Sample Review
Related Disease States

Do retinal hemorrhages occur in infants with convulsions? Ana Isabel Curcoy, Victoria Trenchs, Marta Morales et al. Arch Dis Child 2009;94:873-875. (from Barcelona, Spain)

Review from Winter 2010 issue

This is a prospective study of patients aged 15 days to two years who were admitted for first convulsion to determine the prevalence of retinal hemorrhages. Children suspected of abuse secondary to signs of external trauma were excluded. Of 389 children seen in A+E, 182 were admitted. Of those, 54.9% were boys and 45.1% girls. On admittance, 52% had afebrile generalized seizures, 7.1% had afebrile partial seizure, 31.9% had complex febrile seizures, and 8.2% had simple febrile seizures. Three or more episodes of seizure were seen in 29.5% of the samples during the first 24 hours of hospitalization. In 28.6%, the duration of the seizure was for more than five minutes and 3.8% were admitted with status epilepticus. All patients had a fundus examination by an ophthalmologist. Neuroimaging and an electroencephalogram were performed when necessary.

Of the 182 patients admitted, two were found to have retinal hemorrhages. Both were girls, ages 6 months and 18 months. In each case a subdural hematoma was found. Subsequent investigation supported a diagnosis of shaken baby syndrome for each child.

Since no hemorrhages were found in any of the other children admitted with convulsions, the authors used Hanley’s rule of three to calculate that the chance of a retinal hemorrhage occurring as a result of convulsions alone was at most 0.0167. Thus, the authors conclude that convulsions alone are unlikely to cause a retinal hemorrhage in children less than two years of age and, therefore, if retinal hemorrhages are detected, further investigations into non-accidental injury are warranted.

    Reviewed by Mitchell Strominger, M.D.


Reviewer’s Note: The authors provide convincing evidence to support their conclusion although I am left desiring more information. Of those patients admitted, what was the criterion for obtaining neuroimaging? The authors do not state how many of their patients had neuroimaging nor, of those imaged, how many had subdural hemorrhages. Thus, if all patients had neuroimaging and only those with subdural hematoma had retinal hemorrhages, it might be cost effective to only screen those patients with intracranial hemorrhage for ophthalmic findings. Also, only 47% of patients were admitted under this study. How many patients were admitted with suspected non-accidental injury and thus excluded, and what were the criteria to exclude from admission and thus from the study?