Methods
The study population was the resident population (in 2003, an average of 148,128 inhabitants; 46.9% men) of the HD of Fe, a well-defined area of 724km2 in the province of Ferrara, Italy, which is the primary service area of the University Hospital (UH) of Fe (the only general hospital in this HD). The study period was from 1 January 2003 to 31 December 2003. In the HD of Fe, neurological patients are routinely referred to the UH of Fe,^9–11 where an on-call neurologist is available 24 hours a day, seven days a week. The HD territorial emergency service protocol for SE management recommends administering benzodiazepines intravenously (IV) and moving the patient as soon as possible to the UH emergency room. The purpose of the study was to identify SE cases in the study population in 2003 through an intensive survey of multiple sources of cases. All hospital ICD-9 discharge codes in 2003 concerning epilepsy in all positions were abstracted, and the clinical files were obtained from each hospital ward. Other sources were the UH electroencephalography (EEG) and paediatric EEG services archives. Further sources were the emergency room unit and the territorial emergency service computerised archives. The neurologists, the physicians in the paediatric neurological service and the physicians in the emergency room unit and the intensive care units of the UH of Fe were interviewed.

SE was defined as a single seizure lasting for more than 30 minutes or repeated seizures lasting for a period of more than 30 minutes without intervening recovery between seizures.¹² Charts of any possible SE case in 2003 selected through the above sources were carefully reviewed to identify those with SE and to classify SE by seizure type, duration and aetiology. SE seizure type was classified as: generalised from onset (no indication of anatomical localisation and no clinical evidence of focal onset), with the subgroups convulsive (tonic-clonic, tonic, clonic), non-convulsive (absence status) and myoclonic; partial, without generalisation, with the subgroups simple (no consciousness impairment) and complex (consciousness impairment); secondarily generalised; and unclassified.¹² The physicians involved in the management of SE patients were interviewed. Only the first SE episode was taken into account.^3,5 Each SE case was reviewed according to the Hauser criteria¹³ to determine whether it was acute symptomatic (in close association – within a week – with an acute brain insult) or unprovoked (no association with an acute brain insult); the latter cases were further categorised as progressive symptomatic (in association with a non-static progressive brain lesion), remote symptomatic (in association with a prior brain insult resulting in a static brain lesion; time between SE and the prior brain insult >1 week) or idiopathic/cryptogenic (no acute precipitating factor and no brain lesion), concerning SE of no clear aetiology – idiopathic was reserved for certain partial or generalised epileptic syndromes with particular clinical and EEG characteristics, while cryptogenic referred to cases where no factor associated with increased risk of seizures was identified. The only case of febrile SE (a special class of acute symptomatic SE occurring during a febrile illness among children in the absence of another acute symptomatic cause, such as central nervous system [CNS] infection)^2,5 was not excluded from the statistical analysis.