Development of a Clinical Prediction Rule to Determine Walking Independence in Older Adults With Hip Fractures

Accurate determination of walking independence in older adults after hip fracture surgery is crucial for selecting appropriate walking strategies and providing fall prevention guidance. We developed and validated a clinical prediction rule (CPR) to determine level-surface walking independence and community walking independence in older adults with hip fractures.

In a multicenter, cross-sectional study, we investigated 289 older inpatients with hip fractures. A backward stepwise logistic regression analysis was performed to develop a CPR for determining level-surface walking independence and community walking independence. The independent variables incorporated the patients' walking and balance evaluations at discharge, including the Berg Balance Scale (BBS), the Timed Up and Go (TUG) test, and maximum walking speed (MWS), as well as age and cognitive function evaluations. We used a bootstrap internal validation for the CPR's internal validation.

At their hospital discharge, 171 patients had achieved level-surface walking independence, and 51 had achieved community walking independence. As the level of walking independence decreased, patients tended to be older, with lower BBS scores and slower TUG times and walking speeds. As diagnostic factors, the level-surface walking model included age, BBS, and cognitive decline; the community walking model included BBS, TUG, and MWS. The diagnostic accuracy, represented by the area under the curve (confidence interval), was 0.88 (0.84-0.92) for the level-surface walking model and 0.81 (0.74-0.87) for the community walking model. Internal validation confirmed that the models' discrimination accuracy was good, and no model overfitting was observed.

We developed a moderately accurate CPR to determine walking independence in hip fracture patients. CPR can be a useful indicator for determining an individual's walking independence at present, but external validations with other samples are necessary.