Abstract
Background
Marital status has been clearly linked to subsequent health outcomes in those with established coronary heart disease (CHD). This robust association may be because of both pathophysiological and behavioural mechanisms.
Design
We employed meta-regression to examine the association between marital status and attendance at outpatient cardiac rehabilitation (CR) in published studies.
Methods
We searched electronic databases, for example, Medline and Science Citation Index, for published studies that reported an association between a measure of marital or partnered status and CR attendance in patients with diagnosed CHD.
Results
Eleven studies were identified which incorporated 6984 CHD patients. Being married/partnered was associated with significantly higher odds of attending CR. Using a fixed effects model, the pooled odds ratio of CR attendance was 1.72 [95% confidence interval (CI) 1.50–1.97] for those who were married/partnered. There was no evidence of heterogeneity of effects (P = 0.42) or publication bias (P = 0.12).
Conclusion
CHD patients who are married or have a partner are between 1.5–2 times more likely to attend CR. Associations between marital status and CR attendance may partly explain CHD outcomes.
Introduction
Participation in structured cardiac rehabilitation (CR) has been found to be associated with a 15–28% reduction in all-cause mortality, a 26–31% reduction in cardiac mortality and a 26–29% reduction in myocardial infarction recurrence for those with established coronary heart disease (CHD) [1–4]. Other clinical benefits of CR include reduced atherosclerotic progression, improved cardiac risk factor modification, improved health-related quality of life, improved functional capacity and reduced depression and anxiety [5–9].
Given the strength of this evidence base, CR is now recognized as an important part of the post-discharge treatment of suitable CHD patients [10]. The UK National Institute for Health and Clinical Excellence guidelines state that CR should be equally accessible and relevant to all patients, and should be actively promoted by healthcare professionals [11]. Similar statements have been issued by the American Heart Association [12].
CR attendance, however, has been found to be low, particularly among women and older patients [13–16]. Suaya et al. [17] has recently shown that CR was used in 13.9% of patients hospitalized for acute myocardial infarction in their analysis of Medicare claims in the US. Numerous factors have been found to play a role in predicting referral to and attendance at CR including demographic, psychosocial, clinical and physician factors [13].
There is substantial evidence that a lack of social support is associated with morbidity and mortality in those with CHD. A recent review suggests that low social support confers a risk of 1.5–2.0 for future CHD events in CHD patients [18]. CHD patients with greater structural social support have also been shown to be more likely to attend CR than those with less structural social support [19]. It is possible that this association may be partly explained by a direct relationship between social support and pathophysiological mechanisms [20], and also indirectly through behavioural mechanisms such as attendance at CR [19] and other secondary prevention behaviours. There is, however, considerable variation in the definition and operationalization of social support within these studies. Lett et al. [18] describe how social support is generally disaggregated into two categories: (i) ‘structural or network social support’ that refers to the type, size, density and frequency of contact within an individual's social network and (ii) ‘functional social support’ that refers to the support provided by an individual's social network. The most common measure of structural social support is marital or relationship status. Unmarried or unpartnered individuals have been found to be at an increased risk of adverse CHD-related outcomes [18, 21–23].
Objectives
As marital status is the most commonly measured aspect of structural social support in studies of CR attendance, for example, marital status or partnered versus alone, we meta-analyzed those studies that have reported an association between marital status and CR attendance. The aims of this meta-analysis were three-fold. To provide an estimate of aggregate effect size for the association between marital status and CR attendance across studies. To identify methodological issues in this area of research. To set out an agenda for future research examining the relationship between structural social supports and CR attendance.
Methods
Study selection and data extraction
A search was performed using MEDLINE and Science Citation Index from 1950 until July 2007. The key search terms used were ‘cardiac rehabilitation’ in combination with ‘participation', ‘attendance', ‘uptake', ‘marital', ‘living arrangements', ‘partner’ and ‘spouse'. The inclusion criteria were (i) English language full-length publication in a peer-reviewed journal; (ii) quantitative observational studies; and (iii) measures of marital or partner status and attendance at outpatient CR were provided for patients with diagnosed CHD. We searched the reference list of retrieved articles for relevant studies. Measures of CR attendance could include self-report of attendance and/or hospital records of attendance. Attendance was variously defined across studies but for the purpose of the meta-analysis it was defined as whether patients attended some or all of an outpatient CR programme following a diagnosis of CHD. Measurement of marital/partner status and CR attendance, sample size, sex distribution of sample, adjustment for potential confounders and odds ratios with 95% confidence intervals (CIs) were abstracted by one author (CM.) and verified by two others (M.H. and G.R.). Each study was independently assessed for methodological quality and validity by all reviewers and any discrepancies were discussed until agreement was reached. We contacted the authors for additional data for three studies. The quality of the studies was appraised using a scoring system ranging from 0 to 4 points. The scoring system was based on study design (1 point was allocated if the study was prospective observational), measurement of CR attendance (1 point was allocated if CR attendance was confirmed by the hospital) and controlling for potential confounders (1 point was allocated for adjustment for age and sex, 2 points were allocated for adjustment for age, sex, clinical risk and depression/mental health). Quality scores are displayed in Table 1. We used the Moose framework for meta-analysis of observational studies to guide this review [33].
Data synthesis and analysis
When primary studies provided insufficient data, we calculated estimates of effect size from the data presented in the papers. Odds ratios (ORs) were calculated as measures of effect size. In each case ORs were transformed by taking their natural logarithms (ln) and standard errors were calculated from ln (OR) and corresponding 95% CIs. Provided there was sufficient information, we aimed to perform sensitivity analyses according to study quality score. We simultaneously employed the Q-test for homogeneity between studies, which tests whether the between-study variability in effect sizes exceeds that expected from corresponding within-study variability. We employed fixed effects modelling as there was no evidence of heterogeneity of effects [34,35] to meta-analyze the data. We used ORs from multivariate models with the most complete adjustment for potential confounders whenever this was possible. Finally, to detect publication biases, we measured the degree of asymmetry by using Egger's unweighted regression asymmetry test [36]. All analyses were performed using a widely used meta-analysis program [37].
Results
We identified 11 studies from a potentially relevant 237 that reported an association between marital status and CR attendance, which incorporated 6984 CHD patients. The study sample size ranged from 85 to 3536. Participant attendance at CR ranged from 21% to 73%. Table 1 summarizes the 11 studies that were included in the meta-analysis.
Figure 1 presents a summary of effect sizes for the relationship between marital status and CR attendance in these studies. Six of the 11 studies reported a positive statistically significant association between marital status and CR attendance. The odds ratios reported in Table 1 and used in the meta-analysis were adjusted in Molloy et al. [19], Smith et al. [25] and Barber et al. [14]. The pooled odds ratio for the association between marital status and CR attendance across the 11 studies was 1.72 (95% CI 1.50–1.97). A test for heterogeneity of effects was nonsignificant (P = 0.42). The results of Egger's unweighted regression asymmetry test [36] suggested that publication bias was unlikely (P = 0.12). A sensitivity analysis based on study quality did not suggest a different pattern of findings between low and high quality studies with the summary odds ratio for the six studies scoring 2 or more in the quality score being comparable (odds ratio: 1.72, 95% CI 1.47–2.01) to the overall odds. The summary odds ratio for nine prospective observational studies was 1.76 (95% CI 1.51–2.03).
Summary of studies that examined the relationship between marital status and cardiac rehabilitation attendance
CABG, coronary artery bypass graft surgery; CI, confidence interval; CR, cardiac rehabilitation; CVD, cardiovascular disease; OR, odds ratio.
Discussion
The results from this meta-analysis suggest that CHD patients who are married or have a partner are 1.5–2 times more likely to attend CR. This finding is in line with data showing that people with lower levels of social support are at an increased risk of morbidity and mortality from CHD [21,22]. It is possible that the secondary prevention behaviour of CHD patients with lower social support may explain part of the excess risk for this group [18,19].
There are several conceptual and methodological limitations in this body of work, which should be noted. First, most studies failed to discriminate between initiation (attending at least one CR session) and maintenance (i.e. dropping out or not) of CR attendance. There were fewer women in the reviewed studies than would be expected from CHD samples. In addition, there is considerable heterogeneity in the definition and measurement of attendance at CR within the studies reviewed. Some studies utilized self-report using one item (whether or not the participant attended), whereas other studies explored what proportion of CR sessions the participant had attended.
Overall, the quality of the studies included in this review was poor. In particular, studies often failed to present both adjusted and unadjusted models predicting CR attendance. This prevents readers from assessing whether marital status predicts beyond confounders, or whether there is potential for moderation or mediation effects. Future studies should aim to identify the key moderators and mediators of the relationship between structural social supports and rehabilitation attendance. Another important issue that has been neglected concerns the variability of the content of the CR programme being offered.
Summary of effect sizes for the relationship between marital status and cardiac rehabilitation attendance. CI, confidence interval.
It may be useful to further explore the impact of structural social support, in terms of the wider social network, on CR attendance, particularly as the one study that assessed this found a relatively large relationship between structural social support and CR attendance [19]. It is possible that many CHD patients who are unmarried or without a partner may have structural social support that can compensate for this. Future work should also attempt to answer whether marital/relationship status is important beyond the effects of living arrangements, that is, living alone versus living with others. It is possible that living alone may account for a large proportion of the observed associations.
Both structural and functional social support have been related to better health outcomes in patients with established CHD [18]; however, there is also considerable data showing that the presence of dysfunctional social relationships has been associated with subsequent CHD outcomes in both healthy [38,39] and CHD populations [18]. Future work could examine whether the negative aspects of close relationships influence rehabilitation attendance.
The meta-analysis did not explore the impact of age or sex on the relationship between marital status and CR attendance due to the limited number of studies available. This may be important in the context of previous research that has identified that age and sex have considerable predictive efficacy with regard to CR attendance [40]. The studies that did adjust for these variables, however, found that marital status was predictive beyond the effects of these variables, which included the largest study [25] (N = 3536).
Conclusion
CHD patients who are married or have a partner are between 1.5–2 times more likely to attend outpatient CR. As the data continues to accumulate suggesting that aspects of structural social support such as marital status can have significant effects on morbidity and mortality from CHD that are comparable to other more conventional risk factors, clinicians may wish to consider these data in detail in their long-term management of patients with CHD.
Footnotes
Acknowledgements
This research was supported by the British Heart Foundation.