Abstracts

Rationale: Clinical practice variation (CPV) is inversely related to quality of care. To assess CPV in a large epilepsy program, we determined the magnitude of CPV in changes made to antiepileptic drugs (AED) and investigations ordered by epileptologists, considering patient factors.Methods: CPV within an adult outpatient epilepsy program in a large Canadian health region was examined using prospectively gathered data from first assessments. CPV among five epileptologists (E0-E4) involved and two not involved (E5) with VEM (video-EEG monitoring) was assessed, adjusting for patient characteristics [gender, age, generalized versus focal seizures, epilepsy duration, >1-year seizure free, non-epileptic seizures (NES), driving status, having tried >1 AED, AED side effects, and prior and concurrent investigations]. We assessed the likelihood of ordering 1) MRI, 2) EEG, 3) VEM, and 4) changes to AED regimen using Poisson regression with robust variance estimation. Results are expressed as rate ratios (RR, p-value) using E0 as baseline.Results: Among 1123 patients, 50.5%, 9.8%, 7.5%, 11.4%, and 13.6% were assessed by E0 to E5, respectively. MRI was ordered in 33.9%, EEG in 47.3%, VEM in 6.9%, and AED changes were made in 24.1% of patients. MRI ordering was less likely if >1-year seizure free (0.7,p=0.007), prior MRI (0.52,p<0.001), and with NES (0.61,p=0.01). MRI was more likely with VEM (2.1,p<0.001) and EEG referral (3.01,p<0.001). E1 (1.55,p=0.004), E2 (1.43,p=0.020) and E4 (1.66,p<0.001) were more likely to order MRI, as well as E3 if patients were not driving (2.26,p<0.001). EEG ordering was less likely if not currently driving (0.89,p=0.05), with prior EEG (0.65,p<0.001), and VEM referral (0.36,p<0.001), but more likely if MRI was ordered (2.12,p<0.001). EEG was less likely with E5 (0.77,p=0.021), and E1 in patients <1-year seizure free (0.71,p=0.025). Shorter epilepsy duration was related to increased EEG ordering by E2 and decreased EEG ordering by E4. VEM referral did not differ across E1-E5, but VEM was less likely for males (0.55,p=0.027), if EEG was ordered (0.23,p<0.001), and with >1-year seizure free (0.17,p=0.015). VEM referral was more likely with NES (2.17,p=0.006), MRI referral (1.88,p=0.012), prior MRI (1.89,p=0.038), and longer epilepsy duration (p<0.001). AED changes were more likely with E1 (1.50,p=0.038), E4 (2.04,p<0.001), E5 (1.58,p=0.011), and with generalized seizures (1.39,p=0.024). E1 (9.17,p<0.001), E3 (8.67,p<0.001), and E5 (5.42,p=0.002) were more likely to make changes in patients with side effects. Changes were more likely with >1 AED tried (1.28,p=0.047) except with E2 who was more likely to make changes if 1 or no AEDs had been tried (2.57,p<0.001). Conclusions: Previous and current investigations are highly associated with the follow-up investigations ordered; however, significant differences between epileptologists exist after adjusting for patient characteristics. AED management also varied across epileptologists. Further studies are needed to investigate the causes of this CPV and its impact on patient and health system outcomes.