Validation of the Nottingham hip fracture score in a Hong Kong population

Abstract

Background Musculoskeletal problems and falls are major contributors to disabilities, both globally and in East Asia and the Pacific. Osteoporosis and falls are common in the elderly. Their impact lies in the fragility fractures that result. Of these, hip fracture is the most costly, in terms of mortality, morbidity, and the medical and social care costs incurred. With an ageing population, the number of hip fractures is set to rise. Having a prediction model for outcomes will help risk stratification of patients, and aid planning of clinical care, rehabilitation and discharge. The objective of this study is to validate the Nottingham Hip Fracture Score for mortality prediction in Hong Kong.

Methods This is a retrospective study of 2209 consecutive adult patients admitted to Caritas Medical Centre from 1st January 2007 to 31st December 2012. Data was collected from the Clinical Management System, a local rehabilitation unit database, and the Hong Kong Death Registry. Variables included to predict mortality were age, sex, place of residence, presence of ≤2 comorbidities, presence of cancer, and haemoglobin level on admission. Binary logistic regression was performed. A cognitive function test score is needed for full validation of the Nottingham Hip Fracture Score. This score was only available for patients admitted to the rehabilitation unit. A sensitivity analysis was therefore done.

Results Presence of cancer [odds ratio 3.67 (2.26, 5.95)], male sex [odds ratio 2.32 (1.83, 2.93)], presence ≥2 comorbidities [odds ratio 1.38 (1.09, 1.74)] and age [odds ratio 1.05 (1.03, 1.06)] were associated with 1-year mortality. Low haemoglobin level on admission was found to be a poor prognostic factor in other studies, but appeared to be protective in ours [odds ratio 0.74 (0.70, 0.79)]. Admission from old age home was not statistically significant. Sensitivity analysis assuming both best and worst case scenarios for cognitive function test scores found the Nottingham Hip Fracture Score to be a poor prediction model in Hong Kong.

Conclusions The Nottingham Hip Fracture Score was a poor prediction model in Hong Kong, because of the low hip fracture mortality rates in Hong Kong. To be locally useful, the score need to be re-calibrated. To reduce the burden of disability from hip fracture, prevention is important, as is helping patients to recover from their injury. Better understanding of variations in length of stay and discharge destinations between hospitals in Hong Kong may provide some insights into areas for improvement.