Lavand'homme 2005.

Methods	Double‐blinded (participant, outcome assessor), placebo/sham‐controlled, randomized clinical trial Sequence generation by computer‐generated random numbers Follow‐up for 12 months
Participants	Participants: 85 adults in a university setting in Brussels, Belgium Operation: colonic resection (xiphopubic incision) of rectal adenocarcinoma 4 groups, size: 20/20/20/20 Age (group 1, 2, 3, 4): 53 years (SD ± 8), 54 (SD ± 8), 55 (SD ± 8), 53 (SD ± 10) Men/women (total: group 1, 2, 3, 4): 49/31: 12/8, 13/7, 12/8, 12/8 Remarks: intraoperative discovery of an extended tumour resulted in participants' exclusion from the study.
Interventions	Group 1 (IV/IV): epidural catheter at T8, GA (sufentanil 2.5 µg) IV (lidocaine 2 mg/kg + 0.5 mg/kg/h, clonidine 4 µg/kg + 1 µg/kg/h, sufentanil 0.1 µg/kg + 0.07 µg/kg/h) post‐op IV PCA (lidocaine bolus per request 7.5 mg, clonidine bolus per request 15 µg, morphine bolus per request 1.3 mg) (0.75 mL solution per demand, lockout time 7 min, max 15 mL per 4 h) Group 2 (IV/epidural): epidural catheter at T8, GA (sufentanil 2.5 µg); IV (lidocaine 2 mg/kg + 0.5 mg/kg/h, clonidine 4 µg/kg + 1 µg/kg/h, sufentanil 0.1 µg/kg + 0.07 µg/kg/h), before recovery (epidural bolus 7 mL bupivacaine 0.5%, clonidine 1 µg/kg, sufentanil 0.03 µg/kg) post‐op epidural PCEA (bupivacaine 5 mL 0.0675% + 5 mL/h 0.0675%, clonidine 3.5 µg + 3.5 µg/kg/h, sufentanil 0.05 µg + 0.05 µg/h) (continuous infusion of 5 mL and bolus of 5 mL on request, 40 min lockout time) Group 3 (epidural/epidural): epidural catheter at T8, GA (sufentanil 2.5 µg), preincision epidural (bupivacaine 7 mL 0.5% + 5 mL/h 0.125%, clonidine 1 µg/kg + 0.5 µg/kg/h, sufentanil 0.03 µg/kg + sufentanil 0.015 g/kg/h) post‐op epidural PCEA (bupivacaine 5 mL 0.0675% + 5 mL/h 0.0675%, clonidine 3.5 µg + 3.5 µg/kg/h, sufentanil 0.05 µg + 0.05 µg/h) (continuous infusion of 5 mL and bolus of 5 mL on request, 40 min lockout time) Group 4 (epidural/IV): epidural catheter at T8, GA (sufentanil 2.5 µg), preincision epidural (bupivacaine 7 mL 0.5% + 5 mL/h 0.125%, clonidine 1 µg/kg + 0.5 µg/kg/h, sufentanil 0.03 µg/kg + sufentanil 0.015 g/kg/h), post‐op IV PCA (lidocaine bolus per request 7.5 mg, clonidine bolus per request 15 µg, morphine bolus per request 1.3 mg) (0.75 mL solution per demand, lockout time 7 min, max 15 mL per 4 h) Adjuvants: ketamine from skin incision to the end of surgery (0.5 mg/kg bolus followed by continuous infusion at 0.25 mg/kg/h), clonidine as detailed above Immediate post‐op pain control: significantly improved
Outcomes	Dichotomous: pain at 6 and 12 months Continuous: Pain Disability Index at 6 months, Mental Health Inventory‐18 at 6 months Secondary: punctuate wound hyperalgesia was reported for the first 72 h
Notes	We contacted the study authors for missing data and they responded, but with some data inconsistencies that could not be verified or corrected. The study authors reported an unusually high success rate of epidural analgesia with only 2 failures in 60 participants. Funding sources: "support was provided solely from institutional and/or departmental sources." Conflicts of interest: no conflict of interest statement provided
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "according to a computer‐generated table of random number assignments, each patient was assigned to one of four double‐blinded groups." Bias is unlikely.
Allocation concealment (selection bias)	Unclear risk	The timing of allocation and concealment not detailed. Risk of bias is unclear.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "all of the analgesic solutions were prepared by an anesthesiologist who was not involved in the patients' care." Testing the epidural in the PACU "prevented a true double blinding in the postoperative period."
Blinding of outcome assessment (detection bias) All outcomes	Low risk	However, (quote:) "postoperative parameters were recorded by an anesthesiologist who was not aware of the intraoperative treatment administered to the patient", "mobilization assessed by a blinded observer", telephone interviews were "performed by the research nurse." The study author responded: " the research nurse (outcome assessor) was blinded to the group allocation ..." as there was no random code on questionnaire. Bias is unlikely.
Incomplete outcome data (attrition bias) All outcomes	High risk	Adverse effects and attrition were reported with group allocation. “Absence of thermoanalgesia level as well as intraoperative discovery of an extended tumor resulted in the patient’s exclusion from the study.” ”One was excluded during surgery after discovery of widespread neoplastic disease, and two other patients were excluded for postoperative early dislocation of epidural catheter (before 72‐h follow‐up).” “... one who died of a cardiac arrest at home 2 months" before completion. Results reported on a per‐participant basis, with no ITT analysis considered.
Selective reporting (reporting bias)	Low risk	Primary outcomes fully reported on
Null bias	Low risk	Quote: "patients in group 1 (intravenous–intravenous) experienced significantly more severe pain than patients in the three other groups. Cumulative number of satisfied analgesic requirements was significantly higher in group 1 (intravenous–intravenous) than in the other groups "