Background: Nonunion risk factor identification and modification are subjective. We describe and validate a predictive nonunion risk factor model to identify foot and ankle operative patients at risk for nonunion.
Materials and methods: One hundred international experts in foot and ankle surgery were surveyed. Nineteen nonunion risk factors were stratified into 3 categories: more significant than, as significant as, and less significant than smoking 1 pack per day. A nonunion risk assessment model was developed by assigning a weighted score to each risk factor, based on its mean score from the survey. A total nonunion risk (TNR) score was calculated for individual patients. It was retrospectively validated in 2 patient cohorts from a single center's prospectively collected end-stage ankle arthritis patient database: 22 cases of ankle and/or hindfoot fusion nonunion and 40 sex- and procedure-matched controls with bony fusion. Analyses included descriptive statistics, logistic regression, and univariate and multivariate linear regression models.
Results: The mean TNR score was 6.6 ± 5.6 in controls and 13.5 ± 8.2 in the nonunion group (P < .001). Data showed excellent intraobserver and interobserver correlation coefficients. In a logistic regression model, the risk of nonunion exceeded 9% with a TNR score greater than or equal to 10. Multivariate linear regression analysis, adjusted for age and sex, suggested that lack of fusion site stability and obesity (body mass index greater than 30) were significantly predictive of nonunion.
Conclusion: The nonunion risk assessment model provides a reliable, sensitive, and specific method for predicting nonunion based on objective patient assessment. Orthopaedic patients at risk for nonunion could benefit from targeted intervention.
Level of evidence: Level IV, retrospective observational study.
Keywords: ankle fusion; nonunion; prediction model; risk assessment.
© The Author(s) 2015.