Abstracts

Rationale: Seizures in Lennox-Gastaut syndrome (LGS) are extremely difficult to control with existing medical therapy. Corpus callosotomy (CC) and vagus nerve stimulation (VNS) are frequently considered as palliative surgical options, but the relative cost-effectiveness of these treatments is unclear. Meta-analyses suggest that CC is more effective for controlling seizures, but at a higher complication rate than VNS. Meanwhile, VNS implantation is less expensive, less effective, but with fewer complications than CC in the first year. We compare the one-year cost-effectiveness of CC and VNS for seizures in LGS using an exploratory decision analytic model.

Methods: A one-year exploratory decision analytic model was developed considering outcomes and costs after CC or VNS (Figure 1). Seizure reduction rates, seizure freedom rates, procedure costs, complication rates, and complication costs were obtained from available literature (Figure 2A). Seizure reduction and seizure freedom were considered as distinct probabilities. Wide ranges were considered for complication costs not reported in the literature. Greater than 50% reduction in seizure frequency and no complication was considered a positive outcome from the procedures. Infection, reoperation for any reason, permanent neurologic deficit, and no improvement in seizures were considered adverse outcomes. Cost-effectiveness was calculated as cost per positive seizure outcome gained.

Results: At one year, CC had an 15% greater likelihood of a positive seizure outcome, but per patient costs $68,147 more than VNS, or $451,952 per positive seizure outcome gained. One-way sensitivity analyses (Figure 2B) demonstrate that the probabilities of seizure freedom or reduction by VNS or CC and the cost of CC were the most influential variables in cost-effectiveness results. Probability of complication by VNS or CC had lesser influence on model results. When the model was performed specifically for atonic seizures as a scenario analysis, CC had a 27% greater likelihood of a positive seizure outcome, per patient incremental costs of $68,147 more than VNS, but a more favorable cost of $250,556 per positive seizure outcome gained.

Conclusions: In this one-year exploratory model, VNS may be more economically reasonable than CC for drug-resistant seizures in LGS. This finding was remained consistent when considering only atonic seizures. Seizure reduction rates for CC and VNS, as well as CC cost, had the greatest influence on the results of the model. This model suggests that preoperative tests defining probability of VNS or CC effectiveness would assist economically prudent surgical decision making. This exploratory model does not consider: influence distinct seizure types, costs beyond one year (e.g. pulse generator revisions), or changes in baseline healthcare utilization after CC or VNS. Future work will include a more detailed model with a longer time horizon.

Funding: Please list any funding that was received in support of this abstract.: Children's Hospital of Pittsburgh Foundation. NIH NIDCD (1R21DC019217, 2R01DC013315).