Abstract
BACKGROUND: Hip displacement and dislocation are among the most disabling musculoskeletal sequelae of cerebral palsy (CP) yet reported incidence and risk-factor estimates vary widely across studies. We undertook a systematic review and pooled analysis to quantify cumulative incidence across Gross Motor Function Classification System (GMFCS) strata and identify reproducible clinical and radiographic predictors. METHODS: A protocol was registered in PROSPERO (CRD420251026860). MEDLINE (PubMed), Embase, CINAHL and CENTRAL were searched from inception to March 30, 2025. Eligible longitudinal studies enrolled ≥30 children with CP aged 2-18 years, provided ≥2 years' follow-up without confounding hip-directed intervention, and reported migration percentage (MP) data or equivalent permitting derivation. Hip displacement and dislocation were harmonized as MP >30 % and >50 %, respectively. Study quality was appraised with ROBINS I. Proportions were stabilized with the Freeman-Tukey double-arcsine transformation and pooled in random-effects (REML) models; odds ratios (ORs) for candidate predictors were combined using inverse-variance random-effects methods. Heterogeneity (I(2), τ(2)), prediction intervals, influence diagnostics, Hartung-Knapp sensitivity and Egger tests were performed. Certainty was graded with adapted GRADE. RESULTS: Nineteen studies met inclusion; nine natural-history cohorts (n = 1556; median follow-up 5.1 y) contributed extractable incidence data. The pooled cumulative incidence of hip displacement/dislocation was 38.2 % (95 % CI 31.7-45.1 %; I(2) = 77 %; prediction interval 6.0 53.8 %). Incidence was 17.1 % in ambulant children (GMFCS I-III) and 71.9 % in non-ambulant children (IV-V), yielding an OR 3.72 (95 % CI 2.56-5.40) for non-ambulant vs ambulant groups. A baseline MP ≥ 30 % quadrupled subsequent risk (OR 4.48, 95 % CI 2.66-7.54; I(2) = 0 %). Pelvic obliquity ≥10° was associated with increased risk in a single cohort (OR 2.70, 95 % CI 1.34-5.46) and should be regarded as suggestive pending replication. No consistent effects were found for sex, gestational age or CP subtype. CONCLUSIONS: Approximately four in ten children with CP, and more than two thirds of those in GMFCS IV-V, develop clinically important hip displacement without targeted intervention. GMFCS IV-V status and an early MP ≥ 30 % are robust, actionable triggers for intensifying hip surveillance to six-monthly radiography; pelvic obliquity ≥10° may further stratify risk but requires confirmation. Uniform MP thresholds, time-to-event analyses and reporting of modifiable exposures are needed in future multicenter cohorts to refine preventive care.