Risk Factors for Avascular Necrosis in Children With Cerebral Palsy Following Varus Derotation Osteotomy

Varus derotation osteotomy (VDRO) is a well-established surgery for displaced hips in children with cerebral palsy (CP). However, avascular necrosis (AVN) remains a significant postoperative complication that can result in negative outcomes. Previous studies report various AVN rates after VDRO, and the true incidence and associated risk factors remain poorly characterized. This study aimed to determine the incidence of AVN after VDRO in a large cohort and to identify patient-specific and surgical risk factors for developing AVN.

The study included 316 children with CP who underwent VDRO (593 osteotomies, 501 bilateral) for hip subluxation or dislocation [196 male; mean age 8.3 y, SD: 3.2, range: 2.4 to 17.7; 2 Gross Motor Function Classification System (GMFCS) level I, 34 level II, 38 level III, 115 level IV, 127 level V] at a tertiary referral center. Potential determinants of AVN were analyzed using Fisher exact test and univariate and multivariate logistic regressions.

Ten of 593 hips (1.7%) developed AVN. The entire cohort's mean preoperative Reimers MP was 51.1, SD: 28.0, range: 0 to 100, and was 77.6 (SD: 27.3, range 32.2 to 100) in the hips in which AVN developed and 50.6 (SD: 27.9, range 0 to 100) in those in which it did not. In univariate analysis, Reimers MP (P=0.02), bilateral surgery (P<0.01), and pelvic osteotomy (PO) (P<0.01) were significant risk factors for developing AVN. In multivariate analysis, the sole significant risk factors was PO (P<0.01). Lastly, the Rutz and MCPHCS data reveal similar rates of AVN across these femoral hip classification systems, regardless of the change in Rutz (P=0.13) or MCPHCS (P=0.58).

This is the largest analysis of AVN in children with CP who underwent VDRO. PO was a significant risk factor for developing AVN after VDRO. Although the overall incidence is low, the increased odds of AVN in patients undergoing POs is a factor surgeons should consider when undertaking surgery for children with CP and displaced hips. The association with PO may explain why several factors were significant solely in univariate analysis. For example, children functioning at GMFCS level V underwent ~3.5 times the rate of POs than their higher-functioning peers (29.0% vs. 8.2%, P<0.01). Finally, the Rutz and MCPHCS classifications suggest that AVN may not significantly impact postoperative hip sphericity. In combination with the low incidence of AVN, these findings indicate that rather than focusing solely on AVN risk, sphericity preservation may be more critical for optimal health outcomes in children with CP.