Development of Self‐Administered Versions of Modified Baseline and Transition Dyspnea Indexes in COPD

Abstract

In this study we developed self‐administered versions of modified baseline and transition dyspnea indexes and compared the scores obtained by this method with the mean value obtained by two trained interviewers. Twenty‐five patients (14 males/11 females) with chronic obstructive disease who had a chief complaint of “breathlessness” were tested. Age was 66 ± 11 years; forced expiratory volume in one second was 48 ± 23% predicted. The baseline total scores were 5.0 ± 1.8 for the interviewers and 5.4 ± 2.0 for the self‐administered method. For the baseline dyspnea scores the correlations were 0.83 (p < 0.0001) between self‐administration and the mean value of two interviewers and 0.75 (p < 0.0001) between the two interviewers. The transition total scores, obtained an average of 102 days (range, 7–377 days) later, were − 0.1 ± 3.0 for the interviewers and − 0.4 ± 3.0 for the self‐administered method. For the transition dyspnea scores the correlations were 0.94 (p < 0.0001) between self‐administration and the mean value of two interviewers and 0.83 (p < 0.0001) between the two interviewers. The self‐administered dyspnea scores had similar correlations with measures of lung function as did the interview dyspnea scores. We conclude that self‐administered versions of the modified baseline and transition dyspnea indexes provide comparable scores as those obtained by trained and experienced interviewers. The advantages of the self‐administered versions include standardized methodology and computerized scoring.

• Measurement of dyspnea
• Self‐administered computerized method
• Baseline dyspnea index
• Transition dyspnea index
• Chronic obstructive pulmonary disease

Introduction

Dyspnea is the major reason that patients with chronic obstructive pulmonary disease (COPD) seek medical attention. Accordingly, the ability to measure or quantify this “key” symptom has become increasingly important, especially to evaluate the efficacy of specific treatments [[1]][[2]]. In 1959 the five point Medical Research Council scale [[3]] was developed that allowed patients to grade their breathlessness based on a single dimension (i.e., daily tasks). Subsequently, instruments were created to expand the consideration of factors that influence the sensation of dyspnea as experienced by patients [[4]].

In 1984 Mahler and colleagues [[5]] developed the Baseline Dyspnea Index (BDI) as a discriminative instrument to measure dyspnea at a single point in time, and the Transition Dyspnea Index (TDI) as an evaluative instrument to measure changes in dyspnea from the baseline state. These instruments consist of three components—functional impairment, magnitude of task, and magnitude of effort—that contribute to breathing difficulty based on activities of daily living. An interviewer asks the patient various questions as part of a medical history, and then selects a grade (score) using specific criteria for each of the three components. Total scores for the BDI (range, 0 to 12) and for the TDI (range, − 9 to + 9) are obtained by adding the scores for each of the three components. The interview method was decided upon to allow a person with medical training or background the opportunity to grade breathlessness in a simple and brief encounter. Over the past 20 years the psychometric properties of the BDI/TDI have been well established [[6]][[7]][[8]][[9]][[10]]. Moreover, the BDI/TDI instruments have been used widely in clinical trials evaluating zpharmacotherapy [[11]][[12]][[13]], pulmonary rehabilitation [[14]][[15]][[16]], inspiratory muscle training [[17]][[18]], and lung volume reduction surgery [[19]].

However, a criticism of the BDI/TDI has been that the interview process may introduce bias and interpretation by the interviewer rather than simply obtaining a dyspnea rating from the affected patient. To address this issue, we have developed and tested self‐administered versions of the BDI/TDI. This standardized approach has been computerized to ease the collection of the patients' responses and to facilitate statistical analyses. This report describes our initial experience with the self‐administered computerized (SAC) versions of the BDI/TDI. The hypothesis of this study was dyspnea scores for the BDI and TDI obtained by the SAC methods would be similar to and correlate highly with values recorded by experienced interviewers.

Materials and Methods

Subjects

Twenty‐five patients were recruited from the out‐patient clinic at Dartmouth‐Hitchcock Medical Center for participation in this preliminary study of the SAC BDI/TDI. Inclusion criteria were: a diagnosis of COPD based on the ATS standards [[20]]; primary complaint of exertional breathlessness; ability to read and understand English; ability to use the computer display to render judgments; and be expected to return to the clinic for follow‐up testing. Exclusion criteria were: any individuals with a mental or physical disability which precluded their understanding or performing the task (e.g., unable to read the words on the computer screen because of poor eyesight).

Development of the SAC Versions

The development of the SAC versions of the BDI/TDI was a multi‐step process that involved numerous discussions with individual patients and investigators, subsequent revisions, and pilot testing. Initially, two of the investigators (DAM and JCB) created a software program for the SAC versions of the BDI and TDI. For each possible grade for each of the three components of the BDI/TDI, a descriptive sentence was written that included the unique features of that grade.

Three major changes were made in the SAC versions of the BDI/TDI based on comments and feedback from various individuals and organizations (Table 1). One major change was made from the original BDI for the component “magnitude of task”. Based on our discussions with patients who had various respiratory diseases including asthma, COPD, and interstitial lung disease, it was evident that few, if any, patients actually attempted to climb three or more flights of stairs. If presented with the choice of climbing three flights of stairs, the patients indicated that they would take an elevator rather than attempt climbing the stairs. In contrast, many patients stated that they frequently climbed one flight of stairs at home or in public buildings. Thus, for grade 1, considered “slight” dyspnea under magnitude of task, we changed “climbing three of more flights of stairs” to “climbing one or more flights of stairs.” For grade 2, considered “moderate” dsypnea, we changed “climbing less than three flights of stairs” to “climbing less than one flight of stairs.” All of the patients with whom we discussed this modification agreed that the task of climbing up one flight, rather than three flights, of stairs was more relevant to their daily activities.

Table 1.  Major Changes Made in the SAC Versions of the BDI and TDI from the Original Instruments

Baseline Dyspnea Index
• Changed criterion in magnitude of task from “climbing three flights of stairs” to “climbing one flight of stairs.”
Transition Dyspnea Index
• Provided insert on the computer screen of the descriptor that the patient selected for the corresponding component of the BDI in order to remind the patient of his/her original dyspnea status (see Figure 1).
• Created a bidirectional visual analog scale for each component of the TDI. Patient pressed an up or down key to move a vertical bar up for improvement or down for deterioration in dyspnea compared with the baseline dyspnea status (see Figure 1).

A second major change was made in the TDI. For each component of the TDI the corresponding grade and descriptor selected by the patient for the BDI was displayed as an insert on the computer screen. This descriptor served as a reference in order to remind the patient of his/her choice at the baseline visit.

A third modification was made in the method whereby the patient selected a score for each component of the SAC version of the TDI. A limitation of the original TDI was that only whole integer options were available for rating changes in dyspnea (seven possible grades ranging from − 3 to + 3). To remedy this, we created a bidirectional visual analog scale such that patients could indicate changes in dyspnea from the baseline state for each component of the TDI (Figure 1). The descriptors “minor,” “moderate,” and “major” were positioned next to corresponding numbers 2, 4, and 6, respectively, for either improvement or deterioration. By pressing an up or down arrow on the keyboard, the patient moved a vertical bar adjacent to the numbers and descriptors to rate changes in dyspnea from the baseline state. The SAC TDI total score was divided by two in order to compare directly with the TDI total score obtained by the interviewers.

Figure 1. Example of the self‐administered computerized version for the “functional impairment” component (labeled “activities” on the graph) of the transition dyspnea index (TDI). By pressing the up or down arrow on the keyboard, the patient could move the vertical bar up (for improvement) or down (for deterioration) to represent a change in dyspnea. An insert is present in the left portion of the figure which reminds the patient of his/her selection at baseline. (Full color version available online.)

After the SAC versions were completed we then created a practice question about tiredness which the patient completed prior to both the BDI (tiredness at baseline) and the TDI (change in tiredness from the baseline state). The purpose of the practice session was to familiarize the subject with the computer and procedure for selecting a grade.

Finally, pilot testing was performed in 10 patients with a chronic respiratory disease, all of whom experienced exertional breathlessness. Each patient completed the SAC versions of the BDI and, at a later date, the TDI. At each of these visits an investigator interviewed the patient after he/she completed the SAC version in order to inquire about ease of use and to obtain feedback on any recommendations for changing the measurement procedure. No additional changes were made in the SAC instruments as a result of the pilot testing.

Throughout the development of the SAC versions we periodically reviewed and discussed among ourselves the grades and corresponding descriptors. Each investigator (except for JCB) had extensive experience with interviewing patients and recording scores for dyspnea using the original BDI/TDI.

Procedures for Evaluating the SAC Versions of the Modified BDI/TDI

The following procedures were approved by the Institutional Review Board. After giving verbal consent, each patient initially completed a practice session to grade the degree of tiredness on a typical day at the baseline and follow‐up visits (Table 2). To accomplish this task at the baseline visit the patient positioned the cursor, which was visible on the computer screen, in a box next to the grade and then clicked on the mouse. This action resulted in the appearance of a red ‘X’ in the box. The selection could be changed by clicking on another choice. At all times the technician was available to assist the patient, although no patient required assistance once the practice trial was completed. At the follow‐up visit the patient pressed the up or down arrow on the keyboard to move a vertical bar positioned adjacent to a bidirectional visual analog scale for rating changes in tiredness (as described above). Upon completion of the practice, the grades and descriptors for the three components of the SAC BDI appeared sequentially on the computer screen (Table 3). The same procedures were used by the patient as in the practice for grading dyspnea at baseline and at the transition time period.

Table 2.  Grades and Descriptors for the Practice Session Related to the Experience of Being Tired

Grade	Descriptor
NONE	I have not felt tired.
SLIGHT	I have felt tired occasionally.
MODERATE	I have felt tired about half the time.
SEVERE	I have felt tired much of the time.
VERY SEVERE	I have felt tired all of the time.

Table 3.  Grades and Descriptors for each Component of the Self‐Administered and Computerized Version of the BDI

Baseline dyspnea index
Functional impairment
NONE	I can perform usual activities without shortness of breath.
SLIGHT	I am short of breath in at least one activity, but have not completely stopped any activity.
MODERATE	I can't do at least one usual activity due to shortness of breath.
SEVERE	I can't do most or all activities due to shortness of breath.
VERY SEVERE	I can't do most or all activities and would be unable to work due to shortness of breath.
Magnitude of task
NONE	I am short of breath only with extraordinary tasks, such as carrying very heavy loads on level ground, lighter loads uphill, or running. No shortness of breath with ordinary tasks.
SLIGHT	I am short of breath only with major tasks, such as walking up a steep hill, climbing one or more flights of stairs, or carrying a moderate load on level ground.
MODERATE	I am short of breath with average tasks, such as walking up a gradual hill, climbing less than one flight of stairs or carrying a light load on the level.
SEVERE	I am short of breath with light tasks, such as walking on level ground or standing.
VERY SEVERE	I am short of breath at rest, while sitting, or lying down.
Magnitude of effort
NONE	I am short of breath only with the greatest imaginable effort.
SLIGHT	I am short of breath with extreme effort. I must pause only if the activity requires extreme effort.
MODERATE	I am short of breath with moderate effort. I need to pause and it takes longer to complete the task than the average person.
SEVERE	I am short of breath with little effort. I need to pause frequently and it takes at least 50 % longer to complete the task than the average person.
VERY SEVERE	I am short of breath with no effort.

The practice sessions and the SAC BDI and TDI were presented on a desktop PC computer located in the pulmonary function testing laboratory. One of the investigators logged onto the computer, called up the appropriate program and initiated the procedure. The patient was seated in front of the computer, and all instructions were given on the screen.

For each of the three components of the BDI, the subject was required to select one of five available choices (grades with descriptors) displayed on the computer screen. Afterwards, the patient was questioned separately by two of five possible interviewers who provided grades for dyspnea on the BDI based on the individual's responses during the interview. When patients returned for a follow‐up out‐patient visit they were asked to complete the practice session and then the SAC TDI. Next, they were questioned by the same two individuals who had previously interviewed the patient to obtain TDI total scores compared to the baseline state. The order of testing (SAC first; interview second) was based on our previous experience with the BDI which showed no difference between the BDI total score between the SAC version and the mean of two interviewers when the SAC was administered first, but a difference between the BDI total score when the interviews occurred first [[21]].

At each visit the patient performed spirometry using a standard system (Collins model CPL; Braintree, MA). The highest values for forced vital capacity (FVC) and forced expiratory volume in one second (FEV₁) were selected from a minimum of three FVC maneuvers. Inspiratory capacity (IC) was measured, and the highest value of three efforts was selected. Predicted values were taken from Crapo et al. [[22]]. The order of completion of the SAC BDI or TDI and pulmonary function testing was based on convenience for the individual patient.

Data Analysis

For the BDI the patient's selections could range from a score of 0 (severe) to 4 (unimpaired) for each of the three categories (functional impairment, magnitude of task, magnitude of effort). The three ratings for each category were summed to obtain an overall total score (range, 0 to 12). The lower the total score, the worse the severity of breathlessness.

There was no attempt to obtain TDI ratings at any specific time period following the recordings of the BDI scores. For the TDI the grade was determined based on the location of the edge of the bar on the − 6 to + 6 visual analog scale (one decimal place resolution) as adjusted by the patient for each component (Figure 1), and then divided by two for comparison purposes; or by the grade assigned by the interviewer (range, integers from − 3 to + 3) for each component. The grades on each of the three components were added to obtain the TDI total score.

Linear regression analysis was performed to determine the relationship between dyspnea scores obtained by the SAC versions and the mean of two interviewers. Pearson correlation coefficients were used as a measure of the goodness of fit. Values are presented as mean ± SD. A p value < 0.05 (two‐tailed test) was considered statistically significant.

Results

Twenty‐five patients (14 males; 11 females) with COPD who had a chief complaint of breathlessness were tested. Age was 66 ± 11 years. As a group, the patients had moderate dyspnea with activities of daily living at baseline (SAC BDI score 5.4 ± 2.0), and moderate to severe lung impairment (FEV₁ = 48 ± 23% predicted). All patients completed the SAC BDI in 2–3 minutes. Baseline FVC was 78 ± 18 % predicted, and IC was 85 ± 20 % predicted. Correlation coefficients for the BDI were: 0.83 between the mean interview score and the SAC version (p < 0.0001) (Figure 2) and 0.75 between the two interviewers (p < 0.0001). Correlations between the BDI total scores and lung function are shown in Table 4. Significant correlations were observed between: SAC BDI and FVC % pred (r = 0.41; p = 0.04); SAC BDI and IC % pred (r = 0.47; p = 0.02); and between Interviewer BDI and IC% pred (r = 0.41; p = 0.05).

Table 4.  Correlations Between the BDI/TDI Total Scores and Lung Function

Lung function	SAC BDI	Interviewer BDI
FVC	0.41; p = 0.04	0.33; p = 0.11
FEV1	0.28; p = 0.17	0.28; p = 0.17
IC	0.47; p = 0.02	0.41; p = 0.05
Lung function	SAC TDI	Interviewer TDI
Δ FVC	0.40; p = 0.05	0.37; p = 0.07
Δ FEV1	0.38; p = 0.06	0.45; p = 0.02
Δ IC	0.39; p = 0.06	0.45; p = 0.02

Lung function measures are percent predicted.

Correlation values are Pearson coefficients.

*Mean value of two interviewers.

Figure 2. Relationship between baseline dyspnea index (BDI) total scores represented as mean value of the scores obtained by two interviewers (abscissa) and the self‐administered scores (ordinate) in 25 patients with chronic obstructive pulmonary disease. The line represents the best linear fit.

Patients returned an average of 102 days (range, 7–377) after the initial visit. Although there was minimal change in the mean dyspnea scores (SAC TDI = − 0.4 ± 3.0), a wide range of TDI scores was observed (Figure 3). All patients completed the SAC TDI in 2–3 minutes. The changes in lung function values at the transition time period were: ΔFEV₁ = 0.4 ± 10.1 % predicted; Δ FVC = 1.0 ± 9.9% predicted; and ΔIC = 0.7 ± 15.9% predicted. The correlations for the TDI between the mean interview score and the SAC version was 0.94 (p < 0.0001), and between the two interviewers was 0.83 (p < 0.0001). Significant correlations were observed between: SAC TDI and ΔFVC % predicted (r = 0.40; p = 0.05); interviewer TDI and Δ FEV₁ % predicted (r = 0.45; p = 0.02); and interviewer TDI and ΔIC % predicted (r = 0.45; p = 0.02) (Table 4).

Figure 3. Relationship between transition dyspnea index (TDI) total scores represented as mean value of the scores obtained by two interviewers (abscissa) and the self‐administered scores (ordinate) in 25 patients with chronic obstructive pulmonary disease. The line represents the best linear fit.

Discussion

The rationale for the development of the SAC versions of the BDI and TDI was to provide a standardized approach to the measurement of dyspnea. Although the initial instruments created to measure breathlessness, such as the MRC scale and the oxygen‐cost diagram, were self‐administered, these and similar instruments only evaluated how the magnitude of certain tasks caused difficulty breathing [[4]][[6]]. These unidimensional scales were not particularly responsive to detect or demonstrate changes with therapy. Two additional dimensions which affect breathing difficulty—functional impairment and magnitude of effort—were included in the BDI/TDI in order to expand the measurement of dyspnea [[5]]. An interview process was used so that a physician, nurse, or technician could obtain a grade or score for dyspnea as part of the routine medical history. Although this approach enabled more detailed questioning of the patient, it also required some judgment by the interviewer in order to grade dyspnea for each component of the BDI/TDI.

Over the past 20 years the BDI/TDI have become widely used in numerous clinical trials evaluating therapy in patients with chronic respiratory disease [[1]][[2]][[4]][[7]][[9]][[10]][[11]][[12]][[13]][[14]][[15]][[16]][[17]][[18]][[19]]. To enhance the measurement accuracy of the BDI/TDI in multicenter clinical trials, it has been necessary to train individuals in the interviewer process for grading dyspnea. At least four pharmaceutical companies have made videos in order to provide instructions for interviewers on using the BDI/TDI and also to demonstrate an actual interview with a fictitious patient.

To further improve the measurement of dyspnea we modified the BDI/TDI into self‐administered instruments so that an individual patient could provide direct ratings of breathlessness based on activities of daily living. The overall changes in the SAC versions from the original BDI/TDI are summarized in Table 1. Although conversion to a self‐administered and computerized format clearly altered the original instruments, we maintained the three unique dimensions (functional impairment; magnitude of task; and magnitude of effort) of the BDI/TDI as well as the general contents. Williams et al. [[23]] used a similar approach in the development of a self‐reported Chronic Respiratory Questionnaire, a disease specific health status instrument which requires an interview for scoring.

The two major findings of this preliminary study in 25 patients with COPD are: 1) all patients were able to successfully complete the SAC versions of the BDI/TDI within 2–3 minutes for each instrument; and 2) BDI and TDI total scores from the SAC versions were comparable to dyspnea total scores obtained by the interview process.

Prior to grading the severity of dyspnea on the SAC versions of the BDI/TDI, each patient completed a practice session. The patient was required to use the computer, under the supervision of a technician, to select grades for tiredness. This required use of the computer mouse for the BDI and the keyboard for the TDI. As expected, our patients with COPD were elderly with an age of 66 ± 11years. We estimated that about two‐thirds of our patients had some previous experience using a computer. Based on comments from individual patients the practice session proved to be most helpful as it enabled each subject to use the computer in a manner similar to the method for selecting grades for dyspnea.

To evaluate the SAC versions we compared the patient‐reported dyspnea scores with those obtained by trained and experienced interviewers (i.e., the “gold standard” for the original BDI/TDI). Figures 2 and 3 show the accuracy of the SAC versions of the BDI and TDI throughout the range of possible scores compared with the mean scores obtained by the interviewers. The correlations between the SAC dyspnea scores and the interviews (BDI: r = 0.83; TDI: r = 0.94) were slightly better than the correlations between the two different interviewers (BDI: r = 0.75; TDI: r = 0.83). Not only are these correlations high, but the two sets of absolute values (SAC and interviewer scores) are close to the line of identity (not shown on the figures). In the initial testing of the BDI in 1984 agreement was also high between different interviewers for the BDI and TDI total scores [[5]].

In this study five experienced individuals interviewed the patients in order to obtain ratings of dyspnea. The interviewers included two physicians, two cardiopulmonary technicians, and a respiratory therapist/research assistant. This approach was used so that the same two individuals would be available to interview the patient at both the baseline and transition time periods. At the baseline visits the interviewers used a worksheet to record various activities that the individual reported affected his/her breathlessness; at subsequent visits the interviewer referred back to these notes when inquiring about any changes in dyspnea. We believe that these features enhanced the accuracy and consistency of the dyspnea scores among the interviewers.

As expected, the dyspnea scores obtained by either SAC versions or the interview process correlated only modestly with measures of lung function (Table 4). The magnitudes of the correlations were similar between the SAC versions and the interviewer BDI/TDI. Moreover, these correlations are comparable to those observed in other populations of patients with COPD [[5]][[6]][[7]][[8]]. Once again, these findings emphasize that dyspnea scores represent a different construct than do results of lung function testing.

In summary, we believe that the SAC versions of the modified BDI/TDI provide several advantages over the interviewer method for grading dyspnea in patients with COPD. First, the SAC versions use standardized methodology for each patient. This eliminates any judgment required by the individual interviewer when selecting a dyspnea score based on the patient's responses. Second, these instruments are computerized which should improve the collection and analysis of the results compared with the interviewer method. Third, the SAC TDI is a truly continuous measure of any change in dyspnea as the patient presses a key to indicate the magnitude of improvement or deterioration along a visual analog scale. With the SAC version of the TDI there are numerous possible scores between integers, whereas with the interviewer method of the TDI the only possible scores are whole integers. Although the SAC BDI/TDI compare well with the interviewer administered BDI/TDI, they should not be used interchangeably. Further testing will be required in additional patients to more completely evaluate the psychometric characteristics and clinical utility of the SAC versions of the BDI/TDI.