Abstracts

Rationale: Status epilepticus (SE) treatment may be administered with different levels of intensity, from small benzodiazepine doses to coma induction (CI). For different SE subgroups, it is still unclear how the risk of an aggressive therapeutic approach balances with outcome improvement. We recently developed a clinical prognostic score (Neurology 2006;66:1736-8) to provide clinicians with a practical tool to orient therapeutic management. It relies on the assessment of age (0 or 2 points, cutoff at 65), previous history of seizures (1 point if negative), seizure type (0-2 points) and extent of consciousness impairment (1 point if stuporous or comatose), assessed before treatment institution; a score of 3-6 is defined as unfavorable (see Table 1). The aim of this observational study was to validate the score in a heterogeneous cohort, and to analyze its potential impact on the choice of the SE treatment strategy, particularly CI. Methods: Demographical, historical and clinical variables of consecutive adult patients with SE were prospectively collected in 3 tertiary referral hospitals in the USA (BWH and MGH) and Europe (CHUV) over different time spans. The score was calculated after treatment institution but before outcome assessment (at discharge). Chi square, Fisher’s exact and ANOVA tests were used to compare the variables of interest among the centers. The score’s prediction for survival was assessed with the negative predictive value (NPV). A propensity score for CI was calculated using all potential variables, and effect of coma induction on survival estimated using propensity-based matching.Results: The cohort consisted of 135 patients (BWH 61, CHUV 60, MGH 14). As compared to mortality, which occurred in 23% (see Table 2), the score had a sensitivity of 94% (95% CI: 79-98%), specificity of 58% (95% CI: 48-67%), NPV of 97% (95% CI: 89-99%), and an accuracy of 77%. Among surviving patients, likelihood of return to baseline clinical conditions was considerably higher for subjects with a favorable score (P=0.001). CI was more frequent at BWH (P=0.003), but age, previous history of seizures, SE etiology, SE seizure type, extent of consciousness impairment before treatment, time to treatment, score severity, and mortality were statistically similar among the centers. In patients with an unfavorable score, CI was related to a somewhat lower survival as compared to treatment without coma (48% vs. 66%), whereas in subjects with a favorable score, both strategies were associated with high survival (100% vs. 96%). Propensity-based estimation of CI effect on survival was non-significant (95%CI: –0.335 to 0.111).Conclusions: The SE clinical prognostic score reliably identifies patients who will survive, and is a simple tool to use in practice. It also discriminates subjects with a good functional outcome. Our observations also suggest that aggressive treatment (i.e., coma induction) is not routinely warranted in patients with a favorable SE score, who will almost certainly survive their SE episode. A randomized trial using this score is nevertheless needed to confirm this hypothesis.