Abstracts

Rationale: To evaluate the association between the use of continuous infusions for pediatric refractory status epilepticus (RSE) and in-hospital mortality. Methods: This multi-center prospective observational study included children, admitted from 2011 to 2017, in whom the status epilepticus did not stop after ≥2 anti-seizure medications (ASM). As our primary outcome – in-hospital mortality - was rare, we used Firth regression to estimate odds ratios (OR). We created a propensity score with the variables: Age, Gender, Race, Tonic-clonic RSE, Type of RSE (continuous/intermittent), Onset of RSE (pre-hospital/in-hospital), Etiology (structural, genetic/metabolic/other and unknown), Febrile RSE, Acute remote etiology, Neurological history, Delayed first ASM (>10 min), Previous prescription of rectal diazepam. We adjusted for these confounders with propensity scores using three methodologies: propensity score matching, inverse probability of treatment weighting (IPTW), and standardized mortality ratio (SMR). The final analysis Firth regression model was also adjusted for RSE convulsive duration. As our secondary outcomes – intensive care unit (ICU) length of stay and return to baseline function (yes/no)- were not rare in both groups, we used linear and logistic multivariate regression to evaluate these. Results: We studied 255 patients with a median (IQR) age of 4.2 (1.3-9.5) years (140 males, 55%). One hundred and thirty-seven (54%) patients were treated with continuous infusions, while 118 (46%) received repeated doses of non-benzodiazepine (BZD) ASM after failing ≥2 ASM. Nine (4%) patients died during the RSE hospitalization, and all of them received at least one continuous infusion (Odds Ratio (OR) 16.5; 95%CI 2.03-2135.9; p=0.004). However, patients with higher propensity scores had higher odds of receiving continuous infusions (Figure 1).Matching based on propensity score yielded a cohort of 156 patients: 78 patients treated with continuous infusion and 78 with repeated non-BZD ASM, with balanced confounders (Figure 2). Propensity score matching yielded an OR of 10.6 (95%CI 1.1-1405.1; p=0.037) for in-hospital mortality. IPTW yielded an OR of 30.0 (95%CI 3.9-3842.1; p < 0.001); and SMR yielded an OR of 17.1 (95%CI 2.1-2222.4; p=0.004). Patients treated with continuous infusion had a longer ICU length of stay [median (IQR) 8.2 days (2.2-20.5) vs 2.3 days (1-4)] (p < 0.001), and returned to baseline at discharge less often [81 (67%) vs 86 (86%)] (p=0.001). In multivariate analysis, continuous infusion remained associated with ICU length of stay duration (p < 0.001) but not with return to baseline function (p=0.58), after adjusting for the confounders included in the propensity score, RSE convulsive duration and (respectively) return to baseline function or ICU length of stay. Conclusions: Treatment of pediatric RSE with continuous infusions is associated with increased in-hospital mortality and longer ICU length of stay after controlling for known confounders and convulsive RSE duration. Funding: Supported by Pediatric Epilepsy Research Foundation and Epilepsy Research Fund