Summary of findings 1. Add‐on felbamate compared to placebo for drug‐resistant focal epilepsy.

Add‐on felbamate compared to placebo for drug‐resistant focal epilepsy
Patient or population: people with drug‐resistant focal epilepsy Setting: inpatients and outpatients Intervention: felbamate Comparison: placebo
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with felbamate
50% or greater reduction in seizure frequency Follow‐up: 8 weeks	No pooled analysis			83 (1 RCT)	⊕⊝⊝⊝ Very lowa,b	In the one study that reported this outcome (Binelli 1999), 38% of participants who were allocated to add‐on felbamate had a > 50% reduction in seizures. Of these, 11% had complete cessation of seizures. No data were reported for participants randomised to add‐on placebo.
Absolute or percentage reduction in seizure frequency Follow‐up (range): 8 weeks to 10 weeks	No pooled analysis			172 (3 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	One study reported a percentage reduction in seizure frequency of 35.8% for participants randomised to felbamate compared to a percentage increase of 3.3% for those randomised to placebo (Binelli 1999). Another study reported a less striking percentage reduction in seizure frequency with add‐on felbamate (4.24 ± 55.61%) but a much larger increase in seizure frequency with add‐on placebo (‐19.14 ± 79.70%) (Leppik 1991). Notably, the direction of effect was the same for both of these studies. The third study reported that there was no significant difference in seizure reduction between the two treatment groups (Theodore 1991).
Treatment withdrawal Follow‐up (range): 2 weeks to 10 weeks	No pooled analysis			236 (4 RCTs)	⊕⊝⊝⊝ Very lowa,b,d	Three of the studies reported a higher treatment withdrawal amongst participants randomised to felbamate compared to placebo (Bourgeois 1993; Leppik 1991; Theodore 1991). However, one study reported a lower treatment withdrawal rate for participants randomised to felbamate compared to placebo (4 vs 8 participants, respectively) (Binelli 1999). The two cross‐over studies specifically reported that no participants withdrew from treatment during the placebo treatment period (Bourgeois 1993; Leppik 1991). The direction of effect is therefore unclear.
Adverse effects Follow‐up (range): 2 weeks to 10 weeks	No pooled analysis			236 (4 RCTs)	⊕⊝⊝⊝ Very lowa,b	Amongst the adverse effects reported, headache and dizziness were both reported by all four studies whilst diplopia, nausea and vomiting were each reported by three of the included studies. Two of the studies described the incidence of adverse effects for both treatment groups (Bourgeois 1993; Leppik 1991). The number of participants experiencing individual adverse effects was consistently higher amongst those randomised to felbamate compared to placebo. Other reported adverse effects included: ataxia, fatigue and blurred vision.
Quality of life Follow‐up: 4 weeks	No pooled analysis			64 (1 RCT)	⊕⊝⊝⊝ Very lowa,b	One study reported that motor skills and memory skills, as assessed by a Short Neuropsychological Test, remained the same or improved following treatment (Bourgeois 1993). The study did not, however, indicate whether there was any difference in outcome between the add‐on felbamate and placebo treatment groups.
CI: confidence interval; RCT: randomised controlled trial
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.

^aWe downgraded evidence once for risk of bias due to a lack of methodological details provided, incomplete outcome data, incomplete reporting of outcomes, and other potential bias regarding the stability of participants' drug regimen.
^bWe downgraded evidence once for imprecision due to the narrative synthesis conducted and the absence of an estimated effect size. We downgraded evidence again for imprecision because the number of events did not satisfy the optimal information size.
^cWe downgraded evidence once for inconsistency because the magnitude of effect varied greatly between the three studies that reported this outcome.
^dWe downgraded evidence once for inconsistency because one of the studies reported the opposite direction of effect for this outcome.