Abstract
BACKGROUND: The nonunion rate for atypical femoral fractures (AFF) is known to be higher than that for typical fractures of the femur. We performed a meta-analysis to determine the incidence of nonunion necessitating reoperation following fixation for AFF and compare the rates according to the fracture site (subtrochanter or midshaft). METHODS: A total of 742 AFFs from 29 studies were included. A proportion meta-analysis utilizing a random-effects model was conducted to estimate the prevalence of nonunion. The outcomes were the incidence of reoperations that included osteosynthesis. To determine the association of nonunion with patient mean age or average duration of bisphosphonate use, meta-regression analysis was done. RESULTS: In proportion meta-analysis, the estimated pooled prevalence of nonunion was 7% (95% confidence interval [CI], 5%-10%) from all studies. There was a significant difference in nonunion rate between the 2 groups (I(2) = 34.4%, p = 0.02); the estimated prevalence of nonunion was 15% (95% CI, 10%-20%) in subtrochanteric AFFs and 4% (95% CI, 2%-6%) in midshaft AFFs. From meta-regression analysis, significant correlations were identified between nonunion rate and patient mean age (coefficient: -0.0071, p = 0.010), but not in the average duration of bisphosphonate use (coefficient: -0.0024, p = 0.744). CONCLUSIONS: A notable disparity existed in the nonunion rate among subtrochanteric AFFs and midshaft AFFs group. Therefore, it is critical for orthopedic surgeons to consider the complexity and challenges associated with AFF and to estimate the proper possibility of nonunion according to the fracture site.