This study reports population based findings about factors associated with HRQoL impairment in a sample of older persons from across urban-remote areas of NSW, as well as examining the impact of contextual factors and cardiovascular-affective condition comorbidity on these outcomes. Investigation of the moderating effect of cohort membership on these models revealed that, with the exception of factors directly associated with cohort membership (i.e. remoteness) there were few differences in the magnitude or significance of model predictors, supporting the validity of combining data across different cohorts. Current models explained approximately a quarter of the variance in physical and psychological impairment reported by participants, with demographic indices, cardiovascular and affective conditions and other health indicators accounting for most of the model variance. Physical impairment was consistently associated with increased age, male gender, lower education, being unmarried, retirement, a lifetime history of stroke, heart attack/angina, depression/anxiety, diabetes and hypertension, as well as current obesity and low social support. Psychological impairment was consistently associated with lower age, being unmarried, a lifetime history of stroke, heart attack/angina and depression/anxiety, as well as low social support. In the primary analyses, the influence of participant remoteness on HRQoL impairment was relatively small and varied with the model being examined. Remoteness tended to be more strongly associated with lower psychological impairment, reflecting overall urban versus rural differences, rather than more subtle changes in remoteness.
Previous research in the Australian context has also observed a rural advantage for psychological [56, 57], but not physical quality of life  compared to urban participants. In line with previous research regarding the impact of comorbid physical and mental health conditions , trends suggesting that the lifetime occurrence of both affective and cardiovascular conditions was associated with greater psychological impairment than was explained by either diagnosis alone (p = .010) were observed, although they were not significant in the rural sub-analysis (p = .053). Moreover, there was no evidence that the impact of cardiovascular and affective conditions were influenced by remoteness. However, these findings must be interpreted with caution. Several factors suggest that chronic conditions may be more likely to be miss-classified or of greater severity in our rural populations, particularly: the decreased probability of receiving a diagnosis in remote areas where health services are less accessible [58, 59]; the potential for increased severity at diagnosis; the relative infrequency with which rural populations with health conditions consult their physician; the reduced likelihood of surviving an acute health event; and evidence of migration of persons to less remote areas following diagnosis of mental health conditions . Such biases in diagnostic classification (i.e. more persons incorrectly classified as not having the condition) in regional-remote areas would mean that the influence of disease on quality of life would be underestimated in these areas, though it would be difficult to determine the degree to which such an effect could be offset by the increased severity of diagnosed cases. Further, investigations regarding impact of migration patterns on mental health outcomes suggest migration from rural to urban areas to be associated with increased probability of depression, with decreased contact with friends and neighbours a particular burden in this group . Thus, current results may be best characterised as representing the influence of current community remoteness on persons who have received these diagnoses.
Analyses of NSW Adult Population Health Survey data confirm observations that social capital increases  and experiences of health service accessibility decrease  with remoteness, though no differences were observed for either factor between outer regional and remote/very remote groups. Approximately 32% of persons in outer regional and remote/very remote areas reported difficulty accessing health services when needed compared to 12% living in major cities. Conversely, outer regional and remote/very remote participants reported levels of social capital approximately half a SD greater than their major city counterparts. These findings provide some support for the use of remoteness indices as a proxy for health related community characteristics, although they also tend to suggest that a three category classification would be sufficient (i.e., major city, inner regional, and other areas), and that the current remoteness indices lack greater sensitivity.
The impact of these supposedly opposing forces (increased social capital and decreased health service accessibility) upon health outcomes requires further research, though it is possible that in light of their co-variation, the protective effects of social capital reported here are under-estimated. For example, while we have observed no effect of social capital on physical HRQoL outcomes in our ARMHS sub-analyses, it may be that these community effects are offset by poorer health service accessibility. However, current results are in line with previous investigations of the influence of social capital on HRQoL in Australia, with social capital displaying a particular influence of social capital on psychological HRQoL . Further, while the association of social capital with psychological HRQoL has been observed for both urban and rural participants, evidence suggests that social capital is associated with physical HRQoL only in urban populations . This is consistent with the current null finding regarding the relationship of social capital with HRQoL in our rural sample and may be due to limitations on the capacity of social capital to influence physical health related behaviours in rural areas where health resources are limited.
As discussed above, in line with previous research our sub-analysis of ARMHS data revealed a marginal association of social capital with decreased psychological impairment when controlling for individual level variables such as social support. In our replication of the primary analysis (Additional file 1: Table S3), social capital influenced the association of affective conditions with psychological impairment; as social capital increased, persons with a lifetime diagnosis of depression/anxiety reported less psychological impairment. These effects were observed in the replication despite the fact that other major drivers of wellbeing were included in the model, such as personal social support. This effect was not significant in the extended model which included recent adverse life events, perceived financial difficulty and a marginal interaction of financial difficulty with social capital, suggesting that these variables shared a portion of the variance in psychological impairment accounted for by the social capital and affective disorder interaction. Both marginal interactions observed suggest that interrelated psychological burdens, such as affective disorders and financial difficulties, are similarly ameliorated by social capital. The previously observed trend for comorbid lifetime diagnoses of cardiovascular and affective disorder to be associated with psychological impairment was of similar magnitude but not significant in this subsample (p = .053). The ARMHS cohort sub-analysis also confirmed the influence of recent adverse life events and perceived financial difficulty on HRQoL impairment. Evidence for a moderating effect of social capital on the negative effect of financial difficulty on psychological HRQoL impairment was also observed.
Comparisons between the corresponding analyses (Table 4 vs. Additional file 1: Table S3) show an increment in explained variation of approximately six percent with the inclusion of the additional predictors (adverse life events, perceived financial difficulty and the interaction of financial difficulty and social capital). We acknowledge that the individual level measures of social capital used in these analyses may themselves be influenced by each person’s own psychological HRQoL. However, the patterns of social capital in this sample are consistent with those observed in the NSW data and elsewhere [5–7], namely, increased social capital across rural locations, suggesting this is a potentially health-sustaining quality of rural living, particularly for those with a history of affective conditions. These results are consistent with previously hypothesised and observed ameliorating influences of social capital on stressful situations and events . It is possible that community engagement and support plays a greater role in supporting psychological wellbeing of persons with financial difficulties, suggesting that they have greater engagement with the community in maintaining their psychological wellbeing. Given that these variables were only assessed in the rural-remote ARMHS cohort and not the overall xTEND sample, a limitation of these analyses is that they do not include persons from urban areas and thus the effects and interactions reported here are likely be truncated representations of the effects present in the community at large.
Current findings have practical implications for research into the influence of comorbidity and context on health outcomes, particularly in Australia. This report informs concerns raised by the 2012 National Report Card on Mental Health and Suicide Prevention regarding the physical health of persons affected by mental illness , particularly in light of the burden of cardiovascular disease in these populations. Current results build on past observations of an effect of physical-mental comorbidity on increased days out of role and high health service usage , short term disability and suicidal ideation , decreased HRQoL  and general disability beyond that of diagnoses in isolation . Our results tend to suggest that the disability associated with comorbidity may have a stronger association with psychological HRQoL. In any event, all of the analyses demonstrated clear independent linkages between lifetime cardiovascular and affective conditions and current physical and psychological HRQoL impairment (accounting for between 6.9% and 12.3% of the explained variation).
The strengths of this study are its consideration of data from large community based samples and access to a depth of health information from participants across the spectrum of urban-remote communities that is unprecedented in Australia. Our models include a range of bio-psychosocial risk factors that are not only potentially important for understanding the relationship of physical and mental disorders with HRQoL but which also enable us to tease out some of the contextual, rather than behavioural, influences of remoteness on HRQoL outcomes (such as increased rates of smoking). It should be noted that response rates for these surveys were relatively low, particularly for the oldest persons contacted, among whom the impact of disease on participation is likely to be high. Therefore, we infer that the current subset of participants represents a relatively healthy sub-sample of the population at large, and that the impacts of disease on quality of life depicted here are potentially weaker than those which would be observed in the general population.
The study has several other limitations. Firstly, the use of self-reported lifetime diagnoses for health conditions meant that these variables may reflect a range of symptoms that may not be current and do not account for duration or severity. In the current analyses, self-reported life time diagnoses of affective conditions were among the strongest predictors of both physical and psychological impairment. However, the impact of lifetime health conditions may be variable and the effects of current or recent experiences of these conditions on HRQoL impairment may be greater than those represented here. Secondly, apart from the obvious urban versus rural difference, it is unclear what other cohort related factors may have contributed to differences in mean HRQoL impairment. Finally, it should be noted that our urban population was drawn from a major regional industrial city and thus the current observations of factors influencing HRQoL may not generalise to other urban contexts. In particular, differences between characteristics of major urban locations, which are not necessarily delineated by population density or distance from services, and populations residing within these areas, may impact the experiences of social capital and health service accessibility and their association with health between urban centres .
A strength of the current study is that our primary outcome measure, the AQoL-6D, has been shown to display metric invariance across these cohorts , suggesting that the same constructs are tapped by this measure in both groups. Further, the inclusion of cohort membership in the models did not substantially change the significance or magnitude of model variables as predictors of HRQoL outcomes. Some aspects of the greater impairment reported by the HCS cohort may reflect cohort differences not assessed by our current measures. For example, a component of the observed differences may be a result of the HCS’s focus on recruitment of older persons with an interest in feedback about their health and by the ARMHS protocol of screening out participants with poor hearing and cognitive performance. However, these potential influences are likely to be small. Equally, the residual cohort effects observed in Table 2 (Step 7) may still be due to important elements of urban versus rural differences, but which are simply aspects not captured by the existing remoteness indices. Thirdly, the cross-sectional design of the current study means that we cannot assume that the HRQoL impairments observed here were actually caused by the variables under investigation.
The findings from this study support the influence of social capital on HRQoL impairment, with particular focus on co-existing affective and cardiovascular conditions, two of the most common causes of disease burden in the Australian community. Findings suggest that the psychological impairment experienced by persons affected by lifetime affective conditions may be influenced by comorbid cardiovascular conditions (and vice versa) and by low social capital. Awareness of the compounded effects of physical-mental comorbidity on psychological impairment in these populations is necessary to equitably address their experiences of health conditions. Greater remoteness was associated with higher levels of social capital, reflected in overall urban/rural differences in psychological impairment. The findings suggest that personal social capital may ameliorate the psychological impairment associated with affective disorders and financial difficulties. Initiatives with a focus on social support and social engagement may make help to improve the HRQoL of older persons in the Australian community.