Variables
Annual mean volume for surgeon and hospitals were determined by dividing
the total number of procedures by the number of years for which the
individual surgeon or hospital had data collected. Surgeon and hospital
annual procedural volume were then categorised as ≤ 5, 5.1-10, 10.1-20,
and 20.1-45, for surgeons and ≤ 10, 10.1-20, 20.1-30, and 30.1-75, for
hospitals. We chose these points rather than quartiles as these gave
arbitrary cut-off points with minimal clinical applicability.
Regarding procedural variables, we defined ‘CABG’ as any concomitant
coronary artery bypass, irrespective of number of grafts. We defined any
‘major concomitant cardiac surgery’ as any other valve surgery, left
ventricular reconstruction, ventricular septal defect repair, left
ventricular outflow tract myectomy, pericardiectomy, pulmonary
thromboendarterectomy or embolectomy, resection of cardiac tumour, any
aortic procedure and any other surgery for congenital cardiac disease
(excluding atrial septal defect). A third category of ‘atrial arrhythmia
surgery’ was utilised; the specific lesion set was not captured.
Excluded from these three categories of concomitant procedures, and
therefore not defined as variables for logistic regression analysis,
were atrial septal defect closures, minor cardiac procedures such as
insertion of a left ventricular pacing lead, and exclusion/excision of
the left atrial appendage, as well as thoracic operations. These were
not deemed to add significant risk or alteration to surgical
decision-making with respect to the mitral valve surgery. Procedural
urgency was defined as per the ANZSCTS Database as elective, urgent or
emergency; there were no salvage procedures in this cohort [11].