Annals of Internal Medicine - Helmut Horten Stiftung

Article Interval Exercise for Patients with Chronic Obstructive Pulmonary Disease
Table 1. Baseline Characteristics
Characteristic Interval
Exercise
Group
(n � 48)
Continuous
Exercise
Group
(n � 50)
Age, y 69.0 (9.2) 68.9 (9.2)
Male, % 29 (60.4) 36 (72.0)
FEV 1, L 0.87 (0.27) 0.89 (0.29)
FEV 1, % predicted 34.5 (9.0) 34.1 (8.0)
Mean FEV 1/FVC (SD)* 0.49 (0.19) 0.47 (0.21)
Exacerbation within last 8 wk, % 29 (60.4) 30 (60.0)
PaO 2, mm Hg 57.1 (8.9) 59.5 (9.7)
Mean y since diagnosis (SD)* 9.4 (8.4) 9.2 (6.6)
Mean pack y (SD)† 51.4 (31.4) 54.1 (28.3)
Smoking, % 3 (6.3) 6 (12.0)
Mean body mass index (SD)† 25.4 (6.9) 24.0 (5.8)
Chronic Respiratory Questionnaire
Dyspnea 2.83 (0.90) 3.03 (1.05)
Fatigue 3.63 (1.15) 3.92 (1.28)
Emotional function 3.95 (1.20) 4.12 (1.24)
Mastery 4.03 (1.31) 4.16 (1.34)
Total score 3.61 (0.98) 3.81 (0.98)
Feeling thermometer 53.1 (17.9) 52.9 (18.7)
Hospital Anxiety and Depression Scale
Depression domain 7.71 (4.03) 7.45 (3.91)
Anxiety domain 7.13 (3.64) 6.81 (4.36)
6-min walking distance, m 312.8 (107.9) 332.4 (116.0)
Short-time maximum exercise capacity, W 106.3 (48.8) 119.6 (53.5)
Maximum exercise capacity, W 53.2 (26.5) 50.6 (19.0)
Medication, %
Short-acting �-agonist 37 (77.1) 37 (74.0)
Long short-acting �-agonist 44 (91.7) 43 (86.0)
Anticholinergic bronchodilators 26 (54.2) 28 (56.0)
Inhaled steroid 38 (79.2) 39 (78.0)
Oral steroid 27 (56.3) 21 (42.0)
Theophylline 8 (16.7) 7 (14.0)
Antibiotics 12 (25.0) 4 (8.0)
Anabolic steroid 2 (4.2) 1 (2.0)
Comorbid conditions, %
Cardiovascular 25 (52.1) 32 (64.0)
Endocrine 13 (27.1) 8 (16.0)
Musculoskeletal 7 (14.6) 6 (12.0)
* Data missing for 6 patients (2 in interval and 4 in continuous exercise group).
† Data missing for 3 patients (1 in interval and 2 in continuous exercise group).
Finally, we analyzed exercise tolerance by comparing
the number of breaks between groups using the Mann–
Whitney U test and the proportion of patients achieving
target exercise using the chi-square test.
We handled missing data as follows: If CRQ data were
not available at 5-week follow-up, we carried forward the
CRQ data from the 3-week follow-up assessment at the
end of the inpatient rehabilitation. Data were missing for 6
patients using interval exercise (all were included in per
protocol and intention-to-treat analyses) and for 4 patients
using continuous exercise (2 were included in the per protocol
analysis, and all 4 were included in the intention-totreat
analysis). We did not impute any data if neither the
3-week nor the 5-week assessments were available. The per
protocol and intention-to-treat analyses were based on patients
with follow-up data. Four patients (2 from each
group) had no follow-up data and were excluded from all
analyses. In addition, for the per protocol analyses, 1, 1, 2,
and 9 patients had missing CRQ, 6-minute walking distance,
maximum exercise capacity, and short-term maximum
exercise capacity data, respectively. For the intentionto-treat
analyses, in addition to the 4 patients without
follow-up data, 4, 5, 7, and 13 patients had missing CRQ,
6-minute walking distance, maximum exercise capacity,
and short-term maximum exercise capacity data, respectively.
We conducted a sensitivity analysis of the per protocol
analysis to test our method of carrying forward CRQ scores
from the 3-week assessment to follow-up after 5 weeks for
patients with missing values (n � 8). We used several imputations
with chained equations to predict CRQ total
scores after 5 weeks based on available data and repeated
the adjusted between-group comparisons (ice and micombine
commands of STATA for Windows 8.2, College Station,
Texas).
For data on exercise tolerance (breaks and adherence
to protocol), physiotherapists completed all exercise protocols.
We had computerized recordings for 44 patients using
interval exercise and 50 patients using continuous exercise.
We determined that a sample size of 44 patients in
each group would show the noninferiority of interval exercise,
assuming noninferiority margins of 0.5 in CRQ scores
and 45 meters in 6-minute walking distance between
groups, with 90% power at a significance level of 5% (1sided).
With an assumed drop-out rate of 15%, the total
sample size increased to 104 (23). After 30 participants
were enrolled in each arm, we calculated the standard deviation
of CRQ scores and found that data met our assumptions.
However, we noticed fewer drop-outs than anticipated
and, therefore, terminated recruitment after 100
patients. We performed all analyses using SPSS, version
12.0.1 (SPSS, Inc., Chicago, Illinois).
Role of the Funding Sources
AstraZeneca Switzerland, Boehringer Ingelheim Switzerland,
and the Klinik Barmelweid provided unrestricted
grants for this trial. These funding sources did not have
any influence on the planning, conduct, analyses, or publication
of the trial or its results.
RESULTS
Between May 2004 and November 2005, 100 of 103
eligible patients agreed to participate in this study (Figure
2). One patient in each group withdrew informed consent
for unspecified reasons. Forty-three (89.6%) and 44
(88.0%) patients completed the inpatient rehabilitation in
the interval and continuous exercise groups, respectively.
Eleven patients did not complete the rehabilitation because
of COPD exacerbations (3 patients in the interval and 2
patients in the continuous exercise group); musculoskeletal
pain (2 patients in the interval and 1 patient in the continuous
exercise group); and, in the continuous exercise
820 5 December 2006 Annals of Internal Medicine Volume 145 Number 11 www.annals.org