Correction to: British Journal of Cancer 121:827–836, https://doi.org/10.1038/s41416-019-0587-2, published online 15 October 2019

The original version of this article unfortunately contained a mistake. A correction is needed for one of the trials in the meta-analysis (Fig. 3). For Fakih et al. 1989, the relative risk of death (used as an estimate of the overall survival hazard ratio; as the authors did with other trials that did not report hazard ratios directly) was calculated using 8 deaths among 30 patients who had neck dissection versus 16 deaths among 40 patients who had resection only (RR = 0.67). The authors have since noted that there were 9 deaths among the 30 patients, yielding RR = 0.75 95% CI: 0.39–1.46. This trial had a small weight in the meta-analysis, so the corrected pooled hazard ratio across all 5 trials is now RR = 0.70 (95% CI: 0.55–0.88); almost the same as that published RR = 0.69 (95% CI: 0.55–0.87). Nevertheless, the main focus should be on the authors SEND trial and D’Cruz et al. 2015, as they were the most contemporary and highest quality.