Coronary revascularisation rates vary widely and are not explained by geographical variations in the incidence of coronary artery disease (CAD) [1–3]. Previous studies from North America have suggested that variations tend to reflect inappropriately high thresholds in low frequency areas, rather than overuse elsewhere [4]. A recent study suggested that under use of revascularisation results in preventable morbidity and mortality [5].

Following detection of CAD at coronary angiography, two decisions must be made: whether to attempt revascularisation and whether to do this via coronary artery bypass grafting (CABG) or percutaneous coronary intervention (PCI). Both decisions are commonly made by the cardiologist. Some have argued that this represents a conflict of interest for an interventional cardiologist [6, 7] who acts as both "poacher" and "gamekeeper". This issue has become increasingly important as the use of PCI has grown and CABG rates have fallen [8].

We used standardised case scenarios to examine variations in decisions related to coronary revascularisation among cardiologists and cardiac surgeons.

1. coronary artery by-pass grafting, 2. medical treatment initially with a view to CABG if symptoms dictated, 3. coronary angioplasty, 4. medical treatment initially with a view to PCI if symptoms dictated, 5. medical therapy. Cases were presented in succession with voting after each presentation until all 6 had been presented, initially with no discussion. Each case was then discussed in an open forum led by a panel of experts with the audience asked to participate in the discussion. The cases were then presented for a second time and everyone was asked to revote on their treatment category (1–5) for each case. The responses by the participants have also been compared to treatment recommendations based on the New Zealand score [9] (score greater than 35 points merits revascularisation) and the ACRE study which used a multidisciplinary panel to rate the criteria for selecting patients for coronary revascularisation [10].

Surgeons were more likely to choose surgery as a form of treatment (table 2, P = 0.034,) while interventional cardiologists were more likely to choose PCI (P = 0.056). There were no differences in the form of treatment selected by non-interventional cardiologists compared with interventional cardiologists (P = 0.13).

There was no relationship observed between the New Zealand score and the likelihood of agreement between clinicians. Despite a New Zealand score for all six cases which merited revascularisation there were a significant number of decisions to manage patients medically in more than one third of cases (35%). This was consistent between cardiologists (35%) and cardiac surgeons (36%).

There is no widely accepted clinical pathway for assessing CHD patients for revascularisation following coronary angiography. In tertiary centres with cardiac surgery on-site, the cardiologist may have the opportunity to discuss the case and angiographic findings directly with a cardiac surgeon. This can occur on an ad hoc or a formal basis such as at an open clinical meeting. Rarely does this occur in a dynamic, daily, case by case basis. For cardiologists working in non-surgical centres discussion with a cardiac surgeon is often more difficult and dynamic discussion may not take place. In all situations the cardiologist commonly acts as the gatekeeper for angiography and revascularisation. New developments in information technology with high speed electronic digital links may facilitate formal interactive discussion between specialists but such systems are not yet widely available. Such discussions should not take place with the patient lying waiting on the cardiac catheter laboratory table.

This study highlights a number of issues. Firstly, some variation in the use of coronary revascularisation therapy is due to variations between individual clinician's working practices. Secondly, this variation was greatest between surgeons and interventional cardiologists. This supports findings in other areas of clinical medicine where clinical specialists who perform a procedure are more likely to consider it to be appropriate in certain case scenarios [12].

Cardiologists have, at best, only moderate agreement among themselves in the form of treatment chosen for these six cases despite all six scoring highly using the New Zealand system and using the criteria from the ACRE study. Cardiologists clearly add variability to the decision process even before discussion with a surgeon has taken place. Surgeons had even less agreement among themselves at baseline. The combined effect of these variations could account for a significant degree of variability in use and choice of revascularisation therapy in routine clinical practice.

Discussion between rounds appeared to improve agreement among surgeons and cardiologists although the choice of treatments did not change significantly. This indicates that such discussion may, at least, reduce the variability of decision-making but may not change the overall utilisation of revascularisation therapy. Similar discussions might produce more consistent decisions in clinical practice but may be practically difficult within current systems of healthcare. Ideally the main carers for a patient such as the community practitioner, the cardiologist and a cardiac surgeon would represent the professional core of the discussion group with the patient and, ideally, a close family member also being involved. This might be particularly useful in clinical scenarios where indications for revascularisation were strong but where there was significant co-morbidity which influenced the decision-making process. While this could be done for elective cases it would be very difficult to do in emergency situations.

Similar multidisciplinary approaches to clinical decisions are now widely used in oncology [13–15] and have been proposed previously for cardiac revascularisation [16, 17]. As interventional cardiologists increasingly treat patients with multi-vessel disease [18] the need to carefully assess the range of options available to the patient is extremely important.

In this typical case series, there was poor agreement between cardiac clinical specialists in the choice of treatment offered to patients. Open discussion appeared to improve agreement. These findings strongly support the need for decisions to revascularise patients to be made by a multidisciplinary panel using agreed criteria.