INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchiectasis is a chronic lung condition in which damage to the large airways causes abnormal dilation of one or more bronchi leading to poor clearance and pooling of mucus in the affected areas (1). It is similar to chronic obstructive pulmonary disease (COPD), in that cough, sputum production, wheeze, and breathlessness are common symptoms, and undue tiredness and fatigue are often reported (2). Infective exacerbations are common to both conditions, but the volume of sputum produced is usually greater in bronchiectasis, and chronic infection more frequent.

Health-related quality of life (HRQL) questionnaires allow clinicians to measure directly the impact of disease on a patient's daily life and they can be especially valuable in clinical trials designed to assess the benefits, side effects, and costs of different treatments. Increased recognition of the importance and value of understanding and measuring HRQL in chronic respiratory diseases has led to numerous general and disease-specific HRQL measures being developed and/or validated for use in conditions such as COPD, asthma, and cystic fibrosis (3). To date, however, none have been developed for bronchiectasis even though one of the primary aims of management of this disease is to improve the patient's well-being.

This study was carried out to test whether an existing instrument for obstructive airways disease could be applied to patients with this condition. Specifically we wished to examine relationships between a disease-specific measure developed for asthma and COPD, the St. George's Respiratory Questionnaire (SGRQ), and other HRQL measures and markers of disease activity in patients with bronchiectasis, to test its stability, validity and responsiveness in this condition.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The St. George's Respiratory Questionnaire

The SGRQ is a self-administered HRQL measure containing 50 items and 76 weighted responses divided into three components: Symptoms, Activity, and Impacts. The Symptoms component contains items concerned with the level of symptomatology, including frequency of cough, sputum production, wheeze, breathlessness, and the duration and frequency of breathlessness or wheeze. The Activity component is concerned with physical activities that either cause or are limited by breathlessness. The Impacts component covers such factors as employment, being in control of health, panic, stigmatization, the need for medication and its side effects, expectations for health and disturbance of daily life. Scores ranging from 0 to 100 are calculated for each component, as well as a total score which summarizes the responses to all items. A zero score indicates no impairment of quality of life. The questionnaire takes approximately 10 min to complete and to date has been shown to be reproducible, valid and responsive in both COPD and asthmatic populations (5, 13, 14).

Study Population

One hundred and twenty patients with bronchiectasis diagnosed clinically and by high-resolution computed tomography (HRCT) were approached over a 4-mo period to participate in the study in the Host Defence Unit outpatient clinics at the Royal Brompton Hospital, a tertiary referral center. Of these patients, 111 (67 females and 44 males) agreed to participate. Their mean age was 52 ± 13 yr (range 23 to 77). Etiologies included: idiopathic (58%), allergic bronchopulmonary aspergillosis (10%), hypogammaglobulinemia (9%), postinfective (defined as chronic cough and sputum production directly following pneumonia in childhood, e.g., whooping cough and measles; or following adult tuberculosis or pneumonia where bronchiectasis was localized to the area affected by the illness) (9%), primary ciliary dyskinesia and Young's Syndrome (9%), and bronchiectasis associated with inflammatory bowel disease (5%). Of the 111 patients, 14 had previously undergone surgery to remove bronchiectatic lung but had residual bronchiectasis; seven patients had coexisting emphysema and one had mild pulmonary fibrosis on HRCT criteria, but in all cases bronchiectasis was the predominant pathology. Each patient provided signed written consent and the study protocols were approved by the Royal Brompton Hospital Ethics Committee.

Study Protocols

The primary objectives of this study were to assess whether the SGRQ could measure health consistently (stability), to evaluate the validity of the SGRQ in terms of its ability to distinguish between different levels of health in bronchiectasis patients and to test whether it was responsive to changes in health over time. Both cross-sectional and longitudinal comparisons between the SGRQ and other markers of disease activity and HRQL measures were performed.

Patients were studied on two separate days (entry and follow-up), 6 mo apart. Since no new treatment protocols were being tested during the study phase, we chose a 6 mo-period for retesting to allow enough time for their symptoms and condition to change due to disease activity if they were to change at all. At entry, each of the 111 patients completed the SGRQ and a number of other HRQL measures: the United Kingdom version of the SF-36 Health Survey Questionnaire (9), a general health measure validated for use in respiratory conditions; the Medical Research Council (MRC) Dyspnea and Wheeze scales (15), standardized for use in assessing symptoms in respiratory conditions; a 14-item fatigue scale (16) shown to be valid and reliable for use in a general population; and the Hospital Anxiety and Depression scale (HAD) (17). The questionnaires were presented to the patients in a randomized order.

Each patient also performed a shuttle walking test which is a standardized incremental paced walking test around two stationary cones placed 10 m apart to measure symptom-limited maximal performance (18). Comprehensive lung function tests including measurement of arterial blood gases by earlobe sampling (Pa_O2) were made, and the patients were asked to report the number of infective exacerbations they had experienced over the past year.

To provide us with more information on the condition of the study population, a recent HRCT scan of each of the patients was assessed and scored by the same consultant pulmonary radiologist who was blinded to all other details concerning the patient. Each available lobe of both lungs was scored for bronchiectatic changes on a 0 to 3 scale, giving a maximum of 18 points: 0 = no bronchiectasis; 1 = 1 or < 1 bronchopulmonary segment involved; 2 = > 1 bronchopulmonary segment involved; 3 = gross cystic bronchiectasis. In order to adjust the scores of those patients who had had lobectomies, the bronchiectasis score was calculated as the sum of all points divided by the maximum points available for the individual × 100. This scoring system has been used in previous studies and is associated with low interobserver variation (19).

One hundred of the 111 patients recruited at entry were available for follow-up 6 mo later. Of the 11 patients who were not available, two had died, three had moved away, and six did not want to participate in the follow-up study. The protocol for the follow-up day was identical to that at entry (with the exception of the reporting of infective exacerbations). Finally, a subgroup of patients were selected randomly to assess the test-retest reliability of the questionnaire. Twenty-three patients completed the SGRQ on two separate occasions two weeks apart.

Statistical Analysis

The stability, construct validity and responsiveness of the SGRQ were examined in this study. The stability of the SGRQ was estimated by measuring the test-retest reliability and internal consistency of the questionnaire. Test-retest reliability measures the ability of the questionnaire to produce consistent scores over a short period in stable patients while internal consistency measures the degree of association between the questionnaire items. Construct validity is concerned with the extent to which a particular measure relates to other measures consistent with logical relations that should exist (20, 21). Thus, the validation process involved examining relationships between the SGRQ and the reference measures by testing for expected associations (convergent validity) and expected differences (discriminant validity). For example, the Symptoms score would be expected to correlate strongly with the level of respiratory symptomatology such as frequency of wheeze and number of infections, but not with exercise tolerance. On the other hand, dyspnea during daily activity would be expected to be a dominant correlate of the Activity score.

Prior to conducting the study, two general predictions were made on which to base the specific tests of evidence for validating the SGRQ: (1) the three component scores and the Total score of the SGRQ would be positively correlated with greater severity of disease activity and reduced HRQL, and (2) associations would be strong between individual components of the SGRQ and related measures (e.g., Activity component score and MRC Dyspnea score) and weak between individual components of the SGRQ and unrelated measures (e.g., Symptoms component score and the mood state scores). The responsiveness of the SGRQ was tested by comparing changes in the SGRQ scores to changes in the other markers of disease activity and HRQL scores over the six month period between entry and follow-up.

Repeatability was estimated by calculating the intraclass correlation (r_I) for each of the component scores and the Total score of the SGRQ in a sub-group of patients. Internal consistency was measured by applying Cronbach's alpha coefficient to each of the component scores of the SGRQ using the data from the 111 patients studied at entry. Parametric and distribution-independent tests were performed for all correlation analyses between the SGRQ and the selected reference measures. No differences in statistical significance level were found between these two techniques, so correlations are reported using Pearson's correlation coefficient (r). Student's unpaired t-test and one way analysis of variance (ANOVA) were used to compare sub-groups of patients. Statistical significance was accepted at p < 0.01 unless otherwise stated and summary results are presented as mean ± SD.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Details

The mean ± SD and range of some of the measures of disease activity for the study population are summarized in Table 1. There was a large range of scores suggesting that the population covered a wide spectrum of disease activity. The patient's age was not correlated with any of the SGRQ scores (r values < 0.16, p values > 0.05) and patient's sex also did not affect the SGRQ scores (t values < 1.22, p values > 0.05). There were no differences in SGRQ scores between the patients who had had previous lobectomies or who had coexisting emphysema and those who did not (t values < 2.13, p values > 0.05).

The mean SGRQ scores at entry were as follows: Symptoms score, 70.0 ± 19.5; Activity score, 48.5 ± 25.1; Impacts score, 33.9 ± 18.5; Total score, 44.4 ± 18.4. Figure 1 displays a histogram of SGRQ Total scores for the study population. There was a wide range of HRQL that followed a distribution that was approximately normal. There was no evidence of floor or ceiling effects since none of the patients had a score of 0 (best possible score) or 100 (worst possible score).

View larger version (44K): [in this window] [in a new window]   Figure 1.   Histogram of the St. George's Respiratory Questionnaire (SGRQ) Total score for 111 patients with bronchiectasis. 0 is best possible score and 100 is worst possible score.

Repeatability and Internal Consistency of the SGRQ

The intraclass correlation (r_I) for the short-term repeatability of the SGRQ component and Total scores measured in a sub-group of 23 patients two weeks apart were as follows: symptoms, r_I = 0.93; Activity, r_I = 0.98; Impacts, r_I = 0.94; and Total, r_I = 0.97. Cronbach's alpha coefficients for the SGRQ components were 0.90 for the Symptoms component, 0.89 for the Activity component, and 0.92 for the Impacts component.

Comparisons between SGRQ Scores and Physiological Measures

Physiological measures of disease activity were more strongly correlated with the Activity component of the SGRQ than with either the Symptoms or Impacts scores (Table 2). In general, exercise performance was the strongest physiological correlate of the SGRQ (Table 2 and Figure 2). Correlations between Pa_O2 and the SGRQ scores were stronger than the correlations between spirometry and the SGRQ.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Regression plot of the St. George's Respiratory Questionnaire (SGRQ) Total score versus the Shuttle test distance (r = 0.558, p < 0.0001).

Comparisons between SGRQ Scores and Other Measures of HRQL

Overall, the SGRQ scores were more strongly associated with the physical components of the reference measures than the psychological components (Table 2). The various scores for general health, fatigue and mood state were, in general, better correlated with the Impacts component of the SGRQ than the Symptoms or Activity components.

SGRQ Scores Compared with Symptoms and Infection Frequency

To examine the relationship between the SGRQ scores and the level of respiratory symptoms, one-way analysis of variance with post-hoc comparisons between groups was performed to compare the SGRQ scores with responses to the MRC Wheeze and Dyspnea scales and patients' reports of the number of infections over the preceding year (Table 3). The SGRQ Symptoms score was significantly higher in patients who reported having wheeze on most days (n = 54) as compared with patients who had occasional wheeze (n = 40) or no wheeze (n = 17) (p < 0.001). This component score was also significantly higher in patients who reported having three to six infections (n = 41) or more than six infections (n = 32) during the preceding year compared with those patients who had fewer than three infections over the past year (n = 38). The frequency of wheeze and the number of infections also had effects on the Activity score, but these were much weaker than their effects on the Symptoms score (Table 3). Conversely, dyspnea had a greater effect on the Activity score than the Symptoms score. Patients with a MRC Dyspnea grade of one (the least impaired) had a significantly lower SGRQ activity score than those patients with higher MRC Dyspnea grades.

Figure 3 illustrates the relationship between the frequency of infections and the SGRQ Impacts score. Patients who had more infections over the past year scored significantly higher. It is noteworthy that this component of the questionnaire addresses current health and contains no items that relate specifically to cough, sputum, or previous infections. A similar effect was seen on the SGRQ Total score, however this score includes the SGRQ Symptoms component that asks retrospective questions about the level of symptoms over the preceding year.

View larger version (13K): [in this window] [in a new window]   Figure 3.   ANOVA analysis showing the effect of the number of infections the patient experienced over the past year on the SGRQ Impact score (F = 12.832, p < 0.005). Post hoc tests: Comparison of patients with zero-two infections to patients with three-six and more than six infections, p < 0.001; comparison of patients with three-six infections to patients with more than six infections, p = 0.02. The error bars indicate 95% confidence intervals.

Responsiveness of the SGRQ

To assess the responsiveness of the SGRQ, changes in the patients' scores over the 6-mo period between entry and follow-up were calculated and compared with changes in the other HRQL measures and markers of disease activity. Taking the population as a whole, changes over this time were small, for example: SGRQ Symptoms score, 1.1 ± 14.1; SGRQ Total score, 1.3 ± 11.8; SF-36 physical component summary score, 1.0 ± 7.7; HAD depression score, 0.6 ± 2.5; total fatigue score, 0.6 ± 2.7; and shuttle distance, 1.4 ± 121.1 m. A positive score indicates that the relevant variable was lower at follow-up. Despite these small mean changes there was sufficient within-patient variation to detect associations between changes in the different variables (Table 4). Overall, changes in the SGRQ scores correlated more strongly with changes in physical component measures than with changes in mental component measures (Table 4). This was similar to the results of the cross-sectional analyses.

Of the 100 patients who participated in the follow-up study, 15 reported an increase in frequency of wheeze and in these patients the SGRQ Total score rose (i.e., worse health) by 4.0 ± 14.0 units. In 27 patients the frequency of wheeze fell and in them the SGRQ Total score fell (i.e., better health) by 7.0 ± 10.7 units. These changes were significant (ANOVA p < 0.01). A similar picture was seen for the relationship between changes in the MRC Dyspnea score and changes in SGRQ Total score (Figure 4).

View larger version (14K): [in this window] [in a new window]   Figure 4.   Change in St. George's Respiratory Questionnaire (SGRQ) total score and change in Medical Research Council (MRC) dyspnea grade in 100 patients with bronchiectasis over 6 mo. An increase in SGRQ score indicates a deterioration in health. The error bars indicate 95% confidence intervals.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The process of validating a questionnaire designed to measure impaired health is multifactorial. Evidence for the validity of such instruments is built up from a large number of tests of the relationship between the questionnaire and relevant measures of disease activity and its effects on the patient's health and well being. We have shown that the SGRQ correlated with a number of measures of disease severity and activity relevant to patients with bronchiectasis. The pattern of these correlations with the components of the SGRQ suggested that these components were addressing relatively specific areas of impaired health in this population.

The SGRQ Symptoms score was significantly related to items of symptomatology associated with bronchiectasis such as the frequency of wheeze and the number of infections the patient had over the preceding 12 mo (Table 3) but only had weak associations with the physiological measures and the psychological components of the reference HRQL questionnaires (Table 2). The weak correlations between the SGRQ Symptoms component and the mood state scores (Table 2) is similar to previous findings in other patients with chronic airflow limitations (5). The SGRQ Activity score was most strongly associated with exercise performance and the physical component scores of the other HRQL measures. It was also moderately correlated with Pa_O2 and spirometry (Table 2) and the number of infections experienced over the previous year (Table 3). The associations between HRQL measured using the SGRQ and spirometry was relatively weak in this study, but similarly weak correlations have been observed between spirometry measures and other disease-specific HRQL measures in asthma and COPD (22, 23).

Several measures of disease activity contributed to the SGRQ Impacts score, including frequency of wheeze, dyspnea grade, number of infections in the previous year, fatigue and depression (Tables 2 and 3, Figure 3). The Total score also correlated significantly with the HRQL reference measures and other markers of disease activity (Tables 2 and 3, Figure 2). This was expected since this score was designed to reflect all areas of the patient's impaired health (5). It is worth noting that the mean SGRQ Total score and the mean FEV₁ from our study falls on a previously published regression linking mean SGRQ Total score and mean FEV₁ from a number of studies in asthma and COPD (24). Furthermore, the mean SGRQ component and Total scores from our population are similar to the mean SGRQ scores from the COPD and asthma population used to validate the SGRQ (5).

Overall, the SGRQ component scores were more strongly correlated with the physical components of the SF-36 and the fatigue scale rather than the psychological scores. This may be explained by the fact that questions explicitly designed to assess mood state were not included in the SGRQ because of the availability of existing measures for this specific purpose (5). However, since the Impacts section does contain some items expected to relate to feelings of anxiety and depression, such as social stigmatization, panic and being in control of one's health, it would be expected that this component would be more strongly related to psychological aspects of bronchiectasis than the other components of the SGRQ (Table 2).

The associations between the change in SGRQ scores and the change in other HRQL measures and markers of disease activity were not as strong as the associations from the cross-sectional analyses (Tables 2 and 4). This finding was probably due in part to the relatively small changes in the patients' level of disease activity over the 6-mo follow-up period as illustrated by the number of patients who indicated no change in breathlessness (n = 61) or frequency of wheeze (n = 58). However, an examination of those patients who did show improvement or worsening in their breathlessness or frequency of wheeze showed that changes in their SGRQ scores were statistically significant (see RESULTS and Figure 4) and exceeded the 4 unit threshold for a clinically significant difference (25).

In conclusion, data from this study has shown that the SGRQ has good repeatability, producing consistent scores in a sub-group of stable patients over a short period of time, as well as good internal consistency when applied to patients with bronchiectasis. The data also shows that the SGRQ is a valid measure for bronchiectasis in that it can distinguish between different levels of impaired health and appears to be sensitive to spontaneous changes in health over a 6-mo period. Confirmation of its responsiveness to treatment requires a formal study of the effect of a specific therapeutic intervention.