Minimally Clinically Important Difference for the UCSD Shortness of Breath Questionnaire, Borg Scale, and Visual Analog Scale

Abstract

Dyspnea is a primary symptom of chronic lung disease and an important outcome measure for clinical trials. Several standardized measures have been developed to evaluate this important symptom and are being used increasingly in clinical trials. The minimally clinically important difference (MCID) is not well defined for these measures but is important in interpreting the clinical meaning of results of studies in this area. The purpose of this paper is to evaluate the MCID for three commonly used measures to assess dyspnea in chronic lung disease: UCSD Shortness of Breath Questionnaire (SOBQ), Borg Scale (Borg), and Visual Analog Scale (VAS). The analysis is based on a retrospective review of published trials evaluating the response to a pulmonary rehabilitation or exercise intervention that is known to produce modest, but clinically meaningful changes for such patients. Using a distribution-based approach based primarily on effect size, the recommended MCID for these measures are: 5-units for the SOBQ, 1-unit for the Borg scale, and approximately 10 to 20 units for the VAS.

• Dyspnea
• Minimally clinically important difference
• Chronic lung disease
• Pulmonary rehabilitation

Introduction

Dyspnea, the subjective sensation of difficult or labored breathing, is a primary symptom for patients with chronic lung disease and an important outcome measure for clinical trials in this field. Recommended treatment strategies for the management of diseases such as chronic obstructive pulmonary disease (COPD) are focused on the relief and control of symptoms such as dyspnea. An evidence-based review of pulmonary rehabilitation programs for the management of patients with COPD concluded that improvements in exercise tolerance and dyspnea were the most consistent, evidence-based results of pulmonary rehabilitation and provide the strongest scientific justification for the inclusion of such programs in disease management strategies for such patients [1&2].

Since symptoms of dyspnea depend upon the intensity of the stimulus and occur most often with exertion, attempts to measure dyspnea have used various strategies for standardizing the level of physical activity and/or effort [[3]]. This can be accomplished through a questionnaire format that asks patients to recall their perception of the symptom associated with specific types or levels activities. Alternatively, dyspnea can be induced in a controlled, laboratory setting in which patients are asked to rate their perception of breathlessness while performing standardized physical tasks. Relating breathlessness to physical activity is important because patients with chronic lung disease will frequently avoid activities that produce the uncomfortable sensation of dyspnea and may not report symptoms at rest. The so-called “fear-dyspnea” cycle refers to the vicious cycle of inactivity for such patients who become progressively more inactive in order to reduce the sensation of breathing discomfort and the accompanying fear and anxiety.

The purpose of this paper is to evaluate the minimally clinically important difference (MCID) for three commonly used measures to assess dyspnea in chronic lung disease:UCSD Shortness of Breath Questionnaire (SOBQ), Borg Scale (Borg), and Visual Analog Scale (VAS). The analysis is based on a retrospective review of published clinical trials that have used these outcome measures to evaluate response to a pulmonary rehabilitation or exercise interventions that are known to produce modest, but clinically meaningful, changes in dyspnea in patients with chronic lung disease [1&2]. In order to provide a consistent approach to analyzing MCID with the limited published data with these measures, a distribution-based approach was used by calculating the effect size (change after intervention divided by standard deviation of baseline scores). In addition, for the SOBQ, an anchor-based approach was also evaluated using the Transition Dyspnea Index [[4]] as a reference standard.

UCSD Shortness of Breath Questionnaire

The University of California, San Diego (UCSD) Shortness of Breath Questionnaire (SOBQ) is a self-report questionnaire that asks patients to indicate the severity of shortness of breath experienced on a 6-point scale (0 = Not at all, 4 = Severely, 5 = Maximally or unable to do because of breathlessness) during 21 activities of daily living associated with varying levels of exertion [[5]]. Three additional questions ask about limitations caused by shortness of breath, fear of harm from overexertion, and fear of shortness of breath, for a total of 24 items. If patients do not routinely perform the activity, they are asked to estimate the degree of shortness of breath anticipated. Responses are summed to produce a total score ranging from 0 to 120. Originally developed as an easy-to-administer clinical assessment tool for goal setting in pulmonary rehabilitation, the UCSD SOBQ has been found to be useful as an outcome measure in clinical research studies. Its reliability and validity have been evaluated and reported [5&6].

The original version of the UCSD SOBQ used a slightly different scoring system for the 6-point scale (0 = 0% of the time or never, 1 = 25% of the time or sometimes, 2 = 50% or half of the time, 3 = 75% or most of the time, 4 = 100% or all of the time, and NA 5 = not applicable or unable to do [[7]]. In scoring the original questionnaire, it was assumed that 5 (NA) represented the highest level of dyspnea, implying that patients were unable to perform the specified activity because of dyspnea. However, in some cases, the NA category was used by patients to describe activities that they did not consider relevant (e.g., washing the car if they did not have a car). In other cases, patients would not answer questions for such activities. In the revised version, the same 6-point scale was modified to indicate that 5 clearly indicated the highest level of dyspnea and instructions were modified to indicate that all questions were to be answered. For activities that patients do not perform, they are instructed to give their best estimate if that activity was attempted. Overall, results from the two versions of the questionnaire are similar and correlate highly (r = 0.96) [[5]]. Nevertheless, for the purposes of this discussion, results from the old and new version are presented separately.

In order to address the issue of the minimally clinically important difference (MCID) for the SOBQ, data were examined from several studies that utilized the SOBQ as an outcome measure in evaluating response to a pulmonary rehabilitation or exercise intervention in patients with chronic lung disease. Such interventions are known to produce modest improvements in exercise performance and symptoms of breathlessness in such patient populations. Table 1 summarizes the results from several such studies and includes information about the patients evaluated, lung function, and changes in SOBQ expressed as mean ± SD as well as effect size (mean change divided by baseline SD) [7-12]. After pulmonary rehabilitation, the smallest effect size (0.12) was found in the patients in the NETT study with prior rehabilitation experience (mean FEV₁ = 26% predicted, − 2.3 unit change in SOBQ)(unpublished data). The largest effect size (0.48) was observed in the clinical trial of maintenance after pulmonary rehabilitation at UCSD (mean FEV₁ = 45% predicted, − 10.0 unit change in SOBQ) [[8]]. The two multi-center studies involving large numbers of patients with COPD reported similar effect sizes: 0.31 in the California Collaborative Project (FEV₁ = 44% predicted, − 6.8 unit change in SOBQ) [[9]] and 0.25 in the NETT study, for patients with more severe lung disease (FEV₁ = 26% predicted; − 4.8 unit change in SOBQ) who did not have prior pulmonary rehab experience. In the study by Stulbarg and associates that involved dyspnea management strategiesand exercise training, but not in the context of the comprehensive pulmonary rehabilitation program, the changes in SOBQ were more modest (effect size = 0.17, − 2.3 unit change in SOBQ) [[12]].

Table 1.  University of California, San Diego Shortness of Breath Questionnaire (SOBQ).

Reference	Type of study	N	Age, yrs (Range)	Gender %M/F	FEV1 %pred (L)	Baseline	Change	Effect size
Ries (2003)	Rehab	164	67 ± 8	54/46	45 ± 19	55.5 ± 20.8	− 10.0	0.48
California Collaborative Project (2004)	Rehab	522	68 ± 9	42/58	44 ± 19	56.8 ± 21.8	− 6.8	0.31
NETT	Rehab	1218	67 ± 6	61/39	25 ± 7	65.7 ± 19.0	− 3.2	0.17
No prior rehab	Rehab	496			26 ± 7		− 4.8	0.25
Prior rehab	Rehab	265			26 ± 7		− 2.3	0.12
UCSD clinic (2003)	Rehab	99	(55–85)	59/41		55.8 ± 24.8	− 7.5	0.30
Ries# (1995)	Rehab	57	62 ± 8	74/26	(1.21 ± 0.55)	35.8 ± 18.5	− 7.0	0.38
Stulbarg (2002)	Exer	34	66 ± 6	35/65	44 ± 10	54.7 ± 15.1	− 2.3	0.15

*Results expressed as mean ± standard deviation.

^#Data collected utilizing original version of the SOBQ. Other studies used the revised version.

In order to further evaluate the sensitivity of SOBQ in detecting a change in dyspnea, a receiver operator curve (ROC) analysis was performed using change in the SOBQ versus the Transitional Dyspnea Index (TDI) obtained concurrently in 164 patients with chronic lung disease in the rehabilitation maintenance study [[8]][[13]]. The TDI is a questionnaire that is commonly used to assess change in dyspnea in which a score of + 1 (or − 1) represents the minimum threshold for a patient's perceived change in dyspnea from baseline (reflected in the Baseline Dyspnea Index) [[4]]. For this analysis, it was assumed that an 1-unit TDI represented, by definition, a clinically detectable change in symptoms. For the ROC analysis, sensitivity was plotted against 1-specificity for various change scores in the measure being evaluated in detecting change in the standard measure (i.e., 1-unit improvement in TDI). This analysis suggests that a change in SOBQ of − 5 units is associated with a sensitivity and specificity of approximately 69% and 67%, respectively, in detecting an 1-unit change in TDI. For a − 6 units change in SOBQ, sensitivity and specificity were approximately 66% and 73%, respectively. For a − 4 units change in SOBQ, the sensitivity and specificity were approximately 72% and 63%, respectively.

For comparison, a similar ROC analysis was performed using the dyspnea scale of the Chronic Respiratory Disease Questionnaire (CRQ) [[14]] versus the TDI using data collected in the same study in the same patient cohort. This analysis indicated that the commonly accepted MCID for the CRQ of 0.5 was associated with sensitivity and specificity of 75% and 63%, respectively, in detecting a 1-unit change in the TDI. This suggests that a − 4 units change in the SOBQ gives about the same sensitivity and specificity as the previously defined MCID for the CRQ in detecting an 1-unit change in the TDI.

Based on the effect sizes observed for the several studies with a pulmonary rehabilitation intervention in patients with moderate to severe chronic lung diseases (Table 1), it appears that a change in SOBQ of approximately 5 to 7 units is associated with the lower end of a small effect size (0.25 to 0.30 range) as defined by Cohen [[15]]. Mean changes above 7 units that are associated with larger effect sizes would suggest that this level of change is greater than the minimal difference perceived by patients with chronic lung disease. These results come from a variety of studies including two large multi-center studies. In addition, the ROC analysis of the SOBQ and the CRQ in detecting an 1-unit improvement in the TDI suggests that a similar range of 4 to 6 on this scale provides the optimal degree of sensitivity and specificity. In general, based on these results and clinical experience, an MCID of 5 units has been suggested as a reasonable MCID for the SOBQ [[13]].

Borg Scale

The original perceived exertion scale described by Borg comprised a scale from 6 to 20 (corresponding to a normal heart rate range of 60 to 200) used to measure overall exertion during physical activity [[16]]. The scale was subsequently modified to form a 10-point scale including written indicators of severity to anchor specific numbers on the scale [[17]]. Morerecently, these scales have been used more widely to quantify various “perceived symptoms” such as breathlessness and muscle fatigue during exercise. Several studies have evaluated the validity and reliability of Borg ratings of breathlessness during exercise [18&19].

Table 2 presents a summary of results from selected studies that utilized the Borg breathlessness scale during an exercise test to assess symptom response to various interventions [[12]][20-27]. Since perceived symptoms during exercise depend upon the stimulus, ideally for comparison such measures should be performed at the same work rate. In the study by Stulbarg and coworkers, patients with COPD (mean FEV₁ = 44% predicted) performed both an incremental treadmill exercise test and an endurance test at a constant work rate one level below the maximum on the incremental, maximum test [[12]]. This analysis considered Borg ratings at the highest iso-work levels during the incremental test and at iso-times during the endurance test. In this study, changes in Borg dyspnea ratings before and after an exercise training intervention (with dyspnea management) were − 1.8 (effect size = 1.0) for iso-work on the incremental test and − 1.6 (effect size = 0.8) for iso-time on the endurance test. In the study by Foglio and coworkers of 26 patients with COPD (mean FEV₁ = 43% predicted) before and after pulmonary rehabilitation, Borg dyspnea ratings at iso-work level on an incremental cycle ergometry exercise test decreased by 2.0 units (effect size = 1.5) [[20]]. In a follow-up study from the same group, Clini and coworkers evaluated 43 patients with COPD after inpatient pulmonary rehabilitation (mean FEV₁ = 57% predicted) matched with 43 patients with COPD treated with an outpatient rehabilitation program (mean FEV₁ = 53% predicted) [[21]]. Borg dyspnea scores were obtained at rest and at regular intervals during an incremental, maximum exercise test on a cycle ergometer. Borg dyspnea ratings decreased in both groups after pulmonary rehabilitation by a mean of 2.2 units at iso-work levels (effect size = 1.4 and 1.2 for inpatient and outpatient pulmonary rehabilitation groups, respectively). In the study by Gigliotti and coworkers of 20 patients with COPD (mean FEV₁ = 42% predicted), Borg dyspnea ratings decreased by 3.3 units (effect size = 1.3) iso-work and 2.4 units (effect size = 0.9) at iso-ventilation levels [[22]].

Table 2.  BORG Scale.

Reference	Type of study	N	Age, yrs (Range)	Gender %M/F	FEV1 %pred (L)	Baseline	Change	Effect size
Foglio (1999)	Rehab	26	63 ± 8	88/12	43 ± 15	5.6 ± 1.3	− 2.0	1.50
Clini (2001)	Rehab
Inpatient		43	64 ± 8	67/33	57 ± 28	6.4 ± 1.6	− 2.2	1.38
Outpatient		43	64 ± 7	67/33	53 ± 21	8.5 ± 1.9	− 2.2	1.16
Gigliotti (2003)	Rehab	20	64 ± 8	90/10	42 ± 12
Standard VE						6.6 ± 2.6	− 2.4	0.92
Standard Work						6.6 ± 2.6	− 3.3	1.27
UCSD Clinic (iso-work)	Rehab	115	(55–85)	63/37		3.1 ± 1.5	− 1.5	1.00
Stulbarg (2002)	Exer	34	66 ± 6	35/65	44 ± 10
iso-work						4.8 ± 1.8	− 1.8	1.00
iso-time						4.6 ± 1.9	− 1.6	0.84
Eaton (2002)	O2 Rx	41	67 ± 9	70/30	26 ± 8	4.8 ± 1.5	− 0.7	0.47
Jolly (2001)	O2 Rx
Desaturate		11	67 ± 2	91/9	36 ± 4	5.8 ± 1.5	− 2.1	1.40
Non-desaturate		9	70 ± 3	100/0	35 ± 3	4.2 ± 1.3	− 2.0	1.54
Boni (2002)	Exer	20	65 ± 8	65/35
Flow-limited		11			44 ± 19	3.7 ± 1.1	− 1.1	1.00
Not flow-limited		9			50 ± 16	1.8 ± 1.0	− 0.4	0.40
Martinez (1997)	Exer	12	61 ± 2	65/35	26.7 ± 1.8	7.1 ± 2.0	− 3.6	1.80
Geiser (2001)	Resting LVRS	28	63 (51–75)	89/11	< 35% (1.007)	6.6 ± 1.8	− 2.7	1.50

*Results expressed as mean ± standard deviation.

Martinez and coworkers evaluated changes after lung volume reduction surgery in 17 patients with advanced emphysema (mean FEV₁ = 27% predicted) [[26]]. Borg dyspnea ratings at iso-work levels on an incremental cycle ergometry test decreased by 3.6 units (1.8 effect size).

In a study of 6 weeks of ambulatory oxygen therapy (with a crossover of 6 weeks of compressed air treatment), Eaton and coworkers evaluated dyspnea after a 6-minute walk test with a Borg scale measurement of dyspnea [[23]]. Results in 41 patients with COPD (mean FEV₁ = 26% predicted) demonstrated significantly lower exercise dyspnea on oxygen versus compressed air (4.1 versus 4.8; effect size = 0.5). Six-minute walk distance was significantly higher on oxygen (377 versus 337 meters). In this study, a 1-unit change in the Borg scale dyspnea measure was considered to be a clinically significant outcome.

There are several considerations unique to the Borg scale that should be considered in determining an appropriate level for the MCID. First, the scale is not strictly linear and is designed with ratio properties in its scaling. As such, it is possible that larger changes are more likely to be observed at the higher end of the scale in which there are larger numerical intervals between word anchors for symptom severity. This is consistent with the results reported in Table 2 in which the largest changes from rehabilitation were associated with those patient cohorts with the highest baseline symptoms. In comparing it to a questionnaire assessment of dyspnea during an intervention such as pulmonary rehabilitation, it should be noted that the Borg scale is most commonly administered with a stimulus such as exercise rather than reflecting a patient's historical recollection of symptoms. Also, the intervention in rehabilitation programs is typically targeted to dyspnea during exercise training sessions so the effects of the intervention may be more immediately noticed by the patient compared to a score that reflects historical recall of symptoms associated with various activities of daily living.

The findings in Table 2 suggest that for pulmonary rehabilitation and exercise interventions in which Borg breathlessness scales were measured at comparable exercise levels, it appears that changes in the order of 2 units were most commonly observed and associated with large effect sizes above 0.8 (as defined by Cohen [[15]]). Changes greater than 2 units appear to be associated with stronger interventions such as lung volume reduction surgery in emphysema while changes with less intensive interventions such as an acute response to oxygen supplementation or bronchodilator therapy appear to be associated with lower changes on the order of 1.0 unit with more moderate effect sizes.

Visual Analog Scale

The Visual Analog Scale (VAS) consists of a line 100 mm in length with appropriate written anchors at the extremes (e.g., “not breathless at all” to “extremely breathless” when used to evaluate dyspnea) [[3]][[28]]. The line may be presented either horizontally or vertically. The respondent marks the scale, which is scored by measuring the distance from the lower end of the line to the mark. Reliability and validity of the VAS have been reported [[3]]. Like the Borg scale, the VAS is used most commonly to measure dyspnea during exercise.

Several studies have used the VAS as an outcome measure in response to pulmonary rehabilitation and other interventions (Table 3) [[20]][[26]][29-31]. Foglio and coworkers utilized the VAS to assess a subjective rating of dyspnea (not associated with a physical or exercise task) in 26 patients with COPD (mean FEV₁ = 43% predicted) before and after 8 to 10 weeks of pulmonary rehabilitation [[20]]. VAS dyspnea decreased from 25 to 15 (effect size 0.8). de Torres and coworkers evaluated 37 patients with COPD (mean FEV₁ = 0.70 L) before and after 6 to 8 weeks of pulmonary rehabilitation [[30]]. VAS measured at rest decreased from 11 to 9 (NS, effect size = 0.13) while at the end of a 6-minute walk test from 62 to 50 (effect size = 0.48). Six-minute walk distance in this study increased from a mean of 285 to 343 meters, so the exercise stimulus was higher on the post-rehabilitation measurement. Reardon and coworkers used the VAS to measure dyspnea during an incremental, maximum treadmill exercise test in 10 patients with COPD (mean FEV₁ = 35% predicted) before and after 6 weeks of pulmonary rehabilitation [[20]]. Mean dyspnea at maximum exercise decreased 23.9 units from a baseline VAS score of 74.4 (effect size = 1.3). This was associated with an increase in VO2max at maximum exercise from 865 to 954 ml/min.

Table 3.  Visual Analog Scale.

Reference	Type of study	N	Age years	Gender %M/F	FEV1 %pred (L)	Baseline	Change	Effect size
Reardon (1994)	Rehab	10	66 ± 8	50/50	35 ± 10	74.4 ± 18.9	− 23.9	1.26
de Torres (2002)	Rehab	37	63 ± 6	47/53	(0.7 ± 0.2)	62.0 ± 25.0	− 11.9	0.48
Foglio (1999)	Rehab	26	63 ± 8	88/12	43 ± 15	25.0 ± 13.0	− 10.0	0.77
Alvisi (2003)	Pre/PostO2 sup.	10	69 ± 10	100/0	30 ± 6	25.0 ± 18.0	− 10.0	0.56
Martinez (1997)	Ex.	12	61 ± 2	65/35	27 ± 2 39 ± 4	79.6 ± 17.2	− 30.3	1.76

*Results expressed as mean ± standard deviation.

In the study before and after lung volume reduction surgery, Martinez and coworkers used the VAS to evaluate dyspnea at rest and peak exercise on a maximum cycle ergometry exercise test. In 12 patients evaluated at iso-work levels, mean VAS decreased from 79.6 to 49.3 (effect size = 1.8) [[26]].

Alvisi and coworkers utilized the VAS to evaluate resting dyspnea before and after administration of 30% supplemental oxygen in 10 patients with COPD (mean FEV₁ = 42% predicted). Mean VAS decreased from 25 on air to 15 after 15 minutes of oxygen (effect size = 0.6) [[31]].

The VAS has been used less frequently in studies of pulmonary rehabilitation or in similar type interventions. In Table 3, as expected, the largest changes (30 units) were observed after lung volume reduction surgery. With the exception of the study by Reardon and colleagues [[29]], the two other studies of pulmonary rehabilitation found mean changes of 10 to 12 units with effect sizes in the 0.5 to 0.7 range. Of note, the baseline levels for the VAS in these two studies was quite different: 62 in the study by de Torres and coworkers [[30]] and 25 in the study by Foglio and coworkers. A 10-unit change (effect size 0.6) was also noted after oxygen supplementation in patients with bronchiectasis in the study by Alvisi and coworkers [[31]]. In summary, from these few studies with limited data, it appears that a change in VAS of approximately 10 to 20 is associated with a modest level of symptom change in patients with chronic lung disease.

Summary

Given the modest amount of published data to date utilizing these instruments systematically to evaluate dyspnea in patients with chronic lung disease, one cannot define MCID with certainty or precision. Since change in dyspnea has been demonstrated to be an important outcome after pulmonary rehabilitation, and because pulmonary rehabilitation is expected to produce modest, but clinically significant improvement in symptoms, this analysis focused on trials in this area. The analysis used to develop recommendations for MCID for these three instruments was based on the distribution-based method using effect sizes because these data were available most consistently in the studies examined. The recommendations for MCID, then, should be viewed as reasonable estimates based on the consistency of results using this approach. Future studies utilizing measures of dyspnea, particularly in comparison to each other in the same patients, will be important so that these recommendations can be further evaluated and validated.