The answer from this study, a retrospective review of a decade of flumazenil use in California, is clearly yes. However, the "yes" is only 13 seizures out of 904 reviewed cases, most of whom had some sort of co-ingestant that contributed to the pro-convulsant state. The authors also note, for the cases in which data was available, flumazenil was therapeutic (and potentially diagnostic as well) in 53% of administrations, with return of alertness from unresponsiveness or drowsiness.
So, the answer to the clinical question - whether flumazenil use should be as taboo as current dogma - is more complex, and, unfortunately, descends into that dark area where risks must be weighed against benefits. Is the risk of poor clinical outcome secondary to resuscitative efforts in the field, delayed/missed intubations, etc. greater than the 1-2% risk of seizures? Or can the patient be safely observed with minimal intervention in a monitored setting? Or, if flumazenil is effective, how much money was saved by reducing the need for the expansive medical testing performed on unresponsive individuals? I don't believe a single blanket answer suffices to cover each individual clinical situation.
"A poison center's ten-year experience with flumazenil administration to acutely poisoned adults."