Content uploaded by Evgeny G Vaschillo
Author content
All content in this area was uploaded by Evgeny G Vaschillo on Feb 22, 2016
Content may be subject to copyright.
Content uploaded by Evgeny G Vaschillo
Author content
All content in this area was uploaded by Evgeny G Vaschillo on Feb 22, 2016
Content may be subject to copyright.
Content uploaded by Evgeny G Vaschillo
Author content
All content in this area was uploaded by Evgeny G Vaschillo on Feb 22, 2016
Content may be subject to copyright.
Biofeedback Treatment for Asthma*
Paul M. Lehrer, PhD; Evgeny Vaschillo, PhD; Bronya Vaschillo, MD;
Shou-En Lu, PhD; Anthony Scardella, MD, FCCP;
Mahmood Siddique, DO, FCCP; and Robert H. Habib, PhD
Study objectives: We evaluated the effectiveness of heart rate variability (HRV) biofeedback as a
complementary treatment for asthma.
Patients: Ninety-four adult outpatient paid volunteers with asthma.
Setting: The psychophysiology laboratory at The University of Medicine and Dentistry of New
Jersey-Robert Wood Johnson Medical School, and the private outpatient offices of participating
asthma physicians.
Interventions: The interventions were as follows: (1) a full protocol (ie, HRV biofeedback and
abdominal breathing through pursed lips and prolonged exhalation); (2) HRV biofeedback alone;
(3) placebo EEG biofeedback; and (4) a waiting list control.
Design: Subjects were first prestabilized using controller medication and then were randomly
assigned to experimental groups. Medication was titrated biweekly by blinded asthma specialists
according to a protocol based on National Heart, Lung, and Blood Institute guidelines, according
to symptoms, spirometry, and home peak flows.
Measurements: Subjects recorded daily asthma symptoms and twice-daily peak expiratory flows.
Spirometry was performed before and after each weekly treatment session under the HRV and
placebo biofeedback conditions, and at triweekly assessment sessions under the waiting list
condition. Oscillation resistance was measured approximately triweekly.
Results: Compared with the two control groups, subjects in both of the two HRV biofeedback
groups were prescribed less medication, with minimal differences between the two active
treatments. Improvements averaged one full level of asthma severity. Measures from forced
oscillation pneumography similarly showed improvement in pulmonary function. A placebo effect
influenced an improvement in asthma symptoms, but not in pulmonary function. Groups did not
differ in the occurrence of severe asthma flares.
Conclusions: The results suggest that HRV biofeedback may prove to be a useful adjunct to
asthma treatment and may help to reduce dependence on steroid medications. Further evalua-
tion of this method is warranted. (CHEST 2004; 126:352–361)
Key words: airway resistance; alternative and complementary medicine; disease severity; heart rate variability;
oscillation mechanics; psychology; self-regulation
Abbreviations: HRV ⫽ heart rate variability; LOCF ⫽ last-observation-carried-forward; NHLBI ⫽ National Heart,
Lung, and Blood Institute; Zrs ⫽ pulmonary measures derived from the forced oscillation method
A
n effective nonpharmacologic alternative or ad-
junctive treatment of asthma could provide a
potentially useful contribution to asthma care.
1
Ad-
herence to asthma regimens tends to be low,
2
and
the resort to complementary treatments is common
despite the lack of evidence for effectiveness.
3,4
The
long-term use of oral steroids is expensive and can
have undesirable side effects.
5
Although the weight
of empirical evidence strongly indicates that the
positive effects of inhaled corticosteroids in asthma
far outweigh any negative consequences, there is
*From the Department of Psychiatry (Dr. Lehrer), Robert Wood
Johnson Medical School, The University of Medicine and Den-
tistry of New Jersey, Piscataway, NJ; the Department of Neuro-
sciences (Drs. E. Vaschillo and B. Vaschillo), New Jersey Medical
School, The University of Medicine and Dentistry of New Jersey,
Newark, NJ; the Division of Biometrics (Dr. Lu), School of
Public Health, and the Department of Medicine (Drs. Scardella
and Siddique), Robert Wood Johnson Medical School, The
University of Medicine and Dentistry of New Jersey, New
Brunswick, NJ; and Mercy Children’s Hospital (Dr. Habib),
Toledo, OH.
This work was supported by grant No. R01 HL58805 from the
National Heart, Lung, and Blood Institute, National Institutes of
Health. Fluticasone and salmeterol were provided by Glaxo-
SmithKline.
Manuscript received September 9, 2003; revision accepted
March 30, 2004.
Reproduction of this article is prohibited without written permis-
sion from the American College of Chest Physicians (e-mail:
permissions@chestnet.org).
Correspondence to: Paul Lehrer, PhD, Department of Psychiatry
D-335, UMDNJ-Robert Wood Johnson Medical School, 671 Hoes
Ln, Piscataway, NJ 08854; e-mail: lehrer@umdnj.edu
352 Clinical Investigations
some evidence for adverse effects for these medica-
tions as well,
6,7
and, regardless of the weight of
evidence, many asthma patients remain wary of the
potential side effects, which, in turn, leads to non-
adherence.
8–10
Preliminary research has found that biofeedback
training to increase heart rate variability (HRV)
produces a decrease in respiratory resistance
11
and
improves spirometry performance in asthma pa-
tients,
12
although the mechanism of action has not
been proven. HRV tends to be reduced in patients
with asthma
13
and various diseases affecting the
cardiovascular and/or CNS.
14
HRV biofeedback has
been found to increase peak flow and resting barore-
flex gain and high-frequency HRV among healthy
adults,
15
but a relationship between autonomic and
pulmonary changes has not been established. The
purpose of the study was to determine whether this
biofeedback method can serve as an effective non-
pharmacologic alternative or complementary treat-
ment method for asthma.
Materials and Methods
The design for this study was modeled after that used in a study
by Lo¨fdahl et al,
16
evaluating montelukast sodium effects on
tapering inhaled steroids. The study was approved by the Insti-
tutional Review Board of The University of Medicine and
Dentistry of New Jersey-Robert Wood Johnson Medical School.
Subjects were interviewed about the study and gave written
consent at their first study visit.
Sixty-four female and 30 male paid volunteers (mean age, 37.3
years; SD, 10.2 years) were recruited via physician referrals and
advertisements. The inclusion criteria were as follows: age 18 to
65 years; history of asthma symptoms; and, within the past year,
a positive bronchodilator test result (postbronchodilator FEV
1
increase of ⱖ 12%), a positive methacholine inhalation challenge
test result, or a documented recent history (ie, within the past
year) of clinical improvement and FEV
1
increase of ⱖ 12%
following instigation of inhaled steroid therapy among individuals
with a protracted history of asthma. The exclusion criteria were as
follows: a disorder that would impede performing the biofeed-
back procedures (eg, abnormal cardiac rhythm); a negative
methacholine challenge test result; an abnormal diffusing capac-
ity (tested among all subjects ⬎ 55 years old or with ⬎ 20
pack-years of smoking); or a current practice of any relaxation,
biofeedback, or breathing technique.
Before randomization, we stabilized subjects on the lowest
possible dose of controller medication (based on the standard
protocol shown in Table 1, derived from National Heart, Lung,
and Blood Institute [NHLBI] recommendations
17
) that elimi-
nated asthma symptoms. We titrated medications downward
weekly until symptoms reappeared, lung function abnormalities
recurred, or a maximum of 2 months of titration had passed. The
lowest stable dose was treated as the subject’s baseline dose.
We randomized subjects to four treatment groups, balancing
for age, sex, and end-of-stabilization asthma severity (ie, mild
intermittent, mild persistent, moderate, and severe), based on
medication level (Table 1), scored according to NHLBI criteria.
17
The treatment groups were as follows: (1) the “full protocol” used
in previous research on this method,
18
including HRV biofeed-
back and training in pursed-lips abdominal breathing with pro-
longed exhalation (23 patients); (2) HRV biofeedback alone (22
patients); (3) a previously developed placebo biofeedback proce-
dure
19,20
involving bogus “subliminal suggestions designed to
help asthma” (with no further details provided, and no actual
suggestions given) and biofeedback training to alternately in-
crease and decrease frontal EEG ␣-rhythms (24 patients); and (4)
a waiting list control (25 patients). Subjects in the first three
groups each received 10 biofeedback sessions, approximately
weekly, and were each asked to practice at home for 20 min twice
daily. HRV biofeedback subjects were lent a home trainer unit
(KC-3; Biosvyaz; St. Petersburg, Russia). Placebo subjects were
instructed to maintain a state of relaxed alertness during home
practice, using mental strategies developed during biofeedback
sessions, and were given a tape recording with classical music and
supposed “subliminal suggestions” to improve their asthma, for
use during home practice.
We collected data on asthma symptoms and twice-daily home
peak flow readings (Mini-Wright peak flow meter; Clement-
Clarke; Essex, UK) from a daily diary, pulmonary function test
results from each biofeedback laboratory visit, and the results of
monthly physical examinations by a study physician (one of five
pulmonologists and one allergist who were blinded to experimen-
tal condition). Medication was titrated up or down approximately
biweekly based on criteria similar to the 2002 NHLBI recom-
mendations,
17
as shown in Table 2, by the asthma specialists.
(Note that the NHLBI recommends monthly reassessment, but
we increased the speed of this process because of time con-
straints.) We assessed respiratory resistance at sessions 1, 4, 7,
and 10 (or at approximately 3-week intervals in the waiting list
condition), at approximately the same time of day for each
subject, after 12 h of abstinence from albuterol. Spirometry was
performed before each biofeedback laboratory session (triweekly
for the waiting list group and weekly for the other three groups),
using standard procedures
21
with three forced maximal exhala-
tions (Koko pneumotach-based spirometer; PDS Instrumenta-
tion; Louisville, KY), calibrated daily using a 3-L syringe (using
the norms of Crapo et al
22
), and periodically performed by the
asthma physicians as part of monthly examinations. We analyzed
the maximum value of the three trials for each measure. Subjects
rested for approximately 15 min prior to each spirometry record-
ing, during which daily diaries were reviewed and the subjects
chatted with the researcher about their experience in the study.
(Waiting list subjects were only given spirometry at testing
sessions and at physician visits.)
HRV biofeedback was given using a physiograph (model I-330;
J&J Engineering; Poulsbo, WA). ECG data were collected from
the right arm and left leg, and were digitized at 512 Hz. EEG
biofeedback was given using an appropriate device (Alphascan
400 U; Bioscan Corporation; Houston, TX). To assess baroreflex
gain, beat-to-beat BP was recorded (Ohmeda Finapres model
223; Madison, WI), and digitized at a rate of 256 samples per
second. The sensor was placed on the participant’s left middle
finger, and the hand was elevated on a table to approximately the
level of the heart. ␣-Low-frequency baroreflex gain was calcu-
lated by cross-spectral analysis of the heart rate and BP, in which
coherence between the two measures was ⬎ 0.8 (WinCPRS
program; Absolute Aliens Oy; Turku, Finland).
Respiratory system impedance (Zrs) [between 2 and 32 Hz
with 2-Hz increments] was measured using a pseudorandom
noise forced oscillation system built for our laboratory,
23–26
It was
presented in 40 2-s bursts spaced equally throughout each trial
(with tasks or individual rest periods after each task). To mini-
mize the effects of possible partial glottal closure during exhala-
tion, each burst was triggered by the beginning of an inhalation.
www.chestjournal.org CHEST / 126/2/AUGUST, 2004 353
Also, a pair of large earphones was worn on the cheeks to firmly
support the extrathoracic airways to minimize the potential
confounding effects of airway wall flow shunting.
24
Data from
bursts containing artifacts were eliminated through visual inspec-
tion, and edited data were averaged for each task. Three spectral
features of respiratory resistance data (the real part of Zrs) and
reactance data (the imaginary part of Zrs) were used to char-
acterize the underlying respiratory mechanics, as follows:
(1) resistance at 6 Hz (in cm H
2
O/L/s); (2) frequency depen
-
dence of resistance (in cm H
2
O/L/s) calculated as the differ
-
ence between resistance at 6 Hz and at the frequency between
8 and 32 Hz yielding the minimum resistance; and (3) the
resonant frequency (in Hz), defined as the lowest frequency at
which the reactance crossed 0 from negative to positive.
To determine the relative plausibility of the placebo, we gave a
three-item treatment credibility questionnaire to subjects in the
three intervention groups at each of the four testing sessions,
27
comprising three 9-point Likert items anchored at “not at all” and
“very (much),” as follows: (1) How much do you expect your
asthma to improve as a result of participating in this program?
(2) How effective do you think this method is, in general?
(3) Would you be likely to recommend this technique to a friend
or relative suffering from asthma?
Results
Adherence, Dropouts, and Treatment Duration
The self-reported rate of adherence to biofeed-
back practice was ⬎ 70%, and the rate of completion
of daily home questionnaires was ⬎ 80% among
those who completed the questionnaire. Eighteen
subjects dropped out of the study (Table 3), approx-
imately 20% in the three groups receiving a treat-
ment, a rate similar to that of other asthma behav-
ioral intervention studies from our laboratory.
8,15,16
Because of occasional rescheduled sessions caused
by patients’ schedule conflicts, subjects in the wait-
ing list group spent less time in the study than
subjects in the other groups and had a lower drop-
out rate. The reasons given for dropping out of the
study that were related to deterioration in the pa-
tient’s condition occurred only in subjects of the two
control groups.
Table 1—Criteria for Rating Asthma Severity and Stepped Protocol for Asthma Medication*
Severity Class† Symptom Class‡ Pulmonary Function
Medication
Step Medication Protocol§
Mild
intermittent
Symptoms (wheeze/cough/dyspnea)
ⱕ 2 times a week;
Asymptomatic between
exacerbations;
Nighttime asthma symptoms ⱕ 2
times/mo
FEV
1
or PEF ⱖ 80%
predicted and PEF
variability ⬍ 20%
1, 2 No daily medications needed;
Maximum 2 times/wk albuterol
(4 puffs) except for exercise-
induced asthma
Mild persistent Symptoms ⬎ 2 times/wk but ⬍ 1
time/d;
Exacerbations may affect activity;
Nighttime asthma symptoms ⬎ 2
times/mo (3–4/mo)
FEV
1
or PEF ⱖ 80%
predicted;
PEF variability 20–30%
3, 4, 5 Fluticasone, 44 g1pqd(44g);
Fluticasone, 44 g 1p BID (88 g);
Fluticasone, 44 g 2p BID (176 g)
(up to 4 puffs of rescue albuterol/d
can be added)
Moderate
persistent
Daily symptoms;
Daily use of inhaled short-acting

2
-agonist;
Exacerbations: ⱖ 2 times/wk, may
last days, affect activity;
Nighttime asthma symptoms ⬎ 1
time/wk (ⱖ 5 times/mo)
FEV
1
or PEF ⬎ 60%
ⱕ 80% predicted;
PEF variability ⬎ 30%
6, 7, 8 Fluticasone, 110 g 1p bid (220 g);
Salmeterol or montelukast sodium;㛳
Fluticasone, 110 g 2p bid (440 g)
(up to 4 puffs of rescue albuterol/d
can be added)
Severe
persistent
Continuous symptoms;
Limited physical activity;
Frequent exacerbations;
Frequent nighttime asthma
symptoms
FEV
1
or PEF ⱕ 60%
predicted;
PEF variability ⬎ 30%
9, 10, 11,
12, 13
Fluticasone, 110 g 3p bid (660 g);
Fluticasone, 110 g 4p bid (880 g);
Salmeterol or montelukast sodium㛳
(continue the other of these drugs),
up to 4 puffs of rescue albuterol/d;
Fluticasone, 220 g 4p bid (1760 g);
Prednisone burst (40 mg to taper)
*PEF ⫽ peak expiratory flow during forced expiratory maneuver.
†Total asthma severity, determined after stabilization, was categorized according to the dimension (symptoms/pulmonary function/medication)
with the highest severity level. Asthma severity categorization was used as a factor in randomizing subjects to groups.
‡Classes scored 1 to 4 for the four severity categories.
§The medication protocol was derived from recommended dosage levels for each level of asthma severity from the NHLBI categorization.
1,12
In
the event of severe asthma flares, prednisone was prescribed and gradually titrated downward over a period of 3 to 7 days, depending on the
patient’s condition. Afterward, the preflare medical regimen was resumed or altered, depending on the subject’s clinical condition. Two patients
were unable to tolerate fluticasone. One was instead given an equivalent dose of triamcinalone acetonide, and the other was given budesonide
inhalation powder, according to the NHLBI table of medication equivalence.
14
㛳The choice of salmeterol or montelukast sodium was made by the physician for patients with moderate persistent asthma. In level 11 for patients
with severe asthma, both drugs were given.
354
Clinical Investigations
Baseline Asthma Severity
Subjects began the study with a mean poststabili-
zation medication rating in the moderate persistent
asthma range (Table 1). Baseline FEV
1
values were in
the mild asthma range (mean [⫾ SD], 77.2 ⫾ 22.7%
predicted). There were no significant differences be-
tween groups for either measure at the first session.
Statistical Model
We used mixed effects models for repeated mea-
sures (Proc Mixed, SAS; SAS Institute; Cary, NC).
Based on exploratory analyses and the information
criteria of Akaike,
28
we used an autoregressive model
(order ⫽ 1) for prescribed medication, a compound
symmetry model for Zrs and medication data, and a
heterogeneous autoregressive model (order ⫽ 1) for
treatment credibility. Autoregressive models assume
that the correlations are stronger for measurements
closer in time. Heterogeneous autoregressive models
additionally allow the variance to change between
repeated measures for each individual. The com-
pound symmetry model assumes that closeness in
time is unrelated to the correlation among observa-
tions.
We analyzed data in the following two ways:
(1) last-observation-carried-forward (LOCF, intent-
to-treat); and (2), for the primary outcome variable
(medication level), completers (ie, those who com-
pleted the study), with noninformative dropout as-
Table 2—Criteria for Medication*
Criteria for changing medication
Keep medication constant if any one criterion is present in the
past 2 wk;
Increase medication by one step if two or more criteria are
present in the past 2 wk;
Reduce medication by one step if no criteria are present in the
past 2 wk
1. FEV
1
⬍ 80% of stabilization baseline at any time;
2. PEF ⬍ 80% of personal best or PEF variability ⬎ 20% for
ⱖ 3;
3. Nocturnal awakenings ⬎ 2 times due to asthma;
4. Occurrence of one or more asthma flares within the past
2 wk, not resolved by 6 puffs of albuterol within 1 h;
5. Average of 8 puffs of albuterol daily.
Criterion for resolution of asthma flare: achievement of green
condition
No cough, wheeze, shortness of breath, or chest tightness;
Peak flow ⬎ 80% of personal best (defined as the highest level
previously reported during the study);
Can do usual activities
*Medication was titrated up or down by one level (as defined in Ta-
ble 1), using criteria in this table. Medication adjustments were done
at the time of the monthly physical examination and 2 weeks later,
at which time symptom and pulmonary function data were faxed to
the physician. See Table 1 for abbreviations not used in the text.
Table 3—Subject Characteristics, Adherence, and Dropout*
Condition
Time Spent
in Study,† d
Days
Questionnaire,
%
Home bfk
Practice
Days, %
Age,‡
yr
Male
Gender, %
Weight,‡
kg
Height,‡
cm
Dropouts,
No.
Reported Reason for Dropout
Completers
All
Subjects
Schedule
Conflicts
Life
Crises
Clinical
Deterioration
Full protocol 87 ⫾ 5.5 93.6 80.1 39.0 ⫾ 11.9 26.3 79.5 ⫾ 20.6 168.4 ⫾ 12.8 6 5 1
HRV bfk 95.1 ⫾ 8.2 82.3 74 37.9 ⫾ 20.6 27.8 82.9 ⫾ 34.6 168.6 ⫾ 9.4 5 3 2
Placebo 82 ⫾ 4.2 94.2 73 39.1 ⫾ 14.0 40 78.8 ⫾ 21.6 168.4 ⫾ 10.2 5 3 2
Waiting list 67.6 ⫾ 3.0 81.8 38.6 ⫾ 15.3 30.4 74.2 ⫾ 21.3 165.3 ⫾ 8.2 2 1 1
*bfk ⫽ biofeedback.
†Values calculated for the treatment period only (not including the stabilization period). Values given as mean ⫾ SE.
‡Values given as mean ⫾ SD.
www.chestjournal.org CHEST / 126/2/AUGUST, 2004 355
sumptions.
29
Zrs and measures of respiration rate
and tidal volume yielded skewed data, so these
analyses were performed on natural logarithm trans-
formations.
Asthma Severity
Level of Prescribed Controller Medication: Med-
ication levels at the four testing sessions changed
differentially across groups (with the same p val-
ues for completers as in the LOCF analysis)
[LOCF treatment ⫻ session interaction: F
3,267
⫽ 6.36;
p ⬍ 0.0001], and highly significant decreases in med-
ication consumption occurred in the groups receiv-
ing HRV biofeedback (LOCF t
257
⫽ 8.51 and 6.61,
respectively; p ⬍ 0.0001 [for the full protocol and
HRV biofeedback alone]). Decreases also were
significant in the placebo biofeedback condition
(LOCF t
257
⫽ 2.48; p ⬍ 0.02) but not in the wait
-
ing list condition (LOCF t
257
⫽ 0.4). They were
significantly greater in the combined HRV biofeed-
back groups than in the placebo group, according
to the treatment ⫻ session interaction (LOCF
t
3,201
⫽ 5.03; p ⬍ 0.003 [p ⬍ 0.004 in the analysis of
completers]). There were no significant differences
between the full protocol and HRV biofeedback
alone. Medication levels in the HRV biofeedback
groups tended to fall from the upper levels of
moderate persistent asthma to the upper levels of
mild persistent asthma by the last treatment session
(Table 4), while medication levels remained in the
moderate persistent asthma range in the two control
groups. Although comparisons with the waiting list
group may have been influenced by the duration of
treatment, we noted (Table 4) that there was no
tendency toward improvement over time in this group,
although such a tendency was evident in the biofeed-
back groups, so it is unlikely that a longer passage of
time would have produced greater changes.
Respiratory System Effects
Biofeedback produced significant decreases across
sessions in airway resistance at 6 Hz (presession rest
period: median at the first session, 2.2 cm H
2
O/L/s;
median at the last session, 1.7 cm H
2
O/L/s), fre
-
quency dependence of resistance (median at the first
session, 0.9 cm H
2
O/L/s; median at the last session,
0.5 cm H
2
O/L/s), and resonant frequency of the
airways (median at the first session, 18.2 Hz; median
at the last session, 16.4 Hz), compared with the
waiting list and placebo groups, in which no changes
were observed (log values, for normalization, and
probability statistics are in Fig 1 and Table 5). The
significance of these patterns was tested using the
treatment ⫻ session interaction, adjusted for age,
height, and weight, as shown in Table 5. However,
when controlled for tidal volume and respiration rate to
eliminate spurious findings (Zrs measures decrease as
lung volumes increase during respiration
30
), only the
findings for resistance at 6 Hz remained significant. We
found large and highly significant increases in tidal
volume and decreases in respiratory frequency during
biofeedback in the two groups receiving biofeedback
(Fig 2) [treatment ⫻ task: tidal volume, F
36,1057
⫽ 7.51
(p ⬍ 0.0001); respiratory frequency, F
36,1050
⫽ 23.35
(p ⬍ 0.0001)] (within-group comparisons for biofeed-
back vs rest periods were significant at p ⬍ 0.0001 for
the HRV biofeedback groups but were not significant
for the two control groups). Respiratory frequency
dropped to approximately 0.1 Hz, as occurred in our
previous research on this procedure.
11
The baseline
presession respiration rate dropped significantly in the
full protocol group from the first to last sessions (Fig 2),
but not in the group receiving HRV biofeedback alone.
Biofeedback did not appear to have any immediate
effects on Zrs. The groups did not differ significantly
in within-session contrasts (ie, the treatments ⫻ tasks
interaction, contrasts between rest periods and biofeed-
back periods, and contrasts between beginning-of-
session and end-of-session rest periods to test the
within-session carry-over effect of training).
Spirometry
There were no interpretable changes in spirome-
try, either within or between sessions, in any of the
treatment groups, and no significant differences
between groups.
Table 4—Level of Prescribed Medication (13-Point Scale)*
Session
Full Protocol HRV Biofeedback Alone EEG Biofeedback Placebo Waiting List
No. Mean SD No. Mean SD No. Mean SD No. Mean SD
1 19 8.14 1.82 18 7.42 2.69 20 6.95 2.45 23 7.47 2.8
4 19 7.54 2.19 17 6.59 2.35 20 7.13 3.03 23 8 2.67
7 19 6.51 2.1 17 6.06 2.48 19 6.89 2.53 23 7.79 2.79
10 19 5.49 2.41 17 5.12 2.78 19 6.05 2.46 23 7.58 2.68
*Medication score was for the week prior to each of the four testing sessions. The medication score was increased by one level if a subject took
an average of 8 or more doses of albuterol/d during any of these periods.
356
Clinical Investigations
Asthma Symptoms
Asthma symptoms were scored for the four levels
of asthma severity (Table 1). The groups differed
significantly (Fig 2) [groups ⫻ sessions interaction,
F
9,260
⫽ 4.1; p ⬍ 0.0001]. Symptoms decreased sig
-
nificantly from the first to last sessions for the full
protocol (t
260
⫽ 3.05; p ⬍ 0.003), for HRV biofeed
-
back alone (t
260
⫽ 2.39; p ⬍ 0.02), and for the EEG
biofeedback placebo (t
260
⫽ 2.56; p ⬍ 0.02). The
change was not significant for the waiting list group
(t
260
⫽ 1.1).
Treatment Credibility
We separately analyzed each of the three items on
the credibility questionnaire. No significant between-
groups differences emerged across groups on any of
the questions (p ⬎ 0.4). Subjects gave high credibil-
ity ratings on all three questions (mean range, 6.5 to
7.5 [across groups on the 9-point scale]).
Occurrence of Asthma Exacerbations
Despite the decrease in inhaled steroid dosage in
the biofeedback groups, there was no evidence for
increased risk of a severe asthma flare. Throughout
the study, two subjects each in the full protocol and
the HRV biofeedback alone groups required emer-
gency treatment with oral steroids, whereas four
subjects in the placebo group and five patients in the
waiting list required such an intervention. We also
computed a life table analysis
31
of medication levels
during the week prior to each of the four testing
sessions to examine distribution of increases in con-
troller medication over baseline (Fig 3). We found
no such increases in the full protocol group, three
increases in the HRV biofeedback group over ap-
proximately 4 months, six increases in the placebo
group, and seven increases in the waiting list group
(log-rank test, 8.4088; degrees of freedom, 3;
p ⫽ 0.04). Exacerbations began occurring in the
control groups within ⬍ 20 days.
HRV and Baroreflex Gain
HRV increased in response to biofeedback during
training sessions, as had been found previously
among healthy subjects.
14
Full-spectrum HRV
(range, 0.005 to 0.4 Hz) as well as low-frequency
HRV (range, 0.05 to 0.15 Hz) both differed signifi-
cantly among treatment groups (treatment ⫻ task
interaction: full-spectrum HRV, F
36,612
⫽ 1.95
[p ⬍ 0.001]; low-frequency HRV, F
36,612
⫽ 5.30
[p ⬍ 0.0001]), increasing during biofeedback periods
only for the groups receiving the full protocol (full-
spectrum HRV, t
612
⫽ 7.21 [p ⬍ 0.0001]; low-
frequency HRV, t
612
⫽ 12.59 [p ⬍ 0.0001]) and HRV
biofeedback alone (full-spectrum HRV, t
612
⫽ 5.61
[p ⬍ 0.0001]; low-frequency HRV, t
612
⫽ 11.67
[p ⬍ 0.0001]). Baroreflex gain also increased signifi-
cantly within sessions during biofeedback practice, only
for the groups receiving the full protocol (t
589
⫽ 2.95;
Figure 1. Top: log oscillation resistance at 6 Hz. Middle:
frequency-dependent resistance drop. Bottom: log resonant fre-
quency of airways.
www.chestjournal.org CHEST / 126/2/AUGUST, 2004 357
p ⬍ 0.004) and HRV biofeedback alone (t
589
⫽ 4.56;
p ⬍ 0.0001), but the interaction was not significant.
Cardiovascular measures did not change significantly
across sessions, nor were between-session cardiovascu-
lar changes correlated with between-session effects in
medication consumption, asthma symptoms, or forced
oscillation pneumography.
Discussion
HRV biofeedback appears to be promising as an
adjunctive treatment for asthma, and it appears to
maintain the condition of asthma patients with a
reduced dose of inhaled steroids. A decrease of two
to three medication steps occurred in the active-
treatment groups. This change of approximately one
level in asthma severity (from a mean in the upper level
of moderate asthma to a mean in the upper level of
mild persistent asthma, as defined in Table 1 and with
results shown in Table 3) is clinically significant. No
level changes occurred in the two control groups, and
decreases in medication were greater in the HRV
biofeedback groups than in the two control groups.
The results of biofeedback appear to reflect spe-
cific training effects rather than a placebo-like effect
of treatment expectancy or response to increased
therapeutic attention. The placebo condition had a
very similar format to the real biofeedback condi-
tions and was just as credible as an asthma treatment
for subjects as HRV biofeedback, but it produced
negligible effects on asthma severity. Consistent with
the suggestive power of the placebo condition, the
improvement in asthma symptoms was as great as in
the HRV biofeedback groups, despite the lack of
change in measures of pulmonary function or physi-
cians’ medication prescriptions. HRV biofeedback
also affected the physiologic parameters of asthma.
The mostly equivalent effects for the full protocol
vs HRV biofeedback alone suggests that the biofeed-
back procedure, rather than abdominal or pursed-
lips breathing, produced the therapeutic effects.
However, the mechanism for the biofeedback effects
was not proven. Although, as in previous studies of
healthy people, the biofeedback procedure produced
immediate changes in HRV and baroreflex gain, no
longer term changes occurred in these measures that
could explain the asthma improvements. One possi-
ble mechanism may be a long-term bronchodilation
effect. The immediate bronchodilating effects of
practicing the technique were not apparent, how-
ever, and may have been masked by changes in
respiratory pattern. Nevertheless, some subjects in-
formally reported that they had used the slow-
breathing method to stop asthma exacerbations.
Future research is necessary to verify whether such
rescue effects of HRV biofeedback do occur, and to
examine the effects of HRV biofeedback on inflam-
mation and mucus secretion, particularly in view of
evidence of neurogenic links to these processes.
32,33
However, if HRV biofeedback only produces
bronchodilation, the use of the method as a substi-
tute for antiinflammatory medication should be un-
dertaken with caution. Although bronchodilator
treatment may allow a reduction in conjoint antiin-
Table 5—Forced Oscillation Pneumography: Mixed Models Analyses on Between-Session Effects*
Long-term
(Between-Session)
Effects Measure/Group Contrast
LOCF Completers
Controlled for
Tidal Volume and
Respiration Rate
Statistic Value† p Value Statistic Value† p Value Statistic Value† p Value
Log Zrs 6-Hz Resistance Treatment ⫻ session F
9,156
3.94 ⬍ 0.0002 F
9,267
2.65 ⬍ 0.006 F
9,264
2.75 ⬍ 0.005
Full protocol Baseline session 1 vs 10 t
624
⫺ 3.66 ⬍ 0.0003 t
1068
⫺ 3.23 ⬍ 0.002 t
1056
⫺ 3.41 ⬍ 0.0007
HRV bfk alone Baseline session 1 vs 10 t
624
⫺ 2.39 ⬍ 0.02 t
1068
⫺ 3.2 ⬍ 0.002 t
1056
⫺ 3.25 ⬍ 0.002
Placebo bfk Baseline session 1 vs 10 t
624
0.73 NS t
1068
⫺ 0.04 NS t
1056
⫺ 0.04 NS
Waiting list Baseline session 1 vs 10 t
624
⫺ 0.14 NS t
1068
0.09 NS t
1056
0.09 NS
Log Zrs Freq Depend Treatment ⫻ session F
9,156
3.33 ⬍ 0.0009 F
9,267
2.73 ⬍ 0.005 F
9,264
1.24 NS
Full protocol Baseline session 1 vs 10 t
608
⫺ 3.66 ⬍ 0.0003 t
1066
⫺ 3.73 ⬍ 0.0002 t
1054
⫺ 0.19 NS
HRV bfk alone Baseline session 1 vs 10 t
608
⫺ 1.83 ⬍ 0.07 t
1066
⫺ 2.22 ⬍ 0.03 t
1054
0.01 NS
Placebo bfk Baseline session 1 vs 10 t
608
0.02 NS t
1066
⫺ 0.35 NS t
1054
⫺ 1.41 NS
Waiting list Baseline session 1 vs 10 t
608
0.4 NS t
1066
0.6 NS t
1054
0.67 NS
Log Zrs resonant Freq Treatment ⫻ session F
9,156
5.34 ⬍ 0.0001 F
9,267
5 ⬍ 0.0001 F
9,264
1.08 NS
Full protocol Baseline session 1 vs 10 t
624
⫺ 2.36 ⬍ 0.02 t
1068
⫺ 2.58 ⬍ 0.01 t
1056
⫺ 1.17 NS
HRV bfk alone Baseline session 1 vs 10 t
624
⫺ 2.01 ⬍ 0.05 t
1068
⫺ 3.35 ⬍ 0.001 t
1056
⫺ 0.05 NS
Placebo bfk Baseline session 1 vs 10 t
624
0.78 NS t
1068
0.14 NS t
1056
⫺ 1.73 NS
Waiting list Baseline session 1 vs 10 t
624
1.21 NS t
1068
0.78 NS t
1056
1.64 NS
*NS ⫽ not significant; Freq ⫽ frequency. See Table 3 for abbreviations not used in text.
†Minus sign in t tests represents direction of change from session 1 to 10, and from baseline to biofeedback tasks.
358
Clinical Investigations
flammatory treatment, the total elimination of such
treatment increases the risk of asthma exacerba-
tion.
34
Our exacerbation data suggest that fewer
exacerbations may have occurred in the group re-
ceiving HRV biofeedback, despite the decrease in
inhaled steroid dosage. It is possible that biofeed-
back may have a steroid-sparing effect without some
of the long-term side effects of salmeterol, possibly
including ischemic heart disease.
35
The limitations of this study include its relatively
short duration and its lack of follow-up to assess
long-term effects. Also, the placebo condition may
have required less task involvement than the HRV
biofeedback conditions, and thus may have had a
smaller placebo effect, although such differences
were not found in our measures of treatment cred-
ibility. In addition, this highly controlled experimen-
tal protocol may have attracted patients with higher
treatment motivation than would occur in the gen-
eral population, and personality characteristics of the
single biofeedback therapist in this study, who was
not blinded, may have affected the efficacy of the
intervention. Fourth, differences between the wait-
ing list group and other groups in the duration of the
protocol and the frequency of assessment sessions
may have affected the size of the medication changes
in the current study, although we believe it is
unlikely that this played a role, because there was no
perceptible across-session trend in asthma severity in
this group, while such trends did occur in the HRV
biofeedback groups. Additionally, some asthma ex-
acerbations may have been missed during periods
other than the week prior to each testing session.
Finally, because functional residual volume and rest-
ing lung volume were not measured during forced
oscillation pneumography, it is possible that changes
Figure 2. Top: tidal volume. Middle: respiratory frequency.
Bottom: asthma symptoms (from daily diaries).
Figure 3. Survival function of time to treatment failure. See
Table 3 for abbreviation not used in text.
www.chestjournal.org CHEST / 126/2/AUGUST, 2004 359
in Zrs values could have been caused by increases in
these values, particularly during voluntary respira-
tory maneuvers in biofeedback in which the behav-
ioral effects many have overridden the more auto-
matic physiologic control of breathing. We note,
however, that the Zrs effects, particularly resistance
at 6 Hz, persisted during rest periods, during which
no special breathing maneuvers were performed,
after factoring out the effects of residual changes in
respiratory patterns while at rest. The effects of HRV
biofeedback on respiratory resistance thus were in-
dependent of the effects of voluntary respiratory
maneuvers. It is also unlikely that more general
changes occurred in functional residual capacity or
resting lung volume, because spirometry data show
no changes in full vital capacity, suggesting the absence
of air trapping, and subjects had been instructed to
exhale as fully as was voluntarily possible.
The lack of spirometry findings probably reflects
the use of spirometry as a principal criterion for
adjusting medication (ie, improvements directly led
to decreases in medication, which would prevent
further improvement in spirometry values). It is
notable that the study by Lo¨fdahl et al,
16
using this
same experimental design for evaluating montelukast
sodium, similarly found decreases in inhaled steroid
dosage but no changes in spirometry values. Asthma
symptoms, which also were a criterion for medica-
tion adjustment, decreased, although the results
were not as strong as those for medication dosage or
Zrs. Zrs measures were apparently sensitive to as-
pects of lung function that were not assessed by
spirometry. The fact that, of the Zrs measures, only
resistance at 6 Hz remained significant after adjust-
ment for respiratory patterns, suggests an increase in
airway caliber, rather than in other airway and chest
wall tissue properties (eg, tissue compliance).
23
Air-
way caliber would be particularly relevant for
asthma.
Further research is needed to verify whether, as
suggested by our findings, this biofeedback method
can have a safe but significant steroid-sparing effect
in clinical practice. Caution is advised at this time in
using this method for treating asthma, until the
mechanisms of action are better understood and the
long-term protective effect has been documented.
The decrease in the use of steroid medications with
this method did not appear to pose a risk of asthma
exacerbation in the current study, but it is possible
that such effects might become evident in a longer
trial.
ACKNOWLEDGMENTS: Assistance in the clinical treatment of
subjects was given by Catherine Monteleone, MD, Stuart Ho-
chron, MD, Arvind Das, MD, and Donna Klitzman, MD. Robert
Hamer, PhD, designed the randomization routine. Jonathan
Feldman, PhD, assisted with recruitment. Nissy Ann Vorghese,
Ami Doshi, and Jodi Casabianca scored the medication data and
assisted in developing the manual for scoring asthma severity.
Dwain Eckberg, MD, and Tom Kuusela, PhD, assisted in the
calculation and interpretation of HRV and baroreflex data.
References
1 Wong CA, Walsh LJ, Smith CJ, et al. Inhaled corticosteroid
use and bone-mineral density in patients with asthma. Lancet
2000; 355:1399 –1403
2 Kelloway JS, Wyatt RA, Adlis S. Comparison of patients’
compliance with prescribed oral and inhaled asthma medica-
tions. Arch Intern Med 1994; 154:1349–1352
3 Singh V, Sinha HV, Gupta R. Barriers in the management of
asthma and attitudes towards complementary medicine. Re-
spir Med 2002; 96:835– 840
4 Baldwin CM, Long K, Kroesen K, et al. A profile of military
veterans in the southwestern United States who use comple-
mentary and alternative medicine: implications for integrated
care. Arch Intern Med 2002; 162:1697–1704
5 Tsugeno H, Tsugeno H, Fujita T, et al. Vertebral fracture and
cortical bone changes in corticosteroid-induced osteoporosis.
Osteoporos Int 2002; 13:650 –656
6 Matsumoto H, Ishihara K, Hasegawa T, et al. Effects of
inhaled corticosteroid and short courses of oral corticoste-
roids on bone mineral density in asthmatic patients: a 4-year
longitudinal study. Chest 2001; 120:1468–1473
7 Sivri A, Coplu L. Effect of the long-term use of inhaled
corticosteroids on bone mineral density in asthmatic women.
Respirology 2001; 6:131–134
8 Chan PW, DeBruyne JA. Parental concern towards the use of
inhaled therapy in children with chronic asthma. Pediatr Int
2000; 42:547–551
9 Simon RA. Update on inhaled corticosteroids: safety, compli-
ance, and new delivery systems. Allergy Asthma Proc 1999;
20:161–165
10 Hand CH, Bradley C. Health beliefs of adults with asthma:
toward an understanding of the difference between symptom-
atic and preventive use of inhaler treatment. J Asthma 1996;
33:331–338
11 Lehrer PM, Carr RE, Smetankine A, et al. Comparison of
respiratory sinus arrhythmia and neck/trapezius EMG
biofeedback for asthma: a pilot study. Appl Psychophysiol
Biofeedback 1997; 22:95–109
12 Lehrer P, Smetankin A, Potapova T. Respiratory sinus ar-
rhythmia biofeedback therapy for asthma: a report of 20
unmedicated pediatric cases using the Smetankin method.
Appl Psychophysiol Biofeedback 2000; 25:193–200
13 Kazuma N, Otsuka K, Matsuoka I, et al. Heart rate variability
during 24 hours in asthmatic children. Chronobiol Int 1997;
14:597–606
14 European Society of Cardiology and the North American
Society of Pacing and Electrophysiology. Heart rate variabil-
ity: standards of measurement, physiological interpretation,
and clinical use; Task Force of the European Society of
Cardiology and the North American Society of Pacing and
Electrophysiology. Circulation 1996; 93:1043–1065
15 Lehrer PM, Vaschillo E, Vaschillo B, et al. Heart rate
variability biofeedback increases baroreflex gain and peak
expiratory flow. Psychosom Med 2003; 65:796–805
16 Lo¨ fdahl CG, Reiss TF, Leff JA, et al. Randomised, placebo
controlled trial of effect of a leukotriene receptor antagonist,
montelukast, on tapering inhaled corticosteroids in asthmatic
patients. BMJ 1999; 319:87–90
17 National Asthma Education and Prevention Program. Expert
panel report: guidelines for the diagnosis and management of
asthma; update on selected topics–2002. Washington, DC:
360 Clinical Investigations
US Department of Health and Human Services, 2002
18 Lehrer PM, Vaschillo E, Vaschillo B. Resonant frequency
biofeedback training to increase cardiac variability: rationale
and manual for training. Appl Psychophysiol Biofeedback
2000; 25:177–191
19 Lehrer PM, Hochron SM, McCann B, et al. Relaxation
decreases large-airway but not small-airway asthma. J Psycho-
som Res 1986; 30:13–25
20 Lehrer PM, Hochron S, Mayne T, et al. Relaxation and music
therapies for asthma among patients prestabilized on asthma
medication. J Behav Med 1994; 17:1–24
21 American Thoracic Society. Standardization of spirometry.
Am J Respir Crit Care Med 1995; 152:1107–1136
22 Crapo RO, Morris AH, Gardner RM. Reference spirometric
values using techniques and equipment that meet ATS
recommendations. Am Rev Respir Dis 1981; 123:659 –664
23 Nagels J, Landser FJ, van der Linden L, et al. Mechanical
properties of the lungs and chest wall during spontaneous
breathing. J Appl Physiol 1980; 49:408–416
24 Habib RH, Jackson AC. Total respiratory system impedance
with the upper airway wall shunt minimized. J Appl Physiol
1993; 74:1045–1055
25 Lutchen KR, Jackson AC. Reliability of parameter estimates
from models applied to respiratory impedance data. J Appl
Physiol 1987; 62:403–413
26 Lutchen KR, Habib RH, Dorkin HL, et al. Respiratory
impedance and multibreath N2 washout in healthy, asth-
matic, and cystic fibrosis subjects. J Appl Physiol 1990;
68:2139–2149
27 Borkovec TD, Nau SD. Credibility of analogue therapy
rationales. J Behav Ther Exp Psychiatry 1972; 3:257–260
28 Akaike H. A new look at the statistical model identification.
IEEE Trans Autom Control 1974; 19:716–723
29 Rubin DB. Inference and missing data. Biometrika 1976;
63:581–592
30 van den Elshout FJ, van de Woestijne KP, Folgering HT.
Variations of respiratory impedance with lung volume in
bronchial hyperreactivity. Chest 1990; 98:358–364
31 Lee, ET. Statistical methods for survival data analysis. New
York, NY: Wiley, 1992
32 Birrell MA, Crispino N, Hele DJ, et al. Effect of dopamine
receptor agonists on sensory nerve activity: possible therapeu-
tic targets for the treatment of asthma and COPD. Br J
Pharmacol 2002; 136:620–628
33 Barnes PJ. Neurogenic inflammation in the airways. Respir
Physiol 2001; 125:145–154
34 Lemanske RF Jr, Sorkness CA, Mauger EA, et al. Inhaled
corticosteroid reduction and elimination in patients with
persistent asthma receiving salmeterol: a randomized con-
trolled trial. JAMA 2001; 285:2594–2603
35 Martin RM, Dunn NR, Freemantle SN, et al. Risk of
non-fatal cardiac failure and ischaemic heart disease with long
acting beta 2 agonists. Thorax 1998; 53:558 –562
www.chestjournal.org CHEST / 126/2/AUGUST, 2004 361