Most antibiotics are CNS-safe, and neurotoxicity associated with their administration is infrequent.31 There have been anecdotal reports of seizures complicating antibiotic use, however, and such associations are reported more often for certain drugs. Beta-lactams are the antibiotic class most commonly associated with adverse CNS events.

The exact incidence of seizures complicating antibiotic use is not known. Interpretation of empiric studies and meta-analyses of seizure complications of antibiotic use is complicated by the fact that patients given antibiotics often have other seizure risk factors, making it difficult to attribute the cause of seizures to the antibiotic. For example, the background seizure rate in seriously ill patients likely to receive intravenous antibiotics could be as high as 4%.32

General clinical considerations

Although seizures are a relatively unusual complication of antibiotic use, clinical and experimental studies do suggest certain risk. The clinician can minimize this risk by keeping in mind well-defined predisposing conditions:68,69

  • Renal insufficiency: excess dosage relative to renal function and body weight (perhaps the most common explanation for antibiotic-related convulsions)
  • Age: age-related changes in drug metabolism, renal clearance, blood-brain barrier permeability, or a combination of these
  • Pre-existing CNS disease: including epilepsy, intracranial mass, and meningitis
  • Concomitant use of proconvulsant drugs: drug interactions altering pharmacokinetics, pharmacodynamics, or both

To manage antibiotic-related seizure complications, benzodiazepines and barbiturates should be considered for first-line therapy.31


The neurotoxicity of penicillin was first reported in 1945 by Johnson and Walker, who observed myoclonic twitching after intravenous administration. Further reports appeared in the 1960s of a penicillin-related syndrome involving serially appearing hyperreflexia, myoclonic jerks, and sometimes seizures. Semisynthetic penicillins have also been implicated33; for example, oxacillin and dicloxacillin induce epileptogenic changes in rabbits.34

Epileptogenic characteristics of beta-lactams, like their antimicrobial and antigenic properties, are structure-dependent. The beta-lactam ring is a key feature of penicillin epileptogenesis, and convulsant properties are abolished after incubation with penicillinase.35,36 Ring substitution can influence convulsant activity. For example, substitution of a benzylic hydrogen with an amino group (as in ampicillin and amoxicillin, for example) reduces epileptogenic potential. Drugs characterized by the presence of a ureido group (e.g., piperacillin) appear less epileptogenic than penicillin.


Cephalosporins differ from penicillins in having a six-membered thiazolide moiety fused to the beta-lactam ring. Different substitutions at different sites can affect epileptogenesis.37 Accordingly, compounds with similar substituent profiles—ceftriaxone, cefotetan, ceftazidime, and cefepime (heterocyclic ring at position 3 and a heteroaromatic nucleus at position 7 of 7-aminocephalosporanic acid)—show similar evidence of epileptogenesis but less than that of penicillin. Compounds having one heterocyclic ring at position 7, such as cefuroxime and cefixime, show no epileptic activity.

In contrast, two other similarly substituted cephalosporins, cefazolin and cefmetazole, have greater epileptogenic activity (in fact, marked similarity to the proconvulsant drug PTZ).31 Epileptiform electrographic changes similar to those induced by penicillin have been observed after infusion of high doses of cefazolin.38,39 Actual reports of cefazolin-related seizures are unusual, however.40

The incidence of CNS adverse events observed during third-generation cephalosporin therapy remains below 1%.41–44Clinical experience with fourth-generation cephalosporins (e.g., cefepime) is still limited, but no significant seizure associations have been reported. A large survey of more than 2,000 patients treated with cefepime reported only three cases of seizures, and these were found to be unrelated to antibiotic administration.45


Aztreonam possesses greater convulsant properties than some cephalosporins, but much less than penicillin and cefazolin.46,47 In a series of 22 bacterial meningitis patients treated with aztreonam, seizures occurred in the acute phase in 6 cases but were considered more likely related to underlying disease than to aztreonam.48


Seizure incidence with use of imipenem or cilastatin was first reported by the manufacturer to be 0.4%. Although phase I and II trials did not reveal significant seizure potential, phase III clinical trials and subsequent postmarketing surveys demonstrated that imipenem or cilastatin can be associated with seizures.49–52 A maximal incidence of 33% was found in a small series of 21 children with bacterial meningitis.31 Calandra et al. reviewed more than 1,700 patients treated with imipenem or cilastatin to determine carbapenem seizure risk.32 Patients were moderately to severely ill, with multiple organ system dysfunction, and many had independent seizure risks. Seizures occurred in 3% of patients, but in only 0.9% were the seizures judged to be possibly, probably, or definitely related to imipenem or cilastatin. A seizure incidence of 2–3% was noted among patients treated with other antibiotics. Average time of seizure onset was 7 days after imipenem or cilastatin initiation. As with other beta-lactam antibiotics, CNS disorders (including premorbid epilepsy) and renal insufficiency were independent risk factors for seizure occurrence. Imipenem dosage in excess of recommended guidelines, especially in patients with renal insufficiency, was also a seizure risk factor. As is usually the case with such retrospective analyses, high background seizure incidence in the severely ill patient population in which these antibiotics are used makes precise attribution of seizure etiology difficult.32

The proconvulsive activity of imipenem appears to increase when it is used in combination with cilastatin. This could be due merely to elevated imipenem levels in the presence of cilastatin. (Cilastatin inhibits renal tubular secretion of imipenem.) Alternatively, cilastatin might increase CSF levels of imipenem by inhibiting CSF elimination. Cilastatin alone can induce seizures, but at doses much higher than those used clinically.

De Sarro et al. were able to demonstrate that excitatory amino acid antagonists could counteract seizures induced by imipenem, suggesting an involvement of excitatory amino acids in the genesis of seizures.53

In a study of 5,000 patients treated with meropenem, seizure incidence was only 0.08%.54 Comparison data indicated that seizures occurred no more frequently during treatment with meropenem than during treatment with other beta-lactam antibiotics.55 Using the mouse PTZ model, Day et al. found that pretreatment with meropenem did not enhance PTZ-induced seizures, whereas pretreatment with imipenem or cilastatin increased the convulsive activity of PTZ.56


Overall incidence of CNS adverse reactions, including seizure, is less than 1%.57,58 Isolated case reports note the occurrence of seizures with norfloxacin and ciprofloxacin,59–61 but underlying neurologic diseases, renal insufficiency, concomitant drug use, or a combination of these factors known to lower the epileptic threshold is almost always present.

Proconvulsant effects of quinolones have been attributed to direct pharmacodynamic effects and to pharmacokinetic and dynamic interactions with co-administered drugs. Direct pharmacodynamic proconvulsant mechanisms of quinolones may relate to gamma-aminobutyric acid (GABA)–like substituents, which act as GABA-receptor antagonists.62

Pharmacokinetic and pharmacodynamic interactions between quinolones and other drugs are best described for concomitant use of theophylline and nonsteroidal anti-inflammatory drugs (NSAIDs). Several fluoroquinolones potentially interact with theophylline. Vancutsem and colleagues investigated the influence of fluoroquinolones (ciprofloxacin, norfloxacin) on seizure parameters in amygdaloid- kindled rats and found no change in seizure activity when these antimicrobials were administered alone.63 Seizure activity increased, however, when their administration was concurrent with a non–seizure-modulating dose of theophylline. Epileptogenic-pharmacodynamic interactions may include centrally mediated mechanisms.63

Theophylline, a methylxanthine, is metabolized by several cytochrome P-450 pathway isozymes. Some quinolones exert an inhibitory effect on the cytochrome P-450 system (See Table: Anticonvulsant and antimicrobial cytochrome P-450 effects), thereby reducing theophylline clearance and increasing theophylline blood levels.64 Meta-analysis of quinolone-theophylline interactions revealed that enoxacin may be the strongest inhibitor of theophylline metabolism.31

NSAIDs can also enhance fluoroquinolone toxicity. Certain NSAID metabolites (e.g., biphenylacetic acid) can increase the antagonistic potency of certain quinolones on GABAA receptors.65–67

Adapted from: Goldstein MA and Harden CL. Infectious states. In: Ettinger AB and Devinsky O, eds. Managing epilepsy and co-existing disorders. Boston: Butterworth-Heinemann; 2002;83-133.
With permission from Elsevier (

Reviewed By: 
Steven C. Schachter, MD
Monday, March 1, 2004