OUP user menu

Beating heart against cardioplegic arrest studies (BHACAS 1 and 2): quality of life at mid-term follow-up in two randomised controlled trials

Raimondo Ascione, Barnaby C. Reeves, Fiona C. Taylor, Harpreet K. Seehra, Gianni D. Angelini
DOI: http://dx.doi.org/10.1016/j.ehj.2003.11.015 765-770 First published online: 1 May 2004


Aims Off-pump coronary bypass grafting (OPCAB) has short-term benefits compared to conventional bypass grafting using the heart-lung machine (CABG-CPB) but may compromise longer term outcome. We aimed to compare generic and disease specific quality of life (QoL) two to four years after surgery in participants in two randomised controlled trials of OPCAB vs. CABG-CPB.

Methods and results Trial participants were sent four questionnaires (SF-36, EuroQol/EQ5D, Seattle Angina Questionnaire (SAQ) and Coronary Revascularisation Outcome Questionnaire (CROQ)) to assess generic and disease-specific quality of life (QoL). Of 401 participants, 22 (5.5%) had died; of the 379 survivors, 328 responded (86.5%; 159 CABG-CPB and 169 OPCAB). Median duration of follow-up was three years. QoL scores for both groups were very similar and differences between groups were not significant (Math for all questionnaires and dimensions). Summary SF-36 scores showed poorer than normal physical QoL but normal mental QoL. Among all responders, there was a tendency for CROQ scores (core total, physical and psychosocial functioning and satisfaction with treatment) to deteriorate with time after the operation (Math).

Conclusion Two to four years after surgery, patients randomised to OPCAB and CABG-CPB had similar symptoms, generic and disease-specific QoL.

  • Quality of life
  • CABG
  • Off-pump


Coronary artery bypass surgery on the beating heart (OPCAB) is gaining popularity as an alternative to conventional bypass grafting using the heart-lung machine (CABG-CPB). We have previously shown that OPCAB has short-term benefits and no disadvantages compared to CABG-CPB, with respect to survival free from cardiac-related events two years after surgery.1 However, some surgeons question whether coronary anastomoses can be performed to the same level of quality when carrying out OPCAB surgery. Concern remains that operating on a moving field may compromise the quality of distal anastomoses, with adverse consequences on graft patency with increasing duration after surgery.2

Symptoms and quality of life (QoL) potentially offer a more sensitive test of the relative effectiveness of the two surgical techniques than cumulative cardiac-related events. Event rates vary across studies using different definitions of cardiac-related events. The inclusion of `recurrence of angina' as an event (e.g. defined by symptoms or by investigations prompted by symptoms) increases the overall frequency of such events, as illustrated in cohorts of patients undergoing coronary artery bypass grafting (CABG) recruited to RCTs of CABG vs. PTCA.3 Quality of life instruments such as the Seattle Angina Questionnaire (SAQ) provide more sensitive measurements of symptoms and the impact of symptoms on activities of daily living, not least because the outcome is a continuous score rather than an event. On physiological grounds, symptoms of ischaemia are likely to precede events that are themselves attributable to coronary disease.

We aimed to assess symptoms and QoL two to four years after surgery in patients who participated in the Beating Heart Against Cardioplegic Arrest Studies (BHACAS 1 and 2), two randomised controlled trials of CABG-CPB vs. OPCAB surgery.1


Details of the BHACAS trials have been reported elsewhere,1,4 including follow-up for about two years for deaths from any cause and cardiac-related events.1 Patients undergoing CABG at the Bristol Heart Institute were recruited from March 1997 to August 1998 (BHACAS 1) and from September 1998 to November 1999 (BHACAS 2). Patients were randomly allocated either to CABG-CPB or OPCAB.

The two trials differed slightly with respect to eligibility criteria. For BHACAS 1, patients who had had a myocardial infarction in the last month and patients with coronary disease involving branches of the circumflex artery distal to the first obtuse marginal branch were excluded. These patients were eligible for BHACAS 2.

Peri-operative management was similar in the two trials.4 In BHACAS 2, but not in BHACAS 1, patients received tranexamic acid (2 g) as an intravenous bolus prior to opening of the chest. Myocardial protection in CABG-CPB patients was achieved with intermittent anterograde hyperkalaemic warm blood cardioplegia. Otherwise, perfusion techniques were the same in both trials. The method of exposure and stabilisation for OPCAB patients varied between BHACAS 1 and BHACAS 2, as a result of the evolving OPCAB technique.1 Post-operative management was in accordance with the Unit's protocol.4

Data collection and quality of life instruments

For this report, all participants from BHACAS 1 and 2 who had not been lost to follow-up were sent a booklet containing four QoL instruments. Questionnaires were sent out over a period of 12 months. Depending on the original date of surgery, the booklet was sent during year 2, year 3 or year 4 of follow-up. Patients who did not respond initially were sent up to three reminders.

Two instruments were designed to assess disease-specific QoL (the Seattle Angina Questionnaire (SAQ)5 and the Coronary Revascularisation Outcome Questionnaire (CROQ)6) and two assessed generic QoL (SF-36, UK version,7 and EuroQol/EQ5D8). Scores on several QoL dimensions were derived from participants' responses, according to the developers' instructions for each instrument:

  • SAQ – physical function, angina stability, angina frequency, treatment satisfaction and quality of life;

  • CROQ – core total, symptoms, physical, cognitive and psychosocial functioning, satisfaction and adverse effects;

  • SF-36 – physical and mental component summary scores (PCS and MCS)9;

  • EuroQol/EQ5D – health status and utility scores.

Scores are on scales of 0 to1 (utility), scales centred on 50 with a SD of 10 (i.e., `T' scores; CROQ core total, PCS and MCS) or scales from 0 to 100 (all other dimensions). For all dimensions, higher scores represent better quality of life.

The sample sizes for the original BHACAS trials were each calculated on the basis of length of stay.1 Pooling data from the two trials meant that the comparisons reported here had excellent power (90%) to detect a small to moderate standardised difference (0.43) at a significance level of 0.01 (2-tailed), assuming that 80% follow-up could be achieved (Math). Differences in QoL between groups were estimated using multiple regression modelling, including an indicator variable to take into account trial membership (BHACAS 1 vs. BHACAS 2) and duration of follow-up.

Assessment of symptoms and QoL was not specified in the original BHACAS protocols. Specific ethical approval to carry out this additional study was therefore requested from the Local Research Ethics Committee of the United Bristol Healthcare NHS Trust (approval dated 27 June 2000, under the same reference as for the original studies (E3791)).


Twenty-two (5.5%) of the original 401 randomised patients had died by the time of follow-up, 8 OPCAB (4%) and 14 CABG-CPB (7%). Of the remaining 379 patients, responses were obtained from 328 (86.5%; 159 CABG-CPB and 169 OPCAB; see Fig. 1). Median duration of follow-up for respondents was 3.0 years and did not differ between CABG-CPB and OPCAB groups.

Fig. 1

Flow diagram showing status of participants and reasons for not responding at follow-up.

No differences in baseline demographic and clinical characteristics were observed between participants who died or who were lost to follow-up and those who responded. Nor were there any differences between respondents who had had CABG-CPB and those who had had OPCAB, except for a slightly higher proportion of hypertensives in the OPCAB group (48% vs. 60%, Math; see Table 1).

View this table:
Table 1

Demographic and clinical characteristics at recruitment for respondentsa

VariableCABG-CPB (Embedded Image)OPCAB (Embedded Image)
BHACAS 278 (49.1)87 (51.5)
Male129 (81.1)141 (83.4)
Age (year)61.4±11.063.1±8.8
Diabetes33 (20.8)43 (25.4)
Previous myocardial infarction
Embedded Image1 month52 (32.7)60 (35.5)
Embedded Image1 month9 (5.7)15 (8.9)
Hypertension77 (48.4)101 (59.8)
Hypercholesterolemia112 (70.4)128 (75.7)
Smoking history
Current smoker17 (10.7)16 (9.5)
Ex-smoker91 (57.2)107 (63.3)
Non-smoker51 (32.1)46 (27.2)
Operative priority – unstable angina60 (37.7)69 (40.8)
Angina class (CCS)
I22 (13.8)14 (8.3)
II56 (35.2)56 (33.1)
III50 (31.5)58 (34.3)
IV31 (19.5)41 (24.3)
Ejection fraction Embedded Image50%37 (23.3)36 (21.3)
Parsonnet score Embedded Image1022 (13.8)29 (17.2)
Number of distal anastomoses
116 (10.6)17 (10.6)
254 (34.0)17 (44.4)
367 (42.1)17 (36.7)
⩾422 (13.8)17 (8.9)
Combinations of arterial graftsb
No arterial grafts3 (1.9)10 (6.0)
LITA arterial graft121 (76.6)125 (74.4)
LITA and RITA arterial grafts22 (13.9)11 (6.5)
LITA and RA arterial grafts6 (3.8)16 (9.5)
LITA and RITA and RA grafts4 (2.5)4 (2.4)
Other arterial graft combinations2 (1.3)2 (1.2)
  • Data are presented as number of patients (% of group total) or means±SD.

  • a Characteristics of all trial participants at recruitment by trial (i.e., BHACAS 1 vs. BHACAS 2) are reported elsewhere.1

  • b Data about arterial grafts were missing for 1 CABG-CPB patient and 1 OPCAB patient. LITA, left internal thoracic artery; RITA, right internal thoracic artery; RA, radial artery.

Mean scores and SDs, estimates of differences between groups and confidence intervals for each QoL dimension are shown in Table 2, and group means are shown graphically in Fig. 2. The means for both groups were very similar for all dimensions and none of the differences between groups approached significance. The summary scores of the SF-36 showed poorer than normal physical QoL (about 1 SD below the population average) but normal mental QoL.

View this table:
Table 2

Mean scores for the Seattle Angina Questionnaire (SAQ), Coronary Revascularisation Outcome Questionnaire (CROQ), SF-36 and EuroQol for CABG-CPB and OPCAB groups and estimates of differences between groupsa

nbMeanSDnbMeanSDDifference95% CIp
SAQPhysical functionc14475.224.416774.724.30.59−4.86 to 6.040.83
Angina stabilityc13469.630.914066.327.03.37−3.53 to 10.30.34
Angina frequencyc14286.821.915585.122.31.77−3.26 to 6.790.49
Treatment satisfactionc13788.120.214788.615.6−0.52−4.25 to 3.690.81
Quality of lifec14774.824.415474.422.30.45−4.81 to 5.710.87
CROQCore totald15449.77.8616450.16.54−0.41−1.99 to 1.180.61
Symptomsc15178.021.316279.117.8−1.04−5.41 to 3.320.64
Physical functioningc15579.826.016477.726.12.27−3.46 to 7.990.44
Cognitive functioningc15081.323.616484.719.6−3.30−8.12 to 1.520.18
Psychosocial functioningc15382.321.916483.917.6−1.46−5.82 to 2.910.51
Satisfactionc15573.426.216475.224.2−1.67−6.80 to 3.460.52
Adverse effectsc15291.811.516292.611.3−0.68−3.21 to 1.850.60
SF36Physical component summaryd14541.314.415640.614.10.74−2.50 to 3.980.66
Mental component summaryd14552.610.115652.89.60−0.19−2.43 to 2.060.87
EuroQolUtilitye1510.820.251640.810.240.01−0.04 to 0.060.70
Health statusc15177.019.116576.016.11.01−2.89 to 4.910.61
  • a Mean differences between groups estimated from a multiple regression model including type of surgery, trial (BHACAS 1 vs. BHACAS 2) and duration of follow-up. Interactions between type of surgery and trial were also tested. The results shown in the table are from models without interactions because there was no evidence that the effects of type of surgery varied by trial.

  • b Numbers vary because of missing responses to some items that prevented scores being calculated.

  • c Quality of life dimension scored on a scale from 0 (poor) to 100 (good).

  • d Quality of life dimension scored on a scale with a mean of 50 and a SD of 10, i.e., `T' score. T scores are also normalised for the SF-36 component summaries (PCS and MCS; i.e., 50 represents the mean for a normal population), but not for the CROQ Core Total score.

  • e Utility dimension scored on a scale from 0 (poor) to 1 (good).

Fig. 2

Mean symptom and quality of life scores at follow-up for (a) Seattle Angina Questionnaire (SAQ), (b) Coronary Revascularisation Outcome Questionnaire (CROQ), (c) SF-36 Physical and Mental Component Scores (PCS and MCS) and CROQ core total score and (d) EQ5D utility and health status scores. Error bars are upper 95% confidence limits.

Among all responders, i.e., estimated across the whole sample irrespective of type of surgery, there was a tendency for some CROQ dimension scores (core total, physical and psychosocial functioning and satisfaction with treatment) to deteriorate with time after the operation Math. Changes in QoL with duration of follow-up tended in the same direction for 9 of the 12 other symptom and QoL scores (exceptions being SF-36 MCS, SAQ angina frequency and CROQ adverse effects) but the changes with duration did not approach significance (Math).


Quality of life instruments cannot directly measure the quality of distal anastomoses. However, quality of life measures arguably provide the most important evidence about the relative effectiveness of alternative interventions, i.e., the patient's view. Moreover, when patients have been randomised to CABG-CPB or OPCAB, quality of life data are likely to represent strong indirect evidence of the patency of the grafted vessels. Symptoms during follow-up must be presumed to arise from stenoses in coronary vessels that have developed since the operation. Although there is no guarantee that the new stenoses in individual patients arise in the grafted vessels, the difference in `average' symptoms (and therefore in the underlying stenoses) between groups can only be explained by the type of surgery the patients received. It is possible that one or other technique could predispose native vessels to develop new stenoses but it is more plausible that differences in symptoms and QoL between groups reflect differences in the patency of the grafted vessels.

In this paper, we have shown that at mid-term follow-up of patients in BHACAS 1 and 2 there is no difference in self-reported symptoms and QoL in patients who underwent CABG-CPB and OPCAB surgery. This finding extends the time-frame in which CABG-CPB and OPCAB have been demonstrated to be equivalent.1 Since BHACAS 2 recruited participants with more severe disease than those recruited to BHACAS 1, and to another trial of OPCAB vs. CABG-CPB,10 this report also extends the applicability of the finding of equivalence to a wider population.

We have little reason to doubt the validity of these findings. Patients were originally allocated to OPCAB and CABG-CPB randomly and, although some patients were lost to follow-up, the characteristics of respondents from the two groups were still well balanced (see Table 1). Therefore, significant confounding is unlikely. More CABG-CPB than OPCAB patients had died or were lost to follow-up (42 vs. 31). Comparison of the distribution of prognostic factors among trial participants at baseline1 and among responders (Table 1) suggests that, if anything, there has been a selective loss to follow-up of the more severely ill CABG-CPB patients (e.g., with unstable angina and higher CCS class). Formal statistical testing of this difference must await longer follow-up and more events. However, this observation makes it extremely unlikely that we have missed an important difference in symptoms and QoL in favour of CABG-CPB; the nature of the observed loss to follow-up would tend to favour CABG-CPB, not OPCAB (i.e., “informative censoring”11).

The overall loss to follow-up in the present study was considerably higher than for our previous report on the follow-up of BHACAS participants (98.5% after mean duration of follow-up of 25 months,1 compared with 82% after mean duration of follow-up of 36 months). We believe that the increased loss to follow-up observed in this paper arises primarily from the different methods of follow-up used. The previous paper focused on cardiac-related events and obtained information both from patients and from their general practitioners. For obvious reasons, the present paper relied on responses obtained directly from patients.

Patients were blinded to the type of treatment they received and completed the questionnaires independently of the researchers. Therefore, their responses are unlikely to have been affected by bias. It is possible that loss to follow-up could have been associated with the intervention received, but there was no evidence that this was the case (27/201 were lost to follow-up in the CABG-CPB group compared with 22/200 in the OPCAB group; see Fig. 1).

Findings of no difference are often difficult to interpret, given differences of opinion about the size of difference in outcome that may be important to patients. The equivalence of conventional CABG-CPB and OPCAB techniques is well illustrated by the SAQ scores. Spertus reported that SAQ scores need to differ by about 10 points to be perceived as different by patients,5 whereas the observed differences between CABG-CPB and OPCAB groups ranged from 0.4 to 3.3.

Ideally, baseline symptom and QoL measures would have been available but the assessment of symptoms and QoL was not included in the original trial protocols.1 Analyses adjusted for baseline would be expected to have provided an even more sensitive test of the difference between groups. However, it is notable that the point estimates of differences in symptom and QoL scores are extremely close to zero and suggest no tendency for outcomes to be better for one or other surgical technique.

Because patients were recruited over a long period of time, duration of follow-up varied considerably. We therefore included duration of follow-up in the analyses, on the grounds that some symptomatic outcomes might deteriorate over time. A tendency for such deteriorations to occur was only observed for some CROQ dimensions, i.e., for the `core' total score, physical and psychosocial functioning and satisfaction with treatment. Within the time period studied, the greatest deterioration apparently occurred between years 2 and 3 with less change from year 3 to year 4. The CROQ satisfaction score also differed significantly between trials (worse for BHACAS 2), a finding which we suspect arose because participants in BHACAS 2 were on an average more severely ill at baseline.1

The fact that trends over time were observed with the CROQ but not the SAQ or SF36 is interesting. Our findings suggest that the CROQ may be a more sensitive instrument; it includes modifications of items from both the SAQ or SF36 but also includes additional items (based on extensive development work). We also observed that the response rate was slightly higher for the CROQ than for the SAQ and the SF36 (see Table 2), suggesting that cardiac surgery patients in the UK (on whom the CROQ was initially developed) may perceive the questions to be easier to answer and, possibly, more relevant to their circumstances.

It is tempting to ask what the absolute QoL scores `mean' from a descriptive point of view (in contrast to the differences in QoL between groups). Only the SF-36 scores are normalised, allowing a formal comparison with people without heart disease. For the EuroQol/EQ5D it is assumed that people with no health problems and `perfect' quality of life will have a utility of 1 and health status of 100. In this study, SF-36 PCS scores for respondents indicate that they have poorer than average quality of life because of physical health problems. Mean EuroQol utility scores of about 0.80 also suggest modestly reduced QoL but are typical of the general population in the UK (including some respondents with health problems).12

For the SAQ and CROQ, a score of 100 represents the least impaired or asymptomatic response and 0 represents the most impaired or symptomatic response. However, absolute scores depend on how participants interpret the response categories for items within a specific study population. Therefore, it is not meaningful to compare scores across populations. For example, respondents may interpret the questionnaire items in the context of the activities that they want to do which, in turn, may be age or culturally dependent. Nevertheless, mean SAQ scores reported here are broadly consistent with those obtained by other researchers from conventional CABG-CPB patients one year after surgery.13

In summary, two to four years after surgery, patients randomised to OPCAB and CABG-CPB had similar symptoms, generic and disease-specific QoL. We conclude that OPCAB is as effective as conventional CABG-CPB in improving QoL. Although longer term follow-up including survival free from cardiac-related events is desirable, and we continue to follow participants in BHACAS 1 and 2, these findings suggest that concerns about the long term patency of coronary grafts placed on a moving field are unfounded.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
View Abstract