Health-related quality of life in COPD: why both disease-specific and generic measures should be used

In chronic obstructive pulmonary disease (COPD), measures of health-related quality of life (HRQL) are nowadays frequently used as descriptive instruments or as outcome measures 1–6. Since cure is still impossible for most COPD patients, a major goal of care is to improve HRQL. However, limited attempts have been made to identify the factors that are related to the different aspects of HRQL in COPD 7. Such analyses may also shed light on the properties of the questionnaires that are studied and also show their internal relations.

In the case of generic HRQL measures, mainly the Sickness Impact Profile (SIP) 8, three important determinants have been found: exercise tolerance 1–3, symptoms 3, 4, 9, 10 and emotional status 2, 3, 10. Education 2 and social support 10 have also been shown to explain some of the variance in self-reported health status. All the studies 1–4, 9, 10 report little or no correlation between the SIP and spirometry, possibly due to a limited range of ventilatory restriction in the populations studied.

Concerning the disease-specific St George's Respiratory Questionnaire (SGRQ) 11, exercise tolerance 11–13, dyspnoea 11, 13, 14, anxiety 11, 13, wheezing 11, body composition 14, fatigue 15 and coping strategies 12 have all been reported as significantly related. The correlation to forced expiratory volume in one second (FEV₁) is somewhat stronger than to the SIP 11–14. Moreover, oxygen tension in arterial blood (P_a,O2) has been found to be significantly related to total SGRQ in patients with hypoxaemia 16.

Emotional wellbeing has been examined in two studies. In one 17, it was reported to be significantly related to symptoms, especially fatigue, and in the other to coping pattern 12.

Thus, several factors have been reported to relate to HRQL in COPD, but they have not been systematically investigated. The aim of the present work is to identify the factors most highly related to HRQL in a well-defined COPD population, stratified to represent a wide range of disease severity. The variables that were analysed were selected to cover an expected continuum from physiology, disease-specific problems, functional (generic) and emotional (generic).

Patients and methods

Statistical methods

Descriptive statistics were computed for each of the analysed variables. Two methods were used to identify the factors most associated with HRQL. 1) Parametric and nonparametric (Spearman) correlations were calculated to study the relationships between the HRQL measures and other variables. As no major discrepancies were found, Pearson correlation coefficients have been reported throughout. 2) Stepwise multiple regression 27 was used to identify the factors that were most strongly related to HRQL scores. All variables with a p-value of <0.15 were entered in the regression model, and all results with p<0.05 were accounted for. The following measures of HRQL were used as dependents: SIP overall score, SIP physical dimension, SIP psychosocial dimension, SGRQ total score and total MACL score. In order to avoid contamination, the following variables were excluded from the regression analyses: for the three SIP dimensions, the 12 separate SIP categories; for the SGRQ total score, the three separate SGRQ subscales; and for the MACL score, HADS anxiety and depression scores and SIP emotional behaviour. The analyses then followed the same hierarchical sequence of inclusion of independents. In the first regressions, only measures of lung function were included. Walking distance (6MWD) was then added, followed by the measures of emotional distress (HADS and MACL scores), and finally the SIP and/or SGRQ categories. Thus, for each of the five dependents studied, four separate analyses were performed. Note that the specification of the various regression models as dependents and independents does not imply that the variables are conceptually related in a causal sense.

Results

Baseline characteristics are given in table 1⇑. The patients were all former or current heavy smokers with a disease duration of several years. Of the patients, 21 had an FEV₁ of <30% pred, 25 had values 30–50% and 22 had values 50–80% pred. Seventeen patients (25%) had chronic or intermittent respiratory insufficiency. Only 14 of the 37 patients below retirement age (65 yrs) were gainfully employed. Forty per cent of the patients were living alone and the remaining 60% cohabited with a partner.

The impact of the disease was considerable according to the SGRQ and slight to moderate according to the SIP. In the HADS, pathological levels (scores>10) were seen in 13% of the subjects (n=9) on the anxiety scale and 7% of the subjects (n=5) on the depression scale.

Data 12 months later showed only small changes. Statistical significance (p<0.05) was found for: mean FEV₁ % pred, which decreased from 39.9 to 37.2% pred; mean vital capacity (VC), which decreased from 66.6 to 62.2% pred; and the mean of the SIP psychosocial scale score, which increased from 6.8 to 8.7.

Pearson correlations for the overall SIP scores, the SGRQ total scores and FEV₁ at baseline are given in table 2⇓. The corresponding information on MACL scores is shown in figure 1⇓. The SIP overall column in table 2⇓ gives a sequence of correlations from low with lung function, to more substantial with overall MACL, HADS depression scores and 6MWD, to stronger with the SGRQ activity, impact and total scores. SGRQ total scores showed a similar sequence and magnitude of correlations. Here, too, the strongest correlations were to other HRQL measures, in this case the SIP physical categories (mobility, ambulation and home management) and the SIP sleep/rest category.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Pearson correlations between mood scores (Mood Adjective Check List (MACL)) and data representing physiological, functional and psychological status. HADS: Hospital Anxiety and Depression Scale; SIP: Sickness Impact Profile; SGRQ: St George's Respiratory Questionnaire; 6MWD: 6-min walking distance; VC: vital capacity; P_a,O2: oxygen tension in arterial blood; P_a,CO2: carbon dioxide tension in arterial blood; FEV₁: forced expiratory volume in one second.

Of all the HRQL scales, SGRQ activity, the SIP ambulation and SIP home management scores correlated best with FEV₁, r=0.49, 0.49 and 0.46, respectively, ranking equal to or stronger than 6MWD. Of the lung function data, VC % pred correlated best with the SIP overall scores and the SGRQ total scores; r=0.50 and 0.44, respectively. In all series of correlations, 6MWD is located among the variables measuring self-rated functional status.

The FEV₁ column illustrates an orderly sequence of correlations, starting with lung function, SGRQ activity, total and impacts scores, 6MWD, SIP physical and SIP psychosocial scores, and finally scores measuring emotional function. This sequence is almost completely reversed for the MACL scores (fig. 1⇑), with lung function at one end and anxiety and depression scores at the other. The correlation between FEV₁ and the SGRQ and the SIP overall scores were −0.42 and −0.32, respectively.

Little or no correlation was found between age, sex, domestic arrangements and the HRQL scores.

Discussion

A systematic investigation was conducted to identify factors related to HRQL in COPD patients. Three generic instruments covering basic dimensions of HRQL were supplemented with a disease-specific measure and clinical indicators. Together, these measures examined the effects of COPD on lung function, symptoms, physical function, psychosocial function and emotional status. The presented results show that the measures within each of these groups are highly related internally, but only moderately to poorly between each other. However, an orderly sequence of relationships between these groups of measures can be outlined. Measures of lung function, showed moderate correlations with measures of physical function, small with measures of psychosocial function and nonsignificant correlations with measures of emotional status. This continuum is most evident in the correlation series for FEV₁ (table 2⇑) and for MACL (fig. 1⇑). This model of relationships is of importance for guiding the choice of relevant HRQL instruments. For example, an intervention designed to increase FEV₁ can be expected to have an impact on physical function, but less so for the dimensions further along the continuum. In the same manner, a successful psychological intervention should produce effects on the variables measuring emotional status and psychosocial function, but not on those more “distant” from this target. It can be concluded that a comprehensive assessment of the effects of COPD requires a battery of instruments that not only taps the disease-specific effects in lung function and symptoms, but also the overall burden of the disease on everyday functioning and emotional wellbeing.

In this study, a separate dyspnoea scale was not used since the activity section of the SGRQ has been shown to be closely associated with dyspnoea scales 28. The associations found between HRQL and age, sex and marital status were nonsignificant and/or small, and did not confound the main results of this work.

The presented results show, as expected, that overall SIP scores relate most strongly to other SIP scores, the SGRQ total scores to other SGRQ scores, the MACL scores to other psychological scores, and FEV₁ to other lung function data. The SIP work limitations category is an exception in having no correlations to any of the other variables. This is probably due to the relatively small number of patients in working age, as well as to the peculiarities in the scoring algorithm for this category.

There is a strong correlation between the overall SIP and SGRQ total scores, and also between most of their respective subscores. This was also expected since both measures (except SGRQ symptoms) are intended to measure functional aspects of HRQL. In contrast, SGRQ symptoms scores had, at most, only modest correlations with the overall SIP and MACL scores, and no association was found with FEV₁. This is in accordance with the conceptual model of HRQL, where symptoms are distinctly separate from function and wellbeing. It must also be noted that 6MWD was located close to the self-reported data on physical function, correlating most strongly with the SGRQ activity and the SIP physical dimension scores (r=−0.63 and −0.67, respectively). Moreover, 6MWD is the only clinical measurement related to MACL. Thus, 6MWD appears to represent a more multidetermined measure in which physiological, physical and emotional aspects are involved, and consequently, complements other kinds of exercise tests and HRQL measures.

Stronger relationships were found between lung function and SIP and SGRQ scores than most other investigators 1–4, 9–14. This is probably due to differing characteristics of the study populations. The population used in the present study consisted of patients with a very wide FEV₁ % pred (12–79%), while in other studies the spectrum of FEV₁ has been smaller, possibly masking the effects of pulmonary function on HRQL. It was found that FEV₁ related best to the SGRQ activity section (r=0.49) and the SIP ambulation category (r=0.49), while VC related best to the SIP physical dimension (r=0.52). VC also correlated better with overall SIP scores than FEV₁. The reason for this is unclear, however, this finding has also been reported by Jones et al. 3.

The regression analyses showed that although pulmonary function is important, its relation to health-related quality of life is significantly modulated by three of the factors examined in the present study: 6-min walking distance, dyspnoea-related limitation and depression scores (Hospital Anxiety and Depression Scale). The logical approach must be to study the factors that are in turn associated with these main factors. This has been done to some extent, but not systematically. Dyspnoea intensity as measured by the visual analogue or Borg scales, especially the relationships of visual analogue or Borg scales to work load or minute ventilation, has been thoroughly investigated in previous studies. For example, pulmonary hyperinflation has been described as an important predictor 29. The dyspnoea threshold scales, which are conceptually closer to functional status, have been studied to a lesser degree. As a rule, only modest correlations to lung function have been found 28. The finding that body composition influences health-related quality of life as measured by the St George’s Respiratory Questionnaire via dyspnoea is a further step 14. Factors behind 6-min walking distance have not been systematically investigated, but two aspects have been described: body composition and psychological factors 30, 31. Finally, the Hospital Anxiety and Depression Scale depression scores were related to most aspects of health-related quality of life. This was especially true for the emotional wellbeing or Mood Adjective Check List. In the present study, patients with chronic obstructive pulmonary disease had slightly lower Mood Adjective Check List scores compared to healthy controls 5. However, among the patients, lung function did not correlate with Mood Adjective Check List scores. This means that when the disease is established, other factors tend to predominate. Therefore, factors behind low mood and/or depression in chronic obstructive pulmonary disease should be further investigated. One study 12 has described a coping factor, namely emotional reaction (to disease), as one determinant of mood status. In addition, another aspect, fatigue, measured by the Sickness Impact Profile sleep/rest category appears to be related with the St George’s Respiratory Questionnaire and Mood Adjective Check List scores. This corroborates the findings of two other studies 15, 17.