30-day mortality after heart surgery: a pilot project of the Working Party of Directors of Hospital Cardiology Departments

30-day mortality after operation is generally accepted as a central standard of quality, especially in regard to cardiac operations. The Working Party of Directors of Hospital Cardiology Departments (Arbeitsgemeinschaft Leitender Kardiologischer Krankenhausärzte, ALKK) in Germany set up a pilot project to analyse whether by direct communication with patients by a database centre the expenditure incurred in collecting complete data can be decisively reduced and full documentation of outcome can in this way be obtained even for a large multi-centre patient cohort. Between 1.6.1997 and 31.3.1998, data were consecutively collected by questionnaire on all patients registered for a cardiac operation at 85 of the 135 ALKK centres. The questionnaire included data about each patient and the indication for operation as well as the estimate of operative risk, assigned to one of five risk categories by the referring cardiologist either alone or in conjunction with the cardiac surgeon. Until 30.9.1998, the data were obtained on 11,349 patients who had given informed consent (response rate 99.99%), including survival figures. 824 (7.3%) patients had not undergone the planned cardiac surgery, 134 having died before the data of operation. The 30-day postoperative mortality, obtained in 99.99% of the 10,525 patients, was 3.92%. The operative mortality was lowest, at 3.73%, for aortocoronary bypass only (n = 7932), highest for aortocoronary bypass plus valvular operation (n = 785), at 8.04%. There was good agreement between the cardiologists' preoperative risk assessment and the observed mortality. The 30-day mortality after cardiac operation can be obtained almost completely and with reasonable expenditure even for a large patient cohort. The results confirm that hospital mortality data definitely understate the overall operative risk. The methodology used in this pilot project, namely the inclusion of information from the patients by questionnaire, can also be applied to clinical results in other areas.