Mortality in hip fractures: Stratifying the risk of operative delay and quantifying the benefit of early mobilisation

Abstract Introduction: Early mortality following hip fracture surgery remains a significant issue with a much studied, multifactorial aetiology. This study designed to test the variables affecting 30 day mortality in a socially deprived cohort against national models, and secondarily aimed to uncover and quantify new risk factors. Methods: This was a single centre retrospective study based on National Hip Fracture Database (NHFD) data for 3176 hip fracture patients from 1st May 2008 to December 31st 2017. Data was condensed into a single anonymised workbook and logistic regression used to analyse associations with 30 day mortality. Firstly, the 6 casemix variables used by the NHFD were modelled. Secondarily, a new optimised model based on our data was created. Results: Gross mortality was 11.1% since May 2008 (344/3074). There were 1978 patients in our cohort with sufficient data to run the NHFD casemix model. Overall, this proved fair with a similar area under ROC curve to nationally (0.75 vs. 0.76), although the Odds Ratios (OR) of individual variables differed. The optimised casemix model suggested two powerful prognostic indicators for 30 day mortality, namely delay to theatre for clinical reasons (OR =3.98, p value=0.02) and whether the patient was mobilised day one post op (OR=0.21, p value=0.00). Delay to theatre for non clinical reasons conveyed only a marginal and statistically insignificant increase in risk (OR=1.15, p value=0.77). Conclusion: This study has confirmed the NHFD casemix adjusted model is a fair barometer for units treating a socially deprived cohort. It also has shown a clear differentiation between risk conveyed by delay to theatre for clinical reasons and suggests delay for non clinical reasons, although clearly not desired, may not have a significant effect on death rate. Finally, it both amplifies and prompts further investigation into the potential benefit of early mobilisation.