Author / AEDs analyzed / Patients / Method of analysis / Measurement of efficacy* / Measurement of tolerability†
Otoul et al., 2005 / LEV compared with GBP, LTG,OXC, TGB, TPM, ZNS / Adults and children / The estimate of treatment effect (LEV versus another AED) calculated by the difference between the logs of the 2 odds ratios. All doses included. / Time of observation not considered+
Responder rate:
LEV > GBP (2.64 [1.51–4.63])
LEV > LTG (1.86 [1.04–3.34])
LEV = ZNS (1.80 [0.94–3.45])
LEV = TGB (1.40 [0.72–2.73])
LEV = OXC (1.53 [0.88–2.66])
LEV = TPM (1.02 [0.59–1.77]
Seizure freedom rate not calculated. / Withdrawal rate:
LEV < OXC (0.55 [0.33–0.92])
LEV < TPM (0.52 (0.29–0.93)])
LEV =TGB (0.62 (0.35–1.11)])
LEV = ZNS (0.71 (0.36–1.38)])
LEV = LTG (1.09 (0.64–1.83)]
LEV = GBP (1.21 (0.68–2.15)]
Rate of withdrawal due to adverse effectsnot calculated.
Beyenburg et al., 2012 / Enzyme-inducing (ESL, OXC, TPM > 200 mg/d) versus non enzyme inducing new AEDs / Adults and children / Meta-regression analysis
All enzyme-inducing AEDs versus non enzyme-inducing AEDs. All doses included. / Time of observation not considered+
Responder rate:
the ratio of RRs of enzyme-inducing / non enzyme inducing AEDs non-significant: 1.01 (0.77–1.34)
Seizure freedom rate;
the ratio of RRs of enzyme-inducing/non enzyme-inducing AEDs non-significant: 1.38 (0.60–3.16). / Not analyzed.
Costa et al., 2011 / GBP, ESL, LEV, LTG, OXC, PGB, TGB, TPM, ZNS, LCM. / Adults and children / CMA and NMA. Each AED (irrespectively of dose) compared to the pooled mean effect of all other AEDs. / Time of observation not considered+
Responder rate (Odds ratio [95% CI])
TPM > all other AEDs (1.52 [1.06-2.20])
LEV = all other AEDs (1.31 [0.92-1.87])
TGB = all other AEDs (1.26 [0.73-2.16])
PGB = all other AEDs (1.24 [0.74-2.05])
OXC = all other AEDs (1.11 [0.59-2.06])
LTG = all other AEDs (0.75 [0.52-1.09])
ESL = all other AEDs (0.73 [0.52-1.03])
GBP < all other AEDs (0.67 [0.46-0.97])
LCM < all other AEDs (0.66 [0.48-0.92])
Number needed to treat [95% CI]
LEV < all other AEDs (0.78 [0.66-0.92])
TPM < all other AEDs (0.83 [0.72-0.96])
LCM =all other AEDs (0.84 [0.62-1.15])
OXC = all other AEDs (0.88 [0.61-1.26])
PGB = all other AEDs 0.91 [0.71-1.16])
ESL = all other AEDs (1.10 [0.76-1.58])
ZNS = all other AEDs (1.17 [0.93-1.48])
LTG = all other AEDs (1.22 [0.89-1.68])
GBP > all other AEDs (1.90 [1.28-2.80])
TGB > all other AEDs (1.99 [1.57-2.52])
Seizure freedom rates: no significant differences / Withdrawal rate (Odds ratio [95% CI])
GBP < all other AEDs (0.65 [0.42-1.00])
LEV < all other AEDs (0.62 [0.43-0.89])
LTG =all other AEDs (0.78 [0.57-1.06])
ESL =all other AEDs (0.90 [0.5-1.62])
PGB =all other AEDs (1.03 [0.77-1.40])
ZNS =all other AEDs (1.08 [0.72-1.63])
LCM =all other AEDs (1.24 [0.85-1.81])
TGB =all other AEDs (1.39 [0.88-2.19])
OXC > all other AEDs (1.60 [1.12-2.29])
TPM > all other AEDs (1.68 [1.07-2.63])
Number needed to harm (95% CI)
Not calculated for: LTG, GBP, LEV, and ESL because there were no differences between these AEDs and placebo.
PGB =all other AEDs ( 0.69 [0.33-1.44])
TGB =all other AEDs (0.68 [0.36-1.31])
LCM =all other AEDs (0.68 [0.35-1.31])
ZNS =all other AEDs (0.85 [0.22-3.31])
TPM =all other AEDs (0.92 [0.56–1.52])
OXC < all other AEDs (0.33 [0.21–0.52])
Rate of withdrawal due to adverse effects
LEV < all other AEDs (0.40 [0.23-0.82]
No differences between all other AEDs
Martyn-St James et al., 2012 / ESL, LCM,PGB, RTG, TGB, ZNS / Adults / CMA and NMA of RTG performance relative to other AEDs. Doses pooled only between those approved. / Time of observation considered+
Responder rate during maintenance§
RTG =ESL (1.34 [0.80–2.13])
RTG =LCM 1.28 [0.84–1.87])
RTG =TGB 0.83 [0.29–3.23])
RTG =ZNS 0.91 [0.58–1.52])
Responder rate during all double-blind period§
RTG = ESL (1.02 [0.57–1.67])
RTG =TGB (0.63 [0.33–1.30])
RTG < PGB (0.65 [0.41–0.96]
Seizure freedom rate during maintenance phase
RTG = ESL (1.13 [0.17–3.75])
RTG = LCM (1.09 [0.14–3.52])
Seizure freedom rate during double-blind phase
RTG vs PGB (2.59 [0.29–9.94]) and vs ZNS (0.76 [0.20–6.25]) non significant differences. / Withdrawal rate
RTG vs TGB: 1.09 (0.65–1.89), vs LCM: 1.31 (0.79–2.02), vs ZNS: 1.35 (0.79–2.44) and vs PGB: 1.42 (0.91–2.13) non significant differences
RTG > ESL (1.91 [1.18–2.89])
Rate of withdrawal due to adverse effects
RTG vs ZNS: 0.74 (0.34–1.82, vs ESL 0.87 (0.38–1.67) vs PGB: 0.97 (0.51–1.66), vs LCM: 1.03 (0.53–1.83) and vs TGB 1.12 (0.60–2.38) all non-significant differences.
Bodalia et al., 2013 / GBP, GVG, LCM, LEV, LTG, OXC, RTG, TGB, TPM, VPA, ZNS / Adults / CMA and NMA, comparisons between AEDs at Selected doses (those doses closest to that in established use). / Time of observation not considered+
Responder rate (Risk ratio)
From the most to the least efficacious§§: TPM, OXC, PGB, TGB, LEV, VPA, GVG, ZNS, LTG, GBP, LCM
An approximately two-fold difference in short-term efficacy was observed
Seizure freedom rate not estimated / Rate of withdrawal due to adverse effects (Risk ratio)
From the best to the less well tolerated§§: VPA, LEV, GBP, GVG, LCM, ZNS, PGB, TGB, LTG, TPM, OXC
Approximately a 5-fold difference in short-term tolerability was observed
Khan et al., 2013 / ESL, LCM, RTG, PER / Adults / NMA
Each AED compared with the other.
Pooled only doses licensed in UK / Time of observation not considered+
Responder rate
No significant differences between each of the 4 AEDs
Ranking order (from the most to the less efficacious): RTG, PER, ESL, LCM
Patients with secondarily tonic-clonic seizures at baseline (comparisons between PER and LCM with non significant results
Seizure freedom rate: No significant differences between each of the 4 AEDs / Rate of withdrawal due to adverse effects
No significant differences between each of the 4 AEDs
Ranking order (from the least to the better tolerated: LCM, ESL,RTG, PER*
Withdrawal rate: not done
CMA = classical meta-analyses; NMA = network meta-analysis; ESL = eslicarbazepine; GBP =gabapentin; GVG = vigabatrin;LCM = lacosamide;LEV levetiracetam; LTG lamotrigine; OXC = oxcarbazepine; PER = perampanel; PGB = pregabalin; RTG = retigabine; TGB = tiagabine; TPM = topiramate; VPA = valproate; ZNS = zonisamide
*AEDs are reported from the most to the less efficacious
†AEDs are reported from the best to the less well tolerated
+ In some RCTs, responder rate was calculated comparing seizure frequency during baseline with all double blind duration (titration plus maintenance), while in other RCTs only the maintenance phase was considered. This point has not been considered in several meta-analyses. For details, see text
§ data reported only for available comparisons
§§data reported only in a figure
One meta-analysis (21) has not been included in the table because of some issues with its results.