Differences in the time of onset of postoperative upper limb palsy among surgical procedures: a meta-analysis.

Abstract Background Context The time of onset of postoperative upper limb palsy (ULP), also referred to as C5 palsy, varies among patients. Although some studies addressed the relationship between the time of onset of ULP and the potential etiologies, no meta-analysis has been conducted regarding the association between the time of onset of ULP and other factors such as surgical procedure type and treated diseases. Purpose This study aimed to elucidate differences in the time of onset of ULP among spinal surgical procedures and treated diseases to understand its etiology. Study Design This is a meta-analysis. Outcome Measures The time of onset of ULP after cervical decompression surgery. Methods We conducted a meta-analysis via searches of the PubMed, EMBASE, and Cochrane Library databases. Upper limb palsy within 2 days postoperatively or at 3 days or more postoperatively was defined as early- and late-onset, respectively. We calculated the pooled prevalence of early- and late-onset ULP with regard to surgical procedures and diseases using a random effects model. The proportion of early-onset ULP relative to all ULP was also determined. Surgical procedures were categorized into four procedures: (1) anterior cervical discectomy and fusion (ACDF); (2) anterior cervical corpectomy and fusion (ACCF); (3) laminoplasty or laminectomy (LPN); and (4) posterior spinal fusion with decompression (PSF). Treated diseases were classified as ossification of the posterior longitudinal ligament (OPLL) and non-OPLL. Results The pooled prevalence of early-onset ULP was significantly stratified across three groups of surgical procedures (PSF [9.0%]>ACCF [3.7%] and LPN [2.5%]>ACDF [0.6%]). In the pooled analysis of late-onset ULP prevalence, three procedures (PSF [4.3%], ACCF [2.8%], and LPN [2.9%]) were similar, but ACDF (1.0%) was associated with a significantly lower prevalence than LPN. The prevalence of both early- and late-onset ULP was significantly higher with OPLL than without OPLL. However, the proportion of early-onset ULP relative to all ULP was similar between non-OPLL and OPLL (50.8% vs. 49.5%). Conclusions The high prevalence of early-onset ULP in PSF is attributable to a lag correction effect, which is triggered by posterior correction and fusion through comparison with LPN. The combination of our meta-analysis results and previous knowledge facilitates our understanding of the etiology of ULP.