Hallman 2002.

Methods	Randomised, split‐mouth study, 1 year post‐loading follow‐up. No withdrawals
Participants	Patients having less than 5 mm of alveolar bone in the floor of the sinus. Adults treated at the Gävla Hospital, Gävla, Sweden. No specific exclusion criteria were given (unhealthy, systemic or local contraindications such as undergoing radiation therapy). 11 patients were treated in the split‐mouth study and 10 in the preference group
Interventions	2‐stage sinus lift with autogenous particulate bone from the mandibular ramus versus 2‐stage sinus lift with a mixture of 80% of bovine anorganic bone (Bio‐Oss) and 20% of particulate bone from the mandibular ramus left to heal for 6 months. A fibrin glue (Tisseel Duo Quick, Immuno, Wien, Austria) was added to the grafts after thrombin (Thrombin, Immuno). Procedures were performed under local anaesthesia and oral sedation. All implants were turned titanium self tapping (Nobel Biocare, Göteborg, Sweden): Mark II type implants were used in the former 2 groups and Mark III in the latter. All patients were rehabilitated with screw‐retained metal‐ceramic fixed prostheses A third group (not part of this review) was composed by patients who refused to provide autogenous bone but accepted the treatment with a 2‐stage sinus lift with 100% Bio‐Oss
Outcomes	Prosthesis and implant failures, complications at augmented and donor sites; histomorphometry
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Article: No information presented Author's reply: "The randomization was done by a third party as a lottery. In an envelope there were 12 papers with either 6 possibilities for the mixture to be on the left side or the right side. Each patient was allotted by picking up one paper which said mixture on the left or right side"
Allocation concealment (selection bias)	Low risk	Article: No information presented Author's reply: "The surgeon knew the outcome of the surgery at the time of surgery just before making the incision"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Article: No information presented Author's reply: "The study was not blinded"
Incomplete outcome data (attrition bias) All outcomes	Low risk	All data presented, no drop‐outs
Selective reporting (reporting bias)	Low risk	All planned outcomes apparently reported
Other bias	Low risk	None apparent