TY - JOUR T1 - Maintaining good clinical practice: publication of outcomes and handling of outliers JF - Heart JO - Heart SP - 1518 LP - 1519 DO - 10.1136/heartjnl-2016-309452 VL - 102 IS - 19 AU - Simon Ray AU - Rick Nishimura AU - Sarah Clarke AU - Iain Simpson Y1 - 2016/10/01 UR - http://heart.bmj.com/content/102/19/1518.abstract N2 - There is a growing expectation that information about the performance of clinical services and individual doctors is made publicly available. Patients have a legitimate interest in knowing the care they receive is of high quality, but there are some potential risks in public reporting of these data, not least that of risk-averse behaviour by clinicians concerned about loss of reputation and livelihood that potentially denies patients appropriate treatment. The development of reliable metrics to assess clinical performance is complex, evolving and often controversial. In the USA, the Society of Thoracic Surgeons and the American College of Cardiology (ACC) have published principles for public reporting of outcomes central to which is the use of high-quality, robust and validated clinical data.1 National registries such as the National Cardiovascular Data Registries (NCDR) in the USA and the National Institute for Cardiovascular Outcomes Research audits in the UK are fundamental to the quality of clinical data used for performance reporting as is the involvement and oversight of the respective professional societies who are best placed to define what good looks like. Approaches to publication differ. In the UK, National Health Service England has published individual consultant outcomes for 11 specialties including interventional cardiology and cardiac surgery.2 In the USA, the ACC and NCDR have launched a voluntary programme of public reporting of performance data by institution.3 Although in its infancy, it is gaining considerable professional and public support.For interventional cardiology and adult cardiac surgery in the UK, there are two levels of negative outlier on the basis of 3-year rolling analysis of national audit data: alert and alarm. An alert means that observed survival is 2 SD below the expected mean and an alarm that survival is 3 SD below the expected mean. Data review and validation is rigorous, but … ER -