Thirty-seven patients with non-traumatic cerebellar haemorrhage were treated in our department between 1984 and 1999. The outcome was retrospectively analysed according to the neurological status, CT findings and the selected treatment modality (conservative therapy vs. surgery). Our data indicate that in that sample size, the size of the haematoma was the only significant predictor of the outcome (p = 0.04). The prediction of the outcome by a logistic regression model includes both the size of the haematoma and the treatment modality (surgery or conservative) as variables. The data suggest that patients in a good neurological status on admission (Glasgow Coma Scale 13-15) with a small haematoma (<3 cm) should be treated conservatively. In all patients with large haematomas (>3 cm), including those in a good neurological status, and in all comatose patients, surgical evacuation is indicated. A different haematoma size as a cut-off point for choosing surgery as the preferred therapeutic modality, as well as the indication for ventriculostomy as a temporizing measure or a substitute for surgery are yet undetermined and need further investigation in a randomized trial.
Copyright 2002 S. Karger AG, Basel