Douglas S. Ander, MD, Department of Emergency Medicine, 69 Jesse Hill Jr. Drive, SE, Atlanta, GA 30303 E-mail: firstname.lastname@example.org
Patients presenting to the emergency department with heart failure are evaluated based on the subjective sensation of dyspnea. In this study, the authors sought to determine the change in dyspnea, measured by a visual analog scale (VAS), which is associated with a meaningful change in the patient's perception and the effect of dyspnea severity on the VAS. In this prospective, observational study the authors defined a meaningful change in VAS dyspnea as the difference between VAS scores when patients reported “a little less difficulty breathing” or “a little more difficulty breathing.” Seventy-four patients were evaluated, and the mean for a meaningful change in VAS was 21.1 mm (95% confidence interval, 12.3–29.9 mm). Patients that recorded higher index VAS scores had a significantly greater change in VAS. VAS scores and the changes with treatment provide the treating physician with another means to assess the effects of their interventions.
Patients presenting to the emergency department (ED) with acute exacerbations of heart failure (HF) commonly experience dyspnea. During the evaluation and treatment of HF the clinician gauges the severity of the illness and response to therapy based on multiple objective measures and the patient's perception of dyspnea. The sensation of dyspnea is subjective but has been previously quantified using both the Borg scale and visual analog scale (VAS).1,2 Several other studies have demonstrated that measures of breathlessness including VAS correlate with objective measures of airway resistance, but vary depending on the severity of illness.3–5
When using a VAS clinically it is important to define what change in the VAS score is perceived by either the patient or physician as meaningful. The general approach to determining the significance of change was developed and termed the “minimal clinically important difference.”6–8 This approach has been used in several clinical diagnoses such as pain, wound cosmesis, and dyspnea during acute asthma exacerbations to define the meaningful change in VAS score.9–14 In these studies researchers have defined the change in the VAS that the patient or physician perceived as meaningful. However, in HF exacerbations the difference in VAS dyspnea scores considered by the patient to be meaningful has not been determined. Knowledge of what constitutes a meaningful change in VAS dyspnea scores would help clinicians use this measure to objectively monitor response to treatment of HF. Determining the meaningful difference in VAS dyspnea scores would also help researchers calculate the sample sizes needed to detect a clinically significant response to an investigational treatment. Finally, knowing whether the median change in VAS produced by a therapy is, in fact, meaningful would help clinicians evaluate the validity of clinical trials that use VAS dyspnea score as an outcome measure. Therefore, we sought to determine the change in VAS dyspnea scores that is associated with a self-perceived change in dyspnea and determine the effect of dyspnea severity on this VAS dyspnea score.
Study Setting and Population. This study was conducted in the ED of Grady Memorial Hospital, Atlanta's public hospital and only level I trauma center. Grady's ED manages over 100,000 patient visits annually. English-speaking adults with a chief complaint of “dyspnea” or “shortness of breath,” and a history, physical examination, and chest radiograph consistent with an acute exacerbation of HF were eligible for inclusion in the study. Patients who were unable to give informed consent were excluded. Therapeutic interventions were performed at the discretion of the patients' treating physicians.
Study Protocol and Measurements. The study staff consisted of two physician investigators who were knowledgeable in the administration of the VAS. They worked random schedules that included day, evening, night, weekday and weekend shifts. All patients that presented with HF when a study staff member was on duty were assessed for eligibility. After enrollment the study staff documented each patient's age, gender, ethnicity, Killip classification based on explicit criteria, and disposition.15 The disposition decision was made by the treating physician and included admission to a floor, telemetry, step-down, or intensive care unit bed or home.
Prior to initiation of therapy the patient was asked to indicate his or her current degree of breathing difficulty on the dyspnea VAS by making a mark on a 10 cm uncalibrated horizontal line. The left end of the line is labeled “I Can Breathe As I Normally Do” and the right end of the line is labeled “I Can't Breathe At All” (Figure). The patient's mark was then converted to a score by measuring the distance form the left end of the line to the nearest 1 mm. Therefore, a dyspnea VAS score of 0 corresponds to the patient's subjective feeling of “I Can Breathe Normally” and a dyspnea VAS score of 100 corresponds to “I Can't Breathe At All.”
Following this initial assessment, serial VAS measurements were taken at 20-minute intervals, without reference to previous measurements. At the same time each VAS dyspnea score was obtained, patients noted whether they were experiencing 1) “much less difficulty breathing, 2) a little less difficulty breathing, 3) about the same difficulty breathing, 4) a little more difficulty breathing, or 5) much more difficulty breathing” on a 5-point Likert scale, compared with the most recent measurement. This process was repeated until the patient was no longer dyspneic, was discharged, or a total of 2 hours had passed from the first measurement.
Data Analysis. Change in VAS dyspnea score, the principal measure of interest, was defined as the difference between the index VAS score and the immediately following VAS score. The minimum clinically meaningful change in VAS dyspnea score was defined as the mean difference between the first and second VAS scores among patients that reported experiencing either “a little less difficulty breathing” or “a little more difficulty breathing.”9,12,16,17
The two-sample t test was used to compare the mean change in VAS for patients reporting no change in breathing difficulty with those that reported a clinically meaningful change in their breathing difficulty (i.e., “a little less difficulty breathing” or “a little more difficulty breathing”). Patients were divided into two groups based on the index VAS score to determine if this initial VAS measurement influenced the observed mean change associated with a clinically meaningful improvement. Patients with an index VAS of ≤50 mm were compared with those that placed an initial at a position >50 mm. The pooled two-sample t test was used to determine if differences existed between these two cohorts. The mean and 95% confidence interval (CI) are reported to reflect the central tendency of the change in VAS. Data were collected on standard reporting forms and entered into Microsoft Excel (Microsoft Corp., Redmond, WA) and analyzed with SAS Version 8.0 (SAS Institute, Inc., Cary, NC). Emory University's Institutional Review Board approved the study protocol.
A total of 79 patients were enrolled in the study. Five were excluded due to inability to use the VAS; after instruction in use of the VAS, these patients could not adequately perform the task. The remaining 74 patients were included in the data analyses. The mean (SD) patient age was 55.4 years (13.2). Fifty-three percent of the patients were male. African Americans represented 94% of the study population. Fifty-one percent of the patients were classified as Killip II. At disposition 16%, 38%, 20% of the patients were admitted to the intensive care unit, telemetry bed, or were discharged, respectively.
The mean (SD) index VAS was 62.2 mm (29.9 mm), with a median (interquartile range [IQR] 25%, 75%) of 66.0 mm (IQR 39 mm, 88 mm) for all enrolled patients. The median (IQR 25%, 75%) Killip score was 2 (IQR 1, 3). Between the first and second VAS measurement, 12 patients (16%) reported “much less difficulty breathing,” 32 (43%) reported “a little less difficulty breathing,” 26 (35%) reported that their breathing was “about the same,” and 4 (5%) reported “a little more difficulty breathing.” None of the patients reported “much more difficulty breathing.” Among study patients who were hospitalized, 43 (88%) had a final diagnosis of HF, 11 (4%) possible HF, and 4 (8%) another pulmonary process.
The mean change in VAS score associated with a minimum clinically significant change in breathing difficulty was 21.1 mm (95% CI, 12.3 mm–29.9 mm). The mean change in VAS score for those that reported “about the same difficulty breathing” was 2.7 mm (95% CI, −4.2 mm −9.5 mm) (Table I). The mean change in VAS was significantly different for the two groups of patients (t=3.39, p=0.0012).
Table I. Comparison of Mean Change in Dyspnea VAS Between Patients Reporting No Change vs. Those Reporting a Clinically Significant Change in Breathing Difficulty (i.e., “A Little Less Difficulty Breathing” or “A Little More Difficulty Breathing”)
95% Confidence Interval
About the same difficulty breathing
(−4.2 mm to 9.5 mm)
A little less difficulty breathing or a little more difficulty breathing
(12.3 mm to 29.9 mm)
*The mean change in visual analog scale (VAS) was significantly different for the two groups of patients: Satterthwaite two-sample t test, t=3.39, p=0.0012.
For those reporting a minimum clinically significant change in breathing difficulty, a statistical difference in the mean change in VAS scores based on location of the index VAS was observed. Patients that recorded an index VAS >50 mm (n=22) had a mean change of 30.9 mm (95% CI, 19.1 mm–;42.6 mm) while those with an index VAS ≤50 mm (n=14) had a mean change of 5.7 (95% CI, −3.6 mm–15.0 mm) (Table II). This difference was statistically significant (t=−3.19, p=0.0031).
Table II. The Mean and 95% Confidence Interval of VAS Change Among Those Reporting “A Little Less Difficulty Breathing” or “A Little More Difficulty Breathing”
95% Confidence Interval
0 mm–50 mm
(−3.6 mm to 15.0 m)
51 mm–100 mm
(19.1 mm to 42.6 mm)
*The mean change in visual analog scale (VAS) was significantly different for the two strata of Index VAS: pooled two-sample t test, t=−3.19, p value=0.0031.
No statistical difference was detected in patients who reported no change in symptomatology, in patients with a recorded index >50 mm, and in patients with an Index VAS ≤50 mm (Table III).
Table III. The Mean and 95% Confidence Interval of VAS Change Among Those Reporting “About the Same Difficulty Breathing”
95% Confidence Interval
0 mm–50 mm
(−14.2 mm to 9.12 mm)
51 mm–100 mm
(−3.6 mm to 14.4 mm)
*The mean change in visual analog scale (VAS) was not significantly different for the two strata of Index VAS: pooled two-sample t test, t=−1.15, p value=0.2595.
In this study, we found that for all patients presenting with dyspnea, a change in VAS of 21.1 mm corresponded with a patient's perception of a meaningful change in dyspnea. More important, we noted that this perception was not uniform along the VAS. Patients with more severe perception of dyspnea (as indicated on the index VAS) on presentation had a greater change in VAS that they perceived as meaningful (i.e., “a little less difficulty breathing” or “a little more difficulty breathing”).
This difference in dyspnea perception on the basis of the degree of dyspnea is not unique to dyspnea and has been noted in a study using a VAS for pain.18 It is not surprising that a patient who has a higher degree of dyspnea would require a greater change in dyspnea before they perceive a clinically meaningful change. Some evidence exists that asthma patients with preexisting airflow obstruction perceived breathlessness only at a considerably higher degree of obstruction than patients with prior airway obstruction.5 This suggests a decreased sensitivity to airway obstruction which may have effected the perception of our patients' dyspnea over the clinical spectrum. Another potential explanation for the larger degree of change is that patients with more severe dyspnea at baseline may have been treated more aggressively and had a more rapid decline in their dyspnea.
An important component of using a VAS to ascertain dyspnea is whether it correlates to objective measures of airway resistance. In this study of HF patients no objective gold standard exists in the ED to make this determination. This relationship has been studied in asthmatic patients.3–5,13 These studies did reveal a correlation between the perception of breathlessness and objective measures of airway resistance. Extending these findings to HF patients should be done cautiously since these studies were performed on asthma patients, not in the ED setting, and there were typically large variations in breathlessness scores between the patients.
The methodology we employed to determine the VAS dyspnea score that is associated with a self-perceived change in dyspnea has been previously used to identify the minimum change in VAS associated with meaningful improvement in pain,9–12,16,17 dyspnea associated with asthma,13 and wound cosmesis.14 Two independent studies that used VAS to measure pain estimated the median “meaningful” change to be between 10 and 11 mm.9,10, Karras and colleagues13 analyzed serial VAS scores from patients with dyspnea from an acute exacerbation of asthma and determined that the mean change in score when patients reported feeling “a little better” was 22 mm (95% CI, 11 mm–34 mm).
Our study is limited in several respects. Most notably, it is based on a relatively small sample of patients drawn from a single, large, inner-city hospital ED. The study population was predominantly African American and younger on average than HF patients nationwide.19 In one study designed to determine if African-American and white patients perceive breathlessness differently during induced bronchoconstriction, it was noted that the African-American patients used different descriptors for breathlessness, had a lower perception of breathlessness with a 20% fall in FEV1, and VAS was a more sensitive measure to assess the changes in perception of breathlessness.20 Larger and more demographically diverse groups of patients will be needed to confirm that the median change in VAS dyspnea scores reported here as meaningful does not vary in different populations. In addition, we did not enroll any Killip IV patients in this pilot study. Although Killip IV patients were not automatically excluded, it would have been quite difficult to collect serial VAS scores from a patient in cardiogenic shock.
In our ED, the median change in VAS score associated with a perceptible change in HF-associated dyspnea was 21.0 mm on a 100 mm scale. The degree of change associated with a meaningful change in symptoms varied directly with the patient's perception of the degree of dyspnea. This information provides greater insight as to the value of using VAS dyspnea scores for HF.
Acknowledgment and Disclosure: The authors would like to thank Arthur Kellermann, MD, MPH for his continuous mentorship and support during this project. This paper was presented at the Society for Academic Emergency Medicine Annual Meeting, Boston, MA, May 1999.