Ju 2008.

Methods	Double‐blind (participants and outcome assessor), sham epidural‐controlled, clinical RCT Sequence generation was randomized, but not described Follow‐up: 12 months
Participants	Participants: 114 adults in a university setting in Beijing, China Operation: posterolateral thoracotomy for lung and oesophageal disease 2 groups, size: 57/57 Age (group 1, 2): 61.80 years (SD ± 13.78), 61.41 (SD ± 11.78) Men/women (group 1, 2): 41/13, 38/15 (completed the protocol) Remarks: pulmonary/oesophageal operation (group 1, 2): 28/26, 25/28 7 participants with dislodged catheters were excluded.
Interventions	Group 1 (preincision epidural): epidural at T6/7/8, preincision epidural ropivacaine (0.5%, bolus 5 mL to 10 mL), GA (fentanyl), post‐op for 72 h PCEA (0.125% bupivacaine + 0.05 mg/mL morphine + 0.02 mg/mL droperidol, basal 3 mL/h, demand 3 mL, lock out 15 min). Group 2 (control/cryotherapy): sham epidural at T6/7/8, GA (fentanyl), cryoalgesia, post‐op for 72 h PCA through sham epidural (SC, 1 mg/mL morphine, demand 2 mL, lock‐out in 30 min, no basal) Adjuvants: none Immediate post‐op pain control: not significant
Outcomes	Dichotomous: pain at 6 and 12 months Continuous: not reported Secondary: allodynia at 6 and 12 months
Notes	Funding sources: study supported by grants from Research and Development Foundation of Peking University People's Hospital Conflicts of interest: no conflict of interest statement given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients were stratified by disease sites (lung or oesophagus), and blinded randomized to receive either epidural analgesia (Epidural Group, Group E) or intercostal nerve cryoanalgesia (Cryo Group, Group C), in order to ensure that both groups had comparable operation methods." Randomization method not detailed, but otherwise well documented.
Allocation concealment (selection bias)	Low risk	Participants unaware of allocation, concealment of allocation for providers described: "After obtaining ... written informed consent from the prospective patient cases, 114 physical status I or II patients scheduled for posterolateral thoracotomy for lung or oesophagus diseases were enrolled in the study."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Intraoperative anaesthesia providers were not blinded. An effort was made to blind study participants. Quote: "in order to make the patients blinded to the analgesic method, SC infusion catheters were inserted at upper back (T7‐8 level) in Group C." This is acceptable, bias is unlikely.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessor "who was blinded to the postoperative pain management, interviewed patients by telephone, using a standard questionnaire."
Incomplete outcome data (attrition bias) All outcomes	High risk	Attrition was reported, but no ITT analysis was considered.
Selective reporting (reporting bias)	Unclear risk	No protocol was available, but pre‐specified outcomes within manuscript were all reported on.
Null bias	High risk	Quote: "no statistically significant differences were found between the two groups with respect to NRS pain scores at rest or on motion within three days following surgery"