Prediction of 30-day mortality after hip fracture surgery by the Nottingham Hip Fracture Score and the Surgical Outcome Risk Tool

Abstract

The care of the elderly with hip fractures and their outcomes might be improved with resources targeted by the accurate calculation of risks of mortality and morbidity. We used a multicentre national dataset to evaluate and recalibrate the Nottingham Hip Fracture Score and Surgical Outcome Risk Tool. We split 9,017 hip fracture cases from the Anaesthesia Sprint Audit of Practice into derivation and validation data sets and used logistic regression to derive new model co-efficients for death at 30 postoperative days. The area (95% CI) under the receiver operator characteristic curve of 0.71 (0.67-0.75) indicated acceptable discrimination by the Nottingham Hip Fracture Score and acceptable calibration fit (Hosmer-Lemeshow test), p = 0.23, with a similar discrimination by the Surgical Outcome Risk Tool, 0.70 (0.66-0.74), which was miscalibrated to the observed data, p = 0.001. We recommend that studies test these scores for patients with hip fractures in other countries. We also recommend these models are compared with case-mix adjustment tools used in the UK. Copyright © 2016 The Association of Anaesthetists of Great Britain and Ireland.