Singh 2007.

Methods	Triple‐blind (participant/provider/outcome assessor), placebo‐controlled, clinical RCT Sequence generation by a computer‐based, random numbers generator Follow‐up: mean of 4.7 years (range 4.5–5.4 years)
Participants	Participants : 26 adults in a university setting, Houston, Texas, USA Operation: ICBG for spinal arthrodesis 2 groups, size: 11/14 Age (all, 1, 2): 64 (range 34‐84), 66, 63 years Sex: not reported Comorbidities: not reported Remarks: 11 anterior ICBG included in the initial stage were later excluded
Interventions	Group 1 (treatment): GA, at closure continuous wound irrigation (bupivacaine hydrochloride and epinephrine (Marcaine) 0.5% 2 mL/h) for 48 h post‐op + PCA (hydromorphone hydrochloride (Dilaudid)) (basal, bolus and lock‐out time not specified) Group 2 (control): GA, at closure continuous wound irrigation (normal saline, 2 mL/h) for 48 h post‐op + PCA (Dilaudid) (basal, bolus and lock‐out time not specified) Adjuvants: none Immediate post‐op pain control: significantly improved
Outcomes	Dichotomous: graft site pain at around 55 months Continuous: VAS at around 55 months Secondary: pain frequency in days, functional activity score, overall satisfaction with the surgical procedure at around 55 months
Notes	Funding sources: "no funds were received in support of this work" Conflicts of interest: "no benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript."
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "the method used to generate the randomization consisted of a computer‐based number generator. Moreover, to account for the size of the sample groups, randomization attempted to balance baseline characteristics by stratification, such as age."
Allocation concealment (selection bias)	Low risk	Quote: "the participants were randomized and allocated by a different individual than the one who enrolled the patient." "Randomization and allocation to group type was concealed and not made public to the individual enrolling the patients, the treating physician, or to the nursing staff." "Patients were assigned to receive either one or the other (treatment) solutions at the time of surgery based on a coded sequence enclosed within an envelope."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "blinded and identical in appearance, solutions of saline and Marcaine were prepared." "Physicians, patients, nursing staff, and research personnel conducting the statistical analyses were blinded to the infusion solution until the end of the study to minimize potential for performance and detection bias."
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "the physician conducting the telephone interview as well as recording the data were blinded to the treatment group." "Research personnel conducting the statistical analyses were blinded to the infusion solution until the end of the study to minimize potential for performance and detection bias."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Study authors report details of attrition with reference to the groups participants were randomized to. "An intent‐to‐treat analysis was considered to preserve randomization and to offer the best representation of the clinical population." "Even if we assume that any treatment patient that was lost to follow‐up (n = 6 patients) was considered to be a failure (chronic dysesthesias, an ICBG VAS score of 8, 15 days of narcotic usage/mo, functional activity score of 4, and an overall dissatisfaction with the procedure), a statistical difference was still noted in the 2 groups (p = 0.05)."
Selective reporting (reporting bias)	Low risk	Primary outcomes fully reported on
Null bias	Low risk	Quote: "narcotic dosage, demand frequency, and mean VAS pain score were significantly less in the treatment (Marcaine) group at 24 and 48 hours"