OUP user menu

The prognostic significance of declining health-related quality of life scores at 6 months after coronary artery bypass surgery

J.M. Gunn, A.K. Lautamäki, J. Hirvonen, K.T. Kuttila
DOI: http://dx.doi.org/10.1093/qjmed/hct256 369-374 First published online: 28 December 2013


Purpose: Health-related quality of life (HRQoL) measured on the EQ-5D (European quality of life-5 dimensions)-questionnaire has been shown to improve after coronary artery bypass grafting (CABG), this study investigated whether changes in HRQoL predict later morbidity.

Methods: Included were 404 consecutive patients undergoing isolated CABG between 2008 and 2010 who filled the EQ-5D-questionnaire at baseline and 6 months postoperatively. Records were reviewed for later major adverse cardiac and cerebrovascular events (MACCE) after 6 months. Follow-up was 38.6 months (10–58).

Results: Patients who suffered later MACCE more often had suffered an in-hospital postoperative stroke, had a longer in-hospital stay, had lower HRQoL scores at 6 months and deteriorated on several EQ-5D-subscales. Logistic regression showed 6 months visual analogue scale scores and declining function scores to be independent predictors of later MACCE.

Conclusions: Deteriorating function and HRQoL-scores at 6 months as compared to baseline postoperatively predict later adverse cardiovascular events after CABG.


The impact of coronary revascularization on survival and major adverse cardiac and cerebrovascular events (MACCE) is under constant investigation and the effectiveness and efficacy of revascularization in different clinical scenarios have been established, but long term mortality and morbidity remain a concern. A large number of studies on health-related quality of life (HRQoL) after cardiac surgery describe improvement as compared to baseline scores and those of matched general populations even in the elderly.1–4 Quality of life itself is a subjective assessment of health which is defined by the WHO as ‘a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity’. It is thus understandable that several factors have an impact on values for reported quality of life; gender, age and level of education all correlate with scores.5 Known cardiovascular risk factors have been shown to affect HRQoL as described by De Smedt et al.6 It can be argued that from a cardiovascular physicians perspective HRQoL acts as a surrogate of overall risk burden.

The aim of this study was to assess changes in HRQoL as measured by the EQ-5D (European quality of life-5 dimensions) self-report questionnaire in patients treated with CABG and whether these can be used to predict later adverse outcome (MACCE).


This study population included a consecutive series of 699 patients who underwent isolated CABG between 2008 and 2010 at Turku University Hospital, Turku, Finland. Patients undergoing elective and urgent or emergent CABG were included in the study. Patients were asked to complete the EQ-5D before elective surgery or immediately postoperatively for those undergoing urgent and emergent procedures to evaluate HRQoL of the last 2 weeks preoperatively. The patients who were included in the analysis completed the form again at 6 months after surgery. Four hundred and four of 699 patients completed both questionnaires resulting in a 59.2% response rate. Baseline and operative characteristics were reviewed for the 285 non-responders. A total of 18 patients were non-responders due to death prior to 6 months and two had moved abroad, reasons for not filling out both questionnaires remain unknown for 265 patients. Questionnaires were mailed up to three times to survivors. Follow-up time was 38.6 months (range 10–58 months) up to December 2012 or date of death. Follow-up was complete for all 404 patients as the patients belonged to the same hospital catchment area.

The patient records were reviewed up until December 2012 for baseline characteristics and operative data and any late morbidity and mortality. The data for MACCE were obtained from the local hospital registry and the cause and the date of death were retrieved from the Finnish National Registry for Statistics.

The EQ-5D instrument is used at the ICU of Turku University Hospital as a quality control instrument to evaluate HRQoL. EQ-5D is a standardized and validated instrument for use as a measure of health outcome and it consists of two sections.7 The first section measures five core domains: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each domain consists of three levels: no problems (0), some problems (1) and severe problems (2). The second section of the European Quality of life is a vertical visual analogue scale (VAS) ranging from 0 to 100, with 0 representing the worst and 100 the best imaginable health state.

The main outcome studied was late MACCE, defined as a composite score of stroke or transient ischemic attack, myocardial acute coronary syndromes [acute myocardial infarctions (AMIs) and unstable angina] and cardiac death after completion of the second EQ-5D-questionnaire. Stroke was defined as a new, neurologic deficit accompanied by new structural changes in computed tomography or magnetic resonance imaging or otherwise verified by clinical assessment. TIA was defined as a transient neurologic deficit diagnosed as TIA by a neurologist. Myocardial infarction was diagnosed when a rise in the myocardial injury marker level was detected together with symptoms suggestive of acute myocardial ischemia and unstable angina as an ischemic event without rise in marker levels and diagnosed by a cardiologist.

Statistical analysis was performed using SPSS version 17.1 (IBM Corporation, New York, USA).

Baseline differences between patients with and without late MACCE were analyzed using independent samples t-test and values <0.05 were considered significant. For identification of determinants of survival, a Cox proportional hazards model was used. A backward (Wald) stepwise selection procedure was used to build the model. A P-value of <0.05 was considered significant. Included in the model were age, gender, history of stroke, preoperative AF, left ventricular ejection fraction, length of in-hospital stay, postoperative resternotomy, EQ-5D-scores and postoperative stroke.

The study protocol was approved by the ethics committee of the hospital district of Southwest Finland. All authors have read and approved the manuscript.


A total of 35 patients suffered 36 MACCE occurrences after 6 months postoperatively: 17 cases of TIA or stroke, 11 nonfatal acute coronary syndromes (seven AMIs and four cases of unstable angina) and 8 cardiac deaths. One cardiac death was due to acute type A aortic dissection, two due to AMI and for five patients the cause of death was classified under coronary artery disease. All Acute coronary syndrome (ACD) cases were treated with urgent or emergent percutaneous coronary intervention (PCI).

Baseline characteristics are shown in Table 1 for patients with late MACCE (MACCE) and patients with no late MACCE (no MACCE). Differences in EQ-5D-scores and changes are shown in Table 2. At index hospitalization, patients with late MACCE more often had a preoperative left ventricular ejection fraction <30%, had a longer postoperative stay and more often experienced an in-hospital postoperative stroke. There was no difference in history of recent myocardial infarction and no difference in the proportion of emergent and urgent procedures between study groups.

View this table:
Table 1

Baseline characteristics for patients with or without subsequent MACCE over 6 months after CABG

VariableNo MACCE groupMACCE groupP-value
Number of patients36935
Age (years)66.1 ± 8.167.6 ± 9.50.35
Gender (female)86 (23.3%)8 (22.9%)0.95
Hypertension357 (96.7%)35 (100%)0.90
Diabetes102 (27.6%)11 (31.4%)0.63
Extracardiac arteriopathy24 (6.5%)3 (8.6%)0.64
Preoperative AF25 (6.8%)4 (11.4%)0.53
History of stroke11 (3.0%)2 (5.7%)0.60
EF 30–50%49 (13.3%)7 (20.0%)0.27
EF <30%15 (4.1%)5 (14.3%)0.008
Left main stenosis155 (42.0%)15 (42.9%)0.95
No of diseased vessels0.98
    121 (5.7%)2 (5.7%)
    293 (25.2%)8 (22.9%)
    3254 (68.8%)25 (71.4%)
Recent AMI (within 90 days)59 (16.0%)6 (17.1%)0.86
History of PCI65 (17.6%)7 (9.7%)0.73
History of cardiac surgery6 (1.6%)1 (2.9%)0.59
Urgent or emergent surgery144 (39.0%)14 (40.0%)0.91
Off pump-surgery49 (13.4%)7 (20.0%)0.28
No of distal anastomoses2.7 ± 0.82.5 ± 0.80.47
LIMA graft358 (91.3%)34 (97.1%)0.97
In-hospital stay (days)7.6 ± 3.58.8 ± 7.00.08
Postoperative infectiona4 (1.1%)0 (0%)0.54
Postoperative resternotomy31 (8.4%)2 (5.7%)0.58
Postoperative AMI2 (0.5%)0 (0%)0.66
Postoperative stroke2 (0.5%)8 (22.9%)0.001
  • AMI = acute myocardial infarction within 90 days; EF, left ventricular ejection fraction; LIMA, left internal mammary artery.

  • Continuous variables are reported as means and standard deviation.

  • aDefined as mediastinitis or other deep wound infection.

View this table:
Table 2

Differences in EQ-5D-questionnaire scores between patients with or without later MACCE, independent samples t-test

EQ-5D-scoresNo MACCE groupAny MACCE groupP-value
VAS, baseline65.9 ± 21.267.1 ± 14.90.77
VAS, 6 months77.2 ± 16.568.2 ± 20.20.004
Mobility, baseline1.51 ± 0.621.49 ± 0.660.86
Mobility, 6 months1.37 ± 0.501.43 ± 0.560.52
Caring for self, baseline1.13 ± 0.401.12 ± 0.330.90
Caring for self, 6 months1.10 ± 0.331.20 ± 0.470.10
Usual activities, baseline1.19 ± 0.461.08 ± 0.280.22
Usual activities, 6 months1.31 ± 0.521.49 ± 0.660.06
Pain, baseline1.71 ± 0.591.57 ± 0.610.18
Pain, 6 months1.44 ± 0.531.51 ± 0.510.45
Anxiety/depression, baseline1.29 ± 0.511.11 ± 0.320.005
Anxiety/depression, 6 months1.18 ± 0.401.17 ± 0.380.94
EQ-5D, changes between 0 and 6 months
    VAS11.8 ± 24.13 ± 22.60.07
    Mobility−0.14 ± 0.69−0.06 ± 0.680.52
    Caring for self−0.05 ± 0.430.16 ± 0.550.03
    Usual activities0.07 ± 0.600.48 ± 0.710.01
    Pain−0.27 ± 0.69−0.06 ± 0.730.09
    Anxiety/depression−0.11 ± 0.580.06 ± 0.480.10
  • VAS, EQ-5D visual analogue scale (range 0–100). Higher VAS scores and lower scores on the five subscales indicate better outcome. Positive values for changes in VAS and negative values for changes in subscales indicate improvement.

Table 3 shows baseline characteristics for responders and non-responders (to the EQ-5D-questionnaires); excluded are 18 patients who died within 6 months postoperatively. Non-responders more often had a history of recent myocardial infarction, a lowered left ventricular ejection fraction and urgent or emergent surgery.

View this table:
Table 3

Baseline characteristics for patients who were responders or non-responders to the EQ-5D-questionnaire after CABG

Number of patients404267
Age (years)66.2 ± 8.264.4 ± 19.60.14
Gender (female)94 (23.3%)55 (20.6%)0.42
Hypertension391 (96.8%)242 (90.6%)0.001
Diabetes113 (28.0%)77 (28.8%)0.81
Preoperative AF30 (7.4%)22 (8.2%)0.70
Extracardiac arteriopathy27 (6.7%)22 (8.2%)0.45
History of stroke13 (3.2%)14 (5.2%)0.19
EF <50%76 (18.8%)71 (26.6%)0.03
Left main stenosis170 (42.1%)127 (47.6%)0.26
No of diseased vessels0.03
    123 (5.7%)5 (1.9%)
    2101 (25%)55 (20.6%)
    3279 (69.3%)207 (77.5%)
Recent AMI (within 90 days)65 (16.1%)85 (31.8%)<0.001
History of PCI72 (17.8%)39 (14.6%)0.27
History of cardiac surgery7 (1.7%)6 (2.2%)0.64
Urgent or emergent surgery158 (39.1%)185 (69.3%)<0.001
Off pump-surgery56 (13.9%)23 (8.6%)0.04
No of distal anastomoses2.6 ± 0.82.8 ± 0.80.06
LIMA graft392 (97.0%)251 (94.0%)0.06
In-hospital stay (days)7.7 ± 3.98.3 ± 4.90.06
Postoperative stroke10 (2.5%)8 (3.0%)0.68
  • Continuous variables are reported as means and standard deviation.

VAS scores at baseline were otherwise similar but patients with no later MACCE scored slightly worse scores on the anxiety/depression subscale. At 6 months postoperatively patients who later suffered MACCE had lower VAS scores. Patients who suffered late MACCE also deteriorated on the self-care subscale as compared to those who did not. Both groups deteriorated on the usual activities subscale, but the change was greater in patients with later MACCE. There was a trend (P = 0.07) towards lesser improvement in VAS scores in patients with late MACCE.

The hazard ratios (HR) for late MACCE are summarized in Table 4. Independent predictors of late MACCE on the Cox proportional hazards model were length of stay, prior history of stroke, lower 6 months VAS and negative change on VAS and Usual activities. Another hazard model that included postoperative complications (stroke, resternotomy, AMI) was created to identify predictors of negative change on the VAS scale. Resternotomy was a statistically significant predictor of negative change on VAS (HR 10.5, 95% CI 1.4–77.6, P = 0.02) but not MACCE (P = 0.982).

View this table:
Table 4

Predictors of adverse events on the Cox proportional hazards model (Backward selection, Wald)

Hazard ratio95% confidence intervalP-value
History of stroke7.2871.658–32.0180.009
In-hospital stay (day)1.0761.015–1.1410.02
Worsening usual activities score2.5891.278–5.2450.008
VAS score at 6 months0.9620.936–0.9890.02
Negative change per unit on VAS (0–6 months)1.0471.007–1.0890.02

Separate proportional hazard models were created for late ACS, late stroke and cardiac death. No variables reached statistical significance for ACS or cardiac death separately. Independent predictors of late stroke on the Cox proportional hazards model are shown in Table 5.

View this table:
Table 5

Predictors of stroke on the Cox proportional hazard model

Hazard ratio95% confidence intervalP-value
In-hospital stay (day)1.2711.123–1.438<0.001
Postoperative stroke (in-hospital)24.1595.946–98.162<0.001
Worsening usual activities score2.7311.219–6.1190.02


This study suggests that worsening quality of life scores after coronary artery bypass surgery predict later morbidity and mortality. Patients with later MACCE reported lower quality of life at 6 months postoperatively on the EQ-5D VAS. Of note, the VAS values for patients who did not suffer later MACCE were comparable to population norms (74.4 ± 7.6).8 Deteriorating VAS, lower VAS scores at 6 months and loss of function (on the usual activities subscale) between 0 and 6 months independently predicted later MACCE. Declining function was an independent predictor of stroke, although postoperative cerebrovascular events were the strongest predictors of future stroke.

Preoperative quality of life on EQ-5D has been shown to correlate with in-hospital mortality but its usefulness in risk stratification is unclear.9,10 The predictive value of HRQoL and its changes has also been demonstrated for operative cancer patients and in heart failure.11–13 It has already been shown that lower scores on mental health and physical coping on QoL-questionnaires predict future stroke in a non-surgical population.14,15 Depressive symptoms in this study were not correlated with later MACCE overall nor stroke in this study, but lower scores in physical coping as indicated by the self care-, usual activities- and VAS subscales preceded future events.

The risk factors described in the Framingham study are widely in clinical use for risk stratification in primary prevention of atherosclerosis, but these are hardly useful in secondary prevention as these risks are largely present in patients who have undergone coronary revascularization.16 Although traditional risk factors for adverse events (age, presence of cerebrovascular disease and depressed cardiac function) predicted later morbidity and mortality, the results of this study also suggest that changes in HRQoL might mirror subclinical symptomatology that eventually leads up to unstable atherosclerosis, i.e. acute strokes and infarctions or even cardiac death. It is known that a large proportion of coronary disease in general is primarily dormant and in up to two-third cases the first onset is acute in nature.17,18 Consequently, it is conceivable that many patients in need of secondary interventions after CABG or PCI are equally asymptomatic until an acute event emerges.

Earlier detection of patients at risk for adverse outcome with a QoL-questionnaire might help in secondary prevention and catastrophic late events could be avoided by identifying patients in need of early intervention. Patients with unfavorable changes in QoL could be referred for example for early stress testing and noninvasive imaging studies or antiplatelet management could be intensified. Even if the individual patient turns out not to need specific medical interventions, he or she could be in need of supportive care due to declining function.

Patients who have had coronary surgery are not well suited for risk stratification with traditional methods as risk factors are already largely present. Some of them, however, have a low risk and are ‘cured’ whereas others are at further risk of adverse outcomes. Our results suggest that a validated quality of life measure might be useful in risk assessment in these patients who have undergone surgical coronary revascularization.

A limitation of this study is the use of a composite endpoint, although all the studied events were unforeseen according to patient records during prior out- and inpatient contacts. The relatively small patient population also limits the conclusions of this study and further studies with larger populations are required to identify cutoff values and verify the predictive utility of EQ-5D in a postoperative setting. The response rate is a concern, although the 59% rate is as expected for quality of life questionnaires.19 Some patients were unable to fill the first questionnaire during their index hospitalization which is mirrored by the higher proportion of urgent and emergent cases as well as higher incidence of lowered ejection fraction in the non-responder population. It is unlikely that possible differences in HRQoL within the group of non-responders are relevant to the possible clinical applicability of the EQ-5D-questionnaire as a screening tool; an unfilled questionnaire is an inconclusive test and the predictive value of not replying a subject for another study. Within the responders patients who underwent emergent and urgent surgery filled out the EQ-5D-form postoperatively. This might affect results to some degree but the proportion of these patients was identical between the study groups and urgency was not a predictor of MACCE on Cox analysis. Patients who underwent urgent or emergent surgery probably marked worse scores on the EQ-5D thus making the possible negative changes less pronounced resulting in a dilution of the correlation between lowered HRQoL and later MACCE.


In conclusion, lower quality of life scores at 6 months after CABG on the EQ-5D quality of life questionnaire is associated with later adverse cardiovascular events. Increase in self-reported pain and deteriorating function at 6 months postoperatively independently correlated with later acute cardiac and cerebrovascular events after CABG.


Jutta Kotamäki, ICU nurse, was responsible for administering the questionnaire and is greatly appreciated for the effort.

Conflict of interest: None declared.


View Abstract