Abstracts

Rationale: The incidence of epilepsy is highest in older adults. Though there are recommendations for first drug choice in new-onset epilepsy which are known to be inconsistently followed, there are no data on how initial drug prescribed might be associated with important short-term outcomes including injuries.

Methods: This was a retrospective cohort study of adults 50 years or greater with new-onset epilepsy (identified using validated ICD-9-CM or ICD-10-CM codes for epilepsy or convulsion) in 2015-16, sampled from the IBM Marketscan’s Commercial and Medicare Databases (2013-2017). Participants with an epilepsy related ICD code were excluded if they had an epilepsy-specific medication prescribed (e.g. carbamazepine, oxcarbazepine, eslicarbazepine, topiramate, lacosamide, brivaracetam, valproic acid, phenytoin, felbamate, phenobarbital, vigabatrin, rufinamide), an epilepsy or convulsion diagnosis in the two years prior or if they had less than 3 years of continuous enrollment. The primary outcome, injury within one year of anti-seizure medication (ASM) prescription, was defined using healthcare encounters for injuries including trauma, burns, submersions and falls. ASMs were categorized into 1) “recommended/neutral” (carbamazepine, oxcarbazepine, eslicarbazepine, topiramate, lacosamide, brivaracetam, pregabalin, valproic acid), 2) “not recommended” (phenytoin, felbamate, phenobarbital, vigabatrin, rufinamide, cannabidiol, primidone), or “benzodiazepines.” Descriptive statistics were used to quantify and characterize covariates and logistic regression model was built adjusting for age, sex, Elixhauser comorbidity index, history of stroke, brain tumor or dementia (identified at or before the time of epilepsy diagnosis) to examine the relationship between particular drug choices and one-year injury. Planned analyses include a time-to-event analysis of time to injury, sensitivity analyses looking at prior injury rates and prior non-epilepsy specific prescription (including gabapentin and benzodiazepines).

Results: There 14,404 people with new-onset epilepsy in 2015-2016 who were prescribed an ASM within 1 year. Mean age was 66 and 57% were females. The three most commonly prescribed medications were levetiracetam (36%), gabapentin (21%), lorazepam (9%). Approximately 40% of people with new-epilepsy had a healthcare visit coded with an injury ICD code within 1 year after starting ASM. Among people with an injury: the median age was 66 (as compared to 62 in the non-injured group), median Elixhauser comorbidity index was 12 (9 in non-injured group) and 58.5% were female (56% in non-injured). A multivariable model showed an increased odds of injury within 1 year for people prescribed a “not recommended” ASM (see Table 1).

Conclusions: Older adults are still frequently prescribed sub-optimal medications including benzodiazepines. Initial multivariable models suggest that initial prescription of sub-optimal ASMs is associated with an increased risk of 1-year injury.

Funding: Please list any funding that was received in support of this abstract.: LJB receives support from the AES, the Epilepsy Foundation and the Mount Sinai Claude D Pepper Older Americans Independence Center (5P30AG028741-11). NJ is the Bludhorn Professor of International Medicine.