Prevalence of the high-riding vertebral artery depends on how it is defined: a meta-analysis with implications for C2 screw fixation.

The high-riding vertebral artery (HRVA) is a common anatomical variant that increases the risk of vertebral artery injury during C2 screw placement for craniocervical fixation. The reported prevalence of HRVA has varied widely (4-65%), partly due to inconsistent definitions and mixed analytical units. The current study aimed to provide a definition-sensitive, hierarchically stratified pooled prevalence of HRVA. A systematic search of PubMed/MEDLINE, Web of Science, Scopus, EMBASE, and the Cochrane Library (inception through March 2026) identified 687 records. After screening, 51 full-text articles were reviewed, and 21 studies (34 independent cohorts; 3972 patients, 6598 sides) met criteria for quantitative pooling. A hierarchical analytical framework was applied: a primary, strict analysis restricted to patient-level cohorts using the conventional Neo/Bloch definition (C2 isthmus height ≤ 5 mm and/or internal height ≤ 2 mm); an expanded patient-level analysis incorporating all HRVA definitions; and a separate standard-definition side/joint-level analysis. Prevalence was pooled using random-effects models with the Freeman-Tukey double-arcsine transformation, with logit-transformed and binomial-normal mixed-model analyses as sensitivity analyses. The strict patient-level pooled prevalence was 32.2% (95% CI: 23.5-41.5%; I² = 65.7%; k = 7). Restricting to non-pathological/general populations yielded 28.5% (95% CI: 25.3-31.8%; I² = 39.9%), while pathological/selected populations pooled to 40.0% (95% CI: 17.8-64.5%). The expanded patient-level analysis (k = 21) yielded 31.7% (95% CI: 24.4-39.4%; I² = 89.5%). Definition-stratified subgroup analysis revealed a nearly fourfold range in pooled prevalence (16.8% for axial-pedicle-only to 61.5% for isthmus-only definitions). The standard-definition side- or joint-level prevalence was 20.3% (95% CI: 12.4-29.5%; k = 12). Estimates were robust to the logit transformation and a mixed-effects model, and leave-one-out analysis across all 21 cohorts confirmed robustness (range: 29.4-33.1%). Approximately one in four general-population patients and two in five pathological-population patients harbour at least one HRVA. Prevalence estimates are highly sensitive to HRVA definition, population composition, and analytical unit. Preoperative CT evaluation of the vertebral artery course at C2 should be mandatory for all patients planned for craniocervical instrumentation. Future studies should adopt the standardised Neo/Bloch definition and report prevalence by morphometric subtype to enable meaningful cross-study comparison.