Summary:Purpose: Data accrued from clinical trials of five new antiepileptic drugs (AEDs) are compared for efficacy in reducing seizures and self-reported adverse events as a basis of selection among new AEDs. Drawbacks to use of these data also are demonstrated.Methorh: A review of double-blind, placebo-controlled clinical trials of a new AED or placebo added to a standard AED provided data on reduction of complex partial seizures (CPSs). Success is 250% fewer CPSs with a new AED or placebo; Overall Improvement is the success rate with drug minus the success rate with placebo. Adverse events were tabulated from product-labeling lists of COSTART items (incidence, 2 5 % ) . The Summary Coinplaint score is the total number of reports of individual events for each AED.He,tults: Efficacy data demonstrate differences in Overall Improvement rates among five new AEDs and placebos (p = 0.001). However, rates of response to placebo also differed significantly among trials (p = 0.01). Adverse events predominantly affect central nervous system, psychiatric. and general body systems. However, patients in the placebo control groups did not consistently report adverse effects. Summary Coniplaint scores differ among the five new AEDs, but variability in use of COSTART terms nullifies comparisons.Conclu.sions: Comparisons of data for five new AEDs provide information for selection among treatments when ;I second drug is needed to iinprove control of CPSs. However, significant differences among the control groups and other problems make comparisons between trials problematic. The final choice should be based on the need of the individual patient for superior seizure control versus minimal adverse effects. likely efficacy and adverse effects. Selection of data from a single primary trial performed at the recommended dose might allow optimized comparisons among the new AEDs (6). A recent discussion of meta-analyses noted that summarization of data into a ~i n g l e odds ratio may not be the optimal approach for comparisons among clinical trials (73). Nonetheless, the meta-analysis provides valuable information to guide the planning of a future comparison among the new AEDs. A number of reviews described typical profiles of adverse effects. usually based on clinical experience, but without supporting data (9-1 1). Although Brodie's "Star System" (10) (a combination of efficacy, adverse events, and ease of use) is well founded based on clinical experience, it does not satisfy the need for numerically based comparisons from clinical trials. Most reports of clinical trials listed the incidence of common adverse effects based on patient self-report. These lists usually group problems based on an international dictionary of terms, COSTART, that provides a standardized format for the description of adverse events (12). A different approach was taken in the VA Cooperative Studies (1,2). Patients were asked direct questions and were examined for specific adverse effects by using Systemic and Neurotoxicity Rating Scales (13) to determine t...