
Durham, and Novant Health Heart and
with lower scores indicating more severe physical dysfunction) at 3 months. The Vascular Institute, Charlotte (G.R.R.) —
secondary outcome was the 6-month rate of rehospitalization for any cause. all in North Carolina; the Department of
RESULTS Medicine, Sidney Kimmel Medical Col-
lege at Thomas Jefferson University
A total of 349 patients underwent randomization; 175 were assigned to the reha- (D.J.W.), and the Department of Physical
bilitation intervention and 174 to usual care (control). At baseline, patients in each Therapy, Jefferson College of Rehabilita-
group had markedly impaired physical function, and 97% were frail or prefrail; tion Sciences at Thomas Jefferson Uni-
versity (L.A.H.) — both in Philadelphia;
the mean number of coexisting conditions was five in each group. Patient reten- and Inova Heart and Vascular Institute,
tion in the intervention group was 82%, and adherence to the intervention sessions Fairfax, VA (C.M.O.). Address reprint re-
was 67%. After adjustment for baseline Short Physical Performance Battery score quests to Dr. Kitzman at the Department
of Internal Medicine, Sections of Cardio-
and other baseline characteristics, the least-squares mean (±SE) score on the Short vascular Medicine and Gerontology and
Physical Performance Battery at 3 months was 8.3±0.2 in the intervention group Geriatric Medicine, Wake Forest School
and 6.9±0.2 in the control group (mean between-group difference, 1.5; 95% con- of Medicine, 1 Medical Center Blvd.,
fidence interval [CI], 0.9 to 2.0; P<0.001). At 6 months, the rates of rehospitaliza- Winston-Salem, NC 27157-1045, or at
dkitzman@wakehealth.edu.
tion for any cause were 1.18 in the intervention group and 1.28 in the control
group (rate ratio, 0.93; 95% CI, 0.66 to 1.19). There were 21 deaths (15 from car- A list of the investigators in this trial is pro-
vided in the Supplementary Appendix,
diovascular causes) in the intervention group and 16 deaths (8 from cardiovascular available at NEJM.org.
causes) in the control group. The rates of death from any cause were 0.13 and 0.10,
This article was published on May 16,
respectively (rate ratio, 1.17; 95% CI, 0.61 to 2.27). 2021, at NEJM.org.
CONCLUSIONS
N Engl J Med 2021;385:203-16.
In a diverse population of older patients who were hospitalized for acute decom- DOI: 10.1056/NEJMoa2026141
pensated heart failure, an early, transitional, tailored, progressive rehabilitation Copyright © 2021 Massachusetts Medical Society.
A
cute decompensated heart failure tions. The protocol, which is available at NEJM.org,
is the leading cause of hospitalization was approved by the institutional review board
among older persons in the United States1 at each site. An independent data and safety mon-
and is associated with poor health-related qual- itoring committee evaluated patient safety. The
A Quick Take
ity of life, frequent rehospitalizations, high mor- authors vouch for the accuracy and completeness
is available at tality, and costs exceeding $39 billion per year.1,2 of the data and analyses and for the fidelity of
NEJM.org Most intervention trials in acute decompensated the trial to the protocol.
heart failure have had neutral results, which
suggests that outcomes may be driven in part by Patients and Randomization
mechanisms that have been overlooked.3-5 Patients were screened at the time of hospital
Among older patients with acute heart fail- admission and were enrolled before discharge.
ure, physical function is markedly impaired, and Patients were eligible for participation if they
frailty rates and the burden of coexisting condi- were 60 years of age or older, if they had been
tions are high.5-8 Even among older patients with admitted for acute decompensated heart failure
stable and well-compensated heart failure, se- regardless of ejection fraction, if they could walk
vere impairments in physical function are often at least 4 m at enrollment (with or without the
present owing to the combined effects of aging, aid of an assistive device), if they were function-
cardiovascular dysfunction, and skeletal-muscle ally independent before admission, and if they
dysfunction.9,10 As patients with chronic heart were expected to be discharged home. Full de-
failure transition to acute decompensated heart tails of the inclusion and exclusion criteria are
failure, physical function worsens further, and provided in the Supplementary Appendix.17 After
this decline is exacerbated by hospitalization eligible patients provided written informed con-
and bed rest.8 These deficits often persist. Many sent and completed baseline testing, they were
patients never recover baseline function, lose randomly assigned with equal probability to the
independence, and have high risks of rehospital- rehabilitation intervention (intervention group)
ization and death after discharge (sometimes or to usual care (control group) by a centralized,
referred to as “post-hospital syndrome”).4,5,11-14 Web-based system, with the use of block ran-
However, management guidelines do not ad- domization. Randomization was stratified ac-
dress physical dysfunction in patients hospital- cording to ejection fraction (<45% vs. ≥45%) and
ized for heart failure,15 and previous exercise clinical site. The patients in both trial groups
training trials excluded patients with heart fail- received usual care, as recommended by their
ure who had recently been hospitalized.10,16 To medical providers, which could include inpatient
address these issues, we conducted the Rehabili- or outpatient physical therapy and standard car-
tation Therapy in Older Acute Heart Failure Pa- diac rehabilitation.
tients (REHAB-HF) trial, a multicenter, random-
ized, single-blind, controlled trial of an early, Trial Procedures
transitional, tailored, progressive rehabilitation The trial intervention (for the intervention group)
intervention that included multiple physical-func- was an early, transitional, tailored, progressive
tion domains. We hypothesized that the inter- physical rehabilitation program that had been
vention would improve physical function and developed for frail, older patients with acute
reduce rates of rehospitalization for any cause at decompensated heart failure.18,19 The interven-
6 months. tion focused on four physical-function domains
(strength, balance, mobility, and endurance) and
progressed through four prespecified functional
Me thods
levels within each domain (Table S1). The pro-
Trial Design and Oversight gression of exercise intensity and the types of
Details of the trial design and intervention meth- exercises at each session were individualized on
ods have been described previously.17,18 The orga- the basis of the patient’s performance level within
nizational structure is shown in Figure S1 in the each domain.18 A key goal was to increase each
Supplementary Appendix, available with the full patient’s endurance (duration of walking); doing
text of this article at NEJM.org. The steering this safely required first addressing deficits in bal-
committee designed the trial and oversaw opera- ance, strength, and mobility.
The intervention was initiated in the hospital function were assessed at baseline in the hospi-
when feasible and was subsequently transitioned tal and at 3 months.17 Clinical events were ascer-
to an outpatient facility as soon as possible after tained throughout follow-up from monthly in-
discharge. If needed, home-based sessions were terviews and from review of medical records.
provided by interventionists until the patient was The primary outcome was the score on the
physically able to attend the facility-based outpa- Short Physical Performance Battery at 3 months.
tient sessions. Outpatient sessions were 60 min- The Short Physical Performance Battery is a
utes long, occurred 3 days per week for 12 weeks standardized, reproducible measure of global
(or 36 sessions), and were conducted at a 1:1 physical function that has been validated in
interventionist–patient ratio. Outpatient sessions frail, older persons and predicts a wide range of
were complemented by home exercise (low-inten- clinical outcomes.20-22 It has three components: a
sity walking, which was gradually increased to standing balance test, a gait-speed (4-m walk)
up to 30 minutes daily, and strengthening exer- test, and a strength test (as assessed by the time
cises) on nonprogram days. The home exercise needed to rise from a chair five times). Each
component of the intervention was initiated only component is scored on a scale of 0 to 4; the
after a visit to the patient’s home by an interven- sum of the scores ranges from 0 to 12, with
tionist to evaluate the home environment.18 lower scores indicating more severe physical dys-
A key goal of the intervention during the first function.
3 months (the outpatient phase) was to prepare The secondary outcome was the rate of rehos-
the patient to transition to the independent pitalization for any cause at 6 months, with re-
maintenance phase (months 4 through 6). At the hospitalization defined as any hospital stay
3-month visit, patients were provided with indi- longer than 24 hours. The reasons for rehospi-
vidualized exercise prescriptions and were sub- talization were categorized as noncardiovascular
sequently followed every 4 weeks by telephone cause, heart failure, or another cardiovascular
contact. Patient retention in the intervention group cause by an independent adjudicator who was
and adherence to the intervention sessions were unaware of the trial-group assignments.
reviewed and discussed every 2 weeks by a dedi- Additional physical-function outcomes in-
cated committee in accordance with the recom- cluded 6-minute walk distance, frailty status
mendations of the National Institutes of Health (assessed according to modified Fried criteria6),
Behavior Change Consortium Treatment Fidelity hand-grip strength, and gait speed at 3 months.
Workgroup.18 Additional details regarding the Quality of life was assessed at 3 months with the
intervention are provided in the Supplementary use of the Kansas City Cardiomyopathy Ques-
Appendix. tionnaire (KCCQ) and the EQ-5D-5L (also known
Patients who had been randomly assigned to as the European Quality of Life 5-Dimension
the control group received a telephone call every 5-Level questionnaire) visual-analogue scale. Oth-
2 weeks and had in-person clinic visits at 1 month er outcomes included the Geriatric Depression
and 3 months after discharge from the index Scale–15 score and the Montreal Cognitive As-
hospitalization.17 Information regarding the oc- sessment score.
currence of symptoms or clinical events and the
receipt of rehabilitation therapy unrelated to the Statistical Analysis
trial was collected. Patients received no specific On the basis of the results from a pilot study,19
recommendations with respect to exercise, but we estimated that 258 patients who could be
they were encouraged to adhere to prescribed evaluated for efficacy would provide the trial with
usual-care therapy and follow-up appointments. 80% power to detect a 10% difference (equiva-
Additional details regarding the control group lent to a difference of 0.6 points) between the
are provided in the Supplementary Appendix. intervention group and the control group in the
score on the Short Physical Performance Battery
Trial Outcomes at 3 months (the primary outcome); the enroll-
Outcome measures of physical and cognitive func- ment of 334 patients who could be evaluated for
tion were assessed by personnel who were un- efficacy would be needed to detect a 25% differ-
aware of the trial-group assignments. Physical ence in the rate of rehospitalization at 6 months
function, quality of life, depression, and cognitive (the secondary outcome). We planned to enroll
61 Were excluded
17 Did not meet inclusion criteria
11 Were unable to walk 4 m
4 Were not independent with respect to activities of daily living
2 Were admitted for a reason other than ADHF
43 Met exclusion criteria
12 Were unable or unwilling to comply with trial requirements
10 Had risk of noncompliance
6 Were planned to be discharged to nonindependent living facility
5 Had acute myocardial infarction
5 Had impairment that precluded participation, including cognition
4 Had advanced chronic kidney disease or dialysis
1 Had aortic stenosis
1 Died
12 Died 6 Died
14 Were lost to follow-up and were not 13 Were lost to follow-up and were not
included in the analysis of the included in the analysis of the
primary outcome primary outcome
149 Were included in the analysis of the 155 Were included in the analysis of the
primary outcome primary outcome
360 patients in order to allow for approximately egorical variables. To account for deaths and
7% of the patients to withdraw from the trial. loss to follow-up, joint models of continuous and
Baseline characteristics are presented as survival outcomes were used to assess differ-
means and standard deviations for continuous ences between the intervention group and the
variables and as counts and percentages for cat- control group in the 3-month outcomes (includ-
ical appointments and illness, adherence to the and 8 in the control group were from cardiovas-
intervention was 78±3%. Patients generally pro- cular causes (Table S9). Serious and nonserious
gressed to higher functional levels in each do- adverse events are summarized in Tables S10 and
main during the course of the intervention S11. Chest pain, hypertension, dizziness, hyper-
(Fig. 2). A key goal was to increase each patient’s glycemia, and hypoglycemia were more common
exercise endurance (duration of walking); among in the intervention group than in the control
patients who participated in the first and last group, and falls and heart failure were more com-
sessions, the mean (±SD) endurance doubled mon in the control group.
from 10.7±5.9 minutes in the first session to
22.0±11.1 minutes in the last session. Additional Discussion
data regarding exercise during the hospitaliza-
tion, outpatient, and maintenance phases are The REHAB-HF trial examined the effects of an
provided in Tables S5 through S7; information early, transitional, tailored, progressive rehabili-
regarding usual-care exercise therapy not associ- tation intervention that included multiple physi-
ated with the trial is provided in Table S8. cal-function domains in frail, older patients who
After adjustment for the baseline Short Physi- were hospitalized for acute decompensated heart
cal Performance Battery score and other base- failure. The intervention group had significantly
line characteristics, the least-squares mean (±SE) greater improvement in physical function, as as-
score on the Short Physical Performance Battery sessed by the score on the Short Physical Perfor-
at 3 months was 8.3±0.2 in the intervention mance Battery at 3 months, than the control
group and 6.9±0.2 in the control group (mean group. The results of the analyses of 6-minute
between-group difference, 1.5; 95% confidence walk distance, frailty status, quality of life, and
interval [CI], 0.9 to 2.0; P<0.001) (Table 2 and depression also suggested clinical benefits of
Fig. S4). This effect appeared to be relatively the intervention. Over the course of 6 months,
uniform across a wide variety of prespecified the incidence of rehospitalization for any cause,
subgroups (Fig. 3). The results for each of the rehospitalization for heart failure, and death was
three components of the Short Physical Perfor- high in both groups.
mance Battery are shown in Table 2. The benefit Our trial was designed to address several criti-
of the rehabilitation intervention persisted after cal evidence gaps regarding physical rehabilita-
post hoc adjustment for baseline imbalances in tion in patients with heart failure. Most previous
diagnoses of diabetes and peripheral vascular trials excluded patients who had been hospital-
disease; the least-squares mean score was 8.3±0.2 ized within the previous 6 weeks — a period
in the intervention group and 6.8±0.2 in the during which the severity of physical dysfunction
control group (mean between-group difference, and the risk of clinical events are highest; those
1.5; 95% CI, 0.9 to 2.1). trials also involved few older, frail patients with
The secondary outcome, the rate of rehospi- multiple coexisting conditions in whom differ-
talization for any cause at 6 months, showed no ent approaches may be appropriate.3,10 In previous
appreciable difference between the intervention early trials of rehabilitation after heart failure,
group and the control group, with rates of 1.18 enrollment of the patients and initiation of the
and 1.28, respectively (rate ratio, 0.93; 95% CI, intervention began, on average, 7 weeks after hos-
0.66 to 1.19) (Table 2). The exploratory outcomes, pital discharge; traditional endurance exercise
including 6-minute walk distance, gait speed, training was commonly used24,25; the enrolled pa-
hand-grip strength, frailty status, quality of life, tients were younger and much less frail and diverse
cognition, depression, and clinical events includ- than those in our trial25; there was no control
ing falls and rehospitalizations, are shown in group24,26; and the trials were unblinded25,27 and
Table 2 and Figure S5. often small, single-center trials.27 One of the
There were 21 deaths in the intervention largest of such trials involving recently hospital-
group and 16 deaths in the control group; the ized patients, EJECTION-HF (Exercise Joins Edu-
rates of death were 0.13 and 0.10, respectively cation: Combined Therapy to Improve Outcomes
(rate ratio, 1.17; 95% CI, 0.61 to 2.27) (Table 2). in Newly-Discharged Heart Failure), showed no
Among these deaths, 15 in the intervention group benefit of the intervention over usual care with
Intervention Control
Characteristic (N = 175) (N = 174)
Age — yr 73.1±8.5 72.2±7.7
Female sex — no. (%) 85 (49) 98 (56)
Non-White race — no. (%)† 81 (46) 91 (52)
Body-mass index‡ 32.9±8.2 33.0±8.9
Ejection fraction ≥45%, indicating preserved ejection fraction — no. (%) 93 (53) 92 (53)
Heart failure caused by ischemic heart disease — no. (%)§ 66 (38) 56 (32)
NYHA class — no. (%)
II 33 (19) 34 (20)
III 100 (57) 90 (52)
IV 41 (23) 51 (29)
Median B-type natriuretic peptide (IQR) — pg/ml¶ 595 (259–1292) 645 (381–1072)
Median N-terminal pro–B-type natriuretic peptide (IQR) — pg/ml‖ 2527 (1395–4858) 3615 (1874–8637)
Median no. of days hospitalized during index hospitalization (IQR) 4 (3–7) 5 (3–7)
Patients with ≥1 hospitalization in previous 6 mo — no. (%) 76 (43) 80 (46)
Coexisting conditions
Total no. of coexisting conditions 5.4±2.0 5.0±1.9
Hypertension — no. (%) 159 (91) 162 (93)
History of myocardial infarction — no. (%) 31 (18) 32 (18)
History of coronary revascularization, including PCI and CABG 55 (31) 47 (27)
— no. (%)
Atrial fibrillation — no. (%) 89 (51) 87 (50)
Diabetes mellitus — no. (%) 101 (58) 81 (47)
Hyperlipidemia — no. (%) 110 (63) 120 (69)
Depression, according to electronic medical record — no. (%) 29 (17) 33 (19)
Geriatric conditions
Dementia or cognitive impairment, according to electronic medical 6 (3) 4 (2)
record — no. (%)
Frail, as defined by the presence of at least three Fried criteria** 92 (53) 100 (57)
— no. (%)
Prefrail, as defined by the presence of one or two Fried criteria** 77 (44) 68 (39)
— no. (%)
Urinary incontinence — no./total no. (%) 19/144 (13) 21/142 (15)
Patients with falls in previous 3 mo — no./total no. (%) 24/143 (17) 20/146 (14)
* Plus–minus values are means ±SD. CABG denotes coronary artery bypass graft, IQR interquartile range, NYHA New
York Heart Association, and PCI percutaneous coronary intervention.
† Race was reported by the patient.
‡ The body-mass index is the weight in kilograms divided by the square of the height in meters.
§ Ischemic heart disease was determined to be the cause of heart failure if a patient had a history of myocardial infarc-
tion, coronary revascularization, or both.
¶ This analysis included data from 104 patients in the intervention group and 100 patients in the control group.
‖ This analysis included data from 58 patients in the intervention group and 59 patients in the control group.
** The five Fried criteria include weight loss, exhaustion, low physical activity, slow gait speed, and weak hand-grip
strength.
Table 2. (Continued.)
* Baseline data are presented as means ±SD. Follow-up data at 3 months are presented as least-squares means ±SE,
with adjustment for baseline value, clinical site, age, sex, and ejection fraction category. The effect sizes for the
3-month outcomes are shown as between-group differences in the least-squares mean change. The widths of the
confidence intervals (CIs) have not been adjusted for multiple comparisons, so the intervals should not be used to
infer definitive treatment effects for the secondary outcome and other outcomes.
† Total scores on the Short Physical Performance Battery (SPPB) range from 0 to 12, with lower scores indicating more
severe physical dysfunction; each component (the standing balance test, the gait-speed test [as assessed by a 4-m walk],
and the strength test [as assessed by the time needed to rise from a chair five times]) is scored on a scale of 0 to 4.
‡ P<0.001.
§ For the comparison of the baseline and follow-up results in this trial, the Fried criteria were modified to exclude the
weight-loss criterion owing to difficulty in ascertaining weight changes because of changes in fluid status.
¶ Scores on the Kansas City Cardiomyopathy Questionnaire (KCCQ) range from 0 to 100, with higher scores indicating
better health status.
‖ Scores on the EQ-5D-5L (also known as the European Quality of Life 5-Dimension 5-Level questionnaire) visual-
analogue scale range from 0 to 100, with higher scores indicating better health status.
** Scores on the Montreal Cognitive Assessment (MoCA) range from 0 to 30, with higher scores indicating better cogni-
tive function.
†† Scores on the Geriatric Depression Scale-15 range from 0 to 15, with higher scores indicating worse depressive symptoms.
‡‡ The effect size is shown as a rate ratio.
§§ The effect size is shown as an odds ratio.
80 80
60 60
40 40
Percentage of Patients
Percentage of Patients
20 20
The
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Intervention Session Intervention Session
C Mobility D Endurance
100 100
80 80
60 60
n e w e ng l a n d j o u r na l
40 40
Percentage of Patients
Percentage of Patients
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Intervention Session Intervention Session
Figure 2. Progression of Physical Function over Time in the Intervention Group, According to Exercise Domain.
As described previously,17,18 each exercise session comprised four domains (balance, strength, mobility, and endurance) and four stratification levels (level 1, shown in red; level 2,
yellow; level 3, blue; and level 4, purple) corresponding to increasing thresholds of functional ability. For all four domains, as the number of sessions increased during the rehabili-
tation intervention, the percentage of patients who were performing at higher levels of functional ability (levels 3 and 4) generally increased, whereas the percentage of patients
who were performing at lower levels (1 and 2) generally decreased. The black, dark gray, and light gray areas indicate the percentages of patients who died, were lost to follow-up,
Downloaded from nejm.org by MATHEUS OLIVEIRA on July 24, 2021. For personal use only. No other uses without permission.
or discontinued the intervention, respectively.
Physical Rehabilitation for Heart Failure
Control Intervention
Better Better
respect to 6-minute walk distance, rehospitaliza- depression in patients with heart failure have had
tion, and death, but adherence to the intervention neutral results.
was low (43%).25 The greater improvements in physical func-
Physical dysfunction, frailty, and depression tion relative to the control group were seen de-
are often unrecognized clinically in older patients spite the receipt of routine physical or occupa-
hospitalized for heart failure,8,28 are generally tional therapy or traditional cardiac or pulmonary
not addressed in clinical care pathways,10,29 and rehabilitation as part of usual care by 43% of the
probably contribute to delayed, incomplete re- patients in the control group. At 6 months, 83%
covery and high rates of rehospitalization, death, of the patients in the intervention group who
and long-term loss of independence after hospi- were alive and were being followed by telephone
tal discharge.2,11,12,14 Physical-function impairments contact reported regular home exercise, which
in the patients in the REHAB-HF trial were broad- suggested that behavioral change — a requisite
er and more severe than those observed in pa- for long-term adherence — may have occurred.
tients with chronic heart failure.28 For example, The intervention-related benefits may be re-
the mean baseline 6-minute walk distance in the lated to both the severity of the baseline deficits
REHAB-HF trial was half that observed in the and the robust, broad systemic effects of physical
HF-ACTION (Heart Failure: A Controlled Trial In- exercise, which favorably alters energy metabo-
vestigating Outcomes of Exercise Training) trial,30 lism, oxidative stress, inflammation, tissue repair,
and severe leg weakness prevented nearly one third growth-factor response, and regulatory pathways.38
of the patients from standing even once from a Older patients with heart failure can have severe
seated position without the use of their arms. skeletal-muscle myopathy that contributes to phys-
The patients in our trial also had severe deficits ical dysfunction and abates with exercise.9
in balance and mobility, and a history of falls The number of deaths, including deaths from
and other geriatric conditions was common — cardiovascular causes, was higher in the inter-
findings that are not typically seen in patients vention group than in the control group, al-
with chronic heart failure and are not addressed though the numbers and differences were small
by conventional cardiac rehabilitation. The initia- and may have been due to chance. A meta-analy-
tion of standard endurance exercise training in sis of trials of exercise-based rehabilitation for
frail, older patients without first addressing defi- heart failure showed no significant effect on
cits in balance and mobility can limit efficacy31 death from any cause during follow-up for up to
and increase the risk of injuries and falls.32,33 12 months.39 In the EJECTION-HF trial, fewer
The intervention-related benefits seen in the deaths were reported in the intervention group
REHAB-HF trial generally exceeded previously than in the control group at 12 months.25 Given
reported values for the minimal clinically impor- the wide confidence interval for the rate ratio for
tant difference. The mean difference between the death from any cause, we cannot rule out the pos-
groups in the Short Physical Performance Bat- sibility of an increase (or decrease) in risk with an
tery score (1.5 points) was three times as large early exercise regimen among some patients.
as the reported minimal clinically important Our trial has other important limitations.
difference (0.5 points).21,34 All three of the com- First, the results did not show a beneficial effect
ponents of the Short Physical Performance Bat- on clinical events. However, a study that exam-
tery — corresponding to balance, strength, and ined patient preferences in patients with heart
mobility — showed greater improvement in the failure indicated that improving physical func-
intervention group than in the control group. tion and maintaining independence are highly
The apparent benefits in 6-minute walk distance valued, independent of clinical events.40 Second,
(34 m) and KCCQ score (7.1 points) were also although the staff members who assessed the
larger than the reported minimal clinically im- primary outcome were unaware of the trial-
portant differences (30 m and 5 points, respec- group assignments, it was not possible for pa-
tively).35,36 The suggested benefit for depression tients to be unaware of the group to which they
is of interest, since depression is common among had been randomly assigned. Third, the benefits
patients with heart failure and is associated with of the intervention over usual care may have
frequent rehospitalization,37 and trials targeting been moderated owing to the usual-care exercise
therapy received by the control group. Fourth, port, paid to Wake Forest Baptist Health, and advisory board
fees from AstraZeneca and Bayer, advisory board fees from
differences between the groups in the amount of Boehringer Ingelheim and Corvia Medical, consulting fees
caregiver attention could have influenced out- from Keyto and Merck, and grant support, paid to Wake Forest
comes. Fifth, the long-term durability of the Baptist Health, and consulting fees from Novartis; Dr. Whellan,
receiving grant support, paid to Thomas Jefferson University,
benefit of the intervention is uncertain. Finally, from Amgen, end-point committee fees from CVRx and Fibro-
many patients were ineligible or unable or un- Gen, steering committee fees from Cytokinetics, and grant sup-
willing to participate, and some discontinued port, paid to Thomas Jefferson University, and advisory board
fees from Novo Nordisk; Dr. Duncan, owning stock in Care Di-
the intervention. rections; and Dr. Mentz, receiving grant support from American
Among patients who were hospitalized for Regent and Cytokinetics, consulting fees from Amgen, Boeh-
acute decompensated heart failure, a transitional, ringer Ingelheim, and Relypsa, research support and honoraria
from Abbott, American Regent, Amgen, Boehringer Ingelheim,
tailored, progressive rehabilitation intervention Eli Lilly, Boston Scientific, Cytokinetics, FAST BioMedical, Gil-
that included multiple physical-function domains ead Sciences, Innolife, Medtronic, Relypsa, Respicardia, Roche,
and that began during, or early after, hospital- Sanofi, Vifor Pharma, and Windtree Therapeutics, grant sup-
port, research support, and honoraria from AstraZeneca, and
ization and continued for 12 weeks after hospi- research support, consulting fees, and honoraria from Bayer,
tal discharge resulted in significantly greater im- Merck, and Novartis. No other potential conflict of interest
provement in physical function than usual care. relevant to this article was reported.
Disclosure forms provided by the authors are available with
Supported by research grants from the National Insti- the full text of this article at NEJM.org.
tutes of Health (R01AG045551, R01AG18915, P30AG021332, A data sharing statement provided by the authors is available
P30AG028716, and U24AG059624), the Kermit Glenn Phillips with the full text of this article at NEJM.org.
II Chair in Cardiovascular Medicine, and the Oristano Family We thank Alain Bertoni, M.D., Peter Brubaker, Ph.D., and
Fund at Wake Forest School of Medicine. Jack Rejeski, Ph.D. for their critical review and editing of an
Dr. Kitzman reports receiving steering committee fees and earlier version of the manuscript; and Timothy Morgan, Ph.D.
data and safety monitoring board fees from AbbVie, grant sup- (deceased), for his biostatistical expertise.
References
1. Heidenreich PA, Albert NM, Allen LA, atic review and meta-analysis. J Am Heart hospitalization or restricted activity in
et al. Forecasting the impact of heart fail- Assoc 2018;7(23):e008251. older persons. JAMA 2010;2304:1919-28.
ure in the United States: a policy state- 8. Warraich HJ, Kitzman DW, Whellan 15. Forman DE, Arena R, Boxer R, et al.
ment from the American Heart Associa- DJ, et al. Physical function, frailty, cogni- Prioritizing functional capacity as a prin-
tion. Circ Heart Fail 2013;6:606-19. tion, depression, and quality of life in cipal end point for therapies oriented to
2. Cheng RK, Cox M, Neely ML, et al. hospitalized adults ≥60 years with acute older adults with cardiovascular disease:
Outcomes in patients with heart failure decompensated heart failure with pre- a scientific statement for healthcare pro-
with preserved, borderline, and reduced served versus reduced ejection fraction. fessionals from the American Heart As-
ejection fraction in the Medicare popula- Circ Heart Fail 2018;11(11):e005254. sociation. Circulation 2017;135(16):e894-
tion. Am Heart J 2014;168:721-30. 9. Kitzman DW, Haykowsky MJ, Tomc- e918.
3. Fleg JL. Preventing readmission after zak CR. Making the case for skeletal 16. O’Connor CM, Whellan DJ, Lee KL, et
hospitalization for acute heart failure: muscle myopathy and its contribution to al. Efficacy and safety of exercise training
a quest incompletely fulfilled. JACC Heart exercise intolerance in heart failure with in patients with chronic heart failure:
Fail 2018;6:153-5. preserved ejection fraction. Circ Heart HF-ACTION randomized controlled trial.
4. Alahdab MT, Mansour IN, Napan S, Fail 2017;10(7):e004281. JAMA 2009;301:1439-50.
Stamos TD. Six minute walk test predicts 10. Fleg JL, Cooper LS, Borlaug BA, et al. 17. Reeves GR, Whellan DJ, Duncan P, et
long-term all-cause mortality and heart Exercise training as therapy for heart fail- al. Rehabilitation therapy in older acute
failure rehospitalization in African-Amer- ure: current status and future directions. heart failure patients (REHAB-HF) trial:
ican patients hospitalized with acute de- Circ Heart Fail 2015;8:209-20. design and rationale. Am Heart J 2017;
compensated heart failure. J Card Fail 11. Krumholz HM. Post-hospital syn- 185:130-9.
2009;15:130-5. drome — an acquired, transient condition 18. Pastva AM, Duncan PW, Reeves GR,
5. Chiarantini D, Volpato S, Sioulis F, et of generalized risk. N Engl J Med 2013; et al. Strategies for supporting interven-
al. Lower extremity performance mea- 368:100-2. tion fidelity in the rehabilitation therapy
sures predict long-term prognosis in old- 12. Lorenzo M, de la Espriella R, Miñana in older acute heart failure patients
er patients hospitalized for heart failure. G, et al. Clinical profile and 1-year clini- (REHAB-HF) trial. Contemp Clin Trials
J Card Fail 2010;16:390-5. cal outcomes of super elderly patients ad- 2018;64:118-27.
6. Pandey A, Kitzman D, Whellan DJ, et mitted with acute heart failure. Eur J In- 19. Reeves GR, Whellan DJ, O’Connor
al. Frailty among older decompensated tern Med 2020;81:78-82. CM, et al. A novel rehabilitation interven-
heart failure patients: prevalence, associa- 13. Huusko J, Tuominen S, Studer R, et al. tion for older patients with acute decom-
tion with patient-centered outcomes, and Recurrent hospitalizations are associated pensated heart failure: the REHAB-HF pi-
efficient detection methods. JACC Heart with increased mortality across the ejec- lot study. JACC Heart Fail 2017;5:359-66.
Fail 2019;7:1079-88. tion fraction range in heart failure. ESC 20. Pavasini R, Guralnik J, Brown JC, et
7. Yang X, Lupón J, Vidán MT, et al. Im- Heart Fail 2020;7:2406-17. al. Short physical performance battery
pact of frailty on mortality and hospital- 14. Gill TM, Allore HG, Gahbauer EA, and all-cause mortality: systematic review
ization in chronic heart failure: a system- Murphy TE. Change in disability after and meta-analysis. BMC Med 2016;14:215.
21. Soubra R, Chkeir A, Novella JL. A sys- on function among patients with con- tients: data from the Italian Survey on
tematic review of thirty-one assessment gestive heart failure: a randomized con- Cardiac Rehabilitation-2008 (ISYDE-2008)
tests to evaluate mobility in older adults. trolled trial. Heart Views 2011;12:99-103. — official report of the Italian Associa-
Biomed Res Int 2019;2019:1354362. 28. Reeves GR, Whellan DJ, Patel MJ, et tion for Cardiovascular Prevention, Reha-
22. Volpato S, Cavalieri M, Sioulis F, et al. al. Comparison of frequency of frailty bilitation, and Epidemiology. J Gerontol A
Predictive value of the short physical per- and severely impaired physical function Biol Sci Med Sci 2010;65:1353-61.
formance battery following hospitaliza- in patients ≥60 years hospitalized with 34. Perera S, Mody SH, Woodman RC,
tion in older patients. J Gerontol A Biol acute decompensated heart failure versus Studenski SA. Meaningful change and re-
Sci Med Sci 2011;66:89-96. chronic stable heart failure with reduced sponsiveness in common physical perfor-
23. Wang W, Wang W, Mosley TH, Gris- and preserved left ventricular ejection mance measures in older adults. J Am
wold ME. A SAS macro for the joint mod- fraction. Am J Cardiol 2016;117:1953-8. Geriatr Soc 2006;54:743-9.
eling of longitudinal outcomes and mul- 29. Yancy CW, Jessup M, Bozkurt B, et al. 35. Flynn KE, Piña IL, Whellan DJ, et al.
tiple competing risk dropouts. Comput 2013 ACCF/AHA guideline for the man- Effects of exercise training on health sta-
Methods Programs Biomed 2017;138:23- agement of heart failure: a report of the tus in patients with chronic heart failure:
30. American College of Cardiology Founda- HF-ACTION randomized controlled trial.
24. Acanfora D, Scicchitano P, Casucci G, tion/American Heart Association task force JAMA 2009;301:1451-9.
et al. Exercise training effects on elderly on practice guidelines. J Am Coll Cardiol 36. Spertus JA, Jones PG, Kim J, Globe D.
and middle-age patients with chronic 2013;62(16):e147-e239. Validity, reliability, and responsiveness of
heart failure after acute decompensation: 30. Forman DE, Fleg JL, Kitzman DW, et the Kansas City Cardiomyopathy Ques-
a randomized, controlled trial. Int J Car- al. 6-min walk test provides prognostic tionnaire in anemic heart failure patients.
diol 2016;225:313-23. utility comparable to cardiopulmonary Qual Life Res 2008;17:291-8.
25. Mudge AM, Denaro CP, Scott AC, et exercise testing in ambulatory outpatients 37. Johnson TJ, Basu S, Pisani BA, et al.
al. Addition of supervised exercise train- with systolic heart failure. J Am Coll Car- Depression predicts repeated heart failure
ing to a post-hospital disease manage- diol 2012;60:2653-61. hospitalizations. J Card Fail 2012;18:246-52.
ment program for patients recently hospi- 31. Witham MD, Fulton RL, Greig CA, et 38. Contrepois K, Wu S, Moneghetti KJ, et
talized with acute heart failure: the al. Efficacy and cost of an exercise pro- al. Molecular choreography of acute exer-
EJECTION-HF randomized phase 4 trial. gram for functionally impaired older pa- cise. Cell 2020;181(5):1112-1130.e16.
JACC Heart Fail 2018;6:143-52. tients with heart failure: a randomized 39. Davies EJ, Moxham T, Rees K, et al.
26. Scrutinio D, Passantino A, Catanzaro controlled trial. Circ Heart Fail 2012; 5: Exercise training for systolic heart fail-
R, et al. Inpatient cardiac rehabilitation 209-16. ure: Cochrane systematic review and
soon after hospitalization for acute de- 32. Faber MJ, Bosscher RJ, Chin A Paw meta-analysis. Eur J Heart Fail 2010;12:
compensated heart failure: a propensity MJ, van Wieringen PC. Effects of exercise 706-15.
score study. J Cardiopulm Rehabil Prev programs on falls and mobility in frail 40. Reed SD, Fairchild AO, Johnson FR,
2012;32:71-7. and pre-frail older adults: a multicenter et al. Patients’ willingness to accept mi-
27. Babu AS, Maiya AG, George MM, Pad- randomized controlled trial. Arch Phys tral valve procedure-associated risks varies
makumar R, Guddattu V. Effects of com- Med Rehabil 2006;87:885-96. across severity of heart failure symptoms.
bined early in-patient cardiac rehabilita- 33. Giallauria F, Vigorito C, Tramarin R, Circ Cardiovasc Interv 2019;12(12):e008051.
tion and structured home-based program et al. Cardiac rehabilitation in very old pa- Copyright © 2021 Massachusetts Medical Society.