| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Department of Psychology (Dr. Harver), University of North Carolina at Charlotte, NC; Department of Medicine, Dartmouth Medical School (Dr. Mahler), and Department of Psychological and Brain Sciences, Dartmouth College (Dr. Baird), Dartmouth, NH; and Department of Medicine (Dr. Schwartzstein), Harvard Medical School, Boston, MA.
Correspondence to: Andrew Harver, PhD, Department of Psychology, UNC Charlotte, 9201 University City Blvd, Charlotte, NC 28223-0001; e-mail: arharver{at}email.uncc.edu
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Design: Cluster analyses obtained in healthy individuals were compared with those obtained previously in patients who complained of breathing discomfort. In addition, we used multidimensional scaling (MDS) techniques to analyze relationships among descriptors in healthy individuals.
Setting: Public university.
Participants: The participants were 100 healthy individuals (48 men and 52 women) ranging in age between 18 and 65 years (mean, 27.9 ± 11.7 years).
Measurements and results: Participants judged the dissimilarity among pairs of 15 descriptors of breathlessness that were used previously to examine the experience of dyspnea in patients who complained of breathing discomfort. Cluster analysis solutions obtained in the healthy individuals were virtually identical to those obtained previously in patients. Three dimensions (attributes) of breathing discomfort were uncovered with MDS: "Depth and frequency of breathing," "Perceived need, or urge, to breathe," and "Difficulty breathing and phase of respiration." The results did not depend on age, sex, levels of education, or the presence of uncomfortable awareness of breathing with activities.
Conclusions: The relations among descriptors of breathlessness obtained in healthy individuals support the contention that the association of different clusters with different disease states reflects distinct and separable cognitive constructs that are not simply dependent on the presence of an underlying pathophysiology or on a specific disease condition. Our results in healthy individuals also suggest that distinct qualities of breathlessness relate to different physiologic mechanisms underlying respiratory discomfort.
Key Words: cluster analysis • descriptors of breathlessness • dyspnea • multidimensional scaling
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The features (structures) of breathing discomfort in different patient groups were demonstrated initially by Simon and colleagues.14 They used cluster analysis to identify eight clusters of descriptors in 53 patients with shortness of breath, and reported that certain statements (descriptors) of uncomfortable awareness of breathing were more likely to be endorsed by patients with certain conditions. For example, patients with asthma selected "My breath does not go out all the way," "My breathing requires effort," and "My chest feels tight" as descriptors of their breathlessness; patients with COPD chose "My breathing requires effort" and "I feel a hunger for more air" to characterize their experience of breathlessness. Elliot and colleagues13 examined 45 descriptors of breathing discomfort in 169 patients and separated the descriptors into 12 groups (clusters) that appeared to describe different aspects of breathing discomfort. In a later study, evidence that patients with different cardiorespiratory conditions experience distinct qualities of breathlessness was gathered in a large sample of patients (n = 218) who sought care for difficulty in breathing.12 In a subset of those patients, descriptors selected by COPD patients (n = 16) on two occasions approximately 1 week apart were compared to determine the reliability of selections. The agreement for all descriptors was 79% (r = 0.82).
The perceived clinical value of uncovering features (structures) of breathing discomfort stems, to date, from cluster solutions obtained in patients asked to select from lists of descriptors those that describe their "uncomfortable awareness of breathing" with activities. The groups of descriptors that have emerged to represent different aspects of breathing discomfort have never been shown to be independent either of the presence of an underlying pathophysiology or of a specific medical condition. For example, the descriptor "I feel tight" is recurrently associated with asthma but only because patients with asthma—and not patients with other medical conditions—generally select "I feel tight" as a statement that applies to their "uncomfortable awareness of breathing." It is possible, however, that patients use phrases to describe their condition in ways that are not easily interpretable by nonpatients, and there is little reason to assume that the association among descriptors in patients is the same as the association among descriptors in healthy individuals. Differences in the way groups of individuals use words to describe their symptoms could have implications for inferring the health status of patients. For example, suppose patients were to use phrases such as "uncomfortable" and "in pain" interchangeably to describe their symptoms on separate occasions because they perceive no difference (ie, no "distance") between the terms. If health-care providers, on the other hand, were to picture a much larger conceptual distance between the two terms, they might incorrectly infer their patients’ health status.
In the present study, we tested the hypothesis that the descriptors of breathlessness represent distinct and separable constructs, and predicted that the use of descriptors of breathlessness by outpatients is the same as their use by healthy individuals. Healthy individuals rated the dissimilarity between pairs of descriptors used previously by patients to communicate the experience of dyspnea. Cluster outcomes obtained in healthy individuals were compared with those obtained previously in patients who complained of breathing discomfort. In addition, we used MDS techniques to analyze relationships among descriptors in terms of the dimensions of subject ratings. As a control condition, participants first judged the dissimilarity between pairs of animals.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Materials and Procedure
Each participant completed a survey in about 30 min at a single
setting. In the first part of the survey, participants judged the
dissimilarity between pairs of animals created by the combination of
six different animals (canary, chicken, dog, cat, cow, and sheep). This
condition was included to familiarize subjects with the task and to
make sure they were able to perform dissimilarity ratings. Before the
task, subjects were provided the following set of written instructions:
"In this experiment you will practice making judgments of the difference, or dissimilarity, between pairs of objects. Remember that we will ask you to judge not how similar two things seem, but how dissimilar, or different, two things seem. For each animal pair you are to judge how dissimilar two animals are by using a scale ranging from 0, which means no difference at all between pairs, to 10, which means very much difference between pairs. For each pair of animals you should write a number on the line that is provided. On each trial, therefore, you will be presented with a pair of animals. You should judge how dissimilar each pair is by making your response on the line that follows each animal pair."
After participants read the instructions, they rated the dissimilarity between 21 pairs of animals, the lower triangle of all possible pairs including the diagonal (ie, each animal paired with itself). The pairs were listed on a single sheet of paper. For each participant, both the sequence of all possible pairs and the order of animals within each pair were presented in random order.
Following this, participants judged the dissimilarity between pairs of descriptors of breathlessness created by the combination of 15 different descriptors (Table 1 ). The set of descriptors was the same as that examined previously in patients who complained of breathing discomfort.12 14 Before the task, participants were provided the following set of written instructions:
|
After they read the instructions, they rated the dissimilarity between 120 pairs of descriptors, the lower triangle of all possible pairs including the diagonal (ie, each descriptor paired with itself). The pairs were listed on consecutive sheets of paper. For each participant, both the sequence of all possible pairs and the order of descriptors within each pair were listed in random order.
After completion of the second judgment task, participants completed the third, and final, portion of the survey. Here, they first confirmed the absence of chronic disease and then provided their age, sex, and level of education. Next, they responded (yes or no) as to whether they experienced uncomfortable awareness of breathing with activities. Finally, participants rated their perceived level of understanding (knowledge) of the experience of shortness of breath in exercise, in asthma, in cystic fibrosis, and in congestive heart failure. For each condition, participants rated their "perceived familiarity" of shortness of breath as either "very familiar," "somewhat familiar," "somewhat unfamiliar," or "very unfamiliar."
Data Analysis
Mean dissimilarity judgments were computed for each unique pair
of terms (or phrases) for both the animal and descriptor tasks, and
were analyzed by cluster analysis and MDS. We performed these two
complementary analyses to assess the degree of association among pairs
of verbal stimuli. In primary analyses, mean dissimilarity judgments
were computed using data provided by all participants (n = 100). In
secondary analyses, mean judgments were computed for subgroups of
subjects, and solutions obtained for each complementary pair of
subgroups were compared. Subjects were divided into five dichotomous
groups based on age (< or 
30 years); sex (male or female);
randomization; level of education (
or > 12 years); or,
uncomfortable awareness of breathing with activities (yes or no).
Cluster Analysis:
Cluster analysis was used to group terms
or phrases according to their common properties. Cluster analysis is a
set of statistical tools used to organize a collection of objects or
experiences into fewer, meaningful categories.16
17
Like
other techniques designed to simplify data by identifying a small
number of underlying factors or dimensions, cluster analysis represents
a largely descriptive set of techniques that operates with simple
constraints: similar items are grouped together, and each item is
placed in only one cluster. In most procedures, items are grouped
together to minimize the variance within each cluster and to maximize
the variance between clusters. We used the hierarchical average-linkage
technique, which uses information about all pairs of distances to
provide a visual representation (a dendrogram) of the successive links
or connections among clusters made up of progressively increasing
numbers of descriptors. In average clustering, each member of a cluster
has a smaller average dissimilarity with other members of the same
cluster than with members of any other cluster.17
Use of
the average-linkage technique enabled direct comparisons with cluster
analyses performed with patient data.12
A technical problem with uncovering the underlying structure of proximity data involves the general lack of criteria for determining the optimum number of clusters (ie, a stopping rule17 ). Generally, interpretations consistent with recurring clusters, and those that follow knowledge of or theory about underlying stimulus attributes, are the most satisfactory.16 Our solutions were influenced by the primary aims of the study—to compare clusters and their associated descriptors in healthy individuals with those obtained in patients.12 14 In addition, we compared the pattern of associations between clusters and particular disease conditions through reanalysis of patient data collected in the project by Mahler et al.12
MDS:
In our application, the MDS method solves for
an optimal spatial configuration of verbal stimuli based on their rank
order of dissimilarity and treats mean dissimilarity ratings as
distances in a Euclidean space. The scaling procedure attempts to place
objects (words) in this space along a minimum number of interpretable
dimensions to best satisfy all the pair distances existing in the data
matrix. Each point in the space is associated with a fixed, scaled
location. We used Kruskal’s nonmetric monotonic algorithms to estimate
each solution’s goodness of fit, or "stress." A perfect fit among
the distance measures is assumed when stress equals zero, although in
most situations, a stress value of 0.05 indicates a good
fit.18
Acceptable two-dimensional (stress = 0.003) and
three-dimensional (stress = 0.046) solutions were obtained for the
animal and descriptor tasks, respectively, and are presented in the
following results section.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
21 = 0.02;
p > 0.05), and comparable numbers reported experiencing
uncomfortable awareness of breathing with activities
(21 = 1.06;
p > 0.05). On the other hand, more men than women in the sample had
earned college degrees
(21 = 8.04;
p < 0.05). Most participants (73%) were familiar with the
ex- perience of dyspnea in exercise, and most were unfamiliar
with dyspnea as occurs in asthma (82%), in cystic fibrosis (95%), or
in congestive heart failure (84%). Although the expressed familiarity
with dyspnea as occurs in disease in a few subjects was most likely
based on experiences with relatives, or was acquired in other similar
ways, subjects were not asked to provide the exact source of their
experiences.
|
|
|
|
Cluster Analysis:
The dendrogram resulting from the cluster analysis based on mean
dissimilarity ratings is presented in Figure 3
. In this analysis, the successive links between increasingly similar
descriptors, or sets of descriptors, are represented until all the
descriptors (branches) are combined at the bottom of the tree. The
horizontal dotted lines in Figure 3
represent the levels at which 8 and
10 groupings were formed through combinations of similar elements. The
clusters, and associated descriptors of breathlessness, obtained in
healthy individuals were compared with the 8 clusters reported in a
study of 53 patients by Simon et al14
and with the 10
clusters reported in a study of 218 patients by Mahler et
al.12
|
|
|
Comparisons between the relationships among clusters based on the dissimilarity ratings in the healthy individuals and the disease conditions represented among the 218 patients studied by Mahler et al12 are presented in Table 6 . Seventeen of the 18 associations between clusters and disease conditions were replicated. The exception between the two sets of outcomes is the lack of association between the cluster "Work" and deconditioning. Additionally, the cluster "Out of Breath" (which replaced the cluster "Suffocating" in the healthy individuals; Table 5 ) was associated with every condition except neuromuscular weakness; and the cluster "Air Hunger" (which along with "Suffocating" and "Shallow" was not associated with any condition in the original patient analyses) was associated with neuromuscular weakness.
|
|
|
30 years); sex (male or
female); randomization; level of education ( or > 12
years); or, uncomfortable awareness of breathing with activities (yes
or no) (Table 2)
.
Animal Ratings:
The cluster solutions obtained for each complementary pair of
subgroups were indistinguishable from one another. The results of the
two-dimensional MDS analyses for each of the 10 subgroups were
virtually identical (stress = 0.00 to 0.031; mean, 0.005). The
correlations between each complementary pair of subgroups for the
fixed, scaled locations of each item obtained in the two-dimensional
MDS solution ranged between 0.971 and 0.997 (p < 0.001).
Descriptor Ratings:
We compared the cluster solutions obtained in the complementary pairs
of groups when 10 groupings were formed through combinations of similar
elements, or groups of elements. The clusters, and associated
descriptors of breathlessness, were virtually identical between
subgroups and exactly the same as the solution presented for the entire
sample.
The results of the three-dimensional MDS analyses for each of the 10 subgroups were highly similar (stress = 0.047 to 0.068; mean, 0.059). The correlations between each complementary pair of subgroups for the fixed, scaled locations of each item obtained in the three-dimensional MDS solution ranged from 0.419 for levels of education and 0.442 for age (p < 0.05), to between 0.917 and 0.975 for sex, uncomfortable awareness of breathing with activities, and randomization (p < 0.001).
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cluster solutions and MDS outcomes in healthy individuals were largely independent of demographic variables such as age, sex, level of education, and the experience of uncomfortable awareness of breathing with activities. These latter results demonstrate that the conceptual distance between pairs of descriptors is independent of variation in common demographic variables. Such a conclusion would likely be strengthened by an assessment of the use of descriptors in a sample of individuals comprising a greater number of older (> 60 years) healthy individuals. In addition, our overall conclusions would likely have been further supported by formal assessment of lung function in the participants. Although from an epidemiologic perspective it would be highly unusual to see chronic obstructive lung disease, interstitial lung disease, or occult cardiac disorders in our participants, they were determined to be "healthy" only through self-report.
A limit to the interpretation of the similarity in the use of descriptors of breathlessness by healthy individuals and in the use of descriptors by patients with cardiopulmonary disease observed in previous studies relates to possible changes in knowledge about the symptom of dyspnea in the general population that may have occurred since the study by Simon et al.14 Although we have no specific data to address the concern, we believe it is unlikely that knowledge of dyspnea, per se, has changed significantly among the general public. For example, public education courses for cardiac disease have emphasized early recognition of cardiovascular symptoms and the variety of ways in which myocardial ischemia may become manifest (eg, basic and advanced cardiac life support). Recent efforts to increase public awareness about asthma have focused on use of peak flowmeters, control of environmental triggers for asthma, and use of anti-inflammatory agents. Long-standing educational materials for COPD have largely been geared toward smoking cessation programs.
Cluster Analysis
Our applications demonstrate that methodologic advances in the
study of the language of dyspnea mirror advances in the study of the
language of pain, in that healthy individuals rate the dissimilarity
among phrases in a manner comparable to that of patients. Such
applications may be used with increasing confidence to articulate the
subjective experience of breathing discomfort. Current data suggest
that between 8 and 10 categories of sensory experience serve to
adequately differentiate among various states of breathing discomfort.
Subsequent decisions to recommend an exact number of clusters to guide
clinical judgments will likely depend on the perceived importance of
the pattern and number of associations between specific clusters and
particular pathophysiologic conditions. Schwartzstein4
noted that current patterns of relationships between clusters and
patient conditions include evidence of multiplicity (each condition is
characterized by more than one cluster), uniqueness (each condition is
associated with a unique set of clusters), and sharing (some clusters
are associated with more than one condition). It is likely that work
will continue to elucidate the precise set of phrases and combinations
that best reveal patient experiences of breathing
discomfort.18
Future applications of clustering algorithms
to explore not only the sensory but also the cognitive-evaluative and
affective-motivational nature of descriptors used to characterize
breathing discomfort may affect the number of clusters in any final
solution.5
19
20
The likelihood that breathing discomfort varies along a limited number of factors or dimensions has theoretical and clinical implications.4 12 13 Information about different kinds of sensory experiences may contribute to our understanding of the mechanisms of dyspnea.12 21 The prospective use of descriptors of breathlessness may assist health-care providers in identifying or predicting a specific diagnosis, or in distinguishing among types of dyspnea in a patient with two or more diseases. Additionally, qualitative changes in the experience or intensity of shortness of breath may relate to the symptomatic benefit of a therapeutic intervention.7 22 On the other hand, although we reinforce confidence in the common understanding of the problem between patient and health-care provider, it is not yet clear the extent to which such common understanding may serve to improve patient outcomes.
MDS
We articulated three dimensions of experience of breathing
discomfort through application of MDS algorithms to dissimilarity
ratings. These results complement several decades of research involving
both patients and healthy individuals that have underscored the role of
abnormalities in both the mechanics of breathing and in the drive to
breathe in accounting for the experience of dyspnea. For example,
Comroe,2
in summing up his observations of a 1965
symposium, concluded that there may be as many as five or six types or
grades of breathlessness: awareness of increased ventilation, shortness
of breath, hindered breathing, suffocation, and the sensation at the
breaking point of breath-holding. Guz23
described four
different types of respiratory sensations: breath-holding sensations,
irritation of the tracheobronchial tree, obstructed breathing, and the
inability to get in enough air. Campbell and Guz1
concluded that among the elemental sensations of chest tightness or
irritation, excessive ventilation, excessive frequency of breathing,
and difficulty in the act of breathing, the latter experience was
consonant with classic dyspnea. The three dimensions of breathing
discomfort uncovered with MDS—"Depth and frequency of breathing,"
"Perceived need, or urge, to breathe," and "Difficulty breathing
and phase of respiration"—converge on the recurrent sensory
experiences emphasized by these and other researchers.4
More importantly, the MDS solution for the descriptors we used suggests
a modest total number of attributes (three) of uncomfortable awareness
of breathing.
Our interpretation of the first dimension uncovered through MDS—"Depth and frequency of breathing"—relates primarily to the most influential model for characterizing the basis of breathing discomfort. More than 30 years ago, Campbell and Howell24 suggested that the perception of breathlessness resulted from a perceived mismatch between the ventilation demanded and the ventilation achieved. In its early formulations, conscious perception of an imbalance between ventilatory demand and ventilatory supply—between the respiratory cost of breathing and the achieved benefit of ventilation—was envisaged in all instances that could give rise to the sensation of breathing difficulty. Another model of ventilatory inappropriateness is presented in work by Lougheed et al25 and O’Donnell and colleagues.26 They conclude that the perception of breathlessness in airflow limitation "encompasses the conscious awareness of disproportionate inspiratory effort for a given ventilatory output."
Our interpretation of the location of the descriptors along the first dimension is influenced by models of ventilatory inappropriateness and related abnormalities in the mechanics of breathing. We labeled the dimension "Depth and frequency of breathing" because ventilation is defined as the product of tidal volume and respiratory rate. Descriptors related to obstructed or restricted breathing (eg, "My chest is constricted") are located at one end of the continuum and reflect the greatest degree of perceived inappropriateness (the greatest cost for a given level of output). The perceived imbalance to breathing is presumed to be reduced near the middle of the continuum. In other words, the perceived cost of breathing is diminished when only one phase of breathing appears affected (ie, when "My breath does not go in all the way" or "My breath does not go out all the way"). Descriptors located at the other end of the continuum ("I feel that I am breathing more") are presumed to reflect a more favorable ratio, a perceived gain in output for any given cost of breathing.
We connect the second dimension, "Perceived need, or urge, to breathe," with the construct air hunger, which is evident in conditions that result in an increased drive to breathe caused by exercise, congestive heart failure, and pregnancy.4 Acute abnormalities in blood gas valves that are evident in many disease states also give rise to increases in the perceived need or urge to breathe. The sense of air hunger is not necessarily dependent on increases in ventilation but seems to result from involuntary increases in stimulation, as occurs, for example, during a prolonged breath hold.27 We interpret the arrangement of descriptors along dimension 2—ranging from "I feel a hunger for air" and "I cannot get enough air" to "My breathing is shallow" and "My breath does not go out all the way"—as reflecting variations in the perceived intensity of the need to breathe (from manifest to moderate). In this analysis, the perceived drive to breathe is hypothesized to represent a particular state of respiratory awareness. Variations in ventilatory state (dimension 2) are thought to interact with subsequent perceptions of ventilatory inappropriateness (dimension 1). The third dimension—"Difficulty breathing and phase of respiration"—suggests that qualitatively different sensory experiences of ventilatory inappropriateness (dimension 1) arise from difficulty breathing in as compared with difficulty breathing out (dimension 3).
Previous work suggests that there are unique sensory experiences arising from inspiratory difficulty compared with those arising from expiratory difficulty.12 25 26 A majority of patients report that breathlessness occurs not only while breathing in but also while breathing out.12 Patients with asthma, for example, complain mostly of inspiratory difficulty during acute bronchoconstriction, but breathlessness in asthma is described as involving both inspiratory and expiratory difficulty.25 In healthy individuals, "Airflow direction" (inspiration or expiration) was one dimension of experience that emerged in a previous MDS study28 designed to uncover the organization of sensory experiences elicited by various breathing maneuvers (eg, inspiring or expiring through both small and large resistances).
In the present study, the descriptors associated with "Inspiratory difficulty" ("My breathing is shallow" and "My breath does not go in all the way") are readily discernible, but some descriptors associated with "Expiratory difficulty" (eg, "My breathing requires work" and "My breathing requires effort") seem less clear (Fig 5) . We interpret the descriptors related to the increased work or effort of breathing as referring to increases in the intention or purposefulness of breathing out, which is normally accomplished through passive recoil of the lung. The pair of descriptors used previously by Lougheed et al25 and O’Donnell and colleagues26 —"Breathing in requires more effort" and "Breathing out requires more effort"—highlights the value of recognizing separate difficulties associated with inspiration and with expiration.
In summary, we have shown that the features or structures of breathing discomfort uncovered in patients are not simply dependent on the presence of underlying pathophysiology or on a specific disease condition. Our data support conclusions that breathlessness in each condition results from a different composite of sensations.4 Three attributes of breathing discomfort were revealed through application of MDS algorithms to dissimilarity ratings: disturbances in the mechanics of breathing; alterations in the drive to breathe; and difficulty breathing and phase of respiration. In conclusion, our results in healthy individuals support the contention that the association of different clusters with different disease states reflects distinct qualities of breathlessness and possibly different physiologic mechanisms underlying respiratory discomfort.
|
Footnotes |
|---|
Received for publication October 18, 1999. Accepted for publication April 19, 2000.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Han, Y. Zhu, S. Li, J. Zhang, X. Cheng, O. Van den Bergh, and K. P. Van de Woestijne The Language of Medically Unexplained Dyspnea Chest, April 1, 2008; 133(4): 961 - 968. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. von Leupoldt, S. Balewski, S. Petersen, K. Taube, S. Schubert-Heukeshoven, H. Magnussen, and B. Dahme Verbal Descriptors of Dyspnea in Patients With COPD at Different Intensity Levels of Dyspnea Chest, July 1, 2007; 132(1): 141 - 147. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. von Leupoldt and B. Dahme Cortical Substrates for the Perception of Dyspnea Chest, July 1, 2005; 128(1): 345 - 354. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Han, Y. Zhu, S. Li, X. Chen, C. Put, K. P. Van de Woestijne, and O. Van den Bergh Respiratory Complaints in Chinese: Cultural and Diagnostic Specificities Chest, June 1, 2005; 127(6): 1942 - 1951. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. A. Incalzi, A. Corsonello, C. Pedone, F. Corica, P. Carbonin, R. Bernabei, and on Behalf of the GIFA Investigators Construct Validity of Activities of Daily Living Scale: A Clue To Distinguish the Disabling Effects of COPD and Congestive Heart Failure Chest, March 1, 2005; 127(3): 830 - 838. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
