Impact of timing of continuous intravenous anesthetic drug treatment on outcome in refractory status epilepticus

Patients were eligible for inclusion into the study if they received treatment for RSE on the neurological intensive care unit of our institution between January 2001 and January 2015 with continuous intravenous infusion of at least one of the following drugs (cIVADs): midazolam, propofol, thiopental, and ketamine. We used our electronic medical records and electroencephalography databases to gather information on demographics, RSE etiology, severity and duration, treatment, complications, and outcome. Demographics included gender, age at admission, and premorbid functional status. Data were assessed by two independent reviewers using a standardized data extraction form (DM and RUK). In case of disagreement, data were analyzed by a third reviewer (HBH) and consensus was found through discussion. For reasons of comparability with previous research, RSE caused by hypoxic encephalopathy was excluded. Only incidence episodes were considered; that is, in case of recurrent treatment with anesthetics for RSE in our institution during the study period, only the first episode was entered into the study [9].

In accordance with previous studies, SE was defined as clinically or electroencephalographically persisting seizure with duration of at least 5 min or as a series of seizures without interictal recovery [8]. The worst seizure semiology prior to initiation of antiepileptic therapy was used to categorize episodes as simple partial SE (SPSE), complex partial SE (CPSE), generalized convulsive SE (GCSE), or non-convulsive SE (NCSE) in coma [10]. RSE was defined as SE with ongoing seizure activity despite application of two adequately dosed antiepileptic drugs (AEDs) [13]. RSE duration after initiation of cIVAD therapy was defined as time between the beginning of cIVAD treatment and clear and enduring electroencephalographic or clinical seizure cessation or both. Durations of RSE after initiation of cIVADs as well as times on mechanical ventilation were estimated in days. Any portion of one day in RSE or on the respirator, respectively, was counted as one full day. RSE severity was graded with the STESS and was dichotomized into STESS of less than 3 (mild) and STESS of at least 3 (severe) as previously proposed [12]. RSE etiology was categorized in accordance with the guidelines of the International League Against Epilepsy into acute symptomatic, remote symptomatic, progressive symptomatic, and unknown etiology [14]. A potentially fatal etiology was defined when meeting the criteria introduced by Rossetti et al. [15].

Treatment with cIVADs was defined as early when started within 48 h after RSE onset; otherwise, it was defined as late.

The primary outcome of this study was functional status at last available follow-up graded by the modified Rankin Scale (mRS) [16]. Information on outcome was extracted from discharge summaries of rehabilitation facilities or own records if patients had represented to our hospital. Outcome was defined as good in case of complete recovery after RSE (that is, when the mRS at last available follow-up equaled the premorbid mRS) or, in case of new disability, when the mRS at last available follow-up was less than 3. Otherwise, outcome was defined as poor. Secondary outcomes were functional outcome at discharge (with good and poor defined identically as for the primary outcome), in-hospital mortality, RSE termination, induction of burst suppression, use of thiopental, duration of RSE after initiation of cIVADs, duration of mechanical ventilation, and occurrence of super-refractory SE (SRSE).

Statistical analyses were performed by using SPSS Statistics 21.0 (http://​www.​spss.​com). P values of less than 0.05 were considered statistically significant, and all tests used were two-sided. Baseline clinical data and RSE characteristics were compared by using Pearson chi-squared test or, where appropriate, Fisher’s exact tests for categorical data and Mann–Whitney U test for continuous variables. Crude and adjusted risk ratios for primary and secondary outcomes were estimated by Poisson regression with robust error variance. Multivariable analysis adjusted for STESS (as continuous variable) and a potentially fatal etiology. These calculations were performed on the overall cohort as well as subgroups defined by the STESS (cutoff 3 points) and the variables included in it (that is, age, a history of seizures, level of consciousness, and worst seizure type). Time-to-event outcomes were compared with the Kaplan–Meier method with hazard ratios estimated by using Cox proportional hazard analysis. Patients in whom RSE could not be terminated prior to discharge or death were censored from this analysis.

We identified 159 RSE episodes in 131 patients. In 83 (52.2%) episodes, patients received treatment with cIVADs. After exclusion of recurrent episodes (n = 6), 77 patients remained for final analysis. Of those, 53 (68.8%) received cIVAD therapy within the first 48 h of RSE (Fig. 1). A description of the study cohort is presented in Table 1. Patients were treated with up to four cIVADs, but the majority received a maximum of two cIVADs (n = 59, 76.6%). The cIVAD most frequently used first was propofol (n = 47, 61.0%; Additional file 1: Table S1). In-hospital mortality was 24.7%, and RSE termination rate was 85.7%. Follow-up data were available for all but two patients (97.4%); one patient was lost to follow-up in the early and one in the late cIVAD cohort. The median follow-up time was 11 weeks (interquartile range [IQR] 6–25).

Age, SE severity, and SE etiology did not differ significantly between episodes treated with and without early cIVADs (Table 1). In the early cIVAD group, impaired consciousness on admission and GCSE were more frequent while CPSE was rarer. A comparable number of patients had NCSE in coma. No patient had SPSE.

The results of the analysis are summarized in Table 2. After adjustment for confounders, early use of cIVADs was an independent predictor of good outcome at last follow-up. The follow-up time did not differ significantly between the early and late cIVAD cohort (12 [IQR 6–26] versus 9 [IQR 5–18] weeks; P = 0.483).

In regard to secondary outcomes, early use of cIVADs was significantly associated with lower chance of both use of thiopental and induction of burst suppression but did not independently predict functional outcome at discharge, chance of RSE termination, risk of SRSE development, and in-hospital mortality. RSE duration after initiation of therapy was shorter in early cIVAD patients, whereas duration of mechanical ventilation did not differ significantly between cohorts (Fig. 2).

The results of the subgroup analysis for the primary outcome are depicted in Fig. 3. Early use of cIVADs was independently associated with good outcome at last follow-up in patients with STESS less than 3, age less than 65 years, and a history of seizures but not in patients with STESS of at least 3, age of at least 65 years, and no previous seizures. Level of consciousness and worst seizure type did not influence impact of timing of cIVAD therapy on functional outcome. Because among patients with NCSE in coma one outcome had zero observations, calculation of adjusted risk ratios was not possible for this subgroup. The follow-up times did not differ significantly between patients who received cIVADs early compared with those who received treatment late in any of the subgroups (Additional file 2: Table S2).

Our study has several clear limitations that need to be considered. The sample size is small, which is particularly problematic in a disorder as heterogenic as RSE. Data were collected retrospectively, and the medical records did not contain detailed enough information to determine the exact time of RSE cessation after initiation of cIVADs in all cases; therefore, we had to estimate RSE duration in days and not in hours. Furthermore, we could (with certainty) only tell that but not clearly assess the reasons why cIVADs were applied early or late. Some patients may have received cIVADs because they required intubation for airway protection or mechanical ventilation and not primarily for seizure control. Therefore, confounders not considered in our study may have substantially influenced therapeutic decisions and outcome, introducing the risk of bias into our results. Furthermore, we did not assess cIVAD dosing, and treatment of RSE with cIVADs did not follow a specific protocol in our institution during the early years of the study period. Setting the cutoff between early and late cIVAD treatment at 48 h after RSE onset represents an arbitrary definition which may appear inappropriate in light of guidelines advocating rapid treatment escalation. However, in our experience, a conservative approach in the first 48 h may represent a viable therapeutic option in non-convulsive RSE episodes, including those arising from GCSE either spontaneously or after initial treatment. The primary outcome measure of this study was not based on data collected at defined time points but relied on last available follow-up findings. Although follow-up durations were variable, they did not differ significantly between any of the cohorts compared. Furthermore, the validity of the primary outcome measure is supported by a trend toward higher chance of better outcome in the early cIVAD cohort already at time of discharge.