Three RCTs were included in the review (n=155 patients). Study quality scores were 38, 75 and 76. None of the studies applied blinding and two studies concealed random allocation.
Compared with intramedullary nailing, plate fixation gave a significantly lower relative risk of reoperation (RR 0.26, 95% CI 0.07 to 0.88, NNT=10) and a significantly reduced risk of shoulder impingement (RR 0.10, 95% CI 0.03 to 0.42, NNT=6). There was no statistically significant heterogeneity for these analyses. Plate fixation did not yield greater risks of nonunion, infection or radial nerve palsy.
For reoperation, sensitivity analysis using study design, publication status, and nail insertion did not alter the findings.