A retrospective analysis of patients with febrile seizures followed by epilepsy

Dear Sir, With great interest we read the article by Saltik et al. recently published in the June issue of Seizure1. The authors analysed the clinical patterns of febrile seizures (FS) in patients with later development of epilepsy and tried to find any factors predictive for the subsequent epilepsy syndrome. Among the 108 patients with epilepsy and a previous history of FS they found 75 (68.8%) with partial epilepsies and 17 (15.5%) with generalised epilepsies. Patients with partial epilepsies had more often younger age at onset of FS with presence of focal features, febrile status, and a longer interval between the first FS and the first afebrile seizure together with a high incidence of FS in the family history. Whereas those with generalised epilepsies (GE) had a shorter interval between the first FS and the first afebrile seizure, a high incidence of single FS and a family history of epilepsy. These findings broadly confirm the results from a previous study by our group performed at the Universitätsklinik Innsbruck2. We analysed 113 patients with epilepsy and a previous history of FS. Forty-five (39.8%) had temporal lobe epilepsy (TLE), 41 (36.6%) GE, and 27 (23.9%) had extratemporal epilepsy (ETE). Patients with TLE had a significantly longer duration of FS (P ≤ 0.001), more often focal features (P ≤ 0.001), and febrile status epilepticus (P ≤ 0.001) than patients with GE. Though age at FS, number of FS, family history, birth history and neurological status at FS did not differ between groups, we were able to assign 82% of patients with TLE and all of the patients with GE correctly, using a stepwise discriminant model with inclusion of all analysed factors. Selik and co-workers did not use appropriate tests, like multiple logistic regression analysis or discriminant analysis to detect any predictive factors for the later epilepsy syndrome. However, the inherent weakness of Selik’s and our own study is the retrospective design together with a selection bias towards more severe forms of epilepsy, because both studies were hospital based. Any attempt to find such predictive factors for the development of later epilepsy in a child with FS is important for several reasons. Firstly, to identify children at risk for later TLE and to design prospective long-term studies with MRI follow up. Only prospective long-term studies are able to clarify the relationship between certain clinical features of FS and progression to TLE with hippocampal sclerosis. Secondly, in genetic studies only a meticulous description of the phenotypic features of FS may reduce genetic heterogeneity and thereby increase power in linkage analysis3. Results of a recent case control study by Pal et al. suggested recurrent febrile seizures and afebrile seizures as criteria on which to subgroup a linkage sample. However, these subgroups will not be evident at the time of the initial febrile seizure and further prospective collection of clinical data are still needed. Saltik’s data are of special interest because they analysed a selected population, as they have mentioned in their article. We were intrigued by the high number of children with occipital paroxysm (CEOP), which are well known to have a strong genetic background and a high incidence of FS in their history4, 5. Again, the application of appropriate statistical tests is necessary to describe any predictive factors, which in turn could be useful to increase power in linkage analysis.