Chest ACCP Member Benefits
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Guest Access | Sign In via User Name/Password
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rutten-van Mölken, M. P.M.H.
Right arrow Articles by Monz, B. U.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rutten-van Mölken, M. P.M.H.
Right arrow Articles by Monz, B. U.
(Chest. 2006;130:1117-1128.)
© 2006 American College of Chest Physicians

Does Quality of Life of COPD Patients as Measured by the Generic EuroQol Five-Dimension Questionnaire Differentiate Between COPD Severity Stages?*

Maureen P.M.H. Rutten-van Mölken, PhD; Jan B. Oostenbrink, PhD; Donald P. Tashkin, MD, FCCP; Deborah Burkhart, RN and Brigitta U. Monz, MD, MPH

* From the Institute for Medical Technology Assessment (Drs. Rutten-van Mölken and Oostenbrink), Erasmus MC, Rotterdam, the Netherlands; the David Geffen School of Medicine (Dr. Tashkin), University of California-Los Angeles, Los Angeles, CA; Boehringer Ingelheim Pharmaceuticals Inc (Ms. Burkhart), Ridgefield, CT; and Boehringer Ingelheim International (Ms. Monz), Ingelheim am Rhein, Germany.

Correspondence to: Maureen P.M.H. Rutten-van Mölken, PhD, Erasmus MC, Institute for Medical Technology Assessment (IMTA), PO Box 1738, 3000 DR Rotterdam, the Netherlands; e-mail: m.rutten{at}bmg.eur.nl

Abstract

Objective: To assess the discriminative properties of the EuroQol five-dimension questionnaire (EQ-5D) with respect to COPD severity according to Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria in a large multinational study.

Methods: Baseline EQ-5D visual analog scale (VAS) scores, EQ-5D utility scores, and St. George Respiratory Questionnaire scores were obtained from a subset of patients in the Understanding the Potential Long-term Impact on Function with Tiotropium trial, which was a 4-year placebo-controlled trial designed to assess the effect of tiotropium on the rate of decline in FEV1 in COPD patients aged ≥ 40 years, an FEV1 of < 70% predicted, an FEV1/FVC ratio of ≤ 70%, and a smoking history of ≥ 10 pack-years.

Results: A total of 1,235 patients (mean post bronchodilator FEV1, 48.8% predicted) from 13 countries completed the EQ-5D. The EQ-5D VAS and utility scores differed significantly among patients in GOLD stages 2, 3, and 4, also after correction for age, sex, smoking, body mass index (BMI), and comorbidity (p < 0.001). The mean EQ-5D VAS scores for patients in GOLD stages 2, 3, and 4 were 68 (SD, 16), 62 (SD, 17), and 58 (SD, 16), respectively. The mean utility scores were 0.79 (SD, 0.20) for patients in GOLD stage 2, 0.75 (SD, 0.21) for patients in GOLD stage 3, and 0.65 (SD, 0.23) for patients in GOLD stage 4. Effect sizes for the difference in utility scores between patients in GOLD stages 3 and 4 were more than twice as high as those for the difference between patients in GOLD stages 2 and 3. Gender, postbronchodilator FEV1 percent predicted, the number of hospital admissions and emergency department visits in the year prior to baseline measurements, measures of comorbidity, and BMI were independently associated with EQ-5D utility. EQ-5D utility scores also differed between patients from different countries. French patients especially had lower utility scores than US patients. Utility scores calculated with the US value set were on average 5% higher than those calculated with the UK value set.

Conclusions: Increasing severity of COPD was associated with a significant decline in EQ-5D VAS scores and utility scores. These results demonstrate that a generic instrument can assess COPD impact on quality of life and that the scores discriminate between patient groups of known severity. These utility scores will be useful in cost-effectiveness assessments.

Key Words: COPD • disease severity • EuroQol five-dimension questionnaire • Global Initiative for Chronic Obstructive Lung Disease • health status • quality of life • utility

An increasing number of treatments and interventions for COPD have been and will be subjected to cost-effectiveness studies to provide information for priority setting by health-care decision makers. In many European, North American, and other countries, reimbursement authorities and academic bodies have issued methodological guidelines for conducting cost-effectiveness studies. These guidelines advocate the use of quality-adjusted life-years (QALYs) in addition to clinical outcomes.1234 QALYs are calculated as the sum of the products of time spent in certain health states and the utility scores of these health states. Utility scores are the valuations of health states that are anchored on a numeric scale ranging from death (0) to perfect health (1). In Europe, the EuroQol five-dimension questionnaire (EQ-5D) is the instrument that is the most widely used for generating these utility scores,5 and many countries, including the United States, have published their own set of population-based EQ-5D weights to perform the valuation.56

The cost-effectiveness of several COPD interventions has been assessed using cost per QALY ratios,7891011121314 and some of these have used the EQ-5D to obtain the utility scores.813 Most of these interventions directly affect outcomes, like exacerbation and hospitalization rates, that have been shown to influence survival, as a result of which their potential impact in terms of QALYs is large. The gain in QALYs resulting from COPD treatments that merely affect the quality of life of patients may only become apparent when the time horizon of the economic analysis is fairly long, in many instances even lifelong. This makes necessary an extension of the time horizon that is well beyond that of most clinical trials. This is one of the reasons why economic models are extensively used in cost-effectiveness assessments, especially those of chronic diseases. Several models estimating the cost-effectiveness of COPD interventions have been published.14151617 All of these models are so-called state transition models (or Markov models) that simulate the progression of COPD over different stages of disease severity. These models also have in common that they define COPD severity in terms of FEV1 percent predicted and estimate QALYs by assigning EQ-5D utility scores141617 or other utility scores15 to these COPD severity stages. Hence, it is important to study the ability of the EQ-5D to discriminate between different stages of COPD severity. Since the EQ-5D is a generic health-related quality-of-life instrument that captures the impact of comorbidity and other factors on quality of life, the aim of the current analyses is to estimate the association between EQ-5D and disease severity independent of comorbidity and other factors known, or expected, to be related to quality of life. Its ability to discriminate is compared to that of a disease-specific quality-of-life instrument, the St. George Respiratory Questionnaire (SGRQ). No such study has been conducted previously in such a large multinational group of physician-diagnosed COPD patients with a wide range of lung function impairment as the subgroup of patients from the Understanding the Potential Long-term Impact on Function with Tiotropium (UPLIFT) trial18 that were studied here. This multinational group of patients allowed us to address the association between the country of recruitment and COPD utility. It also allowed us to study whether the ability of the EQ-5D to discriminate between COPD severity stages depends on the value set that is used to generate the utility scores.

Materials and Methods

Setting and Patients
This study used data from a subset of 1,235 patients from 13 countries who completed the EQ-5D at baseline of the UPLIFT trial.18 Because the EQ-5D was added via protocol amendment more than midway through the recruitment period, only the last 1,235 patients of those countries with significant enrollment remaining could be included in this study. Patients were enrolled sequentially from the time of ethics committee approval of the protocol amendment in each particular country until the enrollment completion of the UPLIFT trial.18 The UPLIFT trial is a 4-year randomized, double-blind, placebo-controlled, parallel-group trial that was designed to determine whether tiotropium reduces the rate of decline in FEV1 over time. It includes approximately 6,000 COPD patients (postbronchodilator FEV1, < 70% predicted; FEV1/FVC ratio, ≤ 70%); age, ≥ 40 years; cigarette smoking history of at least 10 pack-years).18 The EQ-5D and SGRQ were completed at the randomization visit prior to spirometry and prior to the start of the administration of the study medication.

Besides age, gender, and information on smoking, patient characteristics included the duration of diagnosed COPD in years, body mass index (BMI), self-reported COPD-related health-care utilization prior to baseline, the presence or absence of home oxygen use, and concomitant diagnoses. Smoking was expressed as self-reported current smoking status (smoking or ex-smoking) and smoking history in terms of the number of pack-years. BMI was calculated as weight (in kilograms) divided by the height squared (in meters). COPD-related health-care utilization was measured as the number of emergency department (ED) visits resulting in hospital admission, the number of ED visits not resulting in hospital admission, the number of direct hospital admissions, the number of unscheduled urgent visits to physicians, and the number of scheduled visits to physicians in the year prior to enrollment. Comorbidity was expressed as the presence of any current concomitant diagnosis, the number of concomitant diagnoses, and the Charlson index score.19 The Charlson index contains 19 conditions. Each condition has an associated weight ranging from 1 to 6; the higher the score, the more severe the condition. The score on the Carlson index is calculated as the sum of weights assigned to each condition that a patient has. Concomitant diagnoses were also grouped by 25 system organ classes, which were further collapsed into 10 "medical system" groups to reduce the number of variables in the analyses.

EQ-5D
The EQ-5D is a self-administered, generic, health-related quality-of-life questionnaire that contains two sections, a descriptive section and a valuation section.2021 The descriptive section is a health status classification instrument with the following five dimensions: mobility; self-care; usual activities; pain/discomfort; and anxiety/depression. Each dimension is divided into the following three levels of functioning: no problems; some problems; and extreme problems. Respondents are asked to describe their health status by ticking off one level of functioning for each of the five dimensions. In the second section, respondents are asked to value their overall health status on a visual analog scale (VAS). This VAS is a simple rating scale ranging from 0 (defined as the worst imaginable health state) to 100 (defined as the best imaginable health state). Using a set of weights (ie, the value set), the descriptive information on health status can be converted into a single aggregate utility on a scale anchored at 1, which represents perfect health, and 0, which represents death. A number of countries have published their own value sets5; in the primary analysis, we employed the most widely used value set, known as the "MVH A1 value set," which was developed in the United Kingdom in the Measurement and Valuation of Health study.22 This set of utility weights was chosen because it is recommended by the EuroQol Group5 for use in cost-effectiveness/utility studies for comparison with other studies, and because these weights were generated from a large sample of the general public in the United Kingdom using the time-tradeoff (TTO) technique. The TTO is a preference-based method asking the respondent to trade length of life for quality of life. In an alternative analysis, the recently published US value set was used.6 This US set was obtained in a replication of the Measurement and Valuation of Health study, using the same TTO protocol. There is little evidence on the minimum clinically meaningful difference in EQ-5D score. A recent study,23 analyzing data from eight studies in 11 different patient groups, reported a mean minimally important difference of 0.074, but there was a wide variation in estimates of the minimally important difference between the studies. In the remainder of this article, EQ-5D VAS will be used to indicate the patient valuation on the VAS, EQ-5D utility will be used to indicate the utility scores obtained after applying a value set, and EQ-5D scores will be used to indicate both.

SGRQ
The SGRQ is a self-administered disease-specific questionnaire that is designed to measure the impact of pulmonary disease on health-related quality of life and well-being. The questionnaire contains 50 items that can be aggregated into an overall score and three subscores for "symptoms," "activity," and "impact." Scores range from 0 to 100, with a lower score representing a better quality of life. A change in score of 4 units is consistent with a clinically relevant change in the patient.2425 The SGRQ is the most frequently used quality-of-life questionnaire in patients with COPD, in whom it was reported to be valid, reliable, and responsive to change due to pharmacologic and nonpharmacologic therapy.

Lung Function and COPD Severity
Both pre-bronchodilator therapy and post-bronchodilator therapy FEV1 values were measured in the morning using calibrated spirometers. After the pre-bronchodilator therapy FEV1 was measured, four inhalations of 20 µg of ipratropium bromide were administered, followed by four inhalations of 100 µg of salbutamol 60 min later. The post-bronchodilator therapy FEV1 was obtained 30 min after salbutamol inhalation. The FEV1 percent predicted was calculated using European Community for Steel and Coal/European Respiratory Society equations.26 Using the post-bronchodilator therapy FEV1 percent predicted, patients were classified into categories of moderate COPD (FEV1, < 80% and ≥ 50% predicted), severe COPD (FEV1, < 50% and ≥ 30% predicted), and very severe COPD (FEV1, < 30% predicted), conforming to the boundaries of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages 2, 3, and 4, respectively.27

Statistical Analysis
Differences in patient characteristics between GOLD stages were tested using the Pearson {chi}2 test for categoric variables, one-way analysis of variance (ANOVA) for continuous variables with normal distributions, and the Kruskal-Wallis test for variables lacking normal distributions. Differences in the percentages of patients reporting any problems vs no problems on the EQ-5D dimensions were tested with the Pearson {chi}2 test. To study the presence of ceiling and floor effects in the EQ-5D dimensions, for the EQ-5D VAS, the EQ-5D utility, and the SGRQ total and individual domain scores the proportions of patients having the best possible score (ceiling effect) and the worst possible score (floor effect) were reported.

Discriminative validity is a specific type of construct validity that is defined here as the capacity of the EQ-5D VAS, the EQ-5D utility, and the SGRQ total and domain scores to differentiate between the GOLD stages of COPD severity. First, differences among the three GOLD stages were tested using ANOVA without adjustments for patient characteristics. Post hoc tests of the three pairwise differences in severity stage were performed using the Bonferroni correction for multiple testing. Second, multivariate analysis of covariance (ANCOVA) was performed with adjustments for age, gender, current smoking status, smoking history, BMI, and either the number of concomitant diagnoses or the Charlson index of comorbidity.

To assess the magnitude of the difference in scores between the GOLD stages, effect sizes were calculated as the mean differences between two consecutive GOLD stages divided by the pooled SD. Differences in utility scores calculated with the UK value set and the US value set were assessed using a t test.

The association between the number of concomitant diagnoses and EQ-5D scores or SGRQ scores was assessed using ANCOVA with two factors (ie, GOLD stage and the presence or absence of any concomitant diagnosis). The same analyses were performed with the number of concomitant diagnoses grouped according to the 25th and 75th percentile, and with the Charlson index grouped by a score of 0, 1, or ≥ 2. Interaction terms were included to study whether the impact of comorbidity on EQ-5D scores was higher in more severely ill COPD patients. To gain further understanding, concomitant diagnoses were grouped by medical system, and differences in EQ-5D scores between patients with or without comorbidity in a certain medical system were tested using t tests.

Finally, the association between EQ-5D utility scores and all patient characteristics, known or expected to be related to quality of life, including FEV1 percent predicted as a continuous variable, health-care utilization, and country of recruitment, was studied with backward, stepwise, multiple linear regression analyses using the EQ-5D utility as the dependent variable. Variables were removed from the regression equation when their p value was > 0.10. It was prespecified that any two patient characteristics that showed an intercorrelation of > 0.4 were not entered into the regression analyses at the same time.28 Moreover, the collinearity statistics of the independent variables, as measured in terms of tolerance, were checked to prevent high correlations between the variables in the regression model. The tolerance of an independent variable i is defined as 1 – Ri2, where Ri2 is the multiple correlation coefficient when the ith variable is predicted from the other independent variables. The tolerance can vary between 0 and 1, and should be large, otherwise i is almost a linear combination of the other independent variables. The overspecification of the regression models is avoided by using a conservative number of independent variables relative to the number of patients (ie, one or two variables per 100 patients), depending on the model. All analyses were performed using a statistical software package (SPSS, version 12; SPSS; Chicago, IL).

Results

Table 1 shows the characteristics of the 1,235 patients included in the study. The mean duration of COPD was approximately 10 years. Thirty-four percent were active smokers. The mean smoking history was approximately 48 pack-years. Most patients (85.7%) had comorbidity. The median number of concomitant diagnoses per patient was three. Patients most frequently experienced vascular disorders (48%), musculoskeletal disorders (34%), metabolic disorders (32%), GI disorders (26%), and cardiac disorders (25%). The mean score on the Charlson index of comorbidity was 0.51, and the median score was 0. The average BMI was in the normal range, and 3.2% of the patients had a BMI < 18.5. In the year before the baseline measurements were made, 3.7% of patients were hospitalized for COPD at least once, 8.8% of patients had one or more COPD-related ED visits, and 26.5% of patients had one or more unscheduled urgent visits to a physician. The majority of patients were living in the United States (34.5%), followed by the Czech Republic (17.5%), Spain (11.9%), Denmark (8.4%), Germany (4.9%), Poland (4.8%), the Netherlands (4.4%), Italy (4.4%), France (3.1%), Hungary (2.5%), the Russian Federation (1.5%), Belgium (1.4%), and Australia (0.8%). Table 1 also shows that the subset of patients in this study did not differ systematically from the total patient population in the UPLIFT trial.18


View this table:
[in this window]
[in a new window]

  		Table 1. Patient Characteristics at Baseline for the Subset of Patients in this Study and All Patients in the UPLIFT Trial*

 
Table 2 shows patient characteristics grouped by GOLD severity stage. A total of 50.7% of patients had stage 2 COPD, 41.8% of patients had stage 3 COPD, and 7.4% of patients had stage 4 COPD. GOLD stages were comparable with respect to the gender distribution, the number of pack-years smoked, and the duration of COPD, but they differed significantly with respect to age, percentage of current smokers, BMI, and health-care utilization in the year prior to the study. Patients in GOLD stage 4 were younger, had a lower BMI, and a higher number of ED visits resulting in COPD-related hospital admission, and a higher proportion of them were using home oxygen than those in stages 2 and 3. There were more current smokers who were in GOLD stage 2 and 4 than were in stage 3. The frequency of unscheduled urgent visits among patients in GOLD stage 3 and 4 was higher than for those in stage 2. Also, the other indicators of health-care utilization and the number of concomitant diagnoses tended to increase as COPD severity increased. However, the Charlson index and the comorbidity profile that resulted from grouping concomitant diagnoses by medical system did not differ significantly between patients in different GOLD stages of COPD (data on comorbidity profile not shown).


View this table:
[in this window]
[in a new window]

  		Table 2. Patient Characteristics by GOLD Stage*

 
Figure 1 shows that the percentage of patients reporting any problems on the EQ-5D dimensions of mobility, self-care, and usual activities increased with increased COPD severity. Differences on the dimensions pain/discomfort and anxiety/depression were not statistically significant, although the percentage of patients reporting problems on these dimensions was higher for patients in GOLD stage 4 than for patients in GOLD stages 2 and 3.


Figure 1
View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Percentage of patients in each COPD severity stage who reported any problems on the EQ-5D dimensions (collapsing the three levels into two: no problems vs any problems). First bar = moderate COPD; second bar = severe COPD; third bar = very severe COPD; mob = mobility; act = activity; disc = discomfort; anx/dep = anxiety/depression. Footnote below Figure 1. p < 0.001 (by {chi}2 test) for mobility, self-care, and usual activities; differences in pain/discomfort and anxiety/depression are not statistically significant.

 
There are indications of a ceiling effect on the EQ-5D utility scale, since the percentage of patients with the best possible EQ-5D utility was 22.9%, whereas it was only 1.1% on the EQ-5D VAS, 0.2% on the SGRQ total score, 1.1% on the SGRQ symptom score, 0.7% on the SGRQ activity score, and 1% on the SGRQ impact score. The percentage of patients with the best possible EQ-5D utility of 1 decreased from 27.8% among patients in GOLD stage 2 and 19.7% among patients in GOLD stage 3 to 4.4% among patients in GOLD stage 4. None of the scales demonstrated floor effects.

Without adjusting for any differences in patient characteristics, there were statistically significant differences in EQ-5D VAS and EQ-5D utility scores among patients in the three GOLD stages. Post hoc analyses with Bonferroni correction for multiple testing showed that all pairwise comparisons between patients in the three GOLD stages were statistically significant. After adjusting for age, gender, current smoking status, the number of pack-years of smoking, BMI, and the number of concomitant diagnoses, pairwise differences in EQ-5D VAS and utility scores remained highly statistically significant (Table 3 ). These results did not change after adjusting for the Charlson index instead of the number of concomitant diagnoses.


View this table:
[in this window]
[in a new window]

  		Table 3. EQ-5D VAS and Utility Scores and SGRQ Scores by COPD Severity*

 
The importance of the differences among patients in different GOLD stages is investigated by calculating effect sizes, as shown in Figure 2 . Both the EQ-5D utility scores and the SGRQ differentiate much better between patients with very severe and severe COPD (Fig. 2, bottom, b) than between those with moderate and severe COPD (Fig. 2, top, a). There is a more than twofold increase evident in effect size between Figure 2, top, a, and Figure 2, bottom, b. The only exception is the patient’s own valuation on the EQ-5D VAS, in which the effect size of the difference between patients with moderate COPD and those with severe COPD is somewhat larger than the effect size of the difference between patients with severe COPD and those with very severe COPD. Patients with very severe COPD had a 7.4% lower EQ-5D VAS score than did patients with severe COPD, who in turn had a 7.1% lower EQ-5D VAS score than did patients with moderate COPD. Patients with very severe COPD had 13.7% lower EQ-5D utility scores (using the UK value set) than did patients with severe COPD, who in turn had only 4.7% lower scores than did patients with moderate COPD. Differences between GOLD stages in the SGRQ total and domain scores were of a higher order of magnitude than the differences in EQ-5D scores. Patients with very severe COPD had a 23.2% worse SGRQ total score than did patients with severe COPD, who in turn had an 11.0% worse score than did patients with moderate COPD.


Figure 2
View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Effect sizes calculated as the mean difference between moderate and severe COPD divided by the pooled SD (top, a) and the mean difference between severe and very severe divided by the pooled SD (bottom, b); util = utility.

 
Utility scores calculated with the US value set were significantly higher than those calculated with the UK value set. The mean difference was –0.052 (95% confidence interval, –0.055 to –0.048). Table 3 shows that differences between the three GOLD stages of COPD severity remained highly statistically significant when replacing the UK value set with the US value set. Effect sizes of the differences between GOLD stages in EQ-5D utility scores calculated with the UK or US value set were similar (Fig. 2).

Since patients with very severe COPD had more concomitant diagnoses than those with less severe disease, Table 4 addresses the question of whether the association between GOLD stage and quality-of-life scores is influenced by comorbidity. The numbers of concomitant diagnoses were grouped using the 25th and 75th percentiles, which were one and six, respectively. The absolute number of concomitant diagnoses was significantly associated with worse EQ-5D utility scores (p < 0.001) and worse SGRQ total scores (p < 0.001), independent of GOLD stage. Also a higher Charlson index score was associated with a worse EQ-5D utility score, independent of GOLD stage (p = 0.002). However, there was no significant interaction between GOLD stages and the presence or absence of comorbidity, between GOLD stages and the three groups of comorbidities shown in Table 4, or between GOLD stages and the Charlson index score. Neither a higher number of comorbidities nor a higher Charlson index score was associated with a worse EQ-5D VAS score. When grouped by system organ class patients with cardiac, vascular, GI, musculoskeletal, neurologic, psychiatric, renal, and urinary disorders, and disorders of the immune system had lower EQ-5D utility scores than patients without these disorders (all p values were < 0.05). Except for the system organ classes "benign and malignant neoplasms" (p = 0.003) and "reproductive system and breast disorders" (p = 0.006), the EQ-5D VAS scores did not differ between patients with and without these comorbidities.


View this table:
[in this window]
[in a new window]

  		Table 4. Mean EQ-5D and SGRQ Scores in Patients With Different Stages of COPD Severity and Different Number of Comorbidities*

 
Since the EQ-5D utility and not the VAS is the variable to be used in cost-effectiveness analyses, the relative contributions of the severity of lung function impairment, comorbidity, country, and other patient characteristics to this score were investigated in multivariate linear regression analyses (Table 5 ). Gender, post-bronchodilator therapy FEV1 percent predicted, number of hospital admissions, number of ED visits not resulting in hospital admission, number of concomitant diagnoses, and BMI were independently associated with EQ-5D utility. Male patients had a 0.057 higher utility than female patients. For each 10-U decrease in FEV1 percent predicted, EQ-5D utility scores decreased by 0.03. A more frequent use of emergency care (either ED visits or hospital admissions) in the year before study recruitment decreased the utility. Although highly significant, the impact of an increasing number of concomitant diagnoses is small. The model that includes dummy variables for the countries that participated in this study (last two columns of Table 5) showed that, after correction for the other variables, the utility scores of Danish patients were 0.06 higher than the utility scores of US patients, whereas the utility scores of Italian, Czech, Polish, and French patients were between 0.04 and 0.15 lower.


View this table:
[in this window]
[in a new window]

  		Table 5. Multivariate Linear Regression Analyses With EQ-5D Utility (UK Value Set) as Dependent Variable

 
Given the large number of patients (> 1,200), analyses were based on parametric tests, even though EQ-5D scores were not normally distributed. All analyses were repeated using nonparametric tests. The p values were very similar, and the differences between the results of analyses using parametric and nonparametric tests were very small and did not change any of the conclusions. Log-transforming EQ-5D utility scores before running regression analyses did not improve the proportion of variance explained.

Discussion and Conclusion

This study has demonstrated that the GOLD staging of COPD severity corresponds to significant differences in generic health-related quality of life, as assessed by the EQ-5D VAS and utility scores. Importantly, these differences were maintained after correction for other variables that were known to or were expected to affect quality of life, especially comorbidity. This finding demonstrates that GOLD staging of COPD severity corresponds not only to differences in disease-specific quality of life, but also to differences in the general perception of quality of life. Hence, the utility estimates for each severity stage can be used in models, especially models designed to estimate the cost-effectiveness of preventive and therapeutic interventions in COPD.

The analyses showed that the magnitude of the difference in EQ-5D utility score between patients with moderate and severe COPD was rather small, whereas the difference in EQ-5D utility score between patients with severe and very severe COPD was more than twice as large as the difference between patients with moderate and severe COPD. These findings parallel the observation that the differences between patients with severe and very severe COPD in terms of the number of ED visits and hospital admissions prior to trial entry, BMI, duration of COPD, and number of pack-years of smoking (Table 2) were more pronounced than the differences between patients with moderate and severe COPD. As expected, the magnitude of the difference in SGRQ scores is greater than the magnitude of the difference in EQ-5D utility scores, but it is noteworthy that the SGRQ score also showed a greater difference between patients with severe and very severe COPD than between those with moderate and severe COPD. These findings differ from those of a previous study,29 which reported the greatest worsening of health status as taking place between patients with moderate and severe COPD. However, the lack of statistically significant differences between patients in consecutive stages other than those with moderate and severe COPD in that study is probably due to the small sample size. Our findings are in agreement with a modeling study16 that applied EQ-5D utility scores by GOLD stage. This study reported an even smaller difference (< 0.01) between patients with moderate and severe COPD, whereas the difference between patients with severe and very severe COPD was about 0.2. These differences were neither statistically tested nor discussed.

The difference in SGRQ scores between patients in consecutive GOLD stages of COPD severity exceeded the threshold value of the minimal clinically relevant difference of 4 U. The minimal important difference in EQ-5D scores has not been established, although a recent article23 reported a mean of 0.07, but the range of estimates was wide. The difference in EQ-5D utility scores between patients with moderate and severe COPD did not reach this threshold value, whereas the difference between patients with severe and very severe COPD did.

Unlike the EQ-5D utility scores and the SGRQ scores, the EQ-5D VAS shows a similar worsening when moving from moderate to severe COPD, as that moving from severe to very severe COPD. This disparity might reflect the higher sensitivity of a patient’s own valuation of his or her health status. However, the EQ-5D VAS score was not sensitive to comorbidity. Comorbidity is known to affect generic health-related quality of life in COPD patients,30 and measuring the impact of comorbidity is one argument for using generic quality-of-life instruments. In our study, regression analyses showed a higher number of concomitant diagnoses to be associated with worsening EQ-5D utility scores, independent of COPD severity and other patient characteristics. There were no indications of a greater impact of comorbidity in patients in more severe stages of COPD. The unexpected finding that the patient valuation on the EQ-5D VAS was not sensitive to comorbidity might be related to the possibility that patients who volunteer for and are enrolled in a COPD trial focus their health perception primarily on the disease under investigation.

Validation studies of the EQ-5D in COPD have been conducted before in relatively small single-country or single-center studies.313233 These studies have not addressed the discriminative properties of the EQ-5D with respect to the GOLD stages of disease severity. Our study used baseline data from the UPLIFT trial.18 This 4-year trial is designed to assess the impact of tiotropium on the natural course of COPD. It provides a unique opportunity to address the issue of the discriminative validity of the EQ-5D, because of its large sample size, the wide variety of countries from which patients were enrolled, and the complete registration of comorbidity. The UPLIFT trial18 was less restrictive with respect to the inclusion criteria than is common among COPD trials. The only exclusion criteria were a history of asthma or pulmonary resection, an exacerbation or respiratory infection in the month before study entry, the use of oxygen for > 12 h a day or a disease that might influence the results or the ability to participate in the study. There is no upper age limit, and there are no restrictions to the use of concomitant medications other than use of open-label anticholinergic agents during the treatment period. Nevertheless, the patient population may not be entirely representative of all patients with moderate-to-very severe COPD, because enrolled patients may have been more stable and less likely to have life-threatening comorbidity given that this would interfere with the ability to observe them for 4 years. This might explain the absence of a difference in the percentage of patients reporting problems on the anxiety/depression domain between the COPD severity stages. It might also explain the absence of an interaction between GOLD stage and comorbidity, especially when comorbidity is expressed as the Charlson index score, because only very severe conditions in that index get a high weight.

EQ-5D utility scores were found to vary by country. After adjusting for other variables, French, Italian, Czech, and Polish patients had significantly lower utility scores than US patients, whereas Danish patients had significantly higher values. The magnitude of the difference varied between 4% and 15% of the scale of 0 to 1. These between-country differences were noted despite the study-wide use of the same set of weights, namely, those obtained from the UK general population. In addition, utility scores calculated with the US value set were significantly higher than those calculated with the UK value set, with the magnitude of the difference being approximately 5%. Although both set of weights are based on the TTO method, this difference is partly due to methodological differences, such as differences in the statistical models used to estimate the weights or differences in the transformation of negative values. Moreover, the UK weights were obtained in the early 1990s, whereas the US weights were obtained in 2002. Nevertheless, when comparing directly elicited values for EQ-5D health states using exactly the same methods, Johnson et al34 found that the US population assigned consistently higher values to the same health states than the UK population. The difference was on average 0.1 but increased when health states became worse. The UK and US value sets are likely to be used as alternatives to obtain utility scores. Theoretically, the choice of weights to calculate the EQ-5D utility scores in economic analyses, whether empirical cost-effectiveness studies appended to clinical trials or modeling studies, should, whenever possible, be based on the country that is represented in the analyses. However, the consequence of the higher EQ-5D utility scores generated by the US value set is that there is less room for improvement, resulting in less QALY gains and a less favorable cost-effectiveness ratio compared to the UK value set.35

This analysis was conducted to assess whether the most widely used classification system of COPD severity (ie, the GOLD classification) corresponds to differences in EQ-5D scores. The threshold values of this lung function classification were taken as a given, and we acknowledge that the threshold values of any classification instrument are to a certain extent arbitrary. We were not searching for a classification that discriminates best between COPD patients with different levels of quality-of-life impairment. A composite classification instrument that includes variables independently associated with quality of life, analogous to the body mass index, airflow obstruction, dyspnea, exercise capacity (BODE) index,36 would probably better serve this purpose.

The EQ-5D utility had a relatively large proportion of responses in the best category. This proportion decreases over GOLD stages 2 to 4. This confirms the presence of a ceiling effect,34 implying that the EQ-5D is not the preferred instrument for distinguishing between less severe COPD states. Patients with mild COPD, according to the GOLD classification, were not investigated in the present study.

Besides the ceiling effect limiting the sensitivity of the EQ-5D utility to differences in patients with less severe COPD, the EQ-5D has the additional problem of not capturing the impact that COPD exacerbations have on quality of life. This problem applies equally to the EQ-5D, the SGRQ, and other generic or COPD-specific quality-of-life instruments. The EQ-5D has no recall period and asks for a description of a patient’s health "today." Even if there is a recall period, as in the symptoms domain of the SGRQ, these quality-of-life instruments are usually administered during a stable phase of the disease, as a result of which they do not capture the impact of exacerbations. This should be taken into account when using the EQ-5D utility scores from this study in modeling studies. It would be a step forward if we could obtain utility scores for COPD health profiles that combine the description of a patient’s underlying COPD severity stage with the description of that patient’s exacerbation profile in terms of the frequency, severity, and impact of the exacerbation. Such health profiles are particularly suitable for describing episodic or fluctuating diseases such as COPD.37

In conclusion, this large multinational study generated utility scores that can be used in modeling studies. It demonstrated that the EQ-5D VAS and EQ-5D utility are capable of differentiating among moderate, severe, and very severe COPD, as defined by the GOLD criteria. The differences remained after correction for other variables, such as comorbidity. The reduction in generic quality of life is much more apparent when patients progress from severe to very severe COPD than when they progress from moderate to severe COPD. The between-country differences that were found in this study also have implications for the use of EQ-5D utility scores in cost-utility studies of COPD interventions. Given the differences between the UK and US value sets, limiting empirical multinational cost-utility studies to the use of only one set of weights is necessary to ensure that QALYs are comparable. When models are used to adapt trial results to better represent the target country or setting, a country-specific set of weights is to be preferred.

Acknowledgements

We thank our Institute for Medical Technology Assessment colleague Dr. Leida Lamers, who is a member of the EuroQol group, for suggestions on the EQ-5D analyses. We also would like to thank those investigators participating in UPLIFT trial who implemented the administration of the EQ-5D, and all patients who completed the questionnaire. We acknowledge Dominique Julien from Boehringer Ingelheim for providing data management support.

Footnotes

Abbreviations: ANCOVA = analysis of covariance; ANOVA = analysis of variance; BMI = body mass index; ED = emergency department; EQ-5D = EuroQol five-dimension questionnaire; GOLD = Global Initiative for Chronic Obstructive Lung Disease; QALY = quality adjusted life year; SGRQ = St. George Respiratory Questionnaire; TTO = time-tradeoff; UPLIFT = Understanding the Potential Long-term Impact on Function with Tiotropium; VAS = visual analog scale

This study was financially supported by Boehringer Ingelheim International and Pfizer Global Pharmaceuticals.

Drs. Rutten-van Mölken and Oostenbrink were reimbursed by Boehringer Ingelheim for consultancy, and for delivering lectures and courses. They have no financial interest in the EQ-5D. Dr. Tashkin is a recipient of research grants from and has been a consultant for Boehringer-Ingelheim Pharmaceuticals Inc.

Received for publication September 5, 2005. Accepted for publication March 21, 2006.

References

  1. Gold, MR Siegel, JE Russell, LBet al eds. Cost-effectiveness in health and medicine 1st ed. 1996 Oxford University Press. New York, NY:
  2. Canadian Coordinating Office for Health Technology Assessment.. Guidelines for economic evaluation of pharmaceuticals 2nd ed. 1997 Canadian Coordinating Office for Health Technology Assessment. Ottawa, ON, Canada:
  3. Garattini, L, Grilli, R, Scopelliti, D, et al A proposal for Italian guidelines in pharmacoeconomics. Pharmacoeconomics 1995;7,1-6[Medline]
  4. Langley, PC The NICE reference case requirement. Implications for drug manufacturers and health systems. Pharmacoeconomics 2004;22,267-271[CrossRef][ISI][Medline]
  5. Brooks, R Rabin, R De Charro, F eds. The measurement and valuation of health status using EQ-5D: a European perspective; evidence from the EuroQol BIOMED Research Programme 1st ed. 2003 Kluwer Academic Publishers. Dordrecht, the Netherlands:
  6. Shaw, JW, Johnson, JA, Coons, SJ US valuation of the EQ-5D health states: development and testing of the D1 valuation model. Med Care 2005;43,203-220[CrossRef][ISI][Medline]
  7. Ramsey, SD, Patrick, DL, Albert, RK, et al The cost-effectiveness of lung transplantation: a pilot study. Chest 1995;108,1594-1601[Abstract/Free Full Text]
  8. Al, MJ, Koopmanschap, MA, Van Enckevort, PJ, et al Cost-effectiveness of lung transplantation in the Netherlands: a scenario analysis. Chest 1998;113,124-130[Abstract/Free Full Text]
  9. Anon, JM, Garcia de Loranzo, A, Zarazaga, A, et al Mechanical ventilation of patients on long-term oxygen therapy with acute exacerbations of chronic obstructive pulmonary disease: prognosis and cost-utility analyses. Intensive Care Med 1999;25,452-457[CrossRef][ISI][Medline]
  10. Van den Boom, G, Rutten-van Mölken, MPMH, Molema, J, et al The cost effectiveness of early treatment with fluticasone propionate 250 mcg twice a day in subjects with obstructive airway disease: results of the DIMCA program. Am J Respir Crit Care Med 2001;164,2057-2066[Abstract/Free Full Text]
  11. Griffiths, TL, Phillips, CJ, Davies, S, et al Cost effectiveness of an outpatient multidisciplinary rehabilitation programme. Thorax 2001;56,779-784[Abstract/Free Full Text]
  12. National Emphysema Treatment Trial Research Group.. Cost effectiveness of lung-volume-reduction surgery for patients with severe emphysema. N Engl J Med 2003;348,2092-2102[Abstract/Free Full Text]
  13. Groen, H, van der Bij, W, Koëter, GH, et al Cost-effectiveness of lung transplantation in relation to type of end-stage pulmonary disease. Am J Transplant 2004;4,1155-1162[CrossRef][ISI][Medline]
  14. Oostenbrink, JB, Rutten-van Mölken, MPMH, Monz, BU, et al Probabilistic Markov model to asses the cost-effectiveness of tiotropium in COPD patients in different countries. Value Health 2005;8,32-46[CrossRef][ISI][Medline]
  15. Sin, DD, Golmohammadi, K, Jacobs, P Cost-effectiveness of inhaled corticosteroids for chronic obstructive pulmonary disease according to disease severity. Am J Med 2004;116,325-331[CrossRef][ISI][Medline]
  16. Borg, S, Ericsson, A, Wedzicha, JA, et al A computer simulation model of the natural history and economic impact of chronic obstructive pulmonary disease. Value Health 2004;7,153-167[CrossRef][ISI][Medline]
  17. Hoogendoorn-Lips, M, Rutten-van Mölken, MPMH, Hoogenveen, RT, et al A dynamic population model of disease progression in chronic obstructive pulmonary disease. Eur Respir J 2005;26,223-233[Abstract/Free Full Text]
  18. Decramer, M, Celli, B, Tashkin, DP, et al Clinical trial design considerations in assessing long-term functional impacts of tiotropium in COPD: the UPLIFT trial. COPD J Obstruct Lung Dis 2004;1,303-312
  19. Charlson, ME, Pompei, P, Ales, KL, et al A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40,373-383[CrossRef][ISI][Medline]
  20. EuroQoL, Group EuroQol: a new facility for the measurement of health-related quality of life. Health Policy 1990;16,199-208[CrossRef][ISI][Medline]
  21. Brooks, R, EuroQol, Group EuroQol: the current state of play. Health Policy 1996;37,53-72[CrossRef][ISI][Medline]
  22. Dolan, P Modeling valuations for EuroQol Health States. Med Care 1997;35,1095-1108[CrossRef][ISI][Medline]
  23. Walters, SJ, Brazier, E Comparison of the minimally important difference for two health state utility measures: EQ-5D and SF-6D. Qual Life Res 2005;14,1523-1532[CrossRef][ISI][Medline]
  24. Jones, PW, Quirk, FH, Baveystock, CM, et al A self-complete measure for chronic airflow limitation: the St George’s Respiratory Questionnaire. Am Rev Respir Dis 1992;145,1321-1327[ISI][Medline]
  25. Jones, PW, Lasserson, D Relationship between change in St George’s Respiratory Questionnaire scores and patient’s perception of treatment efficacy after one year of therapy with nedocromil sodium [abstract].Am Rev Respir Crit Care Med 1994;149,A211
  26. Quanjer, PH, Tammeling, GJ, Cotes, JE, et al Lung volumes and forced ventilatory flows: Report Working Party Standardization of Lung Function Tests, European Community for Steel and Coal. Official Statement of the European Respiratory Society. Eur Respir J Suppl 1993;16,5-40[Medline]
  27. GOLD. Global Initiative for Chronic Obstructive Lung Disease. Workshop Report: Global Strategy for the Diagnosis, Management and Prevention of COPD. Available at: www.goldcopd.com. Accessed July 6, 2005
  28. Landis, JR, Koch, GG The measurement of observer agreement for categorical data. Biometrics 1977;33,159-174[CrossRef][ISI][Medline]
  29. Antonelli-Incalzi, R, Imperiale, C, Bellia, V, et al Do GOLD stages of COPD severity really correspond to differences in health status? Eur Respir J 2003;22,444-449[Abstract/Free Full Text]
  30. Van Manen, J, Bindels, JE, Dekker, F, et al The influence of COPD on health-related quality of life independent of the influence of comorbidity. J Clin Epidemiol 2003;56,1177-1184[CrossRef][ISI][Medline]
  31. Harper, R, Brazier, JE, Waterhouse, JC, et al Comparison of outcome measures for patients with chronic obstructive pulmonary disease (COPD) in an outpatient setting. Thorax 1997;52,879-887[Abstract]
  32. Stavem, K Reliability, validity and responsiveness of two multiattribute utility measures in patients with chronic obstructive pulmonary disease. Qual Life Res 1999;8,45-54[Medline]
  33. Brazier, J, Roberts, J, Tsuchiya, A, et al A comparison of the EQ-5D and SF-6D across seven patient groups. Health Econ 2004;14,873-884
  34. Johnson, JS, Luo, N, Shaw, JW, et al Valuations of EQ-5D health states: are the United States and United Kingdom different? Med Care 2005;43,221-228[CrossRef][ISI][Medline]
  35. Luo N, Johnson JA, Shaw JW, et al. A comparison of the EQ-5D preference weights predicted by the D1 (US) and the N3 (UK) models. Med Decis Making 2006 (in press)
  36. Celli, BR, Cote, CG, Marin, JM, et al The body mass index, airflow obstruction, dyspnea and excercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004;350,1005-1012[Abstract/Free Full Text]
  37. Bonsel, GJ, Janssen, MF, Birnie, E Mild Diseases & Ailments Study (MiDAS): Ziektelastbepaling van 27 vermoedelijk ‘lichte’ ziekten ten behoeve van rationale beperking van het geneesmiddelenpakket: samenvatting. 2003 Academic Medical Centre, Department of Social Medicine. Amsterdam, the Netherlands:




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rutten-van Mölken, M. P.M.H.
Right arrow Articles by Monz, B. U.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rutten-van Mölken, M. P.M.H.
Right arrow Articles by Monz, B. U.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS