TABLE 4.
Cox Regression Models for Risk of Future Seizure Recurrence of Any Type
| Model | Data Included | Covariates | Risk of a First Recurrence of Any Type, Hazard Ratio (95% CI) |
|---|---|---|---|
| 1 | Illness-associated first seizures (n = 112) | First seizure category | |
| Febrile | 1.00 | ||
| Nonfebrile illness | 0.73 (0.31–1.73) | ||
| 2 | Illness-associated first seizures (n = 112) | First Seizure with accompanying illness | |
| Acute non-GI Illness | 1.00 | ||
| Acute GI illness | 0.28 (0.09–0.80) | ||
| 3 | Illness-associated first seizures (n = 112) | Illness accompanying first seizure | |
| Acute non-GI illness | 1.00 | ||
| Acute GI illness | 0.28 (0.10–0.82) | ||
| First seizure category: | |||
| Febrile | 1.00 | ||
| Nonfebrile illness | 0.94 (0.39–2.24) | ||
| 4 | Illness-associated first seizures with testing performed (n = 89) | Rotavirus-negative/norovirus-negative | 1.00 |
| Rotavirus-positive or norovirus-positive | 0.40 (0.12–1.35) |
Cox regression estimates the risk of a recurrent seizure accounting for subject characteristics and time since first seizure. The interpretation of the hazard ratio is analogous to a relative risk. All models were controlled for use of antiepileptic medication as a time-varying covariate. No subsequent nonfebrile seizures were observed in children with first febrile seizures or in children with acute gastrointestinal illness. GI indicates gastrointestinal.