1. Art and Diseño

Original Research
ASTHMA
Symptom Perception in Pediatric
Asthma*
Resistive Loading and In Vivo Assessment
Compared
Gregory K. Fritz, MD; Sue K. Adams, MA; Elizabeth L. McQuaid, PhD;
Robert Klein, MD; Sheryl Kopel, MSc; Jack Nassau, PhD; and
Anthony Mansell, MD
Background: Inaccurate symptom perception contributes to asthma morbidity and mortality in
children and adults. Various methods have been used to quantify perceptual accuracy, including
psychophysical (resistive loading) approaches, ratings of dyspnea during induced bronchoconstriction, and in vivo monitoring, but it is unclear whether the different methods identify the same
individuals as good or poor perceivers. The objectives of the study were as follows: (1) to compare
in the same asthmatic children two methods of quantifying perceptual ability: threshold detection
of added resistive loads and in vivo symptom perception; and (2) to determine which method best
predicts asthma morbidity.
Methods: Seventy-eight asthmatic children 7 to 16 years of age completed two threshold detection
protocols in the laboratory and recorded their subjective estimates of lung function prior to
spirometry at home twice daily for 5 to 6 weeks. Summary measures from both methods were
compared to each other and to asthma morbidity (as measured with the Rosier asthma functional
severity scale).
Results: Symptom perception ability, as summarized by either method, varied greatly from child
to child. Neither of the resistive load detection thresholds were significantly related to any of the
three in vivo perception scores, nor were they related to asthma morbidity. The three in vivo
scores did show a significant or marginal relationship with morbidity (p < 0.01, p < 0.06, and
p < 0.07, respectively).
Conclusions: Resistive loading techniques may not be useful in assessing symptom perception
ability in children. Measuring estimates of symptoms in relation to naturally occurring asthma can
identify children at risk for greater asthma morbidity.
(CHEST 2007; 132:884 – 889)
Key words: asthma; dyspnea; pediatric
Abbreviations: PEFR ⫽ peak expiratory flow rate; Rrs ⫽ intrinsic respiratory resistance
is increasing recognition that for both chilT here
dren and adults with asthma, inaccurate symptom perception is a contributing factor to morbidity
and mortality. Given the centrality of symptom
*From the Departments of Psychiatry (Drs. Fritz, McQuaid, and
Nassau, Ms. Adams, and Ms. Kopel) and Pediatrics (Drs. Klein and
Mansell), Brown University School of Medicine, Providence, RI.
This work was performed at Rhode Island Hospital, Providence,
RI, and The University of Texas Medical Center, Tyler, TX.
This work was supported by National Heart, Lung, and Blood
Institute grant 2R01- HL45157.
The authors have no conflicts of interest to disclose.
884
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
perception in the identification of asthma symptoms
and subsequent management strategies, effective
methods for quantifying perceptual accuracy are
Manuscript received January 4, 2007; revision accepted May 6,
2007.
Reproduction of this article is prohibited without written permission
from the American College of Chest Physicians (www.chestjournal.
org/misc/reprints.shtml).
Correspondence to: Gregory K. Fritz, MD, Department of Psychiatry, Bradley Hasbro Research Center, Coro West 2.155, 1
Hoppin St, Providence, RI 02903; e-mail: [email protected]
DOI: 10.1378/chest.06-2140
Original Research
needed. Blunted sensitivity to added resistive loads
has been shown to differentiate adults and adolescents with near-fatal asthma episodes from other
asthmatics and from control subjects.1,2
A variety of methods have been developed to
assess perception of respiratory sensation and dyspnea.3 One common method is to present varying
external resistive or elastic loads to subjects who
indicate their subjective sensations through a “detect/nondetect” response or an estimate of the magnitude of the resistance. Such extrinsic loading techniques are readily standardized and allow a number
of trials at different resistances to be completed in a
short period of time. However, this approach poorly
replicates the physiologic, mechanical, or psychological experience of real asthma.4 Techniques in which
intrinsic resistance is varied by inducing bronchoconstriction and where dyspnea is quantified with a
visual analog scale or a logarithmic (Borg) scale are
likely to be less artifactual. Protocols in which bronchoconstriction is induced are also well standardized
and efficient, but the number of observations is
typically limited and the laboratory setting may not
truly represent what happens in an asthma episode.
Techniques to quantify patients’ subjective awareness of naturally occurring episodes are the most
realistic method of assessing respiratory perception,
but also may be cumbersome, time consuming, and
difficult to standardize. Conceptual and methodologic issues in this realm have been described,5 but
all approaches typically require a patient to make
multiple subjective symptom assessments prior to
corresponding objective measures of pulmonary
function.
To date, there are no studies of how perception of
pulmonary function indices in real life relate to
either laboratory method of quantifying perceptual
accuracy in either adults or children with asthma.
The present study was therefore undertaken to
compare two well-tested protocols for assessing
symptom perception in a large number of children
with asthma, and to determine the extent that each
method is associated with asthma morbidity.
Materials and Methods
Sample
The subjects in this study were 78 children aged 7 to 17 years
(mean ⫾ SD, 11.4 ⫾ 2.1 years) recruited for participation in
Rhode Island (n ⫽ 42) and Texas (n ⫽ 36). In Rhode Island,
subjects came largely from physician and emergency department
referrals. In Texas, they were recruited from an asthma summer
camp. This study was reviewed and approved by the appropriate
institutional review board at each site, and subjects and parents
gave informed consent prior to participation. Demographic characteristics of the sample are summarized in Table 1. Asthma
www.chestjournal.org
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
Table 1—Demographic Characteristics of Participants
(n ⴝ 78)
Characteristics
Age, yr
Socioeconomic status
(Hollingshead)
Gender
Female
Male
Race/ethnicity
White, non-Hispanic
Black, non-Hispanic
Hispanic
Multiracial
Asthma severity
Mild intermittent
Mild persistent
Moderate persistent
Severe persistent
% of Sample
Mean
SD
Range
11.4
43.4
2.1
11.8
7–16
12–66
40
60
75
14
4
7
10
57
26
7
severity was assessed according to existing National Heart, Lung,
and Blood Institute guidelines via a consensus rating by two
asthma specialists: a pediatric pulmonologist (A.M.) and a pediatric allergist (R.K.). Data from parental report obtained via the
Rosier asthma functional severity scale,6 and the child’s medication regimen provided the basis for the severity consensus rating
in a process that we have described previously.7 Socioeconomic
status was determined using the Hollingshead four-factor index
of social position8; all socioeconomic levels were represented.
The subject groups from Texas and Rhode Island were compared
on these demographic and severity variables. The Texas sample
was ethnically more diverse than the Rhode Island sample (74%
minority vs 24% in Rhode Island, p ⬍ 0.01), but no other
differences were found. Asthma morbidity was summarized in
the Rosier asthma functional severity scale index, a composite
score calculated as a mean of items assessing episode frequency,
symptom frequency between episodes, and impairment during
and between episodes.6
The 78 subjects constituted a subgroup of participants who had
valid data for the variables included in these analyses. The
subgroup was drawn from a larger sample of 121 participants.
The 43 children who were excluded because of incomplete data
had significantly lower socioeconomic status than the subjects
included in the final data set, but no other differences were
identified.
Procedure
Threshold Detection of Added Resistive Loads: The apparatus
and procedure for the resistive loading protocol has been described in detail elsewhere.9 A subject’s own intrinsic respiratory
resistance (Rrs) was measured (Jaeger Impulse Oscillometer,
model LF1; Ganshorn Electronics; Niederlauer, Germany), and
loads were presented as percentages of the child’s Rrs in
accordance with the Weber law. The apparatus and the details of
the protocol used in this study represented an improved, secondgeneration procedure incorporating attentional, motivational,
and technical modification to increase the assessment validity
over our first series of studies.9
Subjects were oriented to the apparatus and practiced load
recognition until they demonstrated an adequate understanding
of the procedure. Once familiarized, the subjects participated in
two protocols, the order of which was randomly determined. In
CHEST / 132 / 3 / SEPTEMBER, 2007
885
the “tracking protocol,” the first load was 100% of the subject’s
Rrs. If detected, the next load was 20% lower; if not detected, the
next load was 20% higher. A total of 20 loads were presented, and
the “tracking threshold” was calculated as a plateau of five
consecutive and alternating detections and nondetections. In the
“random protocol,” subjects were presented with 33 loads in a
predetermined, random order; loads ranged from 0% to 200% of
Rrs. Here the threshold was defined as the load detected 50% of
the time and was calculated via a linear regression of load
presented and subject response (detect/nondetect). Only if the
regression was statistically significant was the case assigned a
“random threshold.”
In Vivo Symptom Perception: Subjects were instructed to use
a programmable, hand-held spirometer (Jaeger AMII; VIASYS
Healthcare; Yorba Linda, CA) twice daily at home for 5 to 6
weeks. At each assessment, subjects used a screen on the side of
the device to enter their subjective estimates of their respiratory
state at the moment in the form of a guess of their current peak
expiratory flow rate (PEFR). Once these values were entered, the
device allowed the subject to proceed with spirometric assessment, and the best of three “blows” was saved with the corresponding subjective data.
Many of the subjects were familiar with peak flow monitoring
and their own PEFRs prior to participation in the study. All
subjects, regardless of previous PEFR experience, received
thorough, individualized training on the Jaeger AMII from a
project staff member. Under the assumption that there might be
a period of familiarization with the instrument, the data from day
1 were omitted from analysis. Experience7,10 has demonstrated
that the peak flow guess is a meaningful way for the children to
quantify their subjective estimates of current pulmonary function. Human subject requirements dictated that the children see
their actual PEFR values at each blow; thus, the possibility of
learning over the period of the study existed. However, a previous
analysis7 of the subjective/objective correlation in the first half of
the study period compared to the last half revealed no significant
difference.
Data generated during the assessment period were stored in
the device and downloaded in the laboratory. The data were
summarized using the asthma risk grid as previously described11
and illustrated in Figure 1. By this method, for each participant
the PEFR guess and corresponding actual PEFR (both converted
to percentage of personal best units) are plotted on the vertical
and horizontal axes, respectively. Each point in an individual
subject’s data set falls into either the “accurate zone” (subjective
assessment closely approximates objective clinical status), the
“danger zone” (clinically significant compromised function is
missed by the patient) or the “symptom magnification zone”
(reflecting oversensitivity to minor symptoms or exaggeration of
symptoms). Each subject’s data are summarized as the percent of
his/her total observations falling in each of the three zones. Thus
each child’s perceptual ability is characterized by three scores
from the asthma risk grid, and the unit of measure in subsequent
analyses is the individual child (Fig 2).
Data Analysis
All analyses were performed using the statistical software
(Statistical Software Package for Social Science, version 12.0;
SPSS; Chicago, IL). Probit transformations12 were applied to
variables unlikely to conform to assumptions of normality and
homogeneity of variance. Specific data-cleaning procedures were
implemented for threshold detection and in vivo data; details are
available on request. Pearson product-moment correlations between primary study variables and demographic variables were
conducted. Independent-sample t tests and univariate analysis if
variance were used for categorical comparisons between demo886
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
The Asthma Risk Grid
Accurate Zone:
–
boxes 1, 5, 9 & +/- 10% wedge
Danger Zone:
–
boxes 4, 7 & 8
Symptom Magnification Zone:
–
boxes 2, 3 & 6
Figure 1. The asthma risk grid. PEF ⫽ peak expiratory flow.
graphic and primary study variables. A series of hierarchical
linear regression analyses were used to examine the independent
contribution of the primary study variables (the two resistive
loading thresholds and the three in vivo scores from the asthma
risk grid) to the prediction of variation in asthma morbidity.
Asthma severity was controlled for in the first step of the
regression equations given its potential as a confounding variable.
Results
Means and SDs of the key predictor and outcome
variables are presented in Table 2.
Resistive Load Detection
Consistent with our previous research,9 there was
a wide range in the thresholds achieved in both
121
Total Participants
Mastered RL Detection Protocol
Sufficient Home Spirometry Data
Complete Outcome Data
No: 15
Yes: 106
No: 16
Yes: 90
No: 12
Yes: 78
Final Sample
Figure 2. Sample derivation.
Original Research
Table 2—Descriptive Data for Primary Study
Variables (n ⴝ 78)
Variables
Predictor
In accurate zone, %
In danger zone, %
In symptom magnifier zone, %
Random threshold, % intrinsic
resistance
Tracking threshold, % intrinsic
resistance
Outcome
Baseline asthma morbidity
Mean (SD)
Range in Sample
76.4 (2.19)
11.7 (15.4)
12.0 (17.3)
91.6 (16.0)
3.3–100.0
0–78.1
0–96.7
46.0–118.6
104.9 (73.0)
10–330.2
1.4 (1.0)
0–4.3
protocols. By definition of the sample, all 78 subjects
achieved a valid threshold in at least one of the
protocols. For the random protocol, 83% (n ⫽ 65)
attained a threshold, which averaged 92% of intrinsic
resistance (range, 46 to 119%). In the tracking
protocol 94% (n ⫽ 73) of the subjects achieved a
threshold, which averaged 105% of intrinsic resistance (range, 10 to 330%).
Associations among demographic variables and
asthma severity to both random and tracking thresholds were explored, and two significant relationships
were identified. Children who achieved random
thresholds were significantly older than those who
did not (mean age, 11.6 years vs 10.2 years, respectively; p ⬍ 0.05). Children with mild intermittent
asthma had higher random thresholds than those
with other levels of asthma severity (p ⬍ 0.01).
In Vivo Symptom Perception
The subjects averaged 52 subjective/objective data
point pairs (range, 21 to 122) using the portable
spirometer over the assessment period, which were
plotted on the grid for each subject. The scores
reflected considerable between-subject variability in
symptom perception accuracy. On average, across all
subjects, 76% of the points were in the accurate
zone, but this percentage ranged from 3 to 100% for
individual children. Of the points in the accurate
zone, on average 22% were when the patient was
compromised (ie, objective PEFR ⬍ 80% of per-
sonal best). The subjects averaged 12% of points in
the danger zone (range, 0 to 78%) and 12% of points
in the symptom magnification zone (range, 0 to
97%). There were no significant associations between demographic variables and symptom perception scores.
Relation Between Resistive Load Detection
Thresholds and In Vivo Symptom Perception Scales
Table 3 summarizes the correlations between primary study variables. Resistive load detection thresholds for the random and tracking protocols were
significantly and positively intercorrelated. Similarly,
the three grid zone scores were highly intercorrelated in the expected directions. We expected those
with high danger zone scores to have relatively
higher thresholds, those with high symptom magnification zone scores to have low thresholds, and
those with high accurate zone scores to have mid
range thresholds. Neither of the resistive load detection thresholds, however, was related to any of the
three scores from the in vivo method of assessing
symptom perception.
Hierarchical linear regression techniques were
used to determine which method of quantifying
perceptual ability had the best predictive validity in
terms of asthma morbidity. These results are summarized in Table 4. The accurate zone score significantly predicted baseline asthma morbidity, whereas
the danger zone score and symptom magnification
score were marginal predictors of morbidity. In
contrast, neither of the resistive loading thresholds
approached predictive significance. Essentially the
same results were obtained when the analyses were
performed when not controlling for asthma severity.
Discussion
Three studies in the literature have compared
resistive loading techniques with laboratory-induced
bronchoconstriction, with mixed results. Two studies13,14 found a significant relationship between the
methods of assessing perception, and one study15 did
not. The present study is the first to examine the
Table 3—Significant Bivariate Correlations of Primary Study Variables
Variables
1.
2.
3.
4.
5.
1. Random Threshold
2. Tracking Threshold
3. Accurate Zone
4. Danger Zone
5. Symptom Magnifier
0.64*
0.02
⫺ 0.08
⫺ 0.09
0.21
⫺ 0.63*
0.09
⫺ 0.06
⫺ 0.73*
⫺ 0.01
Random threshold
Tracking threshold
Accurate zone
Danger zone
Symptom magnifier zone
*p ⬍ 0.01.
www.chestjournal.org
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
CHEST / 132 / 3 / SEPTEMBER, 2007
887
Table 4 —Individual Multiple Linear Regression
Models for Resistance Loading and In Vivo Symptom
Perception Predicting Asthma Morbidity, Controlling
for Asthma Severity*
Baseline Morbidity
Predictors
Model 1
Asthma severity
Random threshold
Model 2
Asthma severity
Tracking threshold
Model 3
Asthma severity
Accurate zone
Model 4
Asthma severity
Symptom magnifier zone
Model 5
Asthma severity
Danger zone
R2
p Value
0.34
0.10
0.10
0.11
0.01†
0.45
0.35
0.05
0.13
0.13
0.002†
0.69
0.37
⫺ 0.29
0.13
0.22
0.001†
0.006†
0.37
0.20
0.13
0.17
0.001†
0.07
0.37
0.20
0.13
0.17
0.001†
0.06
ß
*Each row represents results of a separate regression. In each case,
asthma severity was entered on the first step.
†p ⬍ 0.01.
relationship between naturalistic and psychophysical
approaches to quantifying symptom perception in
asthmatic children. In a reasonably large sample
using the best equipment available in well-validated
protocols, we found no relationship between the
resistive load detection thresholds and summary
scores from the in vivo method. Moreover, the
thresholds bore no relationship to asthma morbidity,
whereas the in vivo method explained a small but
significant portion of the variance in morbidity.
Clearly, the two assessment approaches, when applied in children, do not measure the same thing.
These findings are consistent with the work of Moy
et al,16 who reported qualitative differences in dyspnea reports when bronchoconstriction was compared to external resistive loading. Despite the methodologic efficiency and theoretical appeal of a
resistive loading paradigm, an approach that measures sensitivity to naturally occurring pathophysiology, whether related to bronchoconstriction or airway inflammation, is likely to be more useful in
quantifying symptom perception in children with
asthma.
This study did not determine the factors behind
the lack of a relationship between the two perceptual
measures. Certainly the intrinsic causes of bronchoconstriction during asthma are different from the
relatively pure stimulation of the mechanoreceptors
in respiratory muscles that occurs with resistive
loading. An asthma episode includes physiologic,
psychological, and situational cues that are absent in
888
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
a laboratory resistive loading protocol. The asthmatic
subjects in the present study were not recruited or
classified for analytic purposes into specific subgroups. Therefore, we cannot rule out the possibility
that the resistive loading thresholds are meaningful
and predictive for a subgroup of the asthmatic
population, such as those who have experienced an
episode of life threatening asthma. However, the full
range of asthma severity was represented in this
sample, making the results more generalizable to the
whole population of children with asthma.
The 78 subjects in this study were much younger
(mean age, 11.4 years) than the subjects in other
studies that have found associations between psychophysical indexes and methacholine-induced bronchoconstriction13–15 or clinical asthma characteristics.1,17 It is possible that our subjects’ relative
cognitive immaturity may have made the resistive
loading paradigm more difficult and prone to artifact
during administration than is the case with adults.
The finding that children who achieved random
thresholds were older than those who did not support this idea.
A number of other study limitations deserve mention. First, only 80% of the children in the final study
group achieved thresholds in both protocols. Despite
several years of effort to refine the apparatus, improve the participant orientation, perfect the administration of the stimuli, and maximize children’s level
of involvement, some children could not produce
meaningful thresholds with this laboratory approach.9 Missing data in this study were not randomly distributed: excluded children had lower socioeconomic status, and the subjects were primarily
white, factors that may limit the generalizability of
the findings. Second, the use of PEFR in the in vivo
protocol, while most familiar to the subjects, constitutes a limitation because it is effort dependent and
may not reflect small airway involvement.
From a clinical perspective and consistent with
other research, this study identified a number of
asthmatic subjects who are poor perceivers of
changes in pulmonary function. These subjects,
when identified by in vivo assessment of symptom
perception, had more asthma-related morbidity than
their peers who were better perceivers. Recent
research18 suggesting that antiinflammatory medications should be administered on an as-needed basis
further highlights the importance of accurate symptom perception. The question of whether perception
of asthma symptoms can be improved with training
remains to be answered, but an intervention aimed
at doing so by improving threshold detection skills
would be appear to be ill advised, given the lack of a
relationship with in vivo symptom perception.
Original Research
References
1 Kikuchi Y, Okabe S, Tamura G. Chemosensitivity and perception of dyspnea in patients with a history of near-fatal
asthma. N Engl J Med 1994; 330:1329 –1334
2 Kifle Y, Seng V, Davenport PW. Magnitude estimation of
inspiratory resistive loads in children with life-threatening
asthma. Am J Respir Crit Care Med 1997; 156:1530 –1535
3 Banzett R, Dempsey J, O’Donnell D, et al. Symptom perception and respiratory sensation in asthma. Am J Respir Crit
Care Med 2000; 162:1178 –1182
4 Kelson S, Prestel T, Cherniack N, et al. Comparison of the
respiratory responses to external resistive loading and bronchoconstriction. J Clin Invest 1981; 67:1761–1768
5 Fritz GK, Yeung A, Wamboldt M, et al. Conceptual and
methodologic issues in quantifying perceptual accuracy in
childhood asthma. Pediatric Psychology 1996; 21:153–173
6 Rosier MJ, Bishop J, Nolan T, et al. Measurement of
functional severity of asthma in children. Am J Respir Crit
Care Med 149:1434 –1441
7 Fritz GK, McQuaid EL, Spirito A, et al. Symptom perception
in pediatric asthma: relationship to functional morbidity and
psychological factors. J Am Acad Child Adolesc Psychiatry
1994; 35:1033–1041
8 Hollingshead A. Four-factor index of social status. New
Haven, CT: Department of Sociology, Yale University, 1975
9 Fritz GK, Rosenblum K, Klein RB, et al. Improved methodology for threshold detection studies in asthmatic children.
J Psychophysiol 2001; 15:190 –197
www.chestjournal.org
Downloaded From: http://journal.publications.chestnet.org/ on 02/06/2015
10 Fritz GK, Klein RB, Overholser JC. Accuracy of symptom
perception in childhood asthma. J Dev Behav Pediatrics 1990;
11:69 –72
11 Klein RB, Walders N, McQuaid EL, et al. The asthma risk
grid: clinical interpretation of symptom perception. Allergy
Asthma Proc 2004; 25:1– 6
12 Cohen J, Cohen P. Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum, 1983
13 Yamamoto H, Inaba S, Nishimura M, et al. Relationship
between the ability to detect added resistance at rest and
breathlessness during bronchoconstriction in asthmatics. Respiration 1987; 52:42– 48
14 Julius SM, Davenport KL, Davenport PW. Perception of
intrinsic and extrinsic respiratory loads in children with
life-threatening asthma. Pediatr Pulmonol 2002; 34:425– 433
15 Bijl-Hofland I, Cloosterman S, vanSchayck C, et al. Perception of respiratory sensation assessed by means of histamine
challenge and threshold loading tests. Chest 2000; 117:954 –
959
16 Moy ML, Weiss JW, Sparrow D, et al. Quality of dyspnea in
bronchoconstriction differs from external resistive loads. Am J
Respir Crit Care Med 2000; 162:451– 455
17 Davenport P, Kifle Y. Inspiratory resistive load detection in
children with life-threatening asthma. Pediatr Pulmonol
2001; 32:44 – 48
18 Rabe KF, Pizzichini E, Stallberg B, et al. Budesonide/
formoterol in a single inhaler for maintenance and relief in
mild-to-moderate asthma: a randomized, double-blind trial.
Chest 2006; 1299:246 –256
CHEST / 132 / 3 / SEPTEMBER, 2007
889