A contemporary risk model for predicting 30-day mortality following percutaneous coronary intervention in England and Wales. (1st May 2016)

• Record Type: Journal Article
• Authors: McAllister, Katherine S.L.; Ludman, Peter F.; Hulme, William; de Belder, Mark A.; Stables, Rodney; Chowdhary, Saqib; Mamas, Mamas A.; Sperrin, Matthew; Buchan, Iain E.
• Is Part Of: International journal of cardiology. Volume 210(2016)
• Page Start: 125
• Page End: 132
• Publication Date: 2016-05-01
• Subjects: AUC area under the receiver operating characteristic curve -- BCIS British Cardiovascular Intervention Society -- LVEF left ventricular ejection fraction -- NHS (UK) National Health Service -- MI myocardial infarction -- NICOR National Institute for Cardiovascular Outcomes Research -- NWQIP North West Quality Improvement Programme -- PCI percutaneous coronary intervention; Angioplasty -- Catheterization -- Coronary disease -- Prognosis -- Risk factors; Cardiology -- Periodicals; Electronic journals; 616.12
• DOI: 10.1016/j.ijcard.2016.02.085
• Languages: English
• ISSNs: 0167-5273

Abstract

Abstract: Background: The current risk model for percutaneous coronary intervention (PCI) in the UK is based on outcomes of patients treated in a different era of interventional cardiology. This study aimed to create a new model, based on a contemporary cohort of PCI treated patients, which would: predict 30 day mortality; provide good discrimination; and be well calibrated across a broad risk-spectrum. Methods and results: The model was derived from a training dataset of 336, 433 PCI cases carried out between 2007 and 2011 in England and Wales, with 30 day mortality provided by record linkage. Candidate variables were selected on the basis of clinical consensus and data quality. Procedures in 2012 were used to perform temporal validation of the model. The strongest predictors of 30-day mortality were: cardiogenic shock; dialysis; and the indication for PCI and the degree of urgency with which it was performed. The model had an area under the receiver operator characteristic curve of 0.85 on the training data and 0.86 on validation. Calibration plots indicated a good model fit on development which was maintained on validation. Conclusion: We have created a contemporary model for PCI that encompasses a range of clinical risk, from stable elective PCI to emergency primary PCI and cardiogenic shock. The model is easy to apply and based on data reported in national registries. It has a high degree of discrimination and is well calibrated across the risk spectrum. The examination of key outcomes in PCI audit can be improved with this risk-adjusted model.