Findings and their relation to other studies
A considerable number of studies have led to conceptual models of HRQOL in relation to CHF that describe the interactive relationships between pathophysiology, symptoms (e.g. dyspnea, fatigue, ankle swelling), functional and psychological aspects [
57]. According to Rector, people with CHF need to perceive symptoms - abnormal states produced by the pathophysiology - before their HRQOL is affected by CHF, either directly or indirectly. The model acknowledges further the influence of other exogenous factors, such as personality traits, lifestyle demands, culture and multimorbidity that might alter the apparent relationships.
Our results regarding the impact of objective measures such as EF and BNP on HRQOL are similar to previous literature: For example, while BNP correlated bivariately with HRQOL scales, it did not remain in the multivariate regression analyses with included (known) correlates such as NYHA functional class. Therefore, decreased EF or elevated BNP seems not to be sensed by the individual, but the associated symptoms or functional status [
13,
17].
Depression severity as assessed by the summary score of the PHQ-9 had by far the greatest impact on HRQOL variance in all six investigated summary scores or domains, a finding that is in line with previous findings: Gott et al. found depression, measured by a geriatric depression scale, determining generic and specific HRQOL (summary scores of SF-36 and KCCQ) in a cross-sectional study with 542 elderly (mean age 77 years) patients in primary care, a sample where the diagnosis of CHF was validated by the GP [
23]. In our own previous study, we found depression severity (measured by PHQ-9) to be by far the strongest determinant of subscales of SF-36 in a primary care-based sample of 167 patients (mean age 68 years) with ascertained systolic HF [
25].
In this study, NYHA functional class determined KCCQ Functional Status, but not the Physical Component summary of the SF-36. There, COPD, a disease characterised also by 'dyspnea' was independently associated. These findings are supportive for the higher specificity of the KCCQ than the SF-36 with regard to the cardinal symptoms of CHF.
Moreover, our results regarding the role of disease severity (NYHA class) and Depression (PHQ-9 summary score) in disease-specific HRQOL (KCCQ) are in line with Faller et al. [
20]. They investigated the impact of disease severity (represented by NYHA class) and depression (represented by the categorical algorithm for minor or major depression for the PHQ-9 score) in a sample of 233 heart failure outpatients of a university hospital (mean EF 43%, NYHA I/II/III in 15.9/39.5/34.8% of patients). Disease severity and Depression impacted on the full range of KCCQ domains and summary scores, while they found significant interaction in the KCCQ domain (mental) Quality of life. The authors discussed confounding due to the structural overlap between the PHQ-9 and the quality of life domain of the KCCQ, and biased patients' perception in the sense of over-reporting of subjective symptoms. In a consecutive study entailing 206 subjects from the same sample, by using structural equation techniques, Faller et al. could determine the independent extent of impact of disease severity and Depression on the domains (mental) Quality of life and Physical limitation. They found that depression influenced not only the psychological (ß = 0.75), but also the physical domain (ß = 0.3), whereas heart failure severity, as measured by NYHA functional class, affected the physical (ß = 0.44), but merely the psychological (ß = 0.12) domain. Our results are at the least coherent with these findings, as depression severity impacted on KCCQ Functional status and (mental) Quality of life, but heart failure severity (NYHA functional class) only on Functional status, but not (mental) Quality of life, even though Faller et al. chose the KCCQ domain Physical limitation and used structural equation techniques, which is more appropriate as it allows for simultaneous analysis of the impact of multiple explanatory variables on several dependent variables, which was not the focus of our study.
Notably, physician-rated overall morbidity burden (multimorbidity) was considerable (CIRS summary score) and determined HRQOL in 4 of 6 investigated models, i.e. in the PCS of the SF-36, but not MCS, and in all KCCQ models, except for Self efficacy. Studies that investigate co-/multi-morbidity in relation to HRQOL of patients with CHF usually account for single diseases and/or count the no. of conditions and rarely used the CIRS instrument: We included the CIRS additionally to certain single conditions as a physician-rated disease severity aggregate accounting for all body systems would better reflect the patients' disease burden. Regarding generic HRQOL, our results are consistent to a study in primary care with a sample of 238 patients with chronic diseases: The CIRS played a role within the PCS (R
2 0.18), but not within the MCS [
58] and is in line with empirical research that showed that scales that load highest on the PCS are most responsive to treatments that change physical morbidity, whereas scales loading highest on the MCS respond most to drugs and therapies that target mental health [
48].
In a study with patients with CHF, the CIRS score explained only a small part of the variance in one subscale of the SF-36 - Bodily pain [
25]. In this study, no summary measures were analysed rendering a comparison across levels of aggregation difficult, but the different extent of impact of multimorbidity on generic HRQOL is striking. Gott et al. could show the negative impact of multimorbidity on generic and disease-specific HRQOL, but multimorbidity was measured by counting the number of conditions [
23]. The impact of the comorbidity COPD on PCS is congruent to Müller-Tasch et al. and Franzén et al. who also found an impact of respiratory diseases on the physical dimensions of HRQOL [
16,
25]. The finding of history of CABG surgery and its impact on the MCS of the SF-36 is difficult to interpret and should be replicated by further studies. One might argue that a definitive therapy with improved patient outcome regarding symptoms (Angina pectoris) might impact also on generic aspects of HRQOL. However, pectoral angina was not a significant correlate in our study. We conclude that the role found for multimorbidity (measured by CIRS) represents appropriately the perspective of primary care, where patients suffer from more than one index disease, and balancing care and treatment together with the patient is crucial. The CIRS measure seems the best available for primary care [
59], and an electronic version of the instrument provides a practical application either for clinical or research use [
60], but future studies need to clarify its value regarding HRQOL and prognosis in general practice [
61,
62].
Socio-economic status (SES) impacted on KCCQ Self efficacy and SF-36 PCS. Little is known regarding this aspect in relation to CHF. In the study of Gott et al. lower SES impacted negatively on MCS and KCCQ overall scores [
23], and education determined aspects of HRQOL [
11,
15,
26] or compliance [
63], in patients with heart failure. There is general knowledge that social inequalities are associated with morbidity and mortality [
64,
65], and also with health behaviour [
66]. Our results may add another additional finding that higher educated people (SES) have lower levels of emotional and physical distress, reduced by way of paid work and economic resources, which are associated with high personal control [
67].
A larger practice list size was associated with worse HRQOL (MCS and Self efficacy), whereas a higher number of GPs per practice counterbalanced this observed association (in MCS), which is a new finding, as there is little evidence for the impact of organizational aspects, i.e. practice factors, on HRQOL in CHF in routine care. Moreover, our variables on organization (e.g. workload, full-time equivalents, skill mix, degree of delegation, use of chronic care services) were not comprehensive.
However, some literature on practice performance and patient satisfaction shows associations within the organization of primary care that cannot be fully explained [
68‐
70]: In a cross-sectional study of 1188 general practices in The Netherlands, large practices showed no clear association with higher assistant volumes and GPs' workload; large practices had lower assistant volumes, but more chronic care services [
68]. In an observational study of 140 practices across Europe [
69], a larger practice size was associated with lower GPs' workload, but not chronic care organisation (according to the Chronic Care Model). In a further study in 239 Dutch general practices [
70], van den Hombergh et al. found that GPs providing more care time in the practice, and more time per patient and experiencing less job stress are all associated with patients' perceptions of better care and better practice performance. In the context of these results, our finding would make sense. Thus, it warrants consideration as a potential determinant of HRQOL, while at the same time it needs to be replicated in future studies together with a more complete set of other explaining organisational variables.
The EHFScBS scores that aggregate the actual patient-reported self-care behaviour (12 items) were associated with KCCQ Self efficacy (2 items), as expected, although the two instruments ask not completely the same concepts: The instruments differ in that the EHFScBS asks for patient's agreement on defined 12 behaviours, and the KCCQ Self efficacy asks about patient's
sureness on what to do if heart failure worsens and about the patient's
understanding of the ability to prevent worsening of CHF (for example, weighing yourself, eating a low salt diet, etc.). According to the European Clinical Practice Guideline for CHF, HF self care can be defined as
action aimed at clinical stability, avoidance of behaviour that can worsen the condition, and early detection of symptoms of deterioration. Self care management is regarded as a key issue of successful treatment and can significantly impact on health outcomes [
2]. According to the author of a systematic review on self-care and HRQOL in patients with CHF, findings from RCTs of self-care, as an intervention, on HF patient HRQOL do not allow strong conclusions about the benefits because of methodological and conceptual issues [
71]. The author advocated large multi-site RCTs with self-care as the primary intervention.