The results of this study show that patients with more breathlessness and depression reported lower physical health. Moreover, those with better lung function but more anxiety and depression reported lower mental health. These results also show that symptoms explain a greater proportion of the variance in subjective health status than do demographics, physiological variables, or physical function. According to the biopsychosocial model, no one single factor explains the subjective health status. Instead, it reflects the complexity of the associations between biological and psychosocial factors, progresses of symptoms, to clusters of symptoms, to syndromes, and finally to diseases with specific pathogeneses and pathology .
This is the first study to explore a multivariate perspective on subjective health status in COPD patients based on Wilson and Cleary's  conceptual model of biopsychosocial relationships to subjective health status. In this study, a conceptual model was established based on Wilson and Cleary's framework and previous COPD-specific studies. In the model, there is a unidirectional relationship between the biological and physiological variables, symptoms, and physical function, which leads to the subjective health status (Figure 1). According to Osoba , there is a reasonably strong correlation between the proximal components of Wilson and Cleary's model (such as symptoms and physical function) and a weaker correlation between the more distant components (such as the physiological variables and subjective health status). There may also be a bidirectional relationship between some components . There is not necessarily a strong association between the objective physiological indicators of the disease and the patients' subjective experience of their health status. In this respect, studies of COPD patients have found weak associations between objective measures of disease, symptoms, physical function, and subjective health status [11, 13, 22, 43].
Relationships between age, sex, and physiological variables
The results of this study show insignificant associations between age, sex, and oxygen saturation. Conflicting results have been found in previous studies. De Torres et al.  found that women suffering from COPD tended to have better oxygen saturation than men. Conversely, Di Marco et al.  found an insignificant association between sex and oxygen saturation. Insignificant associations between age, sex, and oxygen saturation suggest that the women and men studied were at the same stage of COPD [5, 44].
Relationships between age, sex, physiological variables, and symptoms
The observation that older COPD patients report less breathlessness than younger is in contrast to Stavem et al  who not find any such association. This finding may be due to response shift . Patients adapt over time in relation to goals, expectations and values, and their perceptions of symptoms may therefore change. Furthermore, the process of learning to cope with health problems is well-known in chronically ill patients . Older COPD patients may have suffered longer from COPD and anticipate illness as part of growing old. Moreover, health- related stressors may not produce the same reactions in elderly. Although older patients may have difficulties due to breathlessness, they may see physical and functional disability as result in growing older [8, 47]. The fact that women tend to report more anxiety than men is not surprising because there is ample evidence of a higher prevalence of anxiety among woman than among men [48, 49]. That women report more anxiety than men is also consistent with previous studies of COPD patients [13, 43]. In this study, small and insignificant associations were identified between physiological variables and symptoms. These results are in accordance with previous studies of COPD patients, which found small and insignificant associations between physiological measurements and breathlessness, anxiety, and depression[7, 11, 22, 43, 45].
Relationships between age, sex, physiological variables, symptoms, physical function, and subjective health status
Patients with less breathlessness and depression reported better physical health, and those with less anxiety and depression reported better mental health, which is consistent with previous studies of COPD patients [8, 45, 50]. However, it is surprising that lung function was not associated with physical health and that better lung function was associated with worse mental health. The same trend was observed in other studies of COPD patients, although the association was not statistically significant [45, 51]. The results of our study show that the association between symptoms and subjective health status was stronger than the association between physiological variables and subjective health status, and this supports the multidimensional impact of COPD on subjective health status . Furthermore, the fact that subjective health status represents something other than physiological and pathological factors is useful information for consideration in the treatment and care of COPD patients [7, 45, 52].
In this study, age, sex, lung function, oxygen saturation, breathlessness, anxiety, depression, and exercise capacity influenced subjective health status. However, according to previous studies of COPD patients, body mass index, education, social status, sleeping habits, and co-morbidity could be important supplementary factors affecting subjective health status in this sample [10, 12, 13]. This study is limited to some degree. The sample size was quite small, which restricts the number of factors included in the multivariate testing of subjective health status . Because of the cross-sectional design, no absolute conclusions can be drawn about causality or the directions of the relationships between many of the variables . The patients included in this study were awaiting participation in a pulmonary rehabilitation programme, and were thus not a representative sample of all COPD patients. The strength of this study is its multivariate approach to explaining subjective health status. According to the biopsychosocial model, subjective health status is associated with physiological factors as well as symptoms and psychosocial factors .
Implications for clinical practice
The results of this study indicate that symptoms are very important to patients' subjective health status, which in turn supports the view that a pulmonary rehabilitation programme focusing on the management of symptoms, such as breathlessness, anxiety, and depression, is required to alleviate symptoms and increase subjective health status.
A model that explains the relationships between different outcomes is important in clinical practice to correctly interpret the results of outcome assessments [4, 42]. For example, if subjective health status is determined by symptoms and physical function, then symptoms and physical function should be treated . In COPD, symptoms such as breathlessness, anxiety, and depression are usually evident before there is a reduction in subjective health status. However, it is more difficult to determine the causal direction between breathlessness, anxiety, depression, and physical function, and as breathlessness, anxiety, and depression may be caused by a decrease in function [52, 56].