This study applied decision tree analysis to evaluate the sensitivity, specificity, and cost-effectiveness of clinical algorithms that incorporate 18F-FDG PET. Methods: A cohort of 176 patients was studied. The localization rate, accuracy, therapeutic impact on the presurgical decision-making process, and correlation with the postsurgical outcome were assessed for the tests commonly performed for seizure localization. Decision tree sensitivity analysis compared 3 imaging strategies with a baseline strategy of medical therapy for all: video-electroencephalography monitoring (VEM)/MRI strategy, in which patients underwent VEM and brain MRI only, and +SPECT and +PET strategies, in which patients with an indeterminate VEM/MRI result underwent ictal SPECT or interictal 18F-FDG PET, respectively. Results: The localization rates for VEM, MRI, 18F-FDG PET, ictal SPECT, and intracranial electroencephalography (EEG) were 62.2%, 35.8%, 75.0%, 60.0%, and 93.8%. The VEM/MRI strategy had the lowest cost per class I/II outcome, but the additional costs per class I/II outcome for the +PET and +SPECT strategies were always below the minimum reported cost savings for a class I/II outcome. There were no valid conditions in which the +SPECT strategy had a lower cost per class I/II outcome than the +PET strategy. Within the range of cost savings estimated to be associated with a class I/II outcome, all decision strategies produced net cost savings; however, these were significantly higher for the +PET and the +SPECT strategies. Conclusion: 18F-FDG PET is cost-effective in the presurgical evaluation, particularly when used in patients with a nonlocalizing or nonconcordant VEM or MRI result.
- Decision tree analysis
- Epilepsy surgery