Derivation and validation of a risk chart for future ischaemic events and mortality following peripheral bypass surgery

Abstract

Purpose: Major cardiovascular events occur frequently in patients with peripheral arterial disease (PAD). A prediction model to identify determinants and quantify the risk of future ischaemic events could provide a personal risk profile to offer individualised patient care. In the present study, we developed and validated a risk chart in patients that underwent infrainguinal bypass surgery. Method(s): The risk chart was derived from and validated in a clinical trial cohort including 2650 patients with severe PAD treated with infrainguinal bypass surgery between 1995-1998. Long-term follow-up data on the composite primary outcome event of all-cause death, non-fatal myocardial infarction, or non-fatal ischaemic stroke were collected from 1995 until 2009 in a subset of 482 patients, the development cohort. Data were obtained from hospital records, general practitioners, direct contact with patients or family. Determinants and the long-term risk of the primary outcome event were identified with multivariate Cox regression analyses. Validation of the risk chart was performed in the remaining 2168 of patients of the complete clinical trial cohort, the validation cohort. The 10-year follow-up data of the validation cohort were completed by consulting the National Death Registry and the National Registry of Hospital Discharge Diagnoses. Result(s): The follow-up data of the development cohort were complete in 454 of the 482 patients (94%) and in 2083 of the 2168 patients in the validation cohort (96%). The primary outcome event occurred in 321 (66.6%) patients in the development cohort during a median follow-up of 6.6 years and in 1371 (65.8%) patients in the validation cohort during a median follow-up of 6.6 years. Determinants included in the risk chart were age, critical limb ischaemia, diabetes, and a prior vascular intervention. The overall performance of the risk chart was fair with a Brier score of 0.19. ROC analysis showed good discriminatory performance of the risk chart with an area under the curve of 73%. The Hosmer-Lemeshow statistic (chi-squared 13.56; p=0.851) in combination with the calibration curve showed a good calibration of the risk chart. Conclusion(s): Our risk chart proves to be a valid tool for the prediction of mortality and major cardiovascular events in patients after peripheral bypass surgery and may serve as a tool for individualised care of patients with severe PAD.