K. Feil, S. Kindzierski, C. Single, L. Geiger-Primo, D. Schweikert, M. Adolph, J. Kegele, H. Lerche, U. Ziemann, L. Möller, A. Mengel, et al. (2026). Ketogenic Diet in Super-Refractory Status Epilepticus: A Retrospective Cohort Study with Severity-Matched Controls in Critically Ill Adults. Neurocritical care, 44(3), 875-887. https://doi.org/10.1007/s12028-025-02431-w

PubMed 41507676

[Abstract]
BACKGROUND: Super-refractory status epilepticus (SRSE) is a life-threatening neurological emergency with limited treatment options. A ketogenic diet (KD) is increasingly considered as a rescue therapy, but controlled data in critically ill adults remain scarce. This study aimed to evaluate the feasibility, safety, and clinical effects of KD in adult SRSE using a severity-matched control group.

METHODS: A retrospective, severity-matched cohort study compared adult patients with SRSE treated with KD to matched controls. The primary outcome was SRSE resolution. Secondary outcomes included the modified Rankin Scale (mRS) and mortality at 3 and 6 months. Time-dependent and multivariate Cox regression models adjusted for illness severity (including age and Status Epilepticus Severity Score [STESS]) and delayed KD initiation. Despite pragmatic matching, baseline differences in age, STESS, and seizure type were addressed through multivariate adjustment.

RESULTS: A total of 34 adult patients with SRSE were analyzed (18 KD, 16 control). KD was initiated after a mean of 16.6 ± 9.4 days and maintained for 12.9 ± 7.7 days. Ketosis was achieved in 33%, with mild, manageable complications in 28%. SRSE resolution occurred in 61.1% of KD patients vs. 87.5% of controls (p = 0.125), although KD patients had significantly longer status epilepticus duration and higher medication burden. Time-dependent Cox regression showed an association with faster SRSE resolution after KD initiation (hazard ratio [HR] 0.26, 95% confidence interval [CI] 0.07-0.97; p = 0.045). In the multivariate Cox model, KD remained independently associated with SRSE resolution (HR 0.368, 95% CI 0.176-0.756; p = 0.006). Earlier KD initiation was independently associated with improved seizure control (p = 0.015). At 3 and 6 months, KD patients showed significantly better functional outcomes (p = 0.023 and p = 0.021, respectively). Ketosis or ketone levels were not associated with outcome, suggesting that therapeutic effects may be independent of measurable ketosis.

CONCLUSIONS: KD is feasible and safe in adult patients with SRSE. Time-dependent models showed a significant therapeutic association, particularly with earlier initiation. These findings support prospective evaluation of KD as a nonpharmacologic therapy in neurocritical care.