Summary of findings 1. Bioengineered devices compared to standard techniques for peripheral nerve repair of the upper limb.

Repair using bioengineered devices versus standard techniques
Patient or population: people undergoing peripheral nerve repair of the upper limb Setting: upper limb peripheral nerve injury Intervention: bioengineered devices Comparison: standard repair
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with standard repair	Risk with bioengineered devices
Muscle strength at ≥ 24 months assessed with: BMRC Grading (manual muscle testing, score 0–5, where 0 = no movement, 5 = normal)	Not reported
Sensory recovery at ≥ 24 months assessed with: BMRC Grading (score S0–S4, where S0 = no sensation, S4 = normal) Follow‐up: 2 years	The mean sensory recovery assessed with BMRC sensory grading in the standard repair group at 2 years was 2.75 points	The mean sensory recovery assessed with BMRC sensory grading at 2 years with bioengineered devices was 0.03 points higher (0.43 lower to 0.49 higher)	—	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b	There may be no difference in therapeutic effect on sensory recovery with bioengineered devices compared to standard repair at 24 months, but the evidence is very uncertain.
Integrated functional outcome at ≥ 24 months assessed with: RMI (scale from 0 to 3, higher score better) Follow‐up: 2 years	The mean integrated functional outcome (RMI score) in the standard repair group was 1.875	The mean integrated functional outcome (RMI score) with bioengineered devices was 0.17 lower (0.38 lower to 0.05 higher)	—	60 (2 RCTs)	⊕⊕⊝⊝ Lowc,d	There may be little or no difference in RMI with bioengineered devices compared to standard repair at 24 months to 5 years. At 5 years, the RMI may be slightly better after device repair than standard repair (MD 0.23, 95% CI 0.07 to 0.38; 1 RCT, 28 participants).
Touch threshold assessed with: Semmes‐Weinstein monofilament (score 0–1, where higher score is better) Follow‐up: 24 months	Mean touch threshold score in the standard repair group was 0.81	The mean touch threshold score with bioengineered devices was 0.01 higher (0.06 lower to 0.08 higher)	—	32 (1 RCT)	⊕⊝⊝⊝ Very lowa,e	There may be little or no difference in touch threshold measured by Semmes‐Weinstein monofilament test with bioengineered nerve conduits compared to standard repair at 24 months. Semmes‐Weinstein monofilament test contributed to RMI data in 2 studies at 12 months. 1 further study planned to use this outcome measure but found it to be imprecise and did not report data.
Impact on daily living assessed with: DASH PROM Scale from: 0 (good) to 100 (poor) Follow‐up: 24 months	No studies employed DASH PROM.
Adverse events assessed as: adverse events (serious and non‐serious) Follow‐up: range 3 months to 5 years	10 per 1000	68 per 1000 (17 to 280)	RR 7.15 (1.74 to 29.42)	213 participants (5 RCTs)	⊕⊝⊝⊝ Very lowf,g,h	Use of bioengineered devices may increase adverse events compared to standard repair techniques, but the evidence is very uncertain. 2 studies included in this analysis had no adverse events. 1 study provided no information on adverse events in the standard repair group.
Specific serious adverse events: further surgery (device removal or revision)i assessed as: any unplanned secondary surgery to remove device Follow‐up: range 3 months to 5 years	12/129 devices required further surgery (device removal) in the bioengineered devices group; 0/127 procedures required further surgery in the standard repair group		RR 7.61 (1.48 to 39.02)	256 repairs (5 RCTs)	⊕⊝⊝⊝ Very lowf,h	The use of bioengineered devices may require more revision (device removal or revision) than standard repair but the evidence is uncertain. Unplanned removal of 12/44 devices (1/21 poly(DL‐lactide‐caprolactone) (Neurolac) devices, 8/17 silicone devices and 3/6 polyglycolic acid devices. 2 studies included in this analysis required no device removal.
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). BMRC: British Medical Research Council; CI: confidence interval; DASH PROM: Disability of Arm Shoulder and Hand Patient‐Reported Outcome Measure; MD: mean difference; RCT: randomised controlled trial; RMI: Rosén Model Instrument; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded twice for imprecision, because of the very small sample size.
^bDowngraded once for study limitations; outcome assessor blinding was broken beyond the first follow‐up year, representing a high risk of bias, and we judged two domains, including allocation concealment, at unclear risk of bias.
^cDowngraded once for imprecision because of the small sample size and the CIs did not rule out an effect (in favour of standard repair).
^dDowngraded once for study limitations; in one study, outcome assessor blinding was broken beyond the first follow‐up year, representing a high risk of bias. Across both studies, multiple domains, including allocation concealment in both studies, were at unclear risk of bias.
^eDowngraded once for indirectness due to subjective nature of the test, and one study found the test results too heterogeneous to be reported.
^fDowngraded twice for very serious imprecision because of the wide CIs.
^gDowngraded once for indirectness. We planned to report serious adverse events, but the studies did not classify adverse events as serious or non‐serious. 
^hDowngraded once for study limitations. All trials were either at high risk of bias or unclear risk of bias in multiple domains.
ⁱWe added secondary surgeries for unplanned device removal to the summary of findings table as a change from protocol, as this outcome is important in decision‐making.