Stan Grant, PhD; Tom Aitchison, BSc; Esther Henderson, BSc; Jim Christie, BSc; Sharam Zare, PhD; John McMurray, MD; and Henry Dargie, MD

Transcription

1 A Comparison of the Reproducibility and the Sensitivity to Change of Visual Analogue Scales, Borg Scales, and Likert Scales in Normal Subjects During Submaximal Exercise* Stan Grant, PhD; Tom Aitchison, BSc; Esther Henderson, BSc; Jim Christie, BSc; Sharam Zare, PhD; John McMurray, MD; and Henry Dargie, MD Objective: To assess which subjective scale, the visual analogue scale (VAS), the Borg CR10 (Borg) scale, or the Likert scale (LS), if any, is decidedly more reproducible and sensitive to change in the assessment of symptoms. Design: Prospective clinical study. Setting: Exercise laboratory. Participants: Twenty-three physically active male subjects (mean SD age of 30 4 years old) were recruited. Intervention: Each subject attended the exercise laboratory on four occasions at intervals of 1 week. Three subjective scales were used: (1) the VAS (continuous scale); (2) the Borg scale (12 fixed points); and (3) the Likert scale (LS; 5 fixed points). Four identical submaximal tests were given (2 min at 60% maximum oxygen uptake [V O2 max] and 6 min at 70% V O2 max). Two tests were undertaken to assess the reproducibility of scores that were obtained with each subjective scale. Two other tests were undertaken to assess the sensitivity of each scale to a change in symptom perception: a double-blind treatment with propranolol, 80 mg, (ie, active therapy; to increase the sensation of breathlessness and general fatigue during exercise) or matching placebo. The subjective scale scores were measured at 1 min 30 s, 5 min 30 s, and 7 min 15 s of exercise. Reproducibility was defined as the proportion of total variance (ie, between-subject plus within-subject variance) explained by the between-subject variance given as a percentage. Sensitivity was defined as the effect of the active drug therapy over the variation within subjects. Results: Overall, the VAS performed best in terms of reproducibility for breathlessness and general fatigue, with reproducibility coefficients as high as 78%. For sensitivity, the VAS was best for breathlessness (ratio, 2.7) and the Borg scale was most sensitive for general fatigue (ratio, 3.0). The relationships between the respective psychological and physiologic variables were reasonably stable throughout the testing procedure, with overall typical correlations of 0.73 to 0.82 Conclusion: This study suggests that subjective scales can reproducibly measure symptoms during steady-state exercise and can detect the effect of a drug intervention. The VAS and Borg scales appear to be the best subjective scales for this purpose. (CHEST 1999; 116: ) Key words: breathlessness; exercise testing; general fatigue; reproducibility; sensitivity; subjective scales Abbreviations: Borg Borg CR10; CHF chronic heart failure; LS Likert scale; R1 Reproducibility One; R2 Reproducibility Two; VAS visual analogue scale; V co 2 carbon dioxide output; V e minute ventilation; V o 2 oxygen uptake; V o 2 max maximum oxygen uptake; Vt tidal volume While a maximal exercise test may determine the physical work capacity of a chronic heart failure (CHF) patient and a range of other patient groups, a maximal test has limitations. Maximal effort can only be sustained for very short periods of time, and *From the Centre for Exercise Science and Medicine, Institute of Biomedical and Life Sciences (Dr. Grant, Ms. Henderson, and Mr. Christie), and the Department of Statistics (Dr. Zare and Mr. Aitchison), University of Glasgow; and the Department of Cardiology (Drs. McMurray and Dargie), Western Infirmary, Glasgow, UK. Manuscript received September 23, 1997; revision accepted June 8, Correspondence to: S. Grant, PhD, Institute of Biomedical and Life Sciences, University of Glasgow, 64 Oakfield Ave, Glasgow, G12 8LT, UK; S.Grant@bio.gla.ac.uk patient populations dislike exercising at or near maximum. Maximal tests are unlikely to relate to daily activities, whereas a submaximal test that simulates normal everyday life may be of value in monitoring changes in patient status. CHF patients experience symptoms while they undertake ordinary everyday tasks. The most common symptoms experienced by CHF patients are breathlessness and general fatigue. A submaximal test that monitors the subjective feelings of the subjects could be helpful with the assessment of patient symptoms. One of the major problems is selecting the most appropriate scale to monitor changes in patient status. Several subjective scales are available to monitor the feelings 1208 Clinical Investigations

2 of breathlessness and general fatigue. Some of these scales are as follows: (1) the visual analogue scale (VAS), a horizontal line with two anchor points, one at each extreme; (2) the modified Borg CR10 (Borg) scale, which has 12 points, 10 of which have verbal descriptors; and (3) a 5-point Likert scale (LS), with verbal descriptors at each number. A comparison of the responses to these three scales would help determine which scale, if any, is more reproducible or more sensitive to change than the others. No study has compared the VAS, the Borg scale, and the LS. There is also limited information on subjective scales during steady-state exercise. In addition, little research has been done to determine the ability of the above scales to detect changes in status. The monitoring of subjective scales and physiologic variables provides information on the relationship between the variables. It may be expected that any differences in the physiologic stimuli would result in changes in the perception of breathlessness and general fatigue. A linkage of the perception of symptoms to physiologic variables may shed light on how symptoms are provoked. In addition, the establishment of an association between physiologic variables and subjective scales may provide an understanding of the mechanisms by which pharmacologic intervention can alleviate symptoms. If a cause and effect for symptoms could be identified, it may be possible to target a particular area in the treatment of patients. The aims of this study are as follows: (1) to assess which subject scale, the VAS, the Borg scale, or the LS, if any, is decidedly more reproducible and sensitive to change in the assessment of symptoms; (2) to determine the magnitude of the visit and therapy effects of the subjective scales and physiologic variables; and (3) to compare the subjective scales with the physiologic responses to exercise. Normal subjects were chosen for this study because it was considered that they would tolerate the test protocol better than patients and would provide information that could be applied to a patient population. In some instances, it is also difficult or undesirable to change transiently the status of some patient groups. It was hoped that submaximal exercise testing with a reproducible rating scale that was sensitive to change would provide an improved method of assessing therapeutic interventions in patients with CHF. In order to test the sensitivity of the scales, it is necessary to disrupt the normal status of the subjects so that the capability of the scales to detect change can be assessed. It was decided to use -blockade to elicit a change in the perception of breathlessness and general fatigue. Subjects Materials and Methods Twenty-three healthy male volunteers (mean SD age of 30 4 years old) gave informed consent to participate in the study. The subjects were recruited by advertising in the local university and hospital. All of the subjects were physically active; the group included runners, hill-walkers, and soccer players. The subjects were recreational exercisers who exercised, on average, two to three times per week. Twenty-four subjects were recruited, but 1 subject became ill during testing and was unable to complete the study. The study received approval from the local ethical committee. All of the subjects completed a medical history questionnaire, underwent a medical examination, and signed a consent form before entry to the study. Research Design Preliminary Tests: The subjects reported to the laboratory at the same time of day. A submaximal economy test (running economy, defined as the steady state oxygen uptake for a particular running velocity 1 ) and a maximum test were carried out so that relative intensities could be determined. A submaximal economy profile was established for each subject using at least three 4-min periods of treadmill exercise. After a 30-min rest, each subject then undertook a symptom-limited maximal treadmill test to measure maximum oxygen uptake (V o 2 max). Depending on submaximal performance, either a Lamb Normal Subject Treadmill Protocol or a Lamb Athlete Treadmill Protocol 2 was used for this test so that treadmill time was around 10 min. This time period has been shown to be optimal for eliciting V o 2 max values. 3 A number of criteria 4 were used to define the quality of a maximum test. A plateau in oxygen uptake (V o 2 ) was defined as an increase of not 2.1 ml/kg/min or an increase of not 5% in energy cost with an increase in workload; 15 of 23 subjects demonstrated a plateau in V o 2 based on the these criteria. All of the subjects attained their age-predicted maximum heart rate (ie, 220 beats/min minus age). All of the subjects had a respiratory exchange ratio 1.05, and 17 subjects had a respiratory exchange ratio Submaximal Tests: The results of the symptom-limited maximal test and the submaximal economy profile were used to devise a standardized submaximal exercise protocol for each individual. This consisted of an initial work rate of 60% V o 2 max that increased after 2 min to 70% V o 2 max, a rate that was maintained for an additional 6 min. Previous pilot work and publications by this group 5 have shown that this relative intensity is appropriate to evaluate symptom scales. Exercise testing at only moderate intensities (described by the authors 6 as exercise with a heart rate 150 beats/min) has shown that propranolol had no effect on the rating of perceived exertion. Two minutes at 60% V o 2 max provides a warm-up, and the remaining 6 min at around 70% V o 2 max is of sufficient intensity to elicit symptoms. Intensities much 70% V o 2 max may produce marked discomfort to subjects undergoing -blockade. Experimental Protocol Visits and Treatments: Each subject attended the exercise laboratory on four occasions at intervals of 1 week. The order of visits was randomized according to a Latin rectangle design. The visits were termed Reproducibility One (R1), Reproducibility Two (R2), Placebo, and Active (propranolol treatment). On CHEST / 116 / 5/ NOVEMBER,

3 visits R1 and R2, no medication was administered. On the other two occasions, propranolol, 80 mg, or a matching placebo was taken 12 h and 2 h before the exercise test. To test sensitivity, the subjects were randomized to receive either propranolol or a matching placebo prior to the two tests. Propranolol, 80 mg, was administered because it is known to increase the perception of breathlessness and general fatigue. Pilot work using the exercise protocol outlined above showed that this dose given 12 h and 2 h before the test produced an increase in the symptoms of breathlessness and general fatigue. Equipment (Treadmill, ECG, and Gas Analysis): Exercise testing was performed using a treadmill and ECG console (MAC2 Exercise Testing System; Marquette; Jupiter, FL). Heart rate was monitored using a three-lead ECG throughout all tests and was recorded during the last 10 s of each minute in all tests. Expired air was collected using a breathing valve (model 2700; Hans Rudolph; Kansas City, MO) with a mouthpiece that was attached by tubing to a metabolic cart (Classic Exercise System Model 2; Beckman Instruments; Anaheim, CA) that analyzed the expired gases. All subjects wore a noseclip. Before each test, the oxygen and carbon dioxide sensors were calibrated with standard gas mixtures. Volume was calibrated using the procedures outlined by the instruction manual (Beckman Instruments). Gas collection and analysis were continuous, and respiratory values for every 30 s were given on a printout. Symptom Scales: During each test, symptom scales (see below) were administered at 1 min 30 s, 5 min 30 s, and 7 min 15 s. Two different scales were used at each time point. Each scale was used twice at each time point: once to measure breathlessness and once to measure general fatigue. In other words, a total of 12 symptom scores were recorded during each exercise test. The presentation of scales was alternated: VAS/other and other/vas. Twelve subjects were randomized to the LS/VAS group, and 11 subjects were randomized to the Borg scale/vas group. The symptom of breathlessness was always measured before general fatigue. With twice as many observations for the VAS, there will be a more precise estimate of reproducibility and sensitivity of VAS, but this in no way will that influence the magnitude of either reproducibility or sensitivity. Each scale was administered using a computer (BBC Master; Acorn Computers; Cambridge, UK) and displayed on a color television screen placed at eye level in front of the subject while he exercised on the treadmill. The subject recorded his response using finger controls (see below), and the information was stored in the computer. An audible prompt was given each time a new scale appeared on the screen. On each occasion, the subjects had to move the lever before the cursor appeared on the screen (ie, before any score was displayed); therefore, the previous score was not displayed when the new scale was presented. A sliding lever allowed the subject to move the light cursor horizontally in either direction along the scale. Once the subject had chosen the desired score, he pressed a button to record it in the computer. At this point, another scale was displayed. Subjective Scales: The VAS scale consisted of a horizontal line. The word none was placed at the left end of the scale, and very severe was placed at the right end of the scale. The VAS was scored from 0 to 100, but the subjects were unaware of the numbers. The Borg scale consisted of a vertical line labeled 0 to 10, with verbal descriptors at fixed points on the scale. The LS consisted of five boxes placed vertically, with the following verbal descriptors adjacent to each: not at all breathless, slightly breathless, moderately breathless, really quite breathless, and very breathless indeed. No study has used a LS to investigate breathlessness and general fatigue. Two symptoms were measured with the scales. The subjects were given the following instructions before each of the submaximal tests: Breathlessness was described as follows: breathless, out of breath, air hunger, and unable to breathe enough. The subjects were asked to rate their sensation of breathlessness by referring to their common experience of an uncomfortable awareness of breathing and the descriptors outlined above. They were asked to avoid simply observing an increase in breathing and to disregard other sensations such as leg fatigue or general fatigue. General fatigue was described as overall tiredness and overall fatigue. The subjects were asked to quantify general fatigue by referring to their common experience of general fatigue and the descriptors of general fatigue. They were instructed to disregard other sensations such as leg fatigue and breathlessness. Statistical Methods For each symptom scale and physiologic variable at each time point, a generalized linear model was used to investigate possible order of tests (ie, systematic learning effects), differences among the tests, and treatment differences. The model also included components of variability due to between-subject and withinsubject variation. Reproducibility was defined as the proportion of the total variance (ie, between-subject plus within-subject variance) explained by the between-subject variance. Appropriate estimates of these were obtained using a generalized linear model incorporating visit, intervention, time point, and subject effects. In effect, this means that the subject differences are pooled to produce an estimate of the between-subject variance, while the residual sum of squares is the basis for the estimate of the within-subject variance. For example, if the between-subject variance was 278 and the within-subject variance was 63, the reproducibility coefficient would be 82% (ie, 278/[278 63]), a very good percentage. A score of 100% would signify perfect reproducibility (ie, no within-subject variability at all). Sensitivity was defined as the ratio of the estimated effect of the active drug over the placebo divided by the estimated within-subject SD. The within-subject SD is based on the usual SD of the three non- -blocker observations for each subject pooled across all subjects. Ninety-five percent confidence intervals for sensitivity were used to determine if there was a significant intervention effect, ie, whether the active treatment was significantly different from the placebo intervention. The reason why this measure of sensitivity was chosen was to allow an assessment of the effect of the intervention on an individual basis, not on a group basis. For example, a sensitivity ratio of 1.5 for a particular intervention means that the effect of the intervention is to change the average level of the variable by one-and-a-half times the variability in the measurement seen across (within) any individual. Thus, sensitivity ratios 1.0 would be considered to be of limited value because the change in the variable is not very different from the variability in the measurement seen across (within) any individual. All three time points were used to examine in detail any systematic order of visit/learning effect. The relationship between each symptomatic scale and physiologic variable is complicated because it may be influenced not only by differences among subjects but also by the effects of any therapy, intervention or, indeed, the cumulative effect of exercise through the effect of time into a test itself. Accordingly, the strategy adopted was, for each combination of subjective scale and physiologic variable, to produce sample correlation coefficients for each individual on each visit. For each individual, these were pooled across visits by simply taking the median of the four sample correlations from that subject s four visits. To summarize the overall or typical correlation for this subjective scale and physiologic variable, the median across all the individuals pooled correlations was taken Clinical Investigations

4 Results Psychological/Subjective Scales Table 1 gives the mean SD values for the VAS, the Borg scale, and the LS for breathlessness and general fatigue for minute This table gives the impression that the scores on the active therapy tend to be higher than on the other three interventions. Measurement of Breathlessness Reproducibility: The estimated reproducibility coefficients for the three scales at the three different time points are shown in Table 2. At all three time points, there is a clear tendency for the VAS to be better than the other two scales (ie, to have higher reproducibility). At 5.5 min, the VAS was significantly higher than the Borg scale. The Borg scale and the LS showed poor reproducibility throughout, apart from 7.25 min for the LS. Sensitivity: The VAS and LS tend to be more sensitive to change than the Borg scale (Table 3). At 5.5 min and 7.25 min, the VAS is significantly higher in sensitivity than the Borg scale. Measurement of General Fatigue Reproducibility: Reproducibility tended to be better for the VAS than the other two scales, but there were no statistically significant differences (Table 2). Sensitivity: The Borg scale was significantly better than the LS at 5.5 min and 7.25 min and, perhaps surprisingly, better than the VAS at 5.5 min (Table 3). Visit and Therapy Effects The 7.25-min time point was selected to provide a detailed and representative example of visit and Table 1 Summary Statistics for the VAS, the Borg Scale, and the LS for Breathlessness and General Fatigue for Min 7.25* Variables R1 R2 Placebo Treatment Active Treatment Breathlessness VAS Borg scale LS General fatigue VAS Borg scale LS *Data are presented as mean SD; units are the numbers that relate to the specific scale. Table 2 Reproducibility Coefficients and 95% Confidence Intervals for Breathlessness and General Fatigue for the VAS, the Borg Scale, and the LS* Variables Time 1.5 Min 5.5 Min 7.25 Min Breathlessness VAS 59 (46 76) 74 (55 82) 78 (66 89) Borg scale 55 (39 78) 42 (33 66) 50 (36 76) LS 50 (36 76) 55 (39 80) 76 (55 90) General fatigue VAS 63 (50 78) 69 (55 82) 78 (66 89) Borg scale 45 (33 72) 63 (42 83) 66 (45 85) LS 45 (36 72) 50 (36 76) 55 (39 80) *Reproducibility coefficients are presented as % (95% confidence intervals). VAS significantly higher than Borg scale. therapy effects. Analysis of the other time points has shown broadly similar results, and, to save unnecessary duplication, 7.25 min was chosen because its results were typical of the other time points. Using a generalized linear model, point estimates were produced for the visit and therapy effects of each subjective scale separately. Table 4 shows the estimated visit and therapy effects for both breathlessness and general fatigue, respectively. It is important to remember that these estimated effects are all relative to the grand mean (in the sense that all visit effects sum to zero). If the analysis of variance test of a visit effect proved to be significant for the particular variable under question, a Bonferroni-based multiple-comparisons procedure was used to assess where the significant differences, if any, lay. Breathlessness: The VAS showed significant visit effects decreasing through the four visits. Formally, visit three was significantly lower than visit one, Table 3 Sensitivity Coefficients and 95% Confidence Intervals for the VAS, the Borg Scale, and the LS* Variables Time 1.5 Min 5.5 Min 7.25 Min Breathlessness VAS 1.3 ( ) 2.6 ( ) 2.7 ( ) Borg scale 1.3 ( ) 2.0 ( ) 2.0 ( ) LS 1.4 ( ) 2.2 ( ) 2.6 ( ) General fatigue VAS 1.7 ( ) 2.1 ( ) 2.4 ( ) Borg scale 1.6 ( ) 2.9 ( ) 3.0 ( ) LS 1.6 ( ) 2.0 ( ) 2.0 ( ) *Sensitivity coefficients are presented as No. (95% confidence intervals). VAS significantly higher than Borg scale. Borg score significantly higher than LS. Borg score significantly higher than VAS. CHEST / 116 / 5/ NOVEMBER,

5 Table 4 Visit and Therapy Effects for the Subjective Scales for Min 7.25* Variables Visit One Two Three Four p Value SE of Visit Effect Therapy Effect Breathlessness VAS 3.9a 2.5ab 0.8b 5.6c Borg scale 0.30a 0.15a 0.12a 0.33a LS 0.16a 0.16a 0.02a 0.3b General fatigue VAS 5.3a 0.9b 0.4b 4.8c Borg scale 0.32a 0.07b 0.18b 0.21b LS 0.32a 0.14a 0.04a 0.14a *A common symbol (eg, a) denotes visits between which there is no statistically significant difference on that particular variable. Denotes significance. while visit four was significantly lower than any other visit (in simple terms, visit one was 3.9 U on average above the overall mean, while visit four was 5.6 U below the grand mean, producing a significant fall of U in the VAS on average across all individuals for visits one to four. For the LS, the visit four value was significantly lower than the other three visits. There was no visit effect for the Borg scale. All scales (VAS, Borg scale, and LS) showed a significant therapy effect. The therapy effect was generally quite a bit higher than any of the individual visit effects in all scales. General Fatigue: The VAS showed significant visit effects, again decreasing through visits. Formally, visit two and visit three were significantly lower than visit one, and visit four was significantly lower than the other three visits. Visit one for the Borg scale was significantly higher than the other three visits which were not significantly different from each other. The LS showed no significant visit effect. All scales (VAS, Borg scale, and LS) showed a significant therapy effect. Again, the therapy effects were much higher than any of the visit effects. Physiologic Variables Aerobic Power and Submaximal Relative Intensity: The mean V o 2 max of ml/kg/min reflects the high aerobic fitness level of this group of recreationally active male subjects. The projected steady-state relative intensity was 70% V o 2 max, and the measured mean percent V o 2 max for the R1 test was 71 6%. The fact that there was no significant difference between the minute 6 and minute 8 values on R1, R2, and placebo for V o 2, carbon dioxide output (V co 2 ), and minute ventilation (V e) indicates that a steady state was achieved in these treatments. The variables investigated were as follows: V e, V o 2,V co 2, frequency of breathing, tidal volume (Vt), and heart rate. Table 5 gives a flavor of the physiologic data and shows the values for R1, R2, placebo, and active therapy for minute 7.25, suggesting that there differences in some variables between the active therapy and other treatments. An area of interest is the magnitude of change at each minute into the test for each of the physiologic variables due to the intervention of the -blocker (ie, the magnitude of the therapy effect through the test). In general, there were substantive and significant effects on the active therapy, as follows: (1) V e, the active therapy was significantly higher than the other three interventions after minute 3; (2) V o 2, the active therapy was significantly lower than the other three interventions at all times until the last 2 min; (3) V co 2, at minutes 1 and 2, the active therapy was significantly lower, while at minute 8, it was significantly higher than the other three interventions; (4) frequency of breathing, after 2 min, the active Table 5 Summary Statistics of V E, V O2,V CO2, Frequency of Breathing, VT, and Heart Rate for Min 7.25* Variables R1 R2 Placebo Treatment Active Treatment V e, L/min V o 2, ml/kg/min V co 2, L/min Frequency of breathing, breaths/min Vt, L Heart rate, beats/min *Data are presented as mean SD Clinical Investigations

6 therapy was significantly higher than the other three interventions; (5) Vt, at 2 min and 6 min, the active therapy was significantly higher than the second reproducibility replicate only; and (6) heart rate, the active therapy was significantly lower than the other three interventions at all time points during exercise. Visit and Therapy Effects Only V e and frequency of breathing produced significant visit effects, although V o 2 was close to the significance borderline. Table 6 shows that there was a trend for most variables to decrease over time: the variables tended to be highest in test one and decreased through the remaining three visits. Relationship Between Physiologic Variables and Breathlessness and General Fatigue The overall pooled/median group correlations, the pooled/median individual correlations, and the minimum and maximum individual visit correlations for the VAS, the Borg scale, and the LS (breathlessness and general fatigue), and V e, V o 2,V co 2, frequency of breathing, heart rate, and Vt are shown in Table 7. There were high overall group correlations for all the subjective scales and V e, V o 2,V co 2, frequency of breathing, heart rate, and Vt. For V e, V o 2,V co 2, and heart rate, and all the subjective scales, overall pooled correlations were A wide range in the minimum and maximum (individual and pooled) visit correlations was found for all the subjective scales and the physiologic variables. These findings indicate that there is a very good correlation between the subjective scales and V e, V o 2,V co 2, frequency of breathing, heart rate, and Vt. For all three time points, all three subjective scales were significantly different at each time point except for the LS (breathlessness), which showed no change between the time points of 5.5 min and 7.25 min. The difference between 5.5 min and 7.25 min was 3.8 for the VAS (breathlessness). For VAS (general fatigue), the difference between 5.5 min and 7.25 min was 3.2. The Borg scale (breathlessness) was significantly higher (0.25) at 7.25 min compared to 5.5 min. On the Borg scale (general fatigue), the 7.25 min score was 0.22 higher than the 5.5 min score. On the LS (general fatigue), the 7.25 min value was 0.24 higher than the 5.5 min value. From 5.5 min to 7.25 min, the perception of breathlessness and general fatigue increased despite the fact that there was no change in a range of physiologic variables. Discussion Breathlessness: In this study, the VAS was more reproducible for the measurement of breathlessness than the Borg scale. The LS tended to perform better than the Borg scale, but there were no significant differences between the scales. The LS scores tended to be lower than the VAS scores. Several groups have shown that the VAS allows reproducible measurement of breathlessness in the short term in both normal subjects and patients. 7 9 Wilson and Jones 10,11 have shown that Borg scale measurements of breathlessness are also reproducible in both the short and long term. Due to different exercise protocols and statistical methods of evaluation, it is impossible to say from these studies whether one scale is more reproducible than the other. There have, however, been no descriptions of the reproducibility of LS measurements of breathlessness during exercise. In the only other direct comparison of the repeatability of the VAS and the Borg scale, Wilson and Jones 10 reported that the Borg scale was more reproducible over the short term (a 2- to 6-week period) than the VAS. For this reason, and the fact that the Borg scale correlated more closely with V e, Wilson and Jones 10 favored the Borg scale. These authors, however, made repeat- Table 6 Visit and Therapy Effects for the Physiologic Variables for Min 7.25* Variables Visit One Two Three Four p Value for Visit Effect SE of Visit Effect Therapy Effect V e, L/min 2.99a 0.07ab 0.84b 2.08b V o 2, L/min 0.07a 0.03a 0.01a 0.09a V co 2, L/min 0.02a 0.001a 0.02a 0.04a Frequency of breathing, 1.57a 0.05b 0.35b 1.27b breaths/min Heart rate, beats/min 2.65a 0.37a 1.05a 1.23a Vt, L 0.02a 0.02a 0.004a 0.04a *A common symbol (eg, a) denotes visits between which there is no statistically significant difference on that particular variable. Denotes significance. CHEST / 116 / 5/ NOVEMBER,

7 Table 7 Correlations for the Subjective and Physiologic Variables* VAS Borg Scale LS Variables Breathlessness General Fatigue Breathlessness General Fatigue Breathlessness General Fatigue V e EOPC PIVCs, min/max 0.75/ / / / / /1.00 IVCs, min/max 0.01/ / / / / /1.00 V o 2 EOPC PIVCs, min/max 0.79/ / / / / /1.00 IVCs, min/max 0.12/ / / / / /1.00 V co 2 EOPC PIVCs, min/max 0.80/ / / / / /1.00 IVCs, min/max 0.16/ / / / / /1.00 Frequency of breathing EOPC PIVCs, min/max 0.94/ / / / / /0.99 IVCs, min/max 1.00/ / / / / /1.00 Heart rate EOPC PIVCs, min/max 0.60/ / / / / /0.98 IVCs, min/max 0.70/ / / / / /1.00 Vt EOPC PIVCs, min/max 0.34/ / / / / /0.97 IVCs, min/max 0.96/ / / / / /1.00 *Data are presented as No.; EOPC estimated overall pooled correlation; PIVCs pooled individual visit correlations; IVCs individual visit correlations; min/max minimum/maximum. ability measurements of the slope of the relationship between the breathlessness score and V e, and not on scores at fixed time points, as was done in this study. The VAS demonstrated significantly better sensitivity at 5.5 min and 7.25 min than the Borg scale. The LS tended to be higher than the Borg scale and the VAS higher than the LS, but there were no significant differences. Some studies have induced a greater degree of breathlessness using resistive loading and have found that the VAS and the Borg scale to be responsive to this type of intervention. El- Manshawi et al 12 used the Borg scale to quantify the intensity of breathlessness associated with exercise and respiratory resistive loading. They found that the perception of breathlessness increased at any given workload with resistive loading. Using bronchodilator treatment in chronic obstructive airways disease patients, the VAS has been shown to be sensitive to change. 13 In a comparison of the Borg 6 20 scale and the VAS, Muza et al 14 reported that the VAS may be twice as sensitive an indicator of change than the Borg 6 20 scale, and that the VAS may be of particular value in measuring subtle differences in the sensation of respiratory effort. In the present study, the VAS was the most sensitive for detecting change in breathlessness following pharmacologic intervention, though it should be noted that this change was in a different direction (ie, increased symptom intensity) than in the study by Stark et al. 13 General Fatigue: In this study, the VAS tended to be the most reproducible scale. The Borg scale tended to perform better than the LS but not as well as the VAS. There appears to be limited information on the reproducibility of general fatigue. The Borg scale was clearly much more sensitive to change than the other two scales for general fatigue. It is unknown why the Borg scale was the most sensitive of the three scales. Furthermore, it is unclear why the Borg scale for general fatigue is much more sensitive than the Borg scale for breathlessness; ie, why is the Borg scale able to detect change so much better in general fatigue than with breathlessness? The fact that there is such a great difference in the performance of the Borg scale between breathlessness and general fatigue indicates that the subjects were able to differentiate between breathlessness and general fatigue. Other studies have reported mixed findings on the ability of subjective scales to establish differences in the perception of fatigue. Explanations for these different findings may be attributed to the fact that a range of exercise intensities and exercise protocols 1214 Clinical Investigations

8 have been employed and a variety of pharmacologic agents and regimens have been used to promote fatigue. Comparison of Breathlessness/General Fatigue Differences and Scales: Reproducibility coefficients for breathlessness and general fatigue were highest for the VAS throughout the three time points, but they were only statistically significant at minute 5.5 for breathlessness. Indeed, the reproducibility coefficients for breathlessness and general fatigue follow a similar pattern. It could be hypothesized that the subjects were unable to distinguish between the two symptoms. However, all of the subjects reported that they could discriminate between breathlessness and general fatigue. Instructions to the subjects were clearly given before each test, and the subjects confirmed that they could clearly differentiate between breathlessness and fatigue. It may be expected that the VAS would be the most sensitive scale because it offers finer adjustment. In breathlessness, this is the case; but perhaps somewhat surprisingly for general fatigue, the Borg scale is the most sensitive of the three scales. The variation in responses between the various time points may be a result of varying perceptions of sensations over time and the time to adapt to the level of exercise. It is speculated here that it is necessary for large changes in status to occur before subjects are likely to move to another point on the LS. If this were true, it would be anticipated that the LS would be very reproducible in similar conditions. However, any movement in a subject s scores is likely to heavily penalize the LS scale because a change of one is equivalent to 20% of the range of possible scores. Word descriptors did not restrict the use of the LS because a score of four was given for LS (breathlessness and general fatigue) on the active therapy, R1, and placebo. Previous studies using 4-, 5-, and 7-point LSs have produced inconclusive results. Guyatt et al 19 stated that a seven-point LS compared to a VAS did not result in a response difference. However, they concluded that the ease of administration of the LS and the extent to which clinicians intuitively grasp its results pointed to the use of the LS instead of VAS. Comparison of the Visit and Therapy Effects: The VAS (breathlessness and general fatigue) and LS (breathlessness) showed a significant visit effect. These visit effects were small compared to therapy effects, but were fairly large in comparison to the absolute scores. Care has to be taken in the use of this protocol because it has been shown that there is a visit effect for some scales. Despite the fact that the therapy effect was much greater than the visit effect, it is advisable to incorporate a run-in period when this protocol is used. Physiologic Variables: V e was increased from 3 to 8 min compared to placebo. The increase in V e during high-intensity steady-state exercise during the later stages of the exercise bout is in agreement with the literature that indicates that subjects need to exercise at fairly high intensities before a significant increase in V e is found An increase in plasma potassium during exercise after -blockade, due to reduced re-uptake in inactive tissues, has been found. 24 This could result in an increased ventilatory drive via the peripheral chemoreceptors. 25 Twentyman et al 20 concluded that the increase in V e at the end of the 70% steady state cannot be explained by depression of central or peripheral ventilatory control by propranolol or changes in lung mechanics. Twentyman et al 20 suggested that the increase in V e in the latter stages of the 70% V o 2 max steady state was the V e responding to increased levels of lactic acid. Airway resistance and lung mechanics are not usually influenced to any great extent by propranolol at rest or during exercise in normal subjects and cannot explain the increase in V e. 26 On the active therapy, the frequency of breathing was significantly increased from minute 2 onwards, but there were only two time points when Vt was lowered with propranolol. Joyner et al 27 reported a small decrease in Vt after nonselective blockade that was compensated by an increased frequency of breathing, whereas Pearson et al 21 found that Vt was increased with atenolol but not with propranolol. The Relationship Between the Subjective Scales and Physiologic Variables Breathlessness: All of the scales showed high overall group correlations with V e, V o 2,V co 2, frequency of breathing, heart rate, and Vt. However, it must be stressed that a high correlation does not necessarily imply a causal relationship. When using VAS and Borg scales, a good correlation has been reported between V e and the perception of breathlessness. 10,14 However, it would be wrong to suggest that the perception of breathlessness is simply a sensing of V e. With no increase in V e, O Neill et al 7 found that their subjects reported a rise in breathlessness. Other studies suggest that V e per se is not the most important factor impinging on breathlessness. The cause of an increase in V e and the time delay between an increase in breathlessness and an increase in V e should be taken into consideration. 28,29 In this study, V e remained stable between 5.5 min and 7.25 min in the R1, R2, and placebo treatments but rose by 4 L/min in the active therapy. CHEST / 116 / 5/ NOVEMBER,

9 Despite the stability of all respiratory variables on R1, R2, and placebo, the perception of breathlessness increased as measured by the VAS and the Borg scale. The correlation between breathlessness (VAS) and frequency of breathing was 0.85 and 0.89 for the correlation between Vt and the LS was (the highest correlations for frequency of breathing and Vt). The findings of Chronos et al 29 cast doubt on a direct linkage between frequency of breathing and Vt with breathlessness. They found that there was no relationship between the changes in breathlessness and changes in frequency of breathing or Vt. General Fatigue: All three scales showed very high correlations between V e, V o 2,V co 2, heart rate, and general fatigue. A review by Carton and Rhodes 30 reported that it is V e and frequency of breathing that are most closely related to perception of effort. In this study, the correlations for V e and frequency of breathing and the three scales were around 0.92 and 0.80, respectively. The fact that there was no significant difference in V e and frequency of breathing between 5.5 min and 7.25 min in R1, R2, and placebo, as well as the fact that the perception of general fatigue was higher at 7.25 min compared to 5.5 min, suggests that the perception of general fatigue is not directly linked to V e and frequency of breathing. This study found a high correlation between heart rate and the subjective scales. A good relationship has been reported between the rating of perceived exertion and heart rate, 31 but a causal link has been rejected. The fact that heart rate can be changed at a given V o 2 with no effect on the rate of perceived exertion substantiates this hypothesis. 32 Conclusion The results of this study demonstrate a similarity in behavior among the VAS, the Borg scale, and the LS. For reproducibility, the VAS produced the highest ratio estimates for breathlessness and general fatigue. For sensitivity, the VAS was best for breathlessness, and the Borg scale had the highest ratio estimates for general fatigue. The findings of this study indicate that during steady-state exercise, subjective scales can reproducibly measure symptoms and are sensitive to changes that result from pharmacologic intervention. Overall, the VAS appears to be the best scale. References 1 Costill DL, Thomason H, Robert E. Fractional utilization of aerobic capacity during distance running. Med Sci Sports Exerc 1973; 5: Lamb DR. Physiology of exercise. New York, NY: MacMillan, Buchfuhrer MJ, Hansen JE, Robinson TE, et al. Optimizing the exercise protocol for cardiopulmonary assessment. J Appl Physiol 1983; 55: British Association of Sports Science (Sports Physiology Section). Position statement on the physiological assessment of the elite competition. 2nd ed. Leeds, UK: British Association of Sports Science, McLenachan JM, Grant S, Dargie HJ, et al. Submaximal, but not maximal, exercise testing detects differences in the effects of beta-blockers during treadmill exercise: a study of celiprolol and atenolol. Am Heart J 1991; 121(2Suppl): Van Herwaarden CLA, Binkhorst RA, Fennis JFM, et al. Effects of propranolol and metroprolol and hemodynamic and respiratory indices and on perceived exertion during exercise in hypertensive patients. Br Heart J 1979; 41: O Neill PA, Stark RD, Allen, SC, et al. The relationship between breathlessness and ventilation during steady-state exercise. Bull Eur Physiopathol Respir 1986; 22: Stark RD, Gambles SA, Lewis JA. Methods to assess breathlessness in healthy subjects: a critical evaluation and application to analyze the acute effects of diazepam and promethazine on breathlessness induced by exercise or by exposure to raised levels of carbon dioxide. Clin Sci 1981; 61: Stark RD, Gambles SA, Chatterjee SS. An exercise test to assess clinical dyspnoea: estimation of reproducibility and sensitivity. Br J Dis Chest 1982; 76: Wilson RC, Jones PW. A comparison of the visual analogue scale and modified Borg scale for the measurement of dyspnoea during exercise. Clin Sci 1989; 76: Wilson RC, Jones PW. Long-term reproducibility of Borg scale estimates of breathlessness during exercise. Clin Sci 1991; 80: El-Manshawi A, Killian KJ, Summers E, et al. Breathlessness during exercise with and without resistive loading. J Appl Physiol 1986; 61: Stark RD, Morton PB, Sharman P, et al. Effects of codeine on the respiratory responses to exercise in healthy subjects. Br J Clin Pharmacol 1983; 15: Muza SR, Silverman MT, Gilmore GC, et al. Comparison of scales used to quantitate the sense of effort to breathe in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1990; 141: Lees KR, Curzio J, Farish E, et al. The influence of betaadrenoceptor antagonists with and without partial agonist activity on exercise tolerance and muscle lactate production. Eur J Clin Pharmacol 1987; 33: Pearson SB, Banks DC, Patrick JM. The effect of B-adrenoceptor blockade on factors affecting exercise tolerance in normal man. J Clin Pharmacol 1979; 8: Kaiser P, Tesch PA, Thorsson A, et al. Skeletal muscle glycolysis during submaximal exercise following acute B- adrenergic blockade in man. Acta Physiol Scand 1985; 123: Tesch PA, Kaiser P, Komi PV. Effects of B-adrenergic blockade on EMG signal characteristics during progressive exercise. Acta Physiol Scand 1984: Guyatt GH, Townsend M, Berman LB, et al. A comparison of Likert and Visual Analogue scales for measuring change in function. J Chronic Dis 1987; 40: Twentyman OP, Disley A, Gribbin HR, et al. Effect of B-adrenergic blockade on respiratory and metabolic responses to exercise. J Appl Physiol 1981; 51: Clinical Investigations

10 21 Pearson SB, Morrison JFJ, Simpson FG. The effects of B-adrenoceptor blockade on breathing during progressive exercise in normal man. Br J Clin Pharmacol 1987; 24: Wilcox RG, Bennett T, MacDonald IA, et al. The effects of acute or chronic ingestion of propranolol or metoprolol on the physiological responses to prolonged, submaximal exercise in hypertensive men. Br J Clin Pharmacol 1984; 17: Tesch PA, Kaiser P. Effects of B-adrenergic blockade on O 2 uptake during submaximal and maximal exercise. J Appl Physiol 1983; 54: van Baak MA, Jooijm JMV, Wijnen AG, et al. Submaximal endurance exercise performance during enalapril treatment in patients with essential hypertension. Clin Pharmacol Ther 1991: Busse MW, Maassen N, Konrad H. Relation between plasma K and ventilation during incremental exercise after glycogen depletion and repletion in man. J Physiol 1991; 60: Warren JB, Jennings SJ, Clark TJH. Effect of adrenergic and vagal block on the normal human airway response to exercise. Clin Sci 1984; 66: Joyner MJ, Jilka SM, Taylor JA, et al. B-Blockade reduces tidal volume during heavy exercise in trained and untrained men. J Appl Physiol 1987; 62: Lane R, Cockcroft A, Guz A. Voluntary isocapnic hyperventilation and breathlessness during exercise in normal subjects. Clin Sci 1987; 73: Chronos N, Adams L, Guz A. Effect of hyperoxia and hypoxia on exercise-induced breathlessness in normal subjects. Clin Sci 1988; 74: Carton RL, Rhodes CR. A critical review of the literature on ratings scales for perceived exertion. Sports Med 1985; 2: Borg GAV. Psychophysical bases of perceived exertion. Med Sci Sports Exerc 1982; 14: Davies CTM, Sargeant AJ. The effects of atropine and practolol on the perception of exertion during treadmill exercise. Ergonomics 1979; 10: CHEST / 116 / 5/ NOVEMBER,