Number 29, 2006
Bypass surgery for coronary artery disease: a vanishing treatment?

Risk associated with cardiac surgery

Back Back to the Summary

Christopher Blauth
Guy's and St Thomas’ NHS Foundation Trust, London, UK
Correspondence: C. Blauth, Cardiothoracic Centre, St Thomas’ Hospital, London SE1 7EH, UK.
Tel: 020 7188 1076; fax: 020 7188 1006; e-mail: christopher.blauth@gstt.nhs.uk

Abstract

Risk assessment is an essential part of the consent process for surgery. Cardiac surgeons have accumulated large databases of operations and outcomes. These can be used to predict risks for individual patients with increasing accuracy. Despite an ageing patient population requiring more complex surgery, the risks of death and complications from cardiac surgery continue to decline.

Keywords: Cardiac surgery, surgical mortality, risk factor, perioperative stroke


Cardiac surgeons in the UK have been collecting outcome data at a national level since 1977. Data are now collected in considerable detail, and enable preoperative prediction of risks with increasing precision. The United Kingdom Cardiac Surgical Register has now metamorphosed into the National Adult Cardiac Surgical Database, which is becoming a powerful national tool for predicting perioperative risk and recording the success rates for cardiac operations.
The headline news is that a patient's risk of death at the time of cardiac surgery has reduced dramatically over the past 25 years. Crude mortality rates compiled by the United Kingdom Cardiac Surgical Register show that the in-hospital risk of mortality for primary isolated coronary bypass grafting decreased from more than 6% in 1977 to 2% in 2003 [1]. These are pooled figures for all patients, including emergencies and salvage procedures, and illustrate a remarkable success rate for treating a disease that itself carries a high risk of premature death. Figures for primary isolated heart valve operations, and combined operations for valve and coronary disease show a similar trend (Table I).

Table I. UK and Ireland mortality rates for cardiac surgical operations (first-time procedures) in 1977 and 2003.

Not surprisingly, this improvement has come in parallel with a huge expansion in the number of operations performed in the UK and Ireland, from approximately 7000 in 1977 to nearly 37 000 in 2003. This increase in numbers, together with the commitment of cardiac surgeons to meticulous and extensive collection of data, has enabled the development of sophisticated systems for predicting the risk of surgical mortality. The main systems in use are the Parsonnet score, the Euroscore, and Bayesian modeling. The Parsonnet and Euroscore systems are additive models that lend themselves readily to outpatient or bedside use. Various risk factors for mortality, including demographic variables, severity of illness and comorbidity, and the complexity of the proposed procedure, incur numerical scores that are added together. In the Parsonnet system, the total score determines one of five risk bands for the patient, ranging from very low, through moderate, to very high risk. The Euroscore system has been devised so that the additive score equates numerically to the actual predicted risk expressed as a percentage. For this reason, and because it is the most simple to use, Euroscore is the system preferred by most UK cardiac surgeons (Table II).

Table II. The Euroscore additive risk model for cardiac surgery. Preoperative scores for a specific patient are added to obtain a percentage mortality risk for the procedure.

The fundamental flaw with both additive models is their reliance on historic outcomes, which leads to a tendency to overestimate the magnitude of the risk of mortality if outcomes continue to improve with time. An alternative strategy is to use Bayesian modeling, which is complex and requires computer assistance. The advantages are that it should be possible to make predictions for the individual patient on the basis of known outcomes for the hospital, or even for the surgical team. This requires real-time access to large data sets, which are currently being collected as part of the National Adult Cardiac Surgical Database project. Partly because of its complexity, Bayesian modeling is also the most accurate method for the prediction of risk and is sensitive to gradual changes in outcomes. Disadvantages include the needs both to ensure completeness of data entry and to establish a reliable information technology infrastructure, including dedicated staff to administer it. As ever, the limiting factor is cost, but the potential value of this system in terms of quality assurance and detecting poor performance is likely to make it available within a few years.
Death is the easiest surgical complication to measure: the data are unequivocal and binary. Nationally, all deaths are registered locally and eventually reach the Office for National Statistics, which is linked to the National Adult Cardiac Surgical Database. In due course, this will provide longer-term survival data for patients after cardiac surgery and will permit survival analysis and predictions for individual patients according to all the variables used in the Euroscore system, and others such as geographical factors. Some long-term survival data of this kind are already available from the United Kingdom Heart Valve Registry, which has data on valve implantations since 1986.
Inevitably, cardiac surgery entails risks of other adverse outcomes, and many patients are more concerned about the possibility of stroke, prolonged ventilator dependency, or an operation failing to relieve their symptoms. Anxieties are also often expressed about longer-term outcomes, including the need for repeat surgery. Informed consent to surgery also needs to consider minor complications that may be disconcerting but are expected to resolve without longer-term sequelae. These include atrial fibrillation, renal dysfunction, wound infections, sternal malunion, paresthesiae, and others. Quantifying these other risks is much more difficult, because the evidence base is incomplete. In part, this is because of the complexity of collecting the data, particularly as wide variation can exist in the severity and duration of nonfatal complications, which consequently are difficult to quantify and analyze statistically.
Nevertheless, numerous studies have addressed the prediction of risk of stroke. We know that the risk factors for postoperative stroke include increasing age, a history of stroke or transient ischemic attack, known cerebrovascular disease, and open-heart surgery such as valve repair or replacement, as opposed to coronary bypass, which takes place largely on the surface of the heart. The majority of perioperative strokes are presumed to be embolic, either from atheroma dislodged from the diseased intima of the ascending aorta during surgical manipulation or clamping, or from thrombus. Many of these strokes can resolve quite rapidly, sometimes within days. In general, the rates of postoperative stroke have declined in recent years, despite continued increase in the average age of patients presenting for surgery. The reasons for this are not clear, and are probably multifactorial, including the widespread preoperative prescription of statins.
Communicating surgical risks to patients is a delicate art. The right balance of keeping expectations realistic without frightening the patient can be difficult to find, and is always open to legal challenge. To add to the difficulties, patients’ perceptions of the concept and magnitude of risk show wide variation, as does their appetite for information. Current guidance advises full discussion and quantification of the risks of death, stroke, and any other major complication that may affect the decision to consent to surgery, together with mention of minor complications that occur relatively frequently and of which the patient should be forewarned.
For most patients, the risks of cardiac surgery are low, and significantly lower than the risk of not operating.

REFERENCE

1. Society of Cardiothoracic Surgeons of Great Britain and Ireland.
Fifth National Adult Cardiac Surgical Database Report 2003.
Henley-on-Thames, UK: Dendrite Clinical Systems Ltd. www.scts.org.

Although great care has been taken in compiling the information given in this website,
the publisher or the sponsor is not responsible for the continued currency of the information,
for any errors or omissions, or for any consequence arising therefrom.
© 2010 Les Laboratoires Servier