OBJECTIVES: To examine the clinical effectiveness, tolerability and cost-effectiveness of gabapentin (GBP), lamotrigine (LTG), levetiracetam (LEV), oxcarbazepine (OXC), tiagabine (TGB), topiramate (TPM) and vigabatrin (VGB) for epilepsy in adults. DATA SOURCES: Electronic databases. Internet resources. Pharmaceutical company submissions. REVIEW METHODS: Selected studies were screened and quality assessed. Separate analyses assessed clinical effectiveness, serious, rare and long-term adverse events and cost-effectiveness. An integrated economic analysis incorporating information on costs and effects of newer and older antiepileptic drugs (AEDs) was performed to give direct comparisons of long-term costs and benefits. RESULTS: A total of 212 studies were included in the review. All included systematic reviews were Cochrane reviews and of good quality. The quality of randomised controlled trials (RCTs) was variable. Assessment was hampered by poor reporting of methods of randomisation, allocation concealment and blinding. Few of the non-randomised studies were of good quality. The main weakness of the economic evaluations was inappropriate use of the cost-minimisation design. The included systematic reviews reported that newer AEDs were effective as adjunctive therapy compared to placebo. For newer versus older drugs, data were available for all three monotherapy AEDs, although data for OXC and TPM were limited. There was limited, poor-quality evidence of a significant improvement in cognitive function with LTG and OXC compared with older AEDs. However, there were no consistent statistically significant differences in other clinical outcomes, including proportion of seizure-free patients. No studies assessed effectiveness of AEDs in people with intellectual disabilities or in pregnant women. There was very little evidence to assess the effectiveness of AEDs in the elderly; no significant differences were found between LTG and carbamazepine monotherapy. Sixty-seven RCTs compared adjunctive therapy with placebo, older AEDs or other newer AEDs. For newer AEDs versus placebo, a trend was observed in favour of newer drugs, and there was evidence of statistically significant differences in proportion of responders favouring newer drugs. However, it was not possible to assess long-term effectiveness. Most trials were conducted in patients with partial seizures. For newer AEDs versus older drugs, there was no evidence to assess the effectiveness of LEV, LTG or OXC, and evidence for other newer drugs was limited to single studies. Trials only included patients with partial seizures and follow-up was relatively short. There was no evidence to assess effectiveness of adjunctive LEV, OXC or TPM versus other newer drugs, and there were no time to event or cognitive data. No studies assessed the effectiveness of adjunctive AEDs in the elderly or pregnant women. There was some evidence from one study (GBP versus LTG) that both drugs have some beneficial effect on behaviour in people with learning disabilities. Eighty RCTs reported the incidence of adverse events. There was no consistent or convincing evidence to draw any conclusions concerning relative safety and tolerability of newer AEDs compared with each other, older AEDs or placebo. The integrated economic analysis for monotherapy for newly diagnosed patients with partial seizures showed that older AEDs were more likely to be cost-effective, although there was considerable uncertainty in these results. The integrated analysis suggested that newer AEDs used as adjunctive therapy for refractory patients with partial seizures were more effective and more costly than continuing with existing treatment alone. Combination therapy, involving new AEDs, may be cost-effective at a threshold willingness to pay per quality-adjusted life year (QALY) greater than 20,000 pounds, depending on patients' previous treatment history. There was, again, considerable uncertainty in these results. There were few data available to determine effectiveness of treatments for patients with generalised seizures. LTG and VPA showed similar health benefits when used as monotherapy. VPA was less costly and was likely to be cost-effective. The analysis indicated that TPM might be cost-effective when used as an adjunctive therapy, with an estimated incremental cost-effectiveness ratio of 34,500 pounds compared with continuing current treatment alone. CONCLUSIONS: There was little good-quality evidence from clinical trials to support the use of newer monotherapy or adjunctive therapy AEDs over older drugs, or to support the use of one newer AED in preference to another. In general, data relating to clinical effectiveness, safety and tolerability failed to demonstrate consistent and statistically significant differences between the drugs. The exception was comparisons between newer adjunctive AEDs and placebo, where significant differences favoured newer AEDs. However, trials often had relatively short-term treatment durations and often failed to limit recruitment to either partial or generalised onset seizures, thus limiting the applicability of the data. Newer AEDs, used as monotherapy, may be cost-effective for the treatment of patients who have experienced adverse events with older AEDs, who have failed to respond to the older drugs, or where such drugs are contraindicated. The integrated economic analysis also suggested that newer AEDs used as adjunctive therapy may be cost-effective compared with the continuing current treatment alone given a QALY of about 20,000 pounds. There is a need for more direct comparisons of the different AEDs within clinical trials, considering different treatment sequences within both monotherapy and adjunctive therapy. Length of follow-up also needs to be considered. Trials are needed that recruit patients with either partial or generalised seizures; that investigate effectiveness and cost-effectiveness in patients with generalised onset seizures and that investigate effectiveness in specific populations of epilepsy patients, as well as studies evaluating cognitive outcomes to use more stringent testing protocols and to adopt a more consistent approach in assessing outcomes. Further research is also required to assess the quality of life within trials of epilepsy therapy using preference-based measures of outcomes that generate cost-effectiveness data. Future RCTs should use CONSORT guidelines; and observational data to provide information on the use of AEDs in actual practice, including details of treatment sequences and doses.