Prediction of Postoperative Outcomes Following Hip Fracture Surgery: Independent Validation and Recalibration of the Nottingham Hip Fracture Score

Abstract

<jats:p>Objectives: Independent validation of risk scores after hip fracture is uncommon, particularly for evaluation of outcomes other than death. We aimed to assess the Nottingham Hip Fracture Score (NHFS) for prediction of mortality, physical function, length of stay and postoperative complications. Design: Analysis of routinely collected prospective data. Setting and Participants: Consecutive hip fracture patients were identified from the Northumbria hip fracture database between 2014-2018. Patients were excluded if they were not surgically managed or if scores for predictive variables were missing. Methods: C-statistics were calculated to test the discriminant ability of the NHFS, Abbreviated Mental Test Score (AMTS), and ASA grade for in-hospital, 30- and 120-day mortality, functional independence at discharge, 30-days and 120-days, length of stay, and postoperative complications. Results We analysed data from 3208 individuals, mean age 82.6 (SD 8.6). 2192 (70.9%) were female. 194 (6.3%) died during the first 30-days, 1686 (54.5%) were discharged to their own home, 211 (6.8%) had no functional mobility at 120-days, 141 (4.6%) experienced a postoperative complication. The median length of stay was 18 days (IQR 8-28). For mortality, c-statistics for the NHFS ranged from 0.68-0.69, similar to ASA and AMTS. For functional independence, the c-statistics for the NHFS ranged from 0.74-0.83, similar to AMTS (0.61-0.82) and better than the ASA grade (0.68-0.71). Length of stay was significantly correlated with each score (p&amp;lt;0.001 by Jonckheere-Terpstra test); NHFS and AMTS showed inverted U-shaped relationships with length of stay. The best predictor of postoperative complications was ASA grade (0.56-0.64), whilst NHFS (0.49-0.59) and AMTS (0.48-0.57) performed similarly poorly. Conclusions and Implications: The NHFS performed consistently well in predicting functional outcomes, moderately in predicting mortality, but less well in predicting length of stay and complications. There remains room for improvement by adding further predictors such as measures of physical performance in future analyses.</jats:p>

Validation of the Nottingham Hip Fracture Score (NHFS) in Patients with Hip Fracture: A Prospective Cohort Study in the Netherlands

“…20 Lastly, a study by Doherty et al was published recently describing the predictive value of the NHFS for mortality, physical function, length of stay, and postoperative complications. 21 Compared to our multicenter study design with an inclusion period of 3 years, this singlecenter study had an inclusion period of about 10 years, limiting its generalizability. Also, its results would be subject to changes in care over the period in which enrolment of the patients was performed.…”

Mortality Prediction in Hip Fracture Patients: Physician Assessment Versus Prognostic Models

“…8,21 It has undergone external validation in several studies, although recalibration has been required to accommodate for geographical differences. 22,23 The NHFS uses 7 variables to calculate the risk of mortality after hip fracture surgery: age, gender, serum hemoglobin, number of comorbidities, institutionalization, malignancy, and the Abbreviated Mental Test Score (AMTS). The recalibrated NHFS uses the formula 100/{1 + e [5.012 • (NHFS • 0.481)] } to predict 30-day mortality after hip fracture surgery.…”

Preoperative and perioperative factors are related to the early postoperative Barthel Index score in patients with trochanteric fracture

“…Several clinical prediction models have been proposed to predict medium- and long-term functional outcomes, such as ADL function and walking ability, based on preoperative and postoperative characteristics of patients with hip fractures. The preoperative factors include age [7,8], nursing care level [7], fracture type [7], cognitive status [9,10], walking ability [8,10–13], residence [10], comorbidities [8,9,14], and nutritional status [10,15]. The only postoperative functional outcome predictor included in previous clinical prediction models is the ADL function measured using the Barthel Index (BI) score at 2 weeks after surgery [7,10,11], which is the most important predictor for medium- to long-term functional outcomes [7,10,11] among all proposed preoperative and postoperative predictors.…”