INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In patients with chronic obstructive pulmonary disease (COPD), impaired pulmonary function and limited exercise capacity may result in reduced quality of life (1). Previous studies have shown that the relationship between respiratory impairment and health-related quality of life (HRQL) measurements is generally weak (2). It is therefore important to evaluate HRQL in patients with COPD, in addition to evaluating pulmonary function and exercise tolerance. Moreover, a questionnaire evaluating HRQL should have two key properties: one is a discriminative property by which HRQL can be described and distinguished in a cross-sectional study, and the other is an evaluative one by which changes in HRQL over time can be detected in a longitudinal study. It is generally accepted that a disease-specific questionnaire is superior to a generic one in evaluating HRQL among patients with a specific disease (5). In patients with COPD, for example, the Chronic Respiratory Disease Questionnaire (CRQ) (6), a disease-specific questionnaire, was found to be more discriminative and useful than the Nottingham Health Profile, a generic questionnaire (7).

Three major disease-specific HRQL questionnaires for patients with COPD have been introduced: the St. George's Respiratory Questionnaire (SGRQ) (4), the Breathing Problems Questionnaire (BPQ) (8), and the CRQ (6). These three questionnaires have already proved to be precise, valid, and responsive, and they have been used for the evaluation of the HRQL of patients with COPD in several clinical studies (4, 8, 9). However, a comparison of these three questionnaires in the same patients has yet to be done. In this regard, the relevant question is which questionnaire should be chosen for a specific aim. The purpose of the present cross-sectional study was to make comparisons between the SGRQ, the BPQ, and the CRQ in their discriminative properties, and to clarify the characteristics of each HRQL questionnaire. We therefore evaluated correlations between the three questionnaires and various parameters, and the contributions of pulmonary function, exercise capacity, dyspnea ratings, and psychologic status to HRQL in patients with COPD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Patients with stable COPD who visited the outpatient clinic at the Chest Disease Research Institute, Kyoto University, were recruited. The diagnosis of COPD was based on the definition provided by the American Thoracic Society (10). Entry criteria for the study were: (1) chronic airflow limitation, (2) smoking history of more than 20 pack-years, (3) no history suggestive of asthma, (4) no exacerbation of airflow limitation in the preceding 6 wk, and (5) no changes in treatment regimen during the preceding 4 wk. All patients had more than 6 mo of outpatient management before entry into the study to avoid significant changes in the HRQL brought about by new medical interventions. Chronic airflow limitation was defined as a maximal FEV₁/FVC ratio, with FVC less than 0.7 and FEV₁ less than 80% of the predicted value for all measurements made during the prior 6 mo. All eligible patients finished the following examinations in the same day, including pulmonary function tests, progressive cycle ergometer testing, assessment of HRQL, and assessment of dyspnea, anxiety, and depression. Verbal informed consent was obtained from all patients.

Age, height (cm), weight (kg), percent of ideal weight, and smoking habits (current or former), including pack-years, were assessed in each patient. Percent of ideal weight was calculated as weight/[(height  100) × 0.9] × 100 (%) and was expressed as %weight. Smoking habits were confirmed by measuring carbon monoxide (CO) levels in expiratory gas and by measuring plasma cotinine levels. The CO level was measured using a smokerlyzer (Micro Smokerlyzer^®; Bedfont Technical Instruments Ltd, Kent, UK). Plasma continine level was measured by gas liquid chromatography. Smoking habit was validated by a CO level greater than 10 ppm or a plasma cotinine level greater than 100 ng/ml.

Measurement of Outcome

Pulmonary function tests. Pulmonary function tests were established as much as possible such that the conditions of the preassessment period were reproduced. Inhaled and oral corticosteroids and theoplylline were administered in the same doses throughout the study. Inhalation of -receptor agonists and anticholinergic drugs was withheld for at least 12 h before pulmonary function tests were performed. FEV₁ and FVC were assessed before, and 15, and 60 min after the inhalation of 400 µg of salbutamol (four puffs of Sultanol; Nippon Glaxo Ltd., Tokyo, Japan) and 80 µg of ipratropium bromide (four puffs of Atrovent; Nippon Boehringer Ingelheim Co. Ltd., Kawanishi, Japan). These doses represented the average doses of each drug that the patients regularly received. The maximal increase in FEV₁ after the inhalation of bronchodilators was used to represent the increase in FEV₁. Bronchodilators were administered by one of the investigators (T.H.) using a metered-dose inhaler (MDI) with a spacer device (InspirEase^®; Schering-Plough K. K., Osaka, Japan). All spirometric flow-volume curves were recorded according to the method described in the American Thoracic Society 1994 update (10). Patients were tested in the standing position using a spirometer (AUTOSPIRO AS-600; Minato Medical Science Co. Ltd., Osaka, Japan), which was calibrated with a 2.0-L syringe. The predicted values for FEV₁ and VC were calculated according to the proposal of the Japan Society of Chest Diseases (11). The residual volume (RV) was measured by the closed-circuit helium method, and the diffusion capacity for carbon monoxide (DL_CO) was measured using the single-breath technique (CHESTAC-65V; Chest, Tokyo, Japan). Patients with long-term domiciliary oxygen therapy (LTOT) were advised to stop inhaling oxygen for more than 12 h prior to their visits to our institute.

Symptom-limited progressive cycle ergometer exercise test. The progressive exercise test to a symptom-limited maximum was performed as described by Ikeda and colleagues (12). An electrically braked cycle ergometer (Corival WLP-400; Lode, Groningen, the Netherlands) was used to deliver the exercise work load progressively. After unloaded pedaling for 3 min, the work load was increased automatically in increments of 1 W every 3 s to the limit of tolerance. Patients maintained a pedaling frequency above 40 cycles/min throughout the test. A face mask connected to a low resistance unidirectional valve (Rudolph Face Mask Exercise Testing; Hans Rudolph, Inc., Kansas City, MO) was placed on the patient's face without leakage. Exercise data were recorded using an automated exercise testing system (Desktop Diagnostics/CPX; Medical Graphic Corp., St. Paul, MN), which converts breath-by-breath analog input to digital form in an on-line fashion. The testing system included a pneumotachograph with a gas analyzer module, and a personal computer that interfaced with the measuring instrument. Minute ventilation (E) and oxygen and carbon dioxide tensions in the expired air were determined every eight breaths, and from these measurements the mean E, oxygen uptake (O₂), and carbon dioxide production (CO₂) were rapidly calculated. Arterial oxygen saturation (Sa_O2) was measured by pulse oximetry (N-200 pulse oximeter; Nellcor Inc., Hayward, CA), and heart rate (HR) and wave form were measured by electrocardiography (Life Scope 8; Nihon Koden Co., Tokyo, Japan). At the end of each exercise test, maximal work load (Wmax) was calculated and symptoms of leg effort and breathlessness were scored with the Borg scale (0 to 10) (13), which was presented in front of the subject. Patients with untoward clinical signs or electrocardiographic changes suggestive of pathologic arrhythmia or significant myocardial ischemia were excluded from the study. All the exercise tests were performed by the same investigator (M.T.) who did not have any prior knowledge of the results of pulmonary function tests.

Assessment of HRQL. The HRQL was assessed by the Japanese version of the SGRQ (4), the BPQ (8), and the CRQ (6). Permission for translation was obtained from the original investigators. The three questionnaires were translated into Japanese in accordance with the standardized methodology. Briefly, a pulmonologist (H.K.) with 10 yr of experience first translated the original of these three questionnaires into Japanese, which were later retranslated into English by an American who does not have a background knowledge in medical science. Then a pulmonologist (K.N.) who has specialized in COPD and asthma for 15 yr translated and modified the English version into Japanese. The number of items and the dimensions of each questionnaire are described in Table 1. Scores for each questionnaire were calculated following the procedures devised by the developers of the questionnaires (4, 6, 8). SGRQ scores ranged from 0 to 100, with a zero score indicating no impairment. The BPQ was scored using a scale of 1 to 103, with lower scores indicating good quality of life. In the CRQ, each question was presented as a seven-point scale. Each dimension of the CRQ was scored as the sum of the points, and higher scores represented better health. In addition, the total score as the sum of scores of these four dimensions was calculated for comparison with other questionnaires, although Guyatt and associates have not introduced the idea of a total score in their analysis of the CRQ (6).

Assessment of dyspnea, anxiety, and depression. To assess dyspnea, the Japanese version of the Baseline Dyspnea Index (BDI) (14), the Oxygen Cost Diagram (OCD) (15) and the Medical Research Council (MRC) (16) were used. The BDI recognizes five grades for each of the following categories: functional impairment, magnitude of task, and magnitude of effort. The OCD is a visual analog scale that corresponds to oxygen requirements at different activity levels, which was represented as a value ranging from 0 to 100 with a score of 100 indicating no impairment.

The Japanese version of the Hospital Anxiety and Depression Scale (HAD) (17) is used for evaluating patients' anxiety and depression status. The HAD consists of 14 items, seven of which score for anxiety and seven of which score for depression. It was scored in accordance with the method of the original author.

Statistical Analysis

All results are presented as mean ± SD. The relationship between two sets of data was analyzed by Spearman's rank correlation test. A p value less than 0.01 was considered to be statistically significant.

Forward and backward stepwise multiple regression analyses were performed to identify the variables that could best predict HRQL (18). Variables that were judged to be important in COPD were used as independent variables for the regression analyses. Those variables were FEV₁ for airway limitation, RV/TLC for hyperinflation, DL_CO/ VA for gas exchange, O₂max for exercise capacity, BDI score for dyspnea ratings, and anxiety of the HAD for psychologic status. Dependent variables for this statistical model were the HRQL scores such as scores for the symptoms, activity, and impact dimensions, the total score for the SGRQ, the total score for the BPQ, and the scores for dyspnea, fatigue, emotional function and mastery dimensions, in addition to the total score for the CRQ.

All analyses were performed using the Statistical Package for Social Sciences (SPSS) (19).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

A total of 143 consecutive male patients was studied from our outpatient clinic at the institute between September 1995 and February 1997. Pulmonary function tests, exercise performance, and demographic details are shown in Table 2. Four patients received no medication. One hundred thirty-six patients (95.1%) were treated with inhalation of both 400 µg of salbutamol and 80 µg of ipratropium bromide more than four times a day, and three patients were treated with other -agonists and anticholinergic agents. All patients who inhaled these drugs were obliged to use a spacer device (InspirEase) regularly. In addition to inhalation agents, oral theophylline (400 to 800 mg/d) was administered to 17 patients (11.9%) and oral corticosteroids (2.5 to 10 mg/d) to seven patients (4.9%). Seventy-eight patients (54.5%) were treated additionally with high doses (1,600 to 2,400 µg/d) of inhaled corticosteroids. Seven patients were managed with LTOT.

Internal Consistency and the Distribution of Test Scores

The internal consistency of each questionnaire was assessed with Cronbach's coefficient  (Table ) (20). The internal consistency of the total score of the SGRQ was high ( = 0.94) and that of the three dimensions ranged from  = 0.76 to  = 0.97. The internal consistency of the emotional function dimension of the CRQ was higher ( = 0.90) than that of the dyspnea, fatigue, and mastery dimensions (ranging from  = 0.76 to 0.85). The frequency distribution histograms of the three HRQL scores are shown in Figure 1.

View larger version (17K): [in this window] [in a new window]   Figure 1.   Frequency distribution histograms of total scores of health-related quality of life (HRQL) questionnaires. In the St. George's Respiratory Questionnaire (SGRQ) and the Breathing Problems Questionnaire (BPQ), higher scores indicate poor health. The Chronic Respiratory Disease Questionnaire (CRQ) was scored using a 20 to 140 scale, with higher scores indicating good quality of life. Numbers in parentheses represent the theoretical score range.

Correlations between Questionnaires

Spearman's rank correlation coefficients (Rs) between the questionnaires are shown in Table 3. With regard to total scores, the SGRQ showed a significant correlation with the BPQ and the CRQ (Rs = 0.86 and 0.74, respectively), and there was also a significant relationship between the BPQ and the CRQ (Rs = 0.75). The total score for the BPQ showed a significant correlation with the SGRQ score, especially in the activity (Rs = 0.81) and impact (Rs = 0.80) dimensions. Among the dimensions of the CRQ, the fatigue dimension had the highest correlation with the dimensions of the other two questionnaires. The correlation between symptoms of the SGRQ and mastery of the CRQ was weakest (Rs = 0.31, p < 0.01) and all other correlations between dimensions were found to be significant (Rs = 0.35 to 0.81, p < 0.001).

Relationship between HRQL and Physiologic Factors

Spearman's rank correlation coefficients were obtained in order to investigate relationships between the scores of the HRQL questionnaires and various factors as shown in Table 4. The total score and the activity dimension of the SGRQ were significantly correlated with VC, FEV₁, DL_CO, DL_CO/VA, Wmax, O₂max, MRC, Borg score at maximal exercise, BDI, OCD, and anxiety and depression based on the HAD scale (Rs = 0.23 to 0.74). The BPQ was not significantly correlated with VC, and the total score of the CRQ was not correlated with VC, FEV₁, DL_CO/VA or the maximal Borg score. The BDI score had the highest correlation with the total scores of the three questionnaires (Rs = 0.58 to 0.76).

Stepwise Multiple Regression Analyses

Forward and backward stepwise multiple regression analyses were used to identify variables that could best predict HRQL (Table 5). The BDI score, anxiety of the HAD scale, and O₂max accounted for 61% of the variance in the total score for the SGRQ. With the BPQ score as the dependent variable, the BDI score and anxiety of the HAD accounted for 53% of the variance. Of the total score for the CRQ, 53% could be accounted for by variables of the BDI score, and anxiety of the HAD scale. Other results of the analyses of the dimensions of the SGRQ and the CRQ as dependent variables are also shown in Table . 

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This is the first cross-sectional study to directly compare three disease-specific HRQL questionnaires, the SGRQ, the BPQ, and the CRQ, in their discriminative properties. First, a difference in frequency distributions is of interest. Whereas the SGRQ and the CRQ were normally distributed, the BPQ demonstrated its score, which was highly skewed toward the very mild end of the scale. This may suggest that the BPQ had less discriminatory power than the SGRQ and the CRQ in this study population. The skewed score of the BPQ could be explained by the observation that the questions in the BPQ seem to concentrate on the more practical aspects of social handicaps. As Hyland and colleagues (8) did not show the frequency distribution of the BPQ in their study, we were unable to compare our score with theirs. Because patients with heterogeneity of disease severity were examined in our study, it is likely that they would show a wide range of impairment in HRQL. The difference in score distributions, therefore, may imply a discrepancy in the three questionnaires with respect to describing the HRQL or in those features of the HRQL measured by each questionnaire. Scores for the three dimensions of the SGRQ in this study were lower than in previous studies (2, 21, 22), and the CRQ presented relatively higher scores than documented in the previous reports (23, 24) despite the fact that each study population had almost the same overall disease severity. The reasons for these differences may be that only those patients who were able to perform a progressive cycle ergometer test entered into the study, and that they were effectively treated with bronchodilators. Although Wijkstra and associates (9) reported that the dyspnea dimension of the CRQ showed low and unreliable internal consistency, the internal consistency of the dyspnea dimension was as high as that of the other dimensions of the CRQ in the present study. The dyspnea dimension therefore was not excluded from the total score.

With respect to relationships between quality of life and various physiologic parameters, expected information regarding quality of life scores was obtained such as the weak correlation with physiologic findings and the moderate-to-strong correlations with dyspnea ratings and psychologic status. In the SGRQ and the CRQ, each dimension showed different sets of variables in the analyses. With regard to comparisons between questionnaires, it is noteworthy that the SGRQ and the BPQ displayed almost identical sets of variables that showed significant correlations with total scores; so did the impact dimension of the SGRQ and the dyspnea dimension of the CRQ. Mild-to-moderate relationships between exercise capacity and most of the dimensions of the questionnaires were significant and relatively stronger than those demonstrated in other reports (4, 21). Dyspnea ratings correlated better with HRQL than pulmonary function test results did, as shown previously in several studies (4, 25, 26). The CRQ showed relatively lower correlation with assessment of dyspnea than did either the SGRQ or the BPQ, even though Wegner and associates (23) proved that the CRQ and dyspnea ratings comprised the same factor, as evaluated by the factor analysis method. Anxiety and depression as evaluated by HAD were significantly associated with all dimensions of the questionnaires. These correlations confirm that emotional status such as anxiety and depression plays an important role in patients with COPD, as suggested previously (4, 27, 28).

The multiple regression analyses in the present study suggest that no overall difference was found between the three questionnaires. The analyses revealed that the BDI score, the anxiety score of the HAD scale and O₂max were significant covariants of the HRQL scores. Judging from its coefficient, the dyspnea rating played a significant role in the HRQL, as demonstrated in previous studies (4, 22). Because anxiety demonstrated a relatively higher coefficient of determination in the CRQ than in the SGRQ and the BPQ, quality of life measured by the CRQ may be more influenced by psychologic status than quality of life measured by the other two questionnaires. With respect to physiologic measures, only O₂max had some contribution to the activity dimension of the SGRQ and the dyspnea dimension of the CRQ. The main reason for this may be because the questions in these dimensions concern daily activities during which patients feel or suffer from the sensation of dyspnea. FEV₁ did not account for any variance in HRQL in these multiple regression analyses, which is somewhat different from observations in previous reports. Determinants of the SGRQ score have been evaluated previously by several investigators (4, 21, 22). Ketelaars and colleagues (21) reported that FEV₁ was recognized as an independent variable of the activity dimension of the SGRQ based on the multiple regression analyses, and Ferrer and colleagues (22) described a contribution of FEV₁ to all dimensions of the SGRQ in their Spanish population. This discrepancy may be due to differences in the independent variables that were used in those analyses. Regarding the CRQ, Wijkstra and colleagues (3) conducted a regression analysis with the CRQ as the dependent variable, and the CRQ was found not to be a determinant of exercise performance. Regression analysis with the CRQ as the independent variable has yet to be performed.

Some limitations of the present study should be mentioned. Only patients who were able to perform a progressive cycle ergometer test were enrolled in the study, and therefore some of the more severe cases may have been excluded, and HRQL might have been even worse. When comparing disease-specific questionnaires, their responsiveness to change, which is an evaluative property, should be examined because one of the important reasons for developing questionnaires is their use in measuring changes over time. Unfortunately, we encountered difficulties evaluating responsiveness because the present study was performed in a cross-sectional fashion for the purpose of investigating the discriminative properties of questionnaires.

In conclusion, dyspnea and psychologic status impacted the HRQL in patients with COPD. Although no substantial differences between the SGRQ, the BPQ, and the CRQ were evident in the correlations with physiologic parameters and the influential factors by the multiple regression analyses, the BPQ was found to be less discriminatory than the other two questionnaires in evaluating HRQL cross-sectionally.