INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Terms such as "breathlessness," "difficulty breathing," or "dyspnea" are often used by clinicians but do not adequately describe the variety of breathing sensations experienced in different clinical conditions (e.g., asthma versus cardiac insufficiency). Respiratory afferent information arises from several different sources. It is reasonable to suppose that afferent information from different sources such as these should give rise to qualitatively different sensations. Recent evidence shows that subjects and patients can reliably discriminate different sensations evoked by different respiratory stimuli applied experimentally or resulting from disease and that they use different words to describe these sensations (1); however, there have been few attempts to quantify more than one respiratory sensation during experimental manipulations.

A sensation of severe air hunger can be elicited in normal subjects by modest increases in PCO₂ (10 mm Hg) if ventilation is held constant (5, 6). It is an unpleasant sensation that can become intolerable. Air hunger is undiminished by the absence of respiratory muscle activity (5, 7). It has been posited that air hunger is elicited by either direct discharge from chemoreceptors or from corollary discharge from respiratory motor activity in brainstem respiratory centers. Air hunger is relieved if tidal volumes are increased, as they would be by the normal reflex response to CO₂ (6, 8, 9).

Unpleasant sensations can also arise when greater than usual respiratory muscle activity (or efferent drive to respiratory muscles) is required to maintain ventilation; e.g., because of fatigue or increased resistance to breathing (2, 10, 11). The character of these sensations will include descriptors such as work and effort; the descriptors used may depend on the afferent source. Such sensations may arise from respiratory muscle afferents or from corollary discharge from the two sources of central commands to respiratory muscles (cortex and brainstem).

Demediuk and coworkers (12) have provided evidence that qualitatively different sensations can result from voluntary (cortical) versus reflex (brainstem) drives to breathe; consistent with the possibility that corollary discharge from medullary respiratory motor centers gives rise to air hunger (13), whereas corollary discharge from cortical motor centers gives rise to a sense of breathing effort (10, 14, but see also 15).

Many experiments have quantified the response of a single respiratory sensation to an experimental manipulation. Often the experiment alters information in several afferent sources but requires subjects to rate a single broadly defined sensation such as "difficulty breathing." For instance, Chonan and coworkers found that subjects rated substantially greater "difficulty in breathing" when forced to breathe above their spontaneous ventilatory level or when forced to breathe below it (16); intuitively, these seem like quite different tasks. Although these investigators treated the ratings as arising from a single sensation, which they termed "dyspnea," their subjects remarked that low ventilation felt different than high ventilation. We suggest that the subjects in such studies actually use the single rating scale to report two distinct sensations: "air hunger," the feeling of a need to breathe more, of not getting enough air, and "effort or work," the perception of how hard it is to breathe, the exertion required to breathe. Unfortunately, except for one study (12), there have been no investigations in which subjects have been asked to report both kinds of sensation as PCO₂ and ventilation were varied independently.

In the present study, we have separately manipulated two independent stimulus variables, end-tidal carbon dioxide pressure (PET_CO2) and minute ventilation (E), to determine whether the sensation of air hunger is distinct from the sensation of work/effort. We found that the stimulus-response slopes of the two sensations changed their relationship depending on stimulus condition. Our approach did not require the assumption that subjects use the rating scales for air hunger and work/ effort in the same way (that is, we did not assume that a change in rating from slight to moderate is comparable for the two sensations).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

We studied five subjects, three men and two women, ages 21 to 35; each underwent five experimental sessions. (One prospective subject was dropped from the study because of a hypertensive response to the experimental situation; another did not wish to continue after the first session.) The remaining five all were healthy with no history of respiratory disease, except for one subject with mild asthma controlled by occasional use of a -agonist (albuterol). None of the subjects was familiar with the ideas being tested in the study, and none had participated in previous studies of respiratory sensations. These experiments were approved by the Harvard School of Public Health Committee on Human Subjects; informed consent was obtained from each subject.

Experimental Setup

Subjects breathed through a circuit with variable reinspired gas arranged to maintain a constant PCO₂ despite changes in ventilation (17). This system allowed us to set a desired PCO₂ by changing the percentage of inspired CO₂ in "fresh" gas. It would then remain constant, regardless of changes in ventilation. The inspired gas was humidified (Puritan Bennett Cascade 1; Los Angeles, CA). Two gas tanks, one of 50% O₂/50% N₂ and one of 10% CO₂/40% N₂/50% O₂ provided gas flow into the circuit. Regulation of gas flow into the system was accomplished by a flow meter (Air Co. Oxygen Flow Meter) and an air- oxygen mixer (Bennett Air-Oxygen Mixer).

Measurements and Recording

Air flow at the mouth was measured by a pneumotachometer (Fleisch No. 2; Switzerland) and pressure transducer (2 cm H₂O MP45; Validyne, Northridge, CA). Airway pressure (56 cm H₂O MP45; Validyne) and PET_CO2 (Cardiocap II CG-2GS; Datex, Tewkesbury, MA) were derived from fine 1.5-mm catheters inserted into the mouthpiece. Magnetometer coils were employed to obtain volume measures (18). Tidal volume (VT), maximal inspiratory volume, and end-expiratory volume were obtained from the summed output of the magnetometer coils placed in the midline on the rib cage at the lower part of the sternum and on the abdomen slightly above the umbilicus, with paired coils placed at the same levels on the back. To prevent shifting of the magnetometer coils by contact with the chair during vigorous breathing efforts, a foam pad with holes to accommodate the coils was strapped to the subject's back with compliant self-adhesive wrap (Coban 1586; 3M Health Care, St. Paul, MN).We implemented a ratio of 4:1 for the relative volume implication of the rib cage and abdomen diameter change (19). We converted the summed magnetometer voltage to volume by comparing it to the integrated pneumotachometer flow signal over a range of VT. This calibration was accepted only if the correlation index (r²) was 0.80 or above.

Blood oxygen saturation and blood pressure were measured noninvasively at regular 3-min intervals (Datex Cardiocap II CG-2GS). The subject indicated ratings of air hunger and work/effort on a 7-point scale (see PERCEPTUAL MEASURES OF AIR HUNGER AND WORK OR EFFORT) by pushing buttons which registered an electrical signal. All signals were continuously recorded and digitized (Dataq Di-220 PGH/PGL; Akron, OH).

Voluntary breathing targets. There were two kinds of voluntary breathing targets that the subjects were asked to match at different times during the experiments. They were displayed on an oscilloscope viewed by the subject. One, the "VT-f target," required the subject to match a selected VT, frequency (f), and end-expiratory volume. The subject's VT, derived from the summed signal of the rib cage and abdomen signals, was displayed on the oscilloscope, and the subject breathed so as to make the volume signal move up and down between two horizontal target lines inscribed 3 inches apart on the face of the oscilloscope. Manipulation of the subject's voluntary VT was accomplished by surreptitiously changing the gain of the displayed volume signal (20). The subject's frequency of breathing was controlled by a metronome: a crescendo tone sounded throughout the inspiratory time (TI), a silent period defined the expiratory time (TE). TI and TE were equal. The tone was presented through earphones, which also minimized external noise and interception of experimental cues.

The other breathing target used, E target, required the subject to maintain E comparable to those required by the VT-f target but permitted the subject to select his or her own ventilatory pattern of VT, f, and end-expiratory volume. The subject's ventilation, registered as a full-wave rectified and filtered signal of air flow, was displayed as a line on the oscilloscope. The subject was asked to keep this line at the same level as a target line drawn on the screen. E was manipulated by changing the gain of the displayed ventilation signal.

Perceptual Measures of Air Hunger and Work or Effort

Initial instructions to the subject. Before each experimental session, we used the following script to introduce subjects to the concepts of air hunger and work/effort:

We will be asking you to rate the intensity of your breathing sensations. In particular, we want you to notice two different kinds of sensation related to breathing: the sense of air hunger and the sense of work and effort of breathing. We define the sense of air hunger as the discomfort caused by your urge to breathe, a feeling of being starved for air. Have you ever held your breath for a long time? How did that feel? (Similarities were discussed.) You no doubt have used the terms work, effort, and force to describe sensations you experience when performing various tasks with your arms and legs; we are simply asking you to apply those terms in the same way to the use of your breathing muscles. Can you give me some examples of situations in which you might exert work and effort with your arms or legs? (These were then discussed.). . . Sensations of work/effort are not necessarily uncomfortablefor instance, some people actually enjoy exercising very hard, while others find it unpleasant. This is in contrast to air hunger, which everyone finds unpleasant. . . The various sensations may come and go throughout the experiment. The different sensations may change together at times and go in opposite directions at other times. It is possible that you will not feel a sensation throughout the experiment.

Ratings of air hunger and work or effort. At 30-s intervals, subjects were asked to rate the intensity of their sensations of air hunger and work or effort using a 7-point ordinal scale:

"None"felt none of the requested sensation.

"None plus"

"Slight"sure that I felt the sensation but the feeling is not very strong or unpleasant and I could tolerate it for a very long time.

"Slight plus"

"Moderately strong"felt a strong level of the requested sensation, but could continue at this level for several minutes.

"Moderately strong plus"

"Extreme"The requested respiratory sensation has reached a maximum level. (In the case of air hunger, this would be intolerable.) If this is selected, we will immediately make a change to make you more comfortable; it may take two or more breaths to take effect. You can select this any time you need to.

Experimental Protocol

Practice session (1 d). On the first experimental day, subjects practiced controlling their ventilation using the targets and using the rating system. Definitions of air hunger and work/effort were introduced.

Experimental sessions (4 d). Using a script, an investigator reviewed definitions of air hunger and work/effort and the use of the rating scale. The physiologic sensors were applied, and the subject was seated in a semireclining position. While the subject was breathing quietly, PET_CO2 was sampled at the nostril, and rib cage and abdomen magnetometer output were measured to establish FRC. The subject then performed several inspiratory capacity maneuvers to establish a maximal volume reference point against which to set the VT targets.

The order in which targeted breathing, free breathing, and debriefing occurred was the same on each experimental day. The subjects began rating sensations during several minutes of free breathing, after which we instructed the subject to breathe to the target. At the end of each target breathing period, we turned the display off and allowed the subject to breathe freely for 2 to 3 min. After free breathing, the subject was shown a list of descriptors and asked to select those that best applied to his or her sensations in the preceding targeted breathing period. Each day consisted of four different target breathing periods. Each was 5 min long except on occasions when target breathing was below the spontaneous hypercapnic breathing, and subjects were too uncomfortable to continue for the full 5 min. At the end of the day's session, the subject was questioned in more detail about the entire experience.

Experimental Conditions

There were nine experimental conditions (Table 1). They consisted of various combinations of PET_CO2 levels and ventilation levels. In any one condition, either ventilation level was held constant while PET_CO2 was changed, or PET_CO2 was held constant while ventilation was changed. Each day four of these nine conditions were presented in the four periods of targeted breathing, with the order varied over four experimental days.

To account for size difference among subjects, the targets were set to be proportional to the subject's inspiratory capacity (IC). In this group of subjects, the average IC was 3.2 L (range 1.8 to 4 L). For the VT  f target, volume-frequency combinations were: baseline (VT ~ 0.3 IC, f = 22 breaths · min¹); VT = 0.45, f = 22; and VT = 0.8, f = 30. The E target was set to match the VT  f minute ventilation, thus yielding target ventilations of 7, 10, and 24 IC · min¹. The actual achieved values for ventilation and PET_CO2 are given in Table 1.

Data Analysis

An example of the recordings that provided the primary data is shown in Figure 1. Each subject's ratings of air hunger and work/effort were related separately to the E and to the PET_CO2 level. Physiologic data were processed to account for the time lag in sensory effects known to occur after a change in PET_CO2 or a change in ventilation (21, 22). Because the dynamic models used in this processing are likely to be imperfect, we also excluded sensory ratings immediately after intentional step changes in ventilation or PET_CO2 (60 s after a PET_CO2 change of 3 mm Hg or more, and for 45 s after a VT change). We also excluded data during and after substantial inadvertent departures from the required target E or PET_CO2. Ratings were excluded when their associated PET_CO2 departed more than 4 mm Hg from the average level for that subject and session, or when their associated E departed more than 15% from the average. Examples of these exclusions are shown in Figure 1.

View larger version (82K): [in this window] [in a new window]   Figure 1.   A sample recording of ventilation, PETCO2, and ratings of air hunger (crosses) and work/effort (circles) when PCO2 was increased with E held constant, followed by an increase in E with PCO2 held constant. Data were selected for analysis during targeted breathing (white areas) and free breathing (crosshatched areas). Data were excluded (striped areas) when E or PCO2 was intentionally changed or when there were inadvertent departures from the required target. Raw recordings (R tracings) were processed (P tracings) to yield time-averaged VT and PETCO2 data with which ratings could be paired in time. Major steps on the 7-point rating scale were "none" (0), "slight" (S), "moderately strong" (M), and "extreme" (E).

The difference between the sensations of air hunger and work/effort in response to the various stimulus conditions was apparent from inspection of the ratings. To test this statistically, it was necessary to transform the changes in ordinal ratings to an appropriate scale. A nonparametric test on the resulting values took into account variations among individual subjects, stimulus conditions, and experimental days (the procedure is described subsequently). Allowing subjects to choose any combination of VT and f (E target) did not affect sensations of air hunger or work and effort compared with the same E at fixed VT and f (VT  f target). Difference in target type was taken into account only to test whether allowing voluntary drive to follow the intrinsic rhythm reduced perception of work/effort. Because both work/effort and air hunger ratings were unaffected by target type, data from the two targets were merged for other analyses.

To test the hypothesis "the sensation of air hunger is distinct from the sensation of work/effort," we determined whether the two sensations (air hunger and work/effort) differed in their response to the two stimulus variables (PET_CO2 and E). We compared a "slope measure" of these ratings when each type of rating was regressed on changes in E or changes in PET_CO2. For this analysis, we compared the behavior of the "slopes" across all conditions and all experimental sessions. Because of the small number of subjects, the repetition of the experimental conditions, the bivariate responses (air hunger and work/effort ratings), and the ordinal nature of the rating scale, a three-step approach was used in this analysis:

First, each ordinal rating response was rescaled by subtracting from it the sample mean of all subjects' ratings and dividing by the sample standard deviation: each response score is therefore expressed as the number of standard deviations from the sample mean. This results in air hunger and work/effort being scaled in the same manner.

Second, slope estimates for air hunger versus PET_CO2 and E, and for work/effort versus PET_CO2 and E were obtained in two independent ways: (1) Proxy slopes were computed from standardized ratings, pooled in an ordinary regression; (2) A "mixed effects model" regression (23) was used. This model takes into account the repeated nature of the data, which makes the standardized ratings statistically dependent within subject, condition, and day. It also takes into account the effect of response variables (ratings) of a mixture of fixed effects (independent variables fixed by the experimenters) condition and random effects (variations in the way individuals responded to the experimental conditions).

Third, a Wilcoxon signed rank test (24) was used to determine whether the median difference between the air hunger and work/effort slopes was zero; i.e., this was a paired comparison. The proxy slopes obtained in (2) were similar to those in (1). Because there is no standard statistical analysis to compare the effect of experimental conditions on these proxy measures, we employed the nonparametric Wilcoxon test.

To infer which afferent sources might give rise to the two distinct sensory ratings, we observed how ratings of air hunger and work/effort were affected by selected stimulus conditions. Similar to the statistical approach described previously, the first two steps of the analysis obtained appropriate measures of the change in air hunger and work/effort responses ratings versus selected stimulus conditions. This was done for each individual. Then an exact Wilcoxon signed rank test was used to test if the average change in rating across individuals was significantly different from zero. These rating changes were expressed as slopes, as described previously.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Independent Manipulation of Sensations of Work/Effort and Air Hunger

We tested the difference in air hunger and work/effort ratings across conditions in which only PCO₂ was changed compared with conditions in which only ventilation was changed. When only PCO₂ was varied, the air hunger slope was on average greater than the work/effort slope; when only ventilation was varied, the work/effort slope was greater than the air hunger slope (exact Wilcoxon p = 0.0625). Figure 2 shows these slope differences for individual subjects. An example of differences in slope among subjects is shown in Figure 3: with PET_CO2 constant at 41 mm Hg, and ventilation increased for 7 to 20 IC/min.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Slopes derived from measured changes in ratings of air hunger and work/effort when ventilation was changed with PETCO2 held constant (left panel ), and when PETCO2 was changed with ventilation held constant (right panel ). The slopes for air hunger and work/effort were affected differently by the two kinds of stimulus change. The lines connect slope values for individual subjects: Subject 2, square; Subject 3, diamond; Subject 4, circle; Subject 6, triangle; Subject 7, square and cross.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Individual differences in sensation as ventilation was increased with PETCO2 held constant at 41 mm Hg, condition VI. The average ventilation values which evoked ratings of work/effort (circles) and air hunger (crosses) are plotted for each subject. Work/effort ratings increased for all subjects, but the changes in air hunger ratings were variable.

Consistent with the behavior of the ratings, subjects also distinguished a clear qualitative difference in the sensations of air hunger and work/effort and used quite different language in describing the two sensations when debriefed after each stimulus period and after each day's session. When debriefed after each experimental session, subjects reported experiencing a difference in the quality of the sensations they rated as air hunger and those rated as work/effort in 15 of the 20 sessions. When asked to describe in their own words what they felt, the language they used to describe air hunger was very different from the language used to describe work/effort (Table 2). The sensations also differed in discomfort: four of the five subjects did not find that sensations of work/effort were unpleasant, whereas all found air hunger to be unpleasant. After each targeted breathing period in an experimental session, subjects selected the descriptors that best described their sensations of air hunger and work/effort. As seen in Figure 4, the profile of these selections was quite different for the two sensations.

View larger version (34K): [in this window] [in a new window]   Figure 4.   The descriptors subjects selected which best described their sensations of air hunger (upper panel ) and work/effort (lower panel ). Shown is the percentage of time each descriptor was selected in the 16 debriefings held after the experimental periods. The selection profiles were quite different for air hunger and work/effort regardless of whether PETCO2 or E was changed.

Relationship of Work/Effort Sensations to E during Voluntary Hyperpnea

All five subjects increased their rating of work/effort as E increased with PET_CO2 constant (conditions VI through IX). The slope of these ratings was positive and significantly different from zero (exact Wilcoxon p = 0.0625). On average, when ventilation rose from 8 to 23 IC · min¹ the rating of work/ effort sensation rose from "slight" to "moderately strong."

Four of the subjects said that in some sessions their ratings of both air hunger and work/effort included difficulty of keeping on the target. Several subjects said they experienced sensations of work and effort during hypercapnia when they had to suppress their breathing below the spontaneous level.

Relationship of Work/Effort Sensation to PET_CO2 at Constant Voluntary Ventilation

The increase in PET_CO2 from 41 to 52 mm Hg at constant ventilation did not change subjects' ratings of work/effort (tested at both 15 and 22 IC/min). There was a slight increase in work/effort ratings (see positive slope, Figure 2), which did not differ significantly from 0 (exact Wilcoxon p = 0.125). This finding is contrary to the hypothesis that medullary drive can supplement cortical motor commands, thus reducing corollary discharge giving rise to effort (12). This analysis included both the E and VT  f targets. Because the E target would permit the depth and timing of breaths to match automatic medullary drive more closely, automatic motor commands might have been able to exert a greater influence with the E. A separate analysis of the data for the two types of target showed that this did not occur: the slope of the rating change for work/ effort did not differ from zero for either the VT  f or the E target (exact Wilcoxon p = 0.3125 and 0.1875, respectively).

Relationship of Air Hunger to Chemical Drive to Breathe and Achieved Ventilation

When PET_CO2 was varied across the range of 41 to 52 mm Hg with E held constant (conditions I through V), air hunger ratings increased slightly. The slope of the ratings was positive and significantly different from zero (exact Wilcoxon p = 0.0625). The level of voluntary targeted breathing during hypercapnia was 15 or 22 IC · min¹, and thus always exceeded spontaneous ventilation (which was 7.5 IC · min¹ at PET_CO2 = 41 mm Hg and 14 IC · min¹ at PET_CO2 = 52 mm Hg).

In two periods (conditions VIII and IX), we lowered target ventilation gradually until subjects rated "moderately strong" or "extreme" air hunger. When PET_CO2 was constant at 52 mm Hg and targeted ventilation was reduced from 22 to 9 IC/min (below the free breathing level of 14 IC · min¹), air hunger ratings increased from "slight" to "moderately strong" (exact Wilcoxon p = 0.0265). As ventilation fell below the free breathing level, subjects found it difficult to remain on target even with constant coaching from the experimenters.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results show that work/effort and air hunger are distinct sensations that can occur together or in isolation. When subjects were exposed to a wide range of ventilatory and PCO₂ stimuli, sensations of work/effort were most consistently associated with changes in ventilation and those of air hunger with changes in PCO₂. Each of the two sensations has been individually evoked and measured in other studies which did not require their simultaneous rating under the same stimulus condition (2, 5, 7, 10, 11, 25). Our results are the first to demonstrate directly that these two sensations are independent, a conclusion implied by the results of earlier studies.

Work and Effort

The increase in work/effort ratings when only ventilation was changed suggests that the sensation of work/effort arises from respiratory muscle mechanoreceptors or cortical motor drive to respiratory muscles, or both. A caveat with regard to our results is that three subjects (3, 4, and 6) reported a progressive increase of work/effort when target ventilation was reduced below spontaneous level. During debriefing, these subjects invariably described the "work/effort" sensation at very low ventilation as a "mental effort" or "concentration" needed to restrain their ventilation in the face of strong air hunger; they noted this was very different from the quality of work/effort when breathing hard.

DeVito and coworkers (25) also found that subjects reported an increase in sense of effort when PCO₂ was raised with breathing volume and frequency held constant. We also found a slight increase in work/effort under these conditions. Contrary to these findings, Demediuk and coworkers (12) found the "effort" of voluntary hyperventilation component to be diminished by elevated PCO₂. They speculated that in this circumstance reflex hyperpnea might contribute to the total ventilatory drive and thus reduce the volitional component and sense of effort. It is possible that in order for the respiratory centers to "help" in this manner, the respiratory rate must be set by the medullary pacemaker (R. Schwartzstein, personal communication). We think this is unlikely to explain the difference between our findings and theirs because we saw no difference in ratings when subjects were allowed to breathe at any rate they chose (E condition), compared with when the rate was constrained by the metronome (VT  f condition). The rating instructions given to the subjects might have differed in the two studies: Demediuk and coworkers narrowly defined the term "effort" as a willful act, whereas our definition of work/effort was somewhat broader, allowing subjects to include sensations of motion and muscle contraction as well as sensations of exerted effort deriving from corollary discharge. Thus, the findings of both studies may be true, and it is left to the reader to decide which definition best suits further experimental or clinical problems.

Air Hunger

The finding that air hunger ratings were positively correlated with PET_CO2 at constant ventilation is consistent with the proposal that this sensation arises as a result of chemoreceptor afferent activity or medullary corollary discharge (5, 7, 13, 26). Although air hunger can be very strong (6, 27) under these circumstances, the effect was small and variable in the present experiments; we attribute the present result to ventilation being well above the level of spontaneous breathing for the level of PET_CO2 imposed.

In our experiment, restraining ventilation below spontaneous levels produced air hunger, as would be expected if this sensation arises from an imbalance between chemical drive to breathe and achieved ventilation. It is well known that air hunger is relieved as VT, f, or E is increased (8, 9, 30). Our results and those of others show that, although this effect is powerful, breathing at high levels may not completely eliminate air hunger at high PET_CO2. For example, when subjects breathe freely during hypercapnia, they nonetheless report air hunger (31) although the sensation is much less than when breathing is constrained by mechanical ventilation (6).

It is complicated to evaluate whether air hunger is truly independent of voluntary drive at voluntary ventilation above spontaneous level. Most of the present constant PCO₂ studies were performed at higher voluntary ventilation where we predict air hunger will be largely independent of the level of ventilation. There was a small, nonsignificant positive correlation of air hunger with voluntary ventilation. Based on information obtained in debriefings, we attribute this to a cognitive association rather than to afferent mechanisms: As one subject put it in debriefing: "I figured that if I was breathing more, I must need to breathe more" (despite the fact that his breathing increased entirely as a result of the voluntary drive needed to meet the target). The cognitive association was no doubt developed over years of experiencing a coincident rise in urge to breathe and rise in ventilation during circumstances such as exercise. Again, although it is clear that subjects can rate at least two dimensions of respiratory discomfort, many subjects have a difficult time making a clear-cut distinction.

Multiple Dimensions of Dyspnea

Previous studies have examined relationships between E, PCO₂, and respiratory sensation. The results of these studies have been in some disagreement, perhaps because most have examined only one sensory dimension, and perhaps also because of differences in what subjects are instructed to rate. The present results allow us to propose a unified explanation of the previous findings.

Our results show that targeted voluntary breathing below the spontaneous hypercapnic level elicits air hunger and that breathing above it elicits work/effort sensation. This appears to disagree with those who have found that "breathlessness" (32) or "difficulty of breathing" (16, 33, 34) increase when subjects breathe both above and below the spontaneous level. The U-shaped curve derived from these studies relating "difficulty of breathing" to ventilation, with the trough centered at the spontaneous level (16, 32) probably describes the effects of two different sensations: one limb of the curve describes air hunger and the other work/effort. Schwartzstein and coworkers (32) and Chonan and coworkers (34) both acknowledge that these different sensations may have been evoked when subjects breathed above or below their spontaneous level. When provided with an insufficient number of rating dimensions, subjects may choose to lump different sensations into one rating. A study by Adams and coworkers (31) of subjects breathing at or above their spontaneous level demonstrates the merits of more precisely defining the sensation to be rated. They instructed their subjects to rate a specific sensation, "an uncomfortable need to breathe" and to ignore sensations of how much they were breathing. Discomfort was rated high during hypercapnic spontaneous breathing, but low during normocapnia with subjects copying the hypercapnic breathing level: instructions that permitted a rating of either "need to breathe" or "work and effort" may have yielded high ratings under both conditions.

Clinical Relevance

Patients are faced with various combinations of air hunger and work and effort of breathing (as well as other afferent input not mentioned here). In many cases these discomforts may arise from increased drive to breathe, from weak respiratory muscles, or from increased lung impedance. The result of this tradeoff no doubt varies with individual patients and with duration of disease. We have previously shown, for instance, that the air hunger evoked by hypercapnia adapts in a matter of days (35), whereas the sense of work may be subject to a very different adaptive time course. Therefore, a dyspneic patient might experience air hunger, or work and effort, or both, depending on the pathophysiologic changes and on the ventilatory strategy adopted. Asking the patient about the different dimensions of his or her dyspnea might therefore yield useful information.