Gomes 2005.

Study characteristics
Methods	Study design: RCT (split‐mouth) Conducted in: Department of Oral and Maxillofacial Surgery, University of Pernambuco, Camaragibe, Brazil
Participants	Inclusion criteria: patients with bilateral mandibular impacted third molars, and all procedures had to be performed by the same operator Exclusion criteria: patients with medical problems that could contraindicate the procedure were excluded, as were any procedures in which complete fractures of the lingual cortex were likely Age: not stated Number randomised: 55 Number evaluated: 55
Interventions	Lingual nerve protection (Free's retractor) versus none Group A (n = 55 teeth): lingual flap with Free's retractor Group B (n = 55 teeth): without lingual flap Procedures under local anaesthesia or general anaesthesia with local anaesthesia. 1 surgeon Follow‐up: 3 months
Outcomes	Pin‐prick test to confirm nerve injury at 1 and 7 days postoperatively
Notes	Sample size calculation: not reported The authors note in the discussion that the lingual bone plate was preserved in all cases, and that this was responsible for a higher degree of difficulty during the procedure, "especially in deeper impactions". This may affect the homogeneity of the study.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "...all patients were randomly allotted" Comment: method of sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	Not mentioned
Blinding (performance bias and detection bias) patient	Unclear risk	Blinding not mentioned. Participants who underwent general anaesthetic are likely to have been unaware of the procedure they received, but those with only local anaesthetic may have been aware of the procedure used.
Blinding (performance bias and detection bias) assessor	Low risk	Assessor was blinded. Quote: "An oral and maxillofacial surgeon who knew the proposal of the study but did not know which side was an experimental or a control group performed this evaluation"
Incomplete outcome data (attrition bias) All outcomes	Low risk	No withdrawals
Selective reporting (reporting bias)	Unclear risk	Single outcome of nerve injury reported.
Other bias	Unclear risk	The authors note in the discussion that the lingual bone plate was preserved in all cases, and that this was responsible for a higher degree of difficulty during the procedure, "especially in deeper impactions".