Predictors of 30-day hospital readmission after hip fracture: a systematic review.

Background: Early readmission to hospital after hip fracture is associated with increased mortality and significant costs to the healthcare system. There is growing interest in the use of 30-day readmission rates as a metric of hospital performance. Identifying patients at increased risk of readmission after hip fracture may enable pre-emptive action to mitigate this risk and the development of effective methods of risk-adjustment to allow readmission to be used as a reliable measure of hospital performance.

Methods: We conducted a systematic review of bibliographic databases and reference lists up to July 2016 to identify primary research papers assessing the effect of patient- and hospital-related risk factors for 30-day readmission to hospital after hip fracture.

Results: 495 papers were found through electronic and reference search. 65 full papers were assessed for eligibility. 22 met inclusion criteria and were included in the final review. Medical causes of readmission were significantly more common than surgical causes, with pneumonia consistently being cited as the most common readmission diagnosis. Age, pre-existing pulmonary disease and neurological disorders were strong independent predictors of readmission. ASA grade and functional status were more robust predictors of readmission than the Charlson score or individual co-morbidities. Hospital-related risk factors including initial length of stay, hospital size and volume, time to surgery and type of anaesthesia did not have a consistent effect on readmission risk. Discharge location and the strength of hospital-discharge facility linkage were important determinants of risk.

Conclusions: Patient-related risk factors such as age, co-morbidities and functional status are stronger predictors of 30-day readmission risk after hip fracture than hospital-related factors. Rates of 30-day readmission may not be a valid reflection of hospital performance unless a clear distinction can be made between modifiable and non-modifiable risk factors. We identify a number of deficiencies in the existing literature and highlight key areas for future research.