Modelling seizure rates rather than time to an event within clinical trials of antiepileptic drugs

Modelling seizure rates rather than time to an event within clinical trials of antiepileptic drugs

Within epilepsy research prediction models tend to consider outcomes such as 12 months without seizures, or treatment failure. The underlying statistical model is relatively simple to produce, is familiar to clinicians and is particularly useful to answer short-term questions such as what treatment should be prescribed following diagnosis. However, this statistical modelling approach only considers time to a first event rather than considering all seizures after starting treatment for example. This makes assessing changes in seizure rate over time difficult. Alternative statistical models exists which enable repeated medical events, such as seizures, to be modelled. These have the potential to make more efficient use of the collected seizure data. Ultimately, it is thought that these alternative statistical models will facilitate more accurate estimates of changes in seizure rates over time. This in turn will help to improve the effectiveness of future clinical studies for people with multiple seizures. This study aims to demonstrate the differences between the standard and alternative statistical approaches, and to consider the benefits of the alternative approaches for assessing change in seizure rates over time.

[{ "PostingID": 2371, "Title": "UCB-N01061", "Description": "Monotherapy With Levetiracetam in Newly Diagnosed Patients Suffering From Epilepsy" },{ "PostingID": 4828, "Title": "UCB-SP0993", "Description": "A Multicenter, Double-blind, Double-dummy, Randomized, Positive- Controlled Study Comparing the Efficacy and Safety of Lacosamide (200 to 600 mg/Day) to Controlled Release Carbamazepine (400 to 1200 mg/Day), Used as Monotherapy in Subjects (≥ 16 Years) Newly or Recently Diagnosed With Epilepsy and Experiencing Partial-onset or Generalized Tonic-clonic Seizures" }]