Predictors of Failure for Vascularized and Nonvascularized Bone Grafting of Scaphoid Nonunions: A Systematic Review

Objective  This article compares predictors of failure for vascularized (VBG) and nonvascularized bone grafting (NVBG) of scaphoid nonunions. Methods  We conducted a systematic literature review of outcomes after VBG and NVBG of scaphoid nonunion. Fifty-one VBG studies ( N  = 1,419 patients) and 81 NVBG studies ( N  = 3,019 patients) met the inclusion criteria. Data were collected on surgical technique, type of fixation, time from injury to surgery, fracture location, abnormal carpal posture (humpback deformity and/or dorsal intercalated segmental instability [DISI]), radiographic parameters of carpal alignment, prior failed surgery, smoking status, and avascular necrosis (AVN) as defined by punctate bleeding, magnetic resonance imaging (MRI) with contrast, MRI without contrast, X-ray, and histology. Meta-analysis of proportions was conducted with Freeman-Tukey double arcsine transformation. Multilevel mixed-effects analyses were performed with univariable and multivariable Poisson regression to identify confounders and evaluate predictors of failure. Results  The pooled failure incidence effect size was comparable between VBG and NVBG (0.09 [95% confidence interval [CI] 0.05-0.13] and 0.08 [95% CI 0.06-0.11], respectively). Humpback deformity and/or DISI (incidence-rate radios [IRRs] 1.57, CI: 1.04-2.36) and lateral intrascaphoid angle (IRR 1.21, CI: 1.08-1.37) were significantly associated with an increased VBG failure incidence. Time from injury to surgery (IRR 1.09, CI: 1.06-1.12) and height-to-length (H/L) ratio (IRR 53.98, CI: 1.16-2,504.24) were significantly associated with an increased NVBG failure incidence, though H/L ratio demonstrated a wide CI. Decreased proximal fragment contrast uptake on MRI was a statistically significant predictor of increased failure incidence for both VBG (IRR 2.03 CI: 1.13-3.66) and NVBG (IRR 1.39, CI: 1.16-1.66). Punctate bleeding or radiographic AVN, scapholunate angle, radiolunate angle, and prior failed surgery were not associated with failure incidence for either bone graft type ( p > 0.05). Conclusion  Humpback deformity and/or DISI and increasing lateral intrascaphoid angle may be predictors of VBG failure. Time from injury to surgery may be a predictor of NVBG failure. AVN as defined by decreased contrast uptake on MRI may be a marker of increased failure risk for both bone graft types.