MANAGEMENT OF EPILEPSY IN WOMEN CONTEMPLATING PREGNANCY: A DECISION ANALYSIS MODEL
Method: We constructed a Markov model that predicts lifetime maternal and fetal health outcomes, comparing no AEDs with carbamazepine (CBZ), lamotrigine (LTG) and levetiracetam (LEV) that provide good seizure control and low fetal anomaly rates, yet differ in the rates of these and other adverse drug events. Our base case was a 30-year old woman with newly diagnosed generalized tonic-clonic seizures stabilized for at least one year on an AED of choice. The model was populated with probabilities for pregnancy events such as maternal seizure control and fetal congenital anomalies. Two value structures were explored: (1) calculating lifetime maternal and fetal quality-adjusted life years (QALYs) life-years (from the maternal and child’s perspective), and (2) calculated QALYs based on maternal perceptions of maternal and fetal health states only. QALYs were discounted at 5% per year.
Result: For the value structure including both, maternal and child lifetime QALYs, fetal anomalies and the consequent life expectancies of the affected offspring determined the optimal AED, with maternal life-events such as seizure rates and even maternal death proving inconsequential. For the value structure including maternal perceptions only, seizure rates with CBZ, LTG, LEV and no AEDs during pregnancy were 27.7%, 31%, 30.9% and 33%, while congenital anomaly rates were 2.88%, 2.0%, 2.44% and 1.1% respectively. In a woman with good seizure control prior to pregnancy, CBZ yielded discounted maternal QALYs following pregnancy of 14.39 years vs. 14.04 years for no AED and 14.03 years for both LTG and LEV. CBZ remained the preferred choice through a wide range of deterministic sensitivity analyses using plausible ranges for each event. The analysis was not sensitive to any variable in particular.
Conclusion: CBZ was the AED of choice during pregnancy for women with newly diagnosed epilepsy with good pre-pregnancy seizure control. LTG, LEV and no medications were seemingly suitable alternatives. This study highlights the issue of competing risks between mother and fetus and the challenges surrounding the value structure for decision analysis in pregnancy. It identifies the need for systematically eliciting maternal preferences for combined maternal and fetal health states for incorporation into decision analysis studies in pregnancy.