Main outcome indicators
3.1 Postoperative nerve damage
Seven of the studies described postoperative nerve damage15-21, which was transient and resolved during the follow-up. In this meta-analysis, postoperative nerve damage was divided into transient greater auricular nerve damage and other nerve damage, including the marginal mandibular branch of the facial nerve, lingual nerve, and hypoglossal nerve. Transient auricular nerve damage mainly manifested as temporary earlobe numbness. Damage to the marginal mandibular branch of the facial nerve, lingual nerve, and hypoglossal nerve was indicated by crookedness of the corners of the mouth, tongue tip numbness, and tongue deviation on extension, respectively.
3.1.1 Temporary earlobe numbness
Four of the eight studies reported the occurrence of temporary postoperative earlobe numbness15, 16, 18, 21. The studies included 161 patients; 73 and 88 in the retroauricular hairline and conventional approach groups, respectively. There were seven cases of temporary earlobe numbness, all in the retroauricular hairline group. The difference in temporary postoperative earlobe numbness between the two surgical approaches was statistically significant (OR: 7.28; 95% CI: 1.57–33.87; P = 0.01; Fig. 2). There was no significant heterogeneity among the studies (χ2 = 0.16, df = 3, P = 0.98, I2 = 0%; Fig. 2). This indicated that submandibular gland excision through the retroauricular hairline approach increases the risk of temporary earlobe numbness, but this resolves during the follow-up.
3.1.2 Other nerve damage
Transient damage to the marginal mandibular branch of the facial nerve, lingual nerve, and hypoglossal nerve was described in seven of the studies15-21, with a total of 270 patients; 114 and 156 in the retroauricular and conventional approach groups, respectively. There was no statistically significant difference between the two surgical approaches (OR: 0.80; 95% CI: 0.38–1.71; P = 0.57; Fig 3). This indicates that the two surgical approaches had similar incidences for damage to the marginal mandibular branch of facial nerve, lingual nerve, and hypoglossal nerve. The incidences of nerve injury in both the conventional and retroauricular groups in the Carlos 2019 study [20] were close to 50%, however, all symptoms disappeared during the follow-up period. When the results of this study were excluded, the results remained statistically insignificant between the two surgical approaches (OR: 0.77; 95% CI: 0.24–2.40; P = 0.65; Fig 12).