Determining risk of hip fracture in older adults with complex needs in New Zealand: A national population time-to-event study

Abstract

Hip fractures are one of the most common and debilitating injuries in older adults. Older adults who sustain a hip fracture are more likely to have increased mortality and morbidity with reduced quality of life. This, combined with slow recovery times, can lead to a need for entry to aged care facilities. Considerable work has been undertaken to investigate risk factors for hip fracture in the wider clinical research. This study built on that work and aims to identify risk factors for hip fracture in older adults with complex needs in the New Zealand context, based on questions from the interRAI home care (interRAI-HC) assessment. The interRAI-HC assessment is a standardised comprehensive clinical assessment typically given to people aged 65 years and older to assess areas of need that each person has. From the determined risk factors, a hip fracture risk score was developed to identify individuals who are more likely to sustain a hip fracture in the two years following their assessment. Two sets of interRAI-HC data were used in this study. The initial dataset (September 2012 to June 2015) was randomly split into two datasets. Two-thirds of the data was used to explore risk factors for hip fracture and to develop a risk score. A competing risk regression was used to determine which variables were significantly associated with hip fracture and were to be included in the hip fracture risk scores. The remaining one-third of the initial dataset was used to perform cross-validation of the developed scores, evaluating how well the scores predicted hip fracture events not used in the creation of the scores. Separate scores for males and females were created due to their different risk profiles. The predictive power of each score was assessed using Receiver Operator Characteristic (ROC) curves and their associated area under the curve (AUC) at various candidate thresholds. The scores developed were further validated with the second, more recent, set of interRAI-HC assessments (November 2015 to June 2018). Factors associated with hip fracture for the whole interRAI-HC assessment cohort were age, sex, ethnicity, falls, mental function varies, wandering, body mass index (BMI), tobacco use, Parkinson's disease, and dyspnoea (shortness of breath). For males, the risk factors associated with hip fracture were age, Parkinson's disease, and dyspnoea. For females, the factors associated with hip fracture were age, ethnicity, wandering, BMI, tobacco use, and dyspnoea. The male's score had an AUC of 0.586 (95% CI: 0.548 to 0.625), and the female's score had an AUC of 0.615 (95% CI: 0.593 to 0.637). When retesting using the more recent dataset, the male's score had an AUC of 0.611 (95% CI: 0.594 to 0.629) and the female’s score had an AUC of 0.624 (95% CI: 0.612 to 0.636). The scores developed here were modestly predictive of hip fracture risk for a New Zealand interRAI-HC cohort. The results of this thesis provide a good foundation for the development of a more sensitive and specific hip fracture prediction model. With further development, the score could have clinical use for individuals who complete interRAI-HC assessments.