Background
Coronary artery bypass graft surgery (CABGS) is aimed at alleviating patients' morbidity and prolonging their lives. Given the high success rate of such surgery in achieving these aims, it is clear why the assessment of health related quality of life (HRQoL) is of such importance. Longitudinal studies [
1‐
3] have confirmed that most patients report improved HRQoL following surgery through reduced symptoms, improved functioning and increased participation in activities. However, for a significant minority of patients, this improvement does not occur or the patients report a deterioration in HRQoL [
4‐
7]. Thus, potentially there are a number of trajectories a patient may follow after surgery – improve, maintain the same level, deteriorate, or a combination of these. The purpose of the present study was to determine the optimal number of trajectories that best fit the HRQoL data in a sample of CABGS patients. A second aim was to identify patient characteristics that predict these different trajectories.
Most studies that have investigated change over time in CABGS patients [
1,
3,
8‐
10] have typically used linear models for data analyses, and assumed that individuals follow the same mean trajectory. In repeated measure analysis of variance, the most common approach, no adjustment is made for situations where measurement intervals are unequal. Moreover, traditional methods dismiss individual differences in change as random error [
11]. In contrast, growth modelling is a relatively new technique that can be used to estimate parameters and model fit statistics for both linear and non-linear change. Furthermore, modelling packages, such as Mplus [
12] have sophisticated routines that permit the inclusion of individuals who were not assessed at all time points. These techniques are increasingly being used to model longitudinal data in repeated measures studies. However, they have only recently been used to examine HRQoL trajectories [
11,
13].
Wilson and Cleary [
14] provide a useful organizing framework for categorising predictors of HRQoL. They distinguish physiological/biological factors, symptoms (including emotional and cognitive variables), individual characteristics, such as gender or age, and environmental characteristics, such as provision of services.
The biological/physiological or medical characteristics that have been consistently associated with poorer HRQoL outcomes after CABGS include pre-surgical cardiac functional status, such as the New York Heart Association (NYHA) classification of dyspnoea [
1,
15,
16], current smoking [
1,
10], poor left ventricular ejection fraction [
1,
8], and presence of a comorbidity such as diabetes [
9,
10,
16,
17] or pulmonary disease [
1,
10,
17]. Operational variables, such as complications arising from the surgery, may also possibly impact upon HRQoL, but these variables have not been extensively examined [
1].
It is not surprising that symptoms of depression or anxiety have been associated with a marked alteration in mental HRQoL and worse outcomes after CABGS. Among patients scheduled for CABGS, the prevalence of depressive symptoms is high, with a recent Australian study finding that over half of bypass candidates were diagnosed as depressed [
18]. Preoperative anxiety and depression often predict the occurrence of symptoms or psychopathology after surgery [
4,
19,
20]. Hofer and colleagues [
21] provide evidence that HRQoL and depression are distinctive psychological entities, but that depression represents the most important indirect influence on the course of HRQoL in coronary artery disease patients.
Decreased cognitive function has been recognized to be a major, although probably partly reversible, unintended outcome of CABGS [
3]. A growing number of studies have investigated the relationship between neurocognitive functioning and HRQoL in CABGS patients, and reported mixed findings. Studies using a composite cognitive index have found strong associations between neurocognitive functioning and change in HRQoL following CABGS [
22,
23]. In contrast to these findings, a recent longitudinal study [
3] found no significant association between HRQoL and cognitive performance. Further investigation of the association between cognitive functioning and HRQoL is warranted.
Among the individual characteristics that have been associated with poorer HRQoL outcomes after coronary interventions are age and gender. Younger patients [
24,
25] have reported lower mental HRQoL scores, and older patients have reported lower physical HRQoL scores [
1,
10,
24]. There have been mixed findings with gender, but studies that have controlled for age and other relevant variables have reported lower HRQoL in females [
10,
17,
25‐
27]. Other variables such as living alone [
28], unemployment [
26] and lower socio-economic status [
29] have been associated with poorer mental and physical HRQoL.
One possible environmental influence on HRQoL outcomes is cardiac rehabilitation (CR) program attendance. Despite its efforts to improve the psychological, as well as the physical, well-being of patients [
30], CR attendance has produced inconsistent associations with HRQoL outcomes. A number of studies have found CR attenders had no better HRQoL than non-attenders at follow up [
31,
32], while other studies have found more positive effects [
33].
The aims of the present study were:
a.
to identify the general form of change of HRQoL over time, i.e. linear or non-linear, using growth curve modelling (GCM)
b.
to identify the different trajectories of HRQoL over a six month period for both physical health and mental health, using growth mixture modelling (GMM)
c.
to identify the socio-demographic, medical, psychological or cognitive variables that predict group membership of HRQoL trajectories.
Methods
Patients
Eligible patients were adults on the waiting list for CABGS at The Royal Melbourne Hospital, Australia, between July 2001 and April 2004. Patients were excluded from the study if they were under the age of 18 or over 85 years of age; were subsequently assigned to a non-CABGS procedure; or refused consent. Patients could withdraw from the study at any stage between consent to participate and the final follow-up assessment at six months. The following were criteria for withdrawal: failure to return postal questionnaires, death, physical illness or frailty preventing participation, occurrence of a major medical or neurological illness that would independently affect cognitive outcome, refusal, and unavailability for follow-up for other reasons.
Of a consecutive sample of 444 patients to whom a questionnaire package was posted, 220 (49%) returned the questionnaires. The mean interval between completion of the baseline questionnaire and surgery was 33 days (SD = 34 days; median = 26 days). Of the returned questionnaires, 37 were excluded because, based on medical records, it was ascertained that the patient had not undergone CABGS. In order to ascertain reasons for non-completion or non-return of questionnaires, a random sample of one in three patients (n = 78) was contacted by telephone. Amongst these patients, reasons for non-completion of the questionnaire were language difficulties (42%), refusal (38%), death (12%) or disability (8%). The 261 patients who were either excluded or did not return questionnaires were compared with the 183 included patients on all medical variables, gender and age. Excluded patients were less likely to have either high cholesterol (χ2 = 10.6, df = 1, p =.001) or a positive family history of cardiovascular disease (χ2 = 4.6, df = 1, p = .033). There were no significant differences in all other medical variables, gender and age. Of the 183 returned questionnaires, HRQoL data were available for 182 patients.
To determine whether there was any systematic bias to the analyses due to participant dropout over the six months, the socio-demographic and pre-operative medical characteristics of patients who completed the HRQoL measure at all three time points (n = 117) were compared with those of the patients who did not complete it at any of these time points (n = 65). There were no significant differences between dropouts and completers on any pre-operative medical or socio-demographic characteristics.
Data collection
Institutional ethics committee approval was obtained for this study. Names and addresses of patients were obtained from the list of patients waiting for CABGS at The Royal Melbourne Hospital, Australia. Patients were posted the questionnaire package, which included a covering letter, signed by the Head of the Cardiothoracic Surgery Unit, outlining the study and requesting patient consent to participate. Questionnaires were completed prior to surgery, and again at two and six months after CABGS. All questionnaires were returned by reply-paid post. To reduce response bias, the order of presentation of the instruments within the questionnaire was systematically varied across the study population.
Discussion
The present study supports previous findings [
1‐
3,
7,
50] that, on average, HRQoL improves over time following CABGS. However, this study has gone further than others by demonstrating that this improvement is not necessarily linear or applicable to all patients. The overall linear improvement of the PCS to six months conforms with traditional expectations of recovery following surgery. From a physical viewpoint it appears that, overall, patients experience a steady and consistent abatement of their physical symptoms and resumption of activities in the six months following surgery. In contrast, the overall trajectory of patients on the MCS fits a square-root curve rather than the expected linear function. It appears that patients tend to experience a more rapid improvement in emotional status in the early weeks following surgery, but that their return to normal emotional roles and social functioning is much slower than physical functioning in the subsequent months. This relatively slow return to normality for mental functioning, compared with physical functioning, is consistent with findings of other studies [
3,
7,
29].
For the total sample there were significant medical and psychosocial predictors of baseline physical HRQoL including poorer NYHA functional class, having peripheral vascular disease and poorer emotional state. Baseline mental HRQoL was predicted by only previous cardiac surgery among the medical variables and poorer scores on all emotional scales. No cognitive variables significantly predicted baseline HRQoL. For the total sample there were no significant predictors of change in physical HRQoL over time. The apparent contradictory findings that poorer emotional scores predicted greater improvement over time in mental HRQoL may be explained by the fact that these patients had much lower scores at baseline and had greater capacity to change over time. It does appear that having a partner was beneficial for recovery in mental HRQoL and this finding is consistent with previous studies [
26,
51].
It is important to note that these results only relate to the trajectory of the whole sample. These overall trends, may mask the true picture of recovery following CABGS. The findings of the present study support the hypothesis that identifiable sub-groups of patients exists, each described by a different growth curve, and that these different groups may have different outcomes. Indeed, in this sample, there are two distinct groups of patients for both the physical and mental components of HRQoL. The GMM has identified a larger proportion of the patient population who experience rapid and continued improvement in PCS and MCS scores over time. Of serious clinical relevance, however, the analysis has also identified a smaller group of patients who experience little or no improvement during the first six months following surgery.
The findings of this study lend support to investigation of a range of biopsychosocial predictors, such as that advocated by Wilson and Cleary [
14], in explaining the different trajectories patients follow after CABGS. A physiological variable, previous cardiac surgery, was found to be a strong predictor of
non-improver membership for the MCS. Characteristics of the individual, namely not being in the workforce and having a manual occupation, were predictive of
non-improver membership for the PCS and the MCS respectively. These findings are consistent with those of past studies [
26,
52]. It is known that those patients with a manual occupation tend to have a lower socio-economic status, have poorer dietary and exercise patterns, are more likely to smoke, are less likely to have health insurance and usually have more physically and psychosocially demanding jobs [
53]. These factors, combined with the recent physiological finding that lower socio-economic status is associated with higher levels of the stress hormones, cortisol and epinephrine [
54], may explain the poorer HRQoL outcomes found with these patients.
Symptom status was also highly predictive of trajectory classification. The finding that higher NYHA dyspnoea class was predictive of
non-improver membership for the PCS is also supported by previous studies [
1]. Presumably patients who experience dyspnoea are more likely to have impaired left ventricular function and breathlessness, and are restricted in home, social or leisure activities [
55] resulting in lower HRQoL.
It is also not surprising that lower scores on the POMS vigor-activity scale indicated
non-improver membership of the PCS, and that higher scores on the POMS depresssion-dejection scale predicted
non-improver membership for the MCS. These findings are consistent with two recent studies that have also found higher pre-operative depression scores to predict lower HRQoL six months after surgery [
24,
56]. It is known that depressed patients are less medically compliant [
57], are less likely to exercise [
58], and more likely to engage in unhealthy behaviours such as smoking [
58], all factors which contribute to poorer HRQoL outcomes.
This study adds weight to the growing body of evidence that patients' perceived cognitive abilities may influence HRQoL outcomes. Lower scores on EFQ concentration were predictive of
non-improver membership for the MCS. This finding is supported by a previous study which showed that perceived cognitive function, reflecting ability to concentrate, was a major determinant of HRQoL outcomes in a cross-sectional study of Swedish CABGS patients [
59]. It has been suggested that poorer cognitive function impedes recovery, particularly in the context of CR, because it impacts on the patient's ability to learn or effectively respond to new information [
60]. Further, it has been found that cognitive deficits after CABGS are associated with less ability to engage in activities of daily living, higher depression, more self-reported mental difficulties and greater symptom limitations [
23] which result in poorer HRQoL outcomes.
The present study also improved upon previous studies by assessing the impact of operational and post-surgical variables. It has been argued that HRQoL in the months after CABGS may be affected by environmental factors such as processes and structures of care, complications of the surgery, or interim life change or health events [
1,
14]. The present study demonstrated that PCS group membership was adversely affected by surgical variables such as experiencing a new cardiac arrhythmia and the recording of higher pulmonary blood pressure during surgery. MCS group membership was not significantly associated with any of these variables in the final logistic regression. Attendance at CR was significantly associated with MCS
improver group when medical variables were analysed separately, but was not significant in the presence of competing variables. This finding is consistent with other studies that have found little or no difference between CR attenders and non-attenders in HRQoL measured by the SF-36 [
7,
32]. It has been suggested that CR is not sufficiently intensive to influence recovery of HRQoL [
7]. CR was only found to benefit physical function in a recent randomised controlled trial of an 18-session program which compared CR with usual care [
61].
The main limitations of the present study concern the fact that it relied on postal collection of data with over half of the questionnaires not being returned. This limitation raises the possibility that the results may not be generalisable to all CABGS patients. Those who did not return the questionnaire had lower rates of high cholesterol and were less likely to have a positive family history of heart disease. However, in all other medical variables there were no significant differences between questionnaire returners and non-returners. Moreover, there were no differences with regard to gender and age. Other socio-demographic measures such as education level and employment status, were not investigated in excluded patients, so the possibility that non participants differed in these characteristics has not been examined.
The present study relied upon self-report for psychosocial measures and HRQoL rather than formal diagnostic criteria. Although the questionnaire order was varied, the possibility of a self-report bias exists. It has been shown that patients with depressive symptoms may over-report negative aspects of HRQoL and a spurious exposure-outcome association may be generated [
62]. The inclusion of a self-report measure of cognitive functioning rather than an objective test battery may also be considered a limitation of the present study. However, it has been argued that neuropsychological test batteries that are often used in these studies may be too insensitive to measure small but personally, significant cognitive decline and that self-reported data may be of more value [
59].
Another possible limitation of the study is the reported lack of responsiveness of the SF-36. Hawkes and colleagues argue that the SF-36 may not be sufficiently responsive as an outcome measure with CR patients [
32]. Devon [
63] compared the SF-36 with three other commonly used instruments, and found that it did not demonstrate responsiveness to change in functional status. Future studies might need to include a disease-specific measure of HRQoL, such as the MacNew HRQoL instrument [
64], so that better responsiveness to change can be observed following a cardiac event such as CABGS.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MLG drafted the manuscript and analysed the data using SPSS. PE supervised the data analysis, assisted with the interpretation of findings, and analysed the data using M-Plus. MW, BM and AJG participated in the design of the study. CE and RH participated in the coordination of the study. All authors read and approved the final manuscript.