Intention-to-Treat Analysis and Generalizability of Results in Randomized Clinical Trials of Newer Anti-Epileptic Drugs
Abstract number :
3.181
Submission category :
Year :
2000
Submission ID :
3322
Source :
www.aesnet.org
Presentation date :
12/2/2000 12:00:00 AM
Published date :
Dec 1, 2000, 06:00 AM
Authors :
Robert T Fitzgerald, Elena Andresen, Edwin Trevathan, Washington Univ Sch of Medicine, St. Louis, MO; St Louis Univ Sch of Public Health, St. Louis, MO.
RATIONALE: Failure to adhere to intention-to-treat (ITT) methods in randomized clinical trials (RCTs) may introduce bias, as the populations in the treatment arms may not be comparable. The results of RCTs are used to make treatment decisions in the general population of epilepsy patients. METHODS: Fifty four published RCTs of anti-epileptic drugs(AEDs) (41 placebo-controlled, 39 add-on studies, 15 monotherapy trials, 15 crossover studies) were identified for review from MEDLINE searches of clinical trials listed under searches of gabapentin, lamotrigine, vigabatrin, tiagabine, topiramate, felbamate, zonisamide, and oxcarbazepine. RF and ET independently reviewed each study and recorded clinical trial data on specifically designed abstraction forms, with special attention to description of study populations, randomization techniques, and adherence to ITT methodology. A database was created and analyzed using SPSS 8.0 RESULTS: 22 of 54 RCTs indicated that ITT methods were used. 6 of these 22 RCTs did not include data on randomized patients after they dropped out in the efficacy analyses. Therefore, these 6 RCTs used on-treatment analysis rather than ITT methods. None of the 54 RCTs reported data on the patient population from which the study participants were drawn. None of the 54 RCTs reported characteristics of patients from the study sites who qualified for the RCT, but for whatever reason were not selected for the study. CONCLUSIONS: Less than half of published RCTs of newer AEDs used ITT methodology. Without ITT methods, the efficacy analysis may be flawed. A comparison of the RCT study subjects and the patients from the study sites who qualify for the RCT but who do not participate in the RCT, should be included as a component of the RCT published results. Without a comparison of the study population to the general population of patients with the same epilepsy syndrome and/or seizure type, generalization of of the RCT results to the population of epilepsy patients found in most neurology clinics may be invalid.