Background
Chronic obstructive pulmonary disease (COPD) is a common chronic condition that severely affects patients' health-related quality of life (HrQoL). With a prevalence of more than 13% in those aged 40 years and older in Germany, COPD is one of the most frequent causes of morbidity and mortality [
1]. Exacerbations, acute worsenings of symptoms, have serious health consequences and are associated with an increased decline in lung function, hospitalization and even death [
2]. It has been shown that on average, patients with severe or very severe COPD experience about 2.7 exacerbations per year, more than 10% of which require hospitalization [
3].
A number of studies measure HrQoL in patients with COPD during stable phases of the disease using e.g. the generic EuroQol 5 dimension (EQ-5D) [
4,
5], the Short Form 12 (SF-12) [
6], or the disease-specific St George's Respiratory Questionnaire (SGRQ) [
4,
7], but only few examine the effect of acute exacerbations. For stable disease, the EQ-5D's ability to discriminate between GOLD stages has been shown [
4], but ceiling effects have also been reported [
4,
8]. However, properties of this instrument have not been assessed for acute exacerbations.
The most frequently used instruments to measure HrQoL during acute exacerbations are the SGRQ [
9,
10] and the Chronic Respiratory Disease Questionnaire (CRQ) [
10,
11]. As disease-specific instruments such as the SGRQ do not capture aspects of HrQoL unrelated to the specific disease and its consequences, utility values based on the SGRQ can not be compared with those based on generic instruments. To calculate quality-adjusted life-years (QALYs) for health-economic evaluations, only utilities based on generic instruments should be used. Utility values for exacerbated COPD based on generic instruments are essential when the cost-effectiveness of interventions is to be assessed that reduce the frequency or severity of exacerbations in COPD. To date, there is only one study that employs the EQ-5D for this question [
12]. Yet it is not clear whether instruments such as the EQ-5D or the SF-12 are suitable to measure the impact of exacerbations on HrQoL.
Therefore, the aim of this study was to evaluate HrQoL using the two generic instruments EQ-5D and SF-12 and the disease-specific SGRQ in patients with severe and very severe COPD hospitalized for exacerbations, and to compare their results with regard to completeness, proportion with best and worst health state, sensitivity to change and discrimination between groups of different disease severity.
Discussion
One disease-specific (SGRQ) and two generic (EQ-5D, SF-12) instruments were used to measure HrQoL at admission and discharge in patients with severe and very severe COPD hospitalized for acute exacerbations. Objectives of the study were to evaluate HrQoL during acute exacerbations and to compare the 3 instruments with regard to completeness, proportion with best or worst health state, sensitivity to change and discrimination between groups of different disease severities.
The main problem of the SF-12 in this self-administered setting was the high proportion of missing values. For less than 55% of all patients, a comparison of utility values at admission and at discharge was possible. Half of all missing utilities were due only to 2 items of the SF-6D. As one missing item precludes the calculation of a utility, and since the proportion of missing values increases with age, a high percentage of missing utilities was observed in this relatively old patient group. For the SGRQ on the other hand, subscores and a total score can still be calculated with up to 6 missing items per subscore. Therefore, completeness was best for the SGRQ in spite of its length of 50 items.
Worst possible scores were observed for the symptom score of the SGRQ at admission, as well as for the activity subscore at both time points. However, whereas this is tolerable for a disease-specific instrument in this severely ill patient group, the relatively high proportion with full health according to the EQ-5D at discharge poses a more serious problem. Ceiling effects were known to be present in stable phases of less severe COPD stages [
4], but in our study the best possible state in EQ-5D was observed in severe and very severe COPD at discharge, while corresponding SF-6D scores were as low as 0.55 and patients reported health restraints in vitality (100%), mental health (83%) and physical functioning (80%). Therefore, the EQ-5D might not be sensitive enough to capture the health restraints that without doubt are still present in patients with severe COPD at discharge from a hospital-treated exacerbation. However, our results are in line with other studies. In a study on patients from various disease groups, full health in EQ-5D was observed in 9% of patients, of whom 92% reported health restraints in SF-6D dimension vitality, 65% in mental health, 71% in physical functioning [
22]. And in a study on liver transplant patients, full health in EQ-5D was observed in 16%, of whom 94% reported health restraints in vitality, 51% in mental health, 80% in physical functioning [
23]. SF-6D scores in individuals with full health in EQ-5D ranged as low as 0.57 and 0.56, respectively.
Sensitivity to change was generally good in all instruments. However, while the SF-12 PCS improved significantly from admission to discharge, the MCS showed only marginal changes. The reason for this might be that it requires more time for patients' mental state to recover from an exacerbation than it does for their physical condition. Differences between the instruments may be due to different reporting periods: the EQ-5D asks for the patient's immediate situation ("today"), whereas the reporting periods of the SF-12 and the SGRQ were 4 weeks and 3 months, respectively. In exacerbations, which usually show an acute onset of health status deterioration, a recall time of 4 weeks or more may be too long to detect these rapid changes. On the other hand, it is not clear how much patients pay attention to the respective reporting periods, particularly when answering the 3 instruments consecutively.
Differences between disease stages were observed for the MCS, but not for the PCS. While these differences are known to be present in stable phases of the disease [
6], they may be reduced in the physical dimension during acute exacerbations.
As previous studies [
8,
24], we found a higher variance in utility values derived from the EQ-5D compared to the SF-6D. Also, mean EQ-5D utilities were higher than SF-6D utilities. Grieve et al. name the absence of a vitality dimension in the EQ-5D as a possible explanation [
24]. In our study, more than 90% of all patients reported in the SF-12 at admission that they had a lot of energy 'some of the time' or less, at discharge, this still held for more than 80%. The effect of this aspect of HrQoL may not fully be captured by the EQ-5D, which may result in higher utilities.
EQ-5D utilities based on the UK tariff and VAS values at admission in our study were considerably higher compared to those by O'Reilly et al. [
12], and still somewhat higher at discharge. One reason for this might be that those patients who were most impaired could not take part in the study, because they were not able to complete the questionnaire. As we applied 3 instruments instead of the EQ-5D only, our questionnaire was considerably longer, which may have caused more patients to deny participation. This probably underestimates the health impairment by exacerbations. Yet for patients participating in the study, no association between the presence of missing utility values and disease severity was observed. Also, although the time frame for assessing the questionnaire was within 3 days of admittance in both studies, there may have been differences in average time. In our study, about 30% of patients completed the questionnaire on the day of admittance, and about another 50% on the day after. O'Reilly et al. do not specify this issue, but if the majority of patients completed their questionnaire on the day of admittance, this may explain some of the differences observed. Differences in HrQoL might also be due to differences in the countries' health care systems. If patients are admitted earlier in Germany than they are in the UK, this could result in better HrQoL. Another reason might be that our staging was based on lung function at discharge, while O'Reilly used the last recorded FEV
1 in General Practitioner notes which might result in less severe COPD stages. However, HrQoL at discharge as measured by the EQ-5D (UK tariff) and the VAS in our study were only slightly below those observed by Rutten-van Mölken et al. in stable stage III and IV [
4], as were mean values for SF-12 PCS and MCS at discharge compared with Garrido et al. [
6], which might indicate appropriate staging. However, HrQoL at discharge according to all SGRQ scores was still considerably worse than in stable disease phases [
4].
Doll et al. found SGRQ scores of 40 to 80 during exacerbations [
9]. The results of the present study agree with these findings with values between 50 and 80 at discharge. Higher values were observed at admission, ranging from 60 to 90. For patients with chronic bronchitis, Doll et al. found a decrease by 7, 2, 7 and 8 points for the SGRQ total, symptoms, activity and impact score, respectively, from exacerbation to stable phase [
9]. In our study, this reduction was similar with 6, 4, 5 and 8 points.
One limitation of our study is that missing values were more frequent in women and in older patients. Especially in the SF-12, this probably resulted in an overestimation of HrQoL, whereas the proportion with worst health state may be underestimated. Sensitivity to change might also be affected, if those more likely to be missing are also more likely to improve from admission to discharge. Furthermore, the most severely ill patients could not be included in the study since they were not able to answer the questionnaire. This is most likely to lead to an overestimation of HrQoL in our study, but the amount of this bias is not known. Also, the time course of patients' health state after discharge was not observed within this study. This information is useful to calculate QALY loss associated with severe exacerbations more precisely. Results from O'Reilly in a subgroup of patients indicate that utility values had dropped 3 months after discharge. However, further research is needed to confirm these findings for other instruments such as the SF-12.
Also, no information on patients' comorbidities was available. However, Rutten-van Mölken et al. found no significant differences in the number of concomitant diagnoses or the Charlson comorbidity index between GOLD stages [
4], so the differences in HrQoL between disease stages that were observed can be expected to persist after an adjustment for comorbidity.
In all, this study showed that generic instruments as the EQ-5D or the SF-12 are suitable to measure HrQoL during acute exacerbations and show good properties for most criteria.
Competing interests
The authors declare that they have no competing interests. This work was supported by the "Kompetenznetz Asthma/COPD (Competence Network Asthma/COPD)" funded by the Federal Ministry of Education and Research (FKZ 01GI0881-0888).
Authors' contributions
PM participated in the design and the coordination of the study, performed the statistical analysis and wrote the manuscript. NW participated in the design and the coordination of the study. RH supervised the study and assisted in writing the manuscript. All authors read and approved the final manuscript.