Article Text

Download PDFPDF

Coronary artery disease
Socio-economic status and early outcome from coronary artery bypass grafting
  1. P H Gibson1,
  2. B L Croal2,
  3. B H Cuthbertson3,
  4. G Gibson4,
  5. R R Jeffrey4,
  6. K G Buchan4,
  7. H El-Shafei4,
  8. G S Hillis1
  1. 1
    Department of Cardiology, University of Aberdeen and Aberdeen Royal Infirmary, Aberdeen, UK
  2. 2
    Department of Clinical Biochemistry, University of Aberdeen and Aberdeen Royal Infirmary, Aberdeen, UK
  3. 3
    Health Services Research Unit, University of Aberdeen and Aberdeen Royal Infirmary, Aberdeen, UK
  4. 4
    Department of Cardiac Surgery, University of Aberdeen and Aberdeen Royal Infirmary, Aberdeen, UK
  1. Dr G Hillis, Cardiovascular Division, The George Institute for International Health, King George V Building, Royal Prince Alfred Hospital, Missenden Road, Sydney NSW 2050, Australia; ghillis{at}


Objective: To determine the effects of socio-economic status (SES) on the outcome of coronary artery bypass grafting (CABG).

Design: Prospective cohort study.

Setting: Regional cardiac surgical unit.

Patients: 1994 consecutive patients undergoing non-emergency CABG.

Measures: SES was determined from the patient’s postcode using Carstairs tables. The primary end-point was all-cause mortality at 30 days.

Results: There were 50 deaths (2.5%) within 30 days of surgery. A higher Carstairs score demonstrated a trend towards increased 30-day mortality (odds ratio (OR) 1.09 per unit, 95% CI 1.00 to 1.20, p = 0.06). In a backward conditional model, including other predictors of early mortality, Carstairs scores were independently predictive (OR 1.12 per unit, 95% CI 1.01 to 1.24, p = 0.02). In a model including only Carstairs scores and the EuroSCORE, both were independent predictors of this outcome (OR for Carstairs score 1.11 per unit, 95% CI 1.00 to 1.22, p = 0.04). The 30-day mortality increases in each quartile of Carstairs scores, with patients in quartile 4 (most deprived) at significantly higher risk compared with quartile 1 (uncorrected OR 2.53 per unit, 95% CI 1.04 to 6.15; OR corrected for EuroSCORE, 2.56 per unit, 95% CI 1.03 to 6.34, p = 0.04 for both). Similarly, patients in the least affluent quartile were twice as likely to suffer a serious complication as those in the most affluent quartile (OR 2.14 per unit, 95% CI 1.32 to 3.46, p = 0.002). This increased risk was also independent of the EuroSCORE.

Conclusions: Lower SES is associated with a poorer early outcome following CABG and is independent of other recognised risk factors.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Coronary artery bypass grafting (CABG) is an important intervention for selected patients with symptomatic or threatening coronary heart disease. Despite continuing improvements in perioperative care, there remains a small but significant risk of death or major morbidity following surgery, and accurate risk prediction remains a challenge for clinicians. Risk estimation is also increasingly important for the standardisation of data comparing the outcomes of cardiothoracic surgical units. Widely used risk-stratification tools, such as the European System for Cardiac Operative Risk Evaluation (EuroSCORE), are imperfect.1 2 There is, therefore, considerable interest in identifying additional factors that may influence outcome.

Social deprivation is increasingly recognised as an important determinant of cardiovascular morbidity and mortality. It is, however, difficult to measure. In Scotland, social deprivation can be determined using a widely used and well-validated system: Carstairs deprivation scores.3 These use census information from each postcode sector to derive a score that describes the relative affluence or deprivation of its population in comparison with the national mean. Carstairs scores have been used to assess the relationship between deprivation and the prevalence of angina and heart failure,4 5 recurrent ischaemia following acute myocardial infarction,6 prescription rates in primary prevention7 and waiting times for cardiac surgery.8 There are, however, limited data regarding socio-economic status (SES) and outcome from CABG. Furthermore, the few prior studies have produced conflicting results and have had limited ability to correct for confounding factors and/or conventional methods of risk stratification. The aim of the current study is, therefore, to assess the importance of social deprivation in predicting early outcome from surgical revascularisation in a large well-characterised population—and to compare the relative importance of SES with other recognised predictors.


Study population

Between April 2000 and March 2004, 2076 patients underwent CABG in our institution, which is the sole unit undertaking cardiothoracic surgery in north-eastern Scotland. Patients receiving private treatment were included. Seventy-six patients who underwent emergency surgery or who had suffered recent (<1 week) myocardial infarction were excluded from the analysis. Six patients residing outside Scotland were also excluded, as no deprivation scores were available for their postcodes. The study cohort comprised the remaining 1994 patients.

One thousand seven hundred and seventy-five patients (89%) underwent isolated CABG, and the remaining 219 (11%) underwent an additional major operative procedure. These included 197 valve replacement or repair procedures (151 aortic (one with aortic root replacement), 43 mitral, and three combined aortic and mitral). There were nine left ventricular aneurysm resections, five combined thoracic procedures, three thoracic aortic procedures (without valve replacement), two atrial myxoma resections, two ventricular septal defect repairs and one pericardiectomy.


Baseline demographic data were collected prospectively by a full-time, experienced data collector. Relevant clinical details were recorded, including New York Heart Association (NYHA) functional class, creatinine, preoperative medication and the EuroSCORE. The glomerular filtration rate was estimated from the Modification of Diet in Renal Disease equation.9 Carstairs scores were assigned to each patient using their postal code as an indicator of their SES. Where more than one Carstairs score exists for a postcode sector, the score for the larger population was used.

The Carstairs scores were calculated from 2001 UK census information, based on the proportion of the population in each postcode sector having no car, living in overcrowded accommodation, and levels of male unemployment and low social class (according to professional classification). After standardisation of each variable to allow equal weighting, the summed scores reflect deprivation relative to the national mean.3 For each postcode area, a negative score denotes relative affluence, and a positive score implies greater deprivation.


Patients were followed up using hospital records and a vital events search by the General Register Office for Scotland. The primary outcome measure was all-cause 30-day mortality. Secondary outcomes included (a) a requirement for an escalation of care whereby a patient previously discharged to the cardiothoracic ward was readmitted to either the cardiothoracic high-dependency or intensive care unit (or readmission to the latter from the former) and (b) the following postoperative complications: stroke, myocardial infarction, septicaemia or mediastinitis, need for sternal reopening, new dialysis or permanent pacemaker implantation. Cardiac troponin I (cTnI) levels 24 h after surgery and the duration of postoperative hospitalisation were also recorded.

Statistical analyses

Categorical data are summarised using absolute values (percentage). Normally distributed continuous data are presented as mean (SD) or, where skewed, as median (interquartile range, IQR). Binary logistic regression was used to calculate odds ratios (OR) and confidence interval (CI) for 30-day mortality. In addition, a backward conditional model was developed. This included all variables that were either significant univariable predictors of 30-day mortality or that demonstrated a trend (p<0.5) towards predicting this outcome. Retention in the model was set at p<0.1. A further model included only the Carstairs score and EuroSCORE. The odds ratio associated with different quartiles of deprivation score was also assessed using logistic regression, with the lowest quartile as the reference. Comparison of clinical characteristics and preoperative medication for patients with differing quartiles of Carstairs score were compared using the χ2 test for trend for categorical variables. Analysis of variance or Jonckheere–Terpstra tests were used for quartile analysis of normally and non-normally distributed continuous variables respectively. SPSS version 14.0 (SPSS, Chicago) was used for all analyses.


Patient population and outcome

The population was predominantly male (78%) with a median age of 66 (table 1). The median Carstairs score was −2.28, and 1543 patients (77%) had a Carstairs score <0 suggesting that the majority of the study population was resident in postcodes which are affluent relative to the Scottish mean. Vital status information was available for all patients. There were 50 deaths (2.5%) within 30 days of surgery.

Table 1 Clinical characteristics of study cohort and univariable relationship to 30-day mortality

Univariable predictors of 30-day mortality

Pre- and perioperative factors associated with an increased risk of death at 30 days were increasing age, ejection fraction <50%, poorer renal function, the need for an additional major operative procedure, longer cardiopulmonary bypass and/or cross-clamp times and increasing EuroSCORE (table 1). A higher Carstairs score demonstrated a trend towards increased risk of death at 30 days (OR) 1.09 per unit, 95% CI 1.00 to 1.20, p = 0.06).

Multivariable analysis of 30-day mortality

In a backward conditional model Carstairs scores predicted 30-day mortality (table 2). Cross-clamp time is closely related to bypass time (r = 0.90, p<0.001) but is an inferior predictor of mortality and was not included in this model. Likewise, EuroSCORE, which includes many of the univariable predictors, was not included. In a model including only Carstairs scores and EuroSCORE, both were independent predictors of this outcome (OR for Carstairs score 1.11 per unit increase, 95% CI 1.00 to 1.22, p = 0.04: OR for EuroSCORE 1.39 per unit increase, 95% CI 1.27 to 1.52, p<0.001).

Table 2 Multivariable predictors of 30-day mortality

Quartile analysis of Carstairs scores

Patients in quartile 4 (least affluent) were generally younger and more likely to be in NYHA functional class III–IV than those in quartile 1 (table 3). The upper quartile also had a higher proportion of females and current smokers. There was no difference between quartiles for any of the univariable predictors of 30-day mortality, including ejection fraction, renal function and EuroSCORE. Quartiles were also similar with respect to preoperative medication (antiplatelet agents, beta-blockers, ACE inhibitors and vasodilators: table 3). Data regarding statin use are not available.

Table 3 Baseline characteristics according to Carstairs score (by quartile)

The risk of death at 30 days increases in each quartile of Carstairs scores, with patients in quartile 4 demonstrating a significantly higher compared with quartile 1 (table 4). This increased risk is independent of the EuroSCORE. One hundred and fifty-eight patients (8%) suffered at least one important complication perioperatively. Patients in the least affluent quartile were twice as likely to suffer a complication as those in the most affluent quartile (OR 2.14, 95% CI 1.32 to 3.46, p = 0.002). This increased risk is also independent of the EuroSCORE (table 4). No associations were observed between the quartile of Carstairs score and postoperative cTnI release or hospital stay after surgery.

Table 4 Outcomes according to the Carstairs score (by quartile)


These data confirm that social deprivation is associated with a worse early outcome following cardiac surgery. In this cohort, patients in the least affluent quartile are approximately 2½ times more likely to die within 30 days of surgery and are twice as likely to suffer a major, non-fatal, complication, when compared with the most affluent. Importantly, the current study demonstrates, for the first time, that the excess mortality associated with lower SES is independent of conventional risk factors and a widely used method of risk assessment in this setting.

Prior studies

There are limited and conflicting data assessing the effects of SES on outcome following CABG. Studies from Italy and the USA have suggested that the most socially disadvantaged patients have a higher 30-day10 and 3-year11 mortality following CABG. However, in neither of these cohorts were the effects of SES adjusted for global measures of perioperative risk, such as the EuroSCORE. In addition, other authors have found no significant effects of SES on short-term mortality following CABG12 and valve replacement.13 Indeed, one study from the USA has suggested that patients without medical insurance may have a similar short-term and better long-term survival when compared with patients with medical insurance.14 This may, however, be largely influenced by uninsured patients undergoing CABG at a significantly younger age.

Hospital volume has also been suggested as a confounding factor in the USA, where low-volume cardiothoracic units are associated with a poorer outcome, but cater for a more affluent population.15 This has been suggested as a potential explanation for the absence of any relationship between SES and unadjusted in-hospital mortality in a retrospective analysis of data from the California CABG Mortality Reporting Program.15 In high-volume (⩾480 CABG operations per year) centres, such as our own, lower SES was associated with a significantly increased in-hospital mortality. This relationship was not, however, observed in medium and low-volume centres. In this cohort, further adjustment for clinical risk factors had little effect—suggesting that hospital volume was the major confounding factor.15

The seemingly contradictory results of the current study and those of the two previous studies conducted in the UK12 13 are worthy of further comment. Both of these studies were performed in comparable healthcare settings and used similar methods to assess SES. In the study by Taylor and colleagues, although there was no relationship between SES and 30-day mortality, a higher SES was associated with a lower incidence of the combined end-point of death, stroke and/or myocardial infarction at 30 days.12 This cohort was younger and had a lower prevalence of major cardiovascular risk factors such as diabetes. This was reflected in lower 30-day mortality. The observed disparity may, therefore, at least in part, reflect the limited power of this previous study to detect mortality differences. The recent study by Bagger and colleagues assessed the effects of SES on outcome from primary aortic or mitral valve replacement surgery using the UK Heart Valve Registry.13 Although SES was a powerful and independent determinant of long-term (up to 15 year) survival in this setting, no association was observed between SES and mortality at 30 days and 1 year. In this study, no data were available regarding the presence of coronary heart disease or whether patients underwent concomitant CABG. Data from the Euro Heart Study on valvular heart disease suggest that CABG is performed in <25% of patients undergoing surgical intervention on a single left-sided heart valve.16 It seems likely, therefore, that this cohort represents a different population to that in the current study. Likewise, the absence of data on other important and potentially confounding comorbidities in the UK Heart Valve Registry limits exploration of the independence of the effects of SES and comparisons with the current cohort.

Deprivation and cardiovascular outcome in other settings

A higher community prevalence of angina and heart failure has been demonstrated among more deprived patients.4 5 These patients are also less likely to consult their general practitioner.4 5 In addition, among patients with heart failure, lower SES is associated with an increased risk of admission to hospital,17 a higher likelihood of readmission18 and higher mortality.19 Deprivation is also associated with an increased incidence of acute myocardial infarction below the age of 65 years,20 a higher risk of early recurrent ischaemia6 and excess short- and long-term mortality.21

Potential mechanisms

Despite evidence suggesting a link between deprivation and cardiovascular outcomes, the reasons for this are unclear.22 Patients undergoing CABG have advanced coronary heart disease and are already a highly selected population. It is likely that traditional risk factors will, therefore, play a lesser role in predicting survival. Certainly, the current data suggest that the increased hazard associated with lower SES is independent of conventional risk factors and a well-validated global score of perioperative risk. This independence of SES from traditional methods of risk stratification has also been observed in other aspects of cardiovascular disease, such as chronic stable angina.23 24 It seems likely, therefore, that a number of other, less easily measured, factors influence perioperative risk. These might include patient communication with medical and nursing staff, expectations of postoperative care, motivation and compliance with postoperative rehabilitation, psychological response to surgery and levels of support after discharge from hospital.

Another potential factor might be that more deprived patients have reduced access to surgical revascularisation. Certainly, data from the USA suggest that this may be the case. Among 28 698 patients hospitalised with an acute myocardial infarction in New York State during 1995, lower income was associated with a considerable reduction in the use of invasive procedures, including coronary angiography, percutaneous coronary intervention and CABG. Indeed, patients with incomes in the highest quintile were 48% more likely to undergo CABG when compared with the least affluent quintile.25 This discrepancy appears to be related primarily to underutilisation of CABG in deprived patients, rather than overuse among those with higher incomes, and is independent of other demographic and clinical factors.25 Similar discrepancies have been observed even in healthcare systems that guarantee universal access, such as that in the UK.10 26

In addition to the evidence suggesting limited availability of CABG among deprived patients, there are also data indicating that patients in lower socio-economic groups are less likely to be classified as requiring urgent surgery, and therefore wait longer.8 We cannot explore issues relating to access to CABG in the current cohort; nor do we have data regarding waiting times. It is, however, unclear that such issues influence the risk of perioperative complications. One might anticipate that a higher threshold for undertaking CABG would be associated with increased preoperative comorbidity—as only the sickest patients within deprived populations would be offered surgery. Our EuroSCORE data do not support this assumption. Likewise, patients with lower SES do not undergo more complex or longer operations and have similar postoperative cTnI release.

Reduced prescription of secondary preventive medications is another possible explanation for worse outcome among patients with lower SES. Inequalities in the use of such treatments have been described in patients with stable coronary heart disease7 but not those with heart failure.5 No such differences are evident in the current study, although data regarding the prescription of statins are missing, and levels of compliance are unknown.

Study strengths and limitations

This study assesses a large cohort of consecutive patients undergoing coronary artery bypass surgery in a regional cardiothoracic unit, with the exception of those undergoing emergency surgery. The results are likely to be generally applicable to clinically stable patients undergoing elective surgery. Nevertheless, reliance on data from a single centre has inherent limitations. In particular, the socio-economic mix of a surgical population will vary depending on the hospital catchment area, and this will affect the relative hazard observed. Estimations of risk in the current study may be influenced by the relatively small number of patients in this cohort who come from extremely deprived areas, and the results may, therefore, underestimate the “risk gradient” compared with the general population undergoing CABG. In addition, it is important to recognise the limitations of the Carstairs score as a measure of SES. These scores describe the features of a particular geographic population rather than an individual. Postcodes covering large rural districts may contain a more heterogeneous populace compared with smaller city areas. The predictive value of a deprivation score for an individual may, therefore, vary depending on the density of population in their area of residence. Other markers of SES, such as social class based on occupation, may avoid some of these limitations—but such data were not available in the current cohort.

The use of 30-day mortality as the primary outcome measure has the advantage of being entirely objective, but data regarding postoperative complications are more subjective. Finally, data regarding preoperative waiting times and statin prescriptions are missing.

Despite these limitations, the current study demonstrates an independent and clinically relevant relationship between SES and outcome from CABG. The mechanisms whereby SES affects early outcome in this setting are, however, unclear and may be complex. Further work is required to address these issues and, in doing so, develop strategies to improve the outcome of more deprived individuals.


We are grateful to the staff of the Cardiothoracic Surgical Unit, Aberdeen Royal Infirmary.



  • See Editorial, p 785

  • Competing interests: None.

  • Ethics approval: Ethics approval was provided by the Grampian Research Ethics Committee.

Linked Articles