Anterior versus posterior decompression and fusion for multilevel cervical myelopathy: a systematic review and meta-analysis.
TL;DR
For multilevel cervical myelopathy, ADF was associated with modest improvements in neurological function, health-related quality of life, cervical alignment, and shorter hospitalization compared with PDF, although several between-group differences were below minimal clinically important difference thresholds and some were sensitive to influential studies.
Abstract
Purpose
Optimal surgical strategy for multilevel cervical myelopathy remains debated. We compared anterior decompression and fusion (ADF) versus posterior decompression with fusion (PDF) across functional, health-related quality of life, radiographic, perioperative, and safety outcomes.
Methods
We updated a prior review with searches of PubMed, Scopus, and Web of Science (January 2023-24 August 2025). Comparative adult studies of multilevel disease treated with ADF or PDF were pooled using random-effects models to estimate mean differences (MD) or odds ratios (OR) with 95% CIs. Heterogeneity (I²), leave-one-out analyses, and Egger's tests assessed robustness and small-study effects.
Results
Thirty-one studies (2 RCTs, 29 cohorts; 5,613 patients: 2,427 ADF; 3,186 PDF) met criteria. ADF was associated with higher postoperative JOA (MD 0.76; 95% CI 0.36-1.16; I²=81.7%) and greater JOA recovery (MD 9.12%; 1.34-16.89; I²=93.4%), although the recovery benefit was not robust after removing one study or under Hartung-Knapp adjustment. NDI favored ADF (MD - 1.24; -2.30 to - 0.17; I²=72.2%) but attenuated in sensitivity and was sensitive to small-study effects. EQ-5D favored ADF (MD 0.07; 0.03-0.11; I²=0%); SF-36 did not differ overall (MD 3.48; -0.43 to 7.38) but favored ADF on sensitivity. Cervical lordosis improved more with ADF (C2-C7 Cobb MD 4.15°; 1.29-7.00; I²=89.7%). LOS was shorter with ADF (MD - 1.92 days; -2.32 to - 1.52; I²=49.9%), while operative time and blood loss were overall similar. Overall complications did not differ significantly between approaches (OR 1.48; 1.00-2.20; I²=65.9%); dysphagia was higher with ADF (OR 7.28; 4.19-12.64; I²=0%) and C5 palsy lower (OR 0.35; 0.15-0.82; I²=0%) versus PDF. Revision and axial pain were comparable. Egger's tests suggested small-study effects for NDI and LOS only; in meta-regression, OPLL status and study quality explained much of the between-study heterogeneity, and GRADE certainty was low to very low for most outcomes.
Conclusion
For multilevel cervical myelopathy, ADF was associated with modest improvements in neurological function, health-related quality of life, cervical alignment, and shorter hospitalization compared with PDF, although several between-group differences were below minimal clinically important difference thresholds and some were sensitive to influential studies. Trade-offs were approach-specific: dysphagia was higher after ADF, whereas C5 palsy was more frequent after PDF; overall complications did not differ significantly. Given the predominance of observational data and low-to-very-low GRADE certainty, approach selection should remain patient-tailored.