Robust Outcome Models Safeguard Patients and Surgeons Alike

Abstract

The development of procedure or pathology specific models followed on the assumption that this would lead to better outcome prediction. Many different models have been developed for abdominal aortic aneurysms (AAAs), partly because open surgery has relatively high mortality rates in both the elective and emergency setting. 3,4 Modeling mortality, a definitive endpoint, is much easier than modeling morbidity, which is often poorly recorded. In order to identify groups of patients at greater or lesser risk of mortality a minimum number of deaths are required to develop statistically valid models. The main limitation of this study the small patient numbers and consequently the low numbers of deaths in both the development and validation sets. The total of 20 deaths in the entire group leads to the wide confidence intervals. The good performance of the combined model may be explained simply by its ability to distinguish between open and endovascular repair. The usefulness of such a model when the mortality associated with open repair has been established as three times greater than EVAR is debatable. 5 EVAR has low mortality rates and EVAR models should also be able to predict longer-term outcome measures, such as reintervention rates. The current controversies around EVAR relate to durability, and models that include anatomical variables have been demonstrated to predict 3 and 5-year re-intervention rates. 6 There is no doubt that the development of new and more robust outcome models that adjust for case risk in AAA surgery are required. We are living in an age of transparency when hospital AAA mortality rates are already published and in the public