Carbamazepine

What it is

Carbamazepine is a prescription anticonvulsant taken by mouth, supplied as 200 mg tablets, 100 mg chewable tablets, a suspension and extended-release forms. Chemically it is 5H-dibenz[b,f]azepine-5-carboxamide, a dibenzazepine (iminostilbene) related to the tricyclic compounds, a white to off-white powder practically insoluble in water. It is handled by the liver, mainly by CYP3A4, to an active metabolite, carbamazepine-10,11-epoxide. It induces its own breakdown, so its half-life falls from 25 to 65 hours on the first dose to 12 to 17 hours after repeated dosing, a process complete after 3 to 5 weeks.

What the research says

For focal (partial) seizures carbamazepine is one of the two drugs that an individual-participant network meta-analysis of 39 trials and 14,789 people places in the best group on treatment failure, alongside lamotrigine and levetiracetam; the same analysis found lamotrigine better than carbamazepine on treatment failure for any reason (HR 1.26, 95% CI 1.10 to 1.44). For generalised tonic-clonic seizures it was worse than sodium valproate (HR 1.52, 95% CI 1.18 to 1.96) and it does not control absence seizures at all. For trigeminal neuralgia 98% of 100 consecutive patients responded initially at a median 600 mg, but 27% had to stop or cut the dose because of side effects within a mean 8.6 months. For other neuropathic pain, Cochrane found no first-tier or second-tier evidence of efficacy at all. For schizophrenia the randomised evidence does not support it. It carries a two-part boxed warning for fatal skin reactions tied to the HLA-B*1502 allele and for aplastic anaemia and agranulocytosis, and it wrecks the blood levels of a very long list of other medicines, hormonal contraception among them.

Evidence grade: Well established.

How it works

Drug class: Dibenzazepine (iminostilbene) anticonvulsant and specific analgesic for trigeminal neuralgia; a potent inducer of hepatic CYP3A4 and of CYP1A2, 2B6, 2C8/9/19

Carbamazepine damps down repetitive firing in nerve cells. In animal models it reduces polysynaptic responses and blocks post-tetanic potentiation, and it abolishes pain triggered by stimulating the infraorbital nerve, which is the facial nerve involved in trigeminal neuralgia. Its label is blunt that the mechanism in people remains unknown. Separately, it switches on liver enzymes, including CYP3A4, which is why it lowers the blood levels of so many other drugs and eventually of itself. (Source 1)

Boxed warning

SERIOUS DERMATOLOGIC REACTIONS AND HLA-B*1502 ALLELE

(Source 2)

What it is used for

• High-certainty network meta-analysis of individual participant data places carbamazepine with lamotrigine and levetiracetam in the best group for focal onset seizures, though lamotrigine beat it on treatment failure (HR 1.26, 95% CI 1.10 to 1.44). For generalised tonic-clonic seizures sodium valproate did better (HR 1.52, 95% CI 1.18 to 1.96). Evidence: established. (Source 3)
• The label states directly that absence seizures do not appear to be controlled by carbamazepine, and warns that in a mixed seizure disorder including atypical absence seizures carbamazepine has been associated with an increased frequency of generalised convulsions. Evidence: not-supported. (Source 3)
• In a 200-patient tertiary-centre cohort, 98% of those started on carbamazepine responded initially at a median dose of 600 mg a day, and late loss of response was rare (3 of the responders). The cost is tolerability: 27% stopped or reduced the dose to an unsatisfactory level because of side effects in a mean 8.6 months. Evidence: established. (Source 3)
• Cochrane found ten small, mostly cross-over studies in 480 people, most lasting four weeks or less, and judged that no study provided first-tier or second-tier evidence for any efficacy outcome. Carbamazepine was generally better than placebo only on third-tier, poorly defined outcomes. Evidence: limited. (Source 4)
• Off-label for the tablet, chewable tablet and extended-release products whose labels we parsed, all of which list only epilepsy and trigeminal neuralgia. A network meta-analysis of 72 double-blind trials in 16,442 adults found carbamazepine beat placebo on response and on improvement in mania symptoms, but it was not among the four drugs with better acceptability than placebo. Those results apply only to randomised trials of a single oral medicine taken on its own for at least 10 days, with trials that allowed a rescue antipsychotic excluded, and the trials averaged under four weeks. Evidence: limited. (Source 5)
• Ten randomised trials in 283 people found no effect on relapse as sole treatment and no difference on a 50% reduction in Brief Psychiatric Rating Scale score as augmentation (6 RCTs, n = 147, RR 0.86, 95% CI 0.67 to 1.12, low quality evidence). Cochrane concludes carbamazepine cannot be recommended for routine use in schizophrenia. Evidence: not-supported. (Source 6)

Interactions

• Grapefruit juice (label): Grapefruit juice blocks CYP3A4, the enzyme that breaks carbamazepine down, so it pushes carbamazepine blood levels up. The label groups it with the drugs that are shown or expected to raise carbamazepine levels, which is the point at which the drug's dizziness, unsteadiness and double vision appear. (Source 7)
• Niacinamide (nicotinamide), a form of vitamin B3 sold as a supplement (label): Niacinamide appears on the label's list of agents expected to raise carbamazepine blood levels, through the same CYP3A4 route as grapefruit juice. It is the one dietary supplement named anywhere in the interaction section. (Source 7)
• Alcohol (label): The sedative effects add up. Carbamazepine itself causes drowsiness and unsteadiness, especially early on, and the label tells patients to be careful with alcohol because of a possible additive sedative effect. (Source 8)
• Hormonal contraceptives, including the pill and the levonorgestrel implant (case reports): Carbamazepine switches on the liver enzymes that destroy the contraceptive hormones, so blood levels of those hormones fall and the contraceptive can stop working. Breakthrough bleeding and unintended pregnancies have been reported. (Source 9)
• Warfarin and the direct oral anticoagulants rivaroxaban, apixaban, dabigatran and edoxaban (label): Carbamazepine lowers the blood levels of these blood thinners, in the case of the direct oral anticoagulants to a degree the label expects to be insufficient to achieve the intended effect, so the combination is generally to be avoided. (Source 9)
• Valproic acid, quetiapine, loxapine and brivaracetam (label): These block the enzyme that disposes of carbamazepine's active epoxide metabolite, so the metabolite can build up even when the carbamazepine level itself looks acceptable; the label says to adjust the dose or monitor levels. (Source 7)
• Isoniazid (case reports): Two things happen at once: isoniazid raises carbamazepine levels, and the combination has been reported to make isoniazid's liver toxicity worse. (Source 9)
• Lithium (label): Taken together, the combination may raise the risk of neurotoxic side effects. (Source 9)
• Other medicines generally, as the sentence that closes the printed boxed warning (label): The printed box ends with an instruction that is easy to lose: before prescribing carbamazepine the prescriber is told to be thoroughly familiar with the whole prescribing information, and the label singles out use with other drugs that increase the potential for toxicity. (Source 10)

Stopping it

• Carbamazepine must not be stopped abruptly by someone whose seizures it controls. The label warns that abrupt discontinuation of any anticonvulsant in a responsive epileptic patient may lead to seizures or even status epilepticus. (Source 11)
• The label's standing instruction is gradual withdrawal rather than a stop. (Source 12)
• The patient-facing Medication Guide puts the same point first, before any side effect. (Source 13)
• In trigeminal neuralgia the label builds deprescribing into the treatment plan, with a review at least every three months to cut the dose or stop the drug. (Source 14)
• Stopping carbamazepine also reverses its enzyme induction, so the blood levels of other medicines rise again and their doses may need cutting. The label spells this out for aripiprazole and lapatinib. (Source 9)
• Stopping is not complicated by dependence or craving: the label records no abuse potential and no psychological or physical dependence in humans. The reason for tapering is the seizure risk, not withdrawal. (Source 15)
• Some harms are themselves reasons to stop: the label says to discontinue at the first sign of a rash unless it is clearly not drug related, to consider stopping if significant bone marrow depression develops, and to consider stopping in symptomatic hyponatraemia. (Source 2)
• Angioedema is the exception to the never-stop-abruptly rule. Where the face, eyes, lips or tongue swell or swallowing or breathing becomes difficult, the label tells the patient to stop the drug and then contact their prescriber, rather than to continue and wait. (Source 8)

What goes wrong

Fatal skin reactions occur in roughly 1 to 6 per 10,000 new users in mainly Caucasian populations and about ten times that in some Asian countries, and are strongly tied to the HLA-B*1502 allele. (Source 2)

• Official position, Moderate certainty.
• Size: Population-level estimates quoted in the boxed warning.
• Who: New users of carbamazepine.
• How long: Over 90% of cases occur within the first few months of treatment.
• Result: 1 to 6 per 10,000 new users in countries with mainly Caucasian populations; about 10 times higher in some Asian countries.
• Funding: Not applicable.

THESE REACTIONS ARE ESTIMATED TO OCCUR IN 1 TO 6 PER 10,000 NEW USERS IN COUNTRIES WITH MAINLY CAUCASIAN POPULATIONS, BUT THE RISK IN SOME ASIAN COUNTRIES IS ESTIMATED TO BE ABOUT 10 TIMES HIGHER.

Aplastic anaemia and agranulocytosis are 5 to 8 times commoner on carbamazepine than in the general population, but the background rate is very low, so the absolute risk stays small. (Source 2)

• Case-control study, Moderate certainty.
• Size: A population-based case-control study cited in the boxed warning.
• Who: Carbamazepine users compared with the general population.
• How long: Not stated in the warning.
• Result: Relative risk 5 to 8 times greater; untreated general-population background approximately six patients per million per year for agranulocytosis and two per million per year for aplastic anaemia, so the absolute excess remains in the order of tens of cases per million treated per year.
• Funding: Not applicable.

DATA FROM A POPULATION-BASED CASE CONTROL STUDY DEMONSTRATE THAT THE RISK OF DEVELOPING THESE REACTIONS IS 5 TO 8 TIMES GREATER THAN IN THE GENERAL POPULATION. HOWEVER, THE OVERALL RISK OF THESE REACTIONS IN THE UNTREATED GENERAL POPULATION IS LOW, APPROXIMATELY SIX PATIENTS PER ONE MILLION POPULATION PER YEAR FOR AGRANULOCYTOSIS AND TWO PATIENTS PER ONE MILLION POPULATION PER YEAR FOR APLASTIC ANEMIA.

In a Taiwanese screening study no severe skin reaction occurred among carriers, because everyone who tested positive for HLA-B*1502 was given a different drug; the expected cases, judged against the historical rate, did not appear. (Source 16)

• Cohort study, Moderate certainty.
• Size: 4,877 carbamazepine-naive candidates from 23 hospitals in Taiwan.
• Who: People in Taiwan who had not taken carbamazepine.
• How long: Weekly telephone follow-up for 2 months.
• Result: 7.7% tested positive and were steered away from carbamazepine; mild transient rash in 4.3% and more widespread rash in 0.1%; no SJS-TEN in any HLA-B*1502-negative subject taking carbamazepine, against an estimated historical incidence of 0.23% that would have predicted about 10 cases (P<0.001)
• Funding: Funded by the National Science Council of Taiwan and the Taiwan Drug Relief Foundation.

Limit of this finding: This was not a randomised trial. Everyone who tested positive for HLA-B*1502 was steered away from carbamazepine, so there was no group of carriers who took it. The comparison is against the rate of these reactions recorded in the past, not against a concurrent control group, which means the study shows that the expected cases did not appear rather than proving that screening is what stopped them.

SJS-TEN did not develop in any of the HLA-B*1502-negative subjects receiving carbamazepine. In contrast, the estimated historical incidence of carbamazepine-induced SJS-TEN (0.23%) would translate into approximately 10 cases among study subjects (P<0.001).

In people of Northern European ancestry the relevant allele is HLA-A*3101, which lifts the risk of a hypersensitivity reaction from 5.0% to 26.0%. (Source 17)

• Case-control study, Moderate certainty.
• Size: Genome-wide association study of 22 hypersensitivity-syndrome cases, 43 maculopapular-exanthema cases and 3,987 controls, replicated in 145 further cases.
• Who: People of European descent taking carbamazepine.
• How long: Not applicable (genetic association)
• Result: HLA-A*3101 prevalence 2 to 5% in Northern European populations; hypersensitivity syndrome P=3.5x10(-8), odds ratio 12.41 (95% CI 1.27 to 121.03); maculopapular exanthema OR 8.33 (3.59 to 19.36); SJS-TEN OR 25.93 (4.93 to 116.18); absolute risk 5.0% to 26.0% with the allele and 5.0% to 3.8% without it.
• Funding: Funded by the UK Department of Health and others.

Limit of this finding: The odds ratio of 12.41 for the hypersensitivity syndrome has a 95% confidence interval of 1.27 to 121.03 and rests on 22 cases, so the true size of the effect is very uncertain and the point estimate should never be quoted on its own. The 5.0% to 26.0% figures are the authors' calculated risk estimates, not counts of patients. PubMed renders the p-values as "P=3.5×10(-8)" and "P=1.1×10(-6)"; those parenthesised digits are negative exponents, so the values are about 0.000000035 and 0.0000011.

The presence of the allele increased the risk from 5.0% to 26.0%, whereas its absence reduced the risk from 5.0% to 3.8%.

Allele frequency and so screening need vary enormously by ancestry, and a negative genotype does not remove the risk. (Source 18)

• Official position, Moderate certainty.
• Size: Population prevalence figures quoted in the label.
• Who: Patients of differing ancestries.
• How long: Not applicable.
• Result: HLA-A*3101 expected in more than 15% of patients of Japanese, Native American, Southern Indian and some Arabic ancestry, up to about 10% of Han Chinese, Korean, European, Latin American and other Indian ancestry, and up to about 5% of African-Americans and patients of Thai, Taiwanese and Hong Kong Chinese ancestry.
• Funding: Not applicable.

Many HLA-B1502-positive and HLA-A3101-positive patients treated with carbamazepine will not develop SJS/TEN or other hypersensitivity reactions

In 26,179 people with epilepsy, carbamazepine was associated with moderate to severe hyponatraemia in a dose-related pattern, and the rate rose steeply with age. (Source 19)

• Cohort study, Moderate certainty.
• Size: 26,179 patients: 8,598 aged 0-15, 16,476 aged 16-64 and 1,105 aged 65 and over, with sodium measured between 2006 and 2020.
• Who: People with epilepsy in a large Japanese real-world dataset.
• How long: 15 years of records.
• Result: 677 patients (2.6%) developed moderate-severe hyponatraemia, defined as serum sodium below 130 mEq/L; incidence per 1,000 person-years 3.1 in children, 19.8 in adults and 50.4 in older adults; carbamazepine, valproate, phenytoin, phenobarbital, benzodiazepines and antipsychotics were significant risk factors, and in adults carbamazepine acted in a dose-dependent manner.
• Funding: Independent; authors declare no competing interests.

Limit of this finding: This is a database study with no untreated comparison group and no randomisation, so it shows which drugs went with low sodium rather than how much of it each drug caused. People on first-generation antiseizure drugs differ from people on newer ones in ways the analysis cannot fully adjust for. The dose-related pattern strengthens the case but does not establish it.

In adult patients, carbamazepine, benzodiazepine, and antipsychotics induced hyponatremia in a dose-dependent manner.

The label attributes much of the hyponatraemia to inappropriate ADH secretion, says the risk looks dose-related, and names older people and those on diuretics as higher risk. (Source 12)

• Official position, Low certainty.
• Size: Not quantified in this section.
• Who: Patients treated with carbamazepine, especially elderly patients and those on diuretics.
• How long: During treatment.
• Result: No rate given here; symptoms listed are headache, new or increased seizure frequency, difficulty concentrating, memory impairment, confusion, weakness and unsteadiness, which can lead to falls.
• Funding: Not applicable.

Hyponatremia can occur as a result of treatment with carbamazepine. In many cases, the hyponatremia appears to be caused by the syndrome of inappropriate antidiuretic hormone secretion (SIADH). The risk of developing SIADH with carbamazepine treatment appears to be dose-related. Elderly patients and patients treated with diuretics are at greater risk of developing hyponatremia.

Antiseizure drugs as a class roughly double the risk of suicidal thinking or behaviour, which in absolute terms is about one extra case per 530 people treated. (Source 20)

• Meta-analysis, Moderate certainty.
• Size: Pooled analysis of 199 placebo-controlled trials of 11 antiepileptic drugs; 27,863 drug-treated and 16,029 placebo-treated patients.
• Who: Patients taking an antiepileptic drug for any indication.
• How long: Median treatment duration 12 weeks; most trials did not run beyond 24 weeks.
• Result: Adjusted relative risk 1.8 (95% CI 1.2, 2.7); incidence 0.43% on drug versus 0.24% on placebo, an increase of approximately one case of suicidal thinking or behaviour for every 530 patients treated; risk appeared as early as one week; by indication, epilepsy 1.0 versus 3.4 events per 1,000 (risk difference 2.4) and psychiatric 5.7 versus 8.5 per 1,000 (risk difference 2.9)
• Funding: Not applicable (regulatory pooled analysis)

Limit of this finding: The label prints this confidence interval as "95% CI:1.2, 2.7" with no space after the colon; that is the label's own typesetting and has been kept. These are pooled trial data across eleven different antiseizure drugs and all their indications, not a measurement of carbamazepine on its own. The absolute numbers are small: 0.43% against 0.24%, about one extra case for every 530 people treated.

the estimated incidence rate of suicidal behavior or ideation among 27,863 AED-treated patients was 0.43%, compared to 0.24% among 16,029 placebo-treated patients, representing an increase of approximately one case of suicidal thinking or behavior for every 530 patients treated

Carbamazepine induces its own metabolism, so blood levels fall over the first weeks even on an unchanged dose. (Source 21)

• Blood level study, Moderate certainty.
• Size: Not stated in this section.
• Who: People taking carbamazepine repeatedly.
• How long: Autoinduction complete after 3 to 5 weeks of a fixed dosing regimen.
• Result: Initial half-life 25 to 65 hours falling to 12 to 17 hours on repeated doses; plasma levels range 0.5 to 25 mcg/mL with no apparent relationship to daily intake; usual adult therapeutic levels 4 to 12 mcg/mL.
• Funding: Not applicable.

Because carbamazepine induces its own metabolism, the half-life is also variable. Autoinduction is completed after 3 to 5 weeks of a fixed dosing regimen. Initial half-life values range from 25 to 65 hours, decreasing to 12 to 17 hours on repeated doses.

The serious harms reach beyond skin and marrow to the heart, the liver and the pancreas, and some cardiovascular complications have been fatal. (Source 22)

• Case series, Very low certainty.
• Size: Not quantified; a label listing of reported reactions.
• Who: Patients taking carbamazepine.
• How long: Not stated.
• Result: No rates attached. Reported reactions include aplastic anaemia, agranulocytosis, pancytopenia, thrombocytopenia, acute intermittent porphyria; acute generalized exanthematous pustulosis and generalized bullous fixed drug eruption; congestive heart failure, arrhythmias and AV block; cholestatic and hepatocellular jaundice, hepatitis and very rare hepatic failure; pancreatitis.
• Funding: Not applicable.

Limit of this finding: These are reports collected after marketing with no denominator, so no rate can be worked out from them and a listed reaction is not proof the drug caused it. The same label section also mixes in animal findings — testicular atrophy in rats and bladder discoloration in dogs — and says plainly that their relevance to humans is unknown.

Some of these cardiovascular complications have resulted in fatalities.

The commonest reactions, dizziness, drowsiness, unsteadiness, nausea and vomiting, cluster at the start of treatment. (Source 11)

• Official position, Low certainty.
• Size: Not quantified in this section.
• Who: Patients starting carbamazepine.
• How long: Particularly the initial phases of therapy.
• Result: No rates given in this section.
• Funding: Not applicable.

The most frequently observed adverse reactions, particularly during the initial phases of therapy, are dizziness, drowsiness, unsteadiness, nausea, and vomiting.

Carbamazepine is a potent enzyme inducer that lowers the blood levels of a long list of other medicines, including hormonal contraception, warfarin and the direct oral anticoagulants. (Source 9)

• Official position, Moderate certainty.
• Size: Not quantified; a label listing.
• Who: Patients taking carbamazepine with other medicines.
• How long: For as long as the two are combined.
• Result: No numbers; the label names more than fifty affected drugs and states that breakthrough bleeding and unintended pregnancies have been reported with hormonal contraceptives.
• Funding: Not applicable.

Concomitant use of carbamazepine with hormonal contraceptive products (e.g., oral, and levonorgestrel subdermal implant contraceptives) may render the contraceptives less effective because the plasma concentrations of the hormones may be decreased. Breakthrough bleeding and unintended pregnancies have been reported.

The pooled antiseizure-drug analysis gives the absolute risk separately by indication: in the epilepsy trials the rate rose from 1.0 to 3.4 cases per 1,000 patients. (Source 20)

• Meta-analysis, Moderate certainty.
• Size: 199 placebo-controlled trials of 11 antiepileptic drugs; 27,863 drug-treated and 16,029 placebo-treated patients.
• Who: Patients taking an antiepileptic drug for epilepsy, for a psychiatric indication, or for another condition.
• How long: Median treatment duration 12 weeks; the risk was seen as early as one week and the data do not extend beyond 24 weeks.
• Result: Table 1, events per 1,000 patients, placebo versus drug, with relative risk and risk difference: epilepsy 1.0 versus 3.4, relative risk 3.5, risk difference 2.4; psychiatric 5.7 versus 8.5, relative risk 1.5, risk difference 2.9; other 1.0 versus 1.8, relative risk 1.9, risk difference 0.9; total 2.4 versus 4.3, relative risk 1.8, risk difference 1.9. The risk did not vary substantially by age from 5 to 100 years.
• Funding: Not applicable (regulatory pooled analysis)

Limit of this finding: The figures above come from the label's Table 1. This rendering lays a table out as one value per line rather than in columns, so in the quoted block the numbers appear beneath their headings rather than beside them; they are reproduced in reading order and none has been altered. Relative risk and absolute risk pull in different directions here: the multiple is largest in epilepsy, but the extra number of people affected is about the same in epilepsy and psychiatric use.

The relative risk for suicidal thoughts or behavior was higher in clinical trials for epilepsy than in clinical trials for psychiatric or other conditions, but the absolute risk differences were similar for the epilepsy and psychiatric indications.

The patient-facing Medication Guide names allergic and organ reactions that can occur with no rash at all, and lists the symptoms that should prompt an immediate call. (Source 13)

• Official position, Low certainty.
• Size: Not quantified; the patient-facing Medication Guide.
• Who: People taking carbamazepine tablets or chewable tablets.
• How long: Not stated; serious skin reactions are described as more likely in the first four months but able to occur later.
• Result: No rates are given. The blood-problem symptoms listed are fever, sore throat or other infections that come and go or do not go away, easy bruising, red or purple spots, bleeding gums or nose bleeds, and severe fatigue or weakness. The call-immediately list adds swelling of the face, eyes, lips or tongue, painful sores in the mouth or around the eyes, trouble swallowing or breathing, yellowing of the skin or eyes, and severe muscle pain.
• Funding: Not applicable.

Limit of this finding: The Medication Guide's own numbering jumps from item 3 to item 5 — there is no item 4 in the label — and that gap has been kept rather than renumbered. The Guide gives no frequencies, so nothing here says how often any of these happen.

3. Carbamazepine tablets or chewable tablets may cause allergic reactions or serious problems, which may affect organs and other parts of your body like the liver or blood cells. You may or may not have a rash with these types of reactions.

The Medication Guide also warns that carbamazepine can cause abnormal heart rhythms that may be life-threatening, with higher risk in people who already have heart problems. (Source 13)

• Official position, Low certainty.
• Size: Not quantified in the Medication Guide.
• Who: People taking carbamazepine tablets or chewable tablets, particularly those with underlying heart problems.
• How long: Not stated.
• Result: No rate given. The symptoms the Guide tells patients to report at once are a fast, slow or pounding heartbeat, fainting, dizziness, chest pain and shortness of breath. The same Guide puts the risk of suicidal thoughts or actions at about 1 in 500.
• Funding: Not applicable.

6. Carbamazepine tablets or chewable tablets can cause abnormal heart rhythms, which may be life-threatening. This risk may be increased in patients with underlying heart problems.

In children the relationship between dose and blood level is poor, and the active metabolite builds up relatively more in the youngest children. (Source 21)

• Blood level study, Low certainty.
• Size: Not stated in this section; the label cites one report for the metabolite ratio.
• Who: Children, compared with adults.
• How long: Not applicable.
• Result: The ratio of carbamazepine-10,11-epoxide to carbamazepine fell with age, in one report from 0.44 in children below 1 year to 0.18 in children aged 10 to 15 years. The epoxide is described as equipotent to carbamazepine as an anticonvulsant in animal screens.
• Funding: Not applicable.

Limit of this finding: The claim that the metabolite is as potent as the parent drug rests on animal screening tests, as the label says, not on human outcomes.

However, there is a poor correlation between plasma concentrations of carbamazepine and carbamazepine dose in children.

Anaphylactic reactions and angioedema can occur with carbamazepine, and this is one of the few situations in which the label tells the patient to stop the drug before speaking to anyone. (Source 8)

• Official position, Low certainty.
• Size: Not quantified in this section.
• Who: People taking carbamazepine.
• How long: Not stated.
• Result: No rate given. The signs named are swelling of the face, eyes, lips or tongue, or difficulty in swallowing or breathing.
• Funding: Not applicable.

Patients should be advised that anaphylactic reactions and angioedema may occur during treatment with carbamazepine (see WARNINGS). Advise patients to immediately report signs and symptoms suggesting angioedema (swelling of the face, eyes, lips, or tongue, or difficulty in swallowing or breathing) and to stop taking the drug until they have consulted with their healthcare provider.

The label's own list of reported reactions includes hyponatraemia, lowered blood calcium and osteoporosis, none of them with a frequency attached. (Source 22)

• Case series, Very low certainty.
• Size: Not quantified; a label listing of reported reactions with no denominator.
• Who: Patients taking carbamazepine.
• How long: Not stated.
• Result: No rates attached. Reported under Metabolism are fever and chills, hyponatraemia, decreased levels of plasma calcium and osteoporosis; a lupus erythematosus-like syndrome and raised cholesterol, HDL cholesterol and triglycerides are listed separately.
• Funding: Not applicable.

Limit of this finding: This is a list of reactions that have been reported, with no denominator and no comparison group, so it cannot tell you how often any of them happen or whether carbamazepine caused them. It is included because bone and mineral effects appear nowhere else in this entry.

Fever and chills. Hyponatremia (see WARNINGS, General). Decreased levels of plasma calcium have been reported. Osteoporosis has been reported.

The authors of the hyponatraemia study tell clinicians to monitor sodium carefully in people on first-generation antiseizure medicines, and expect the problem to grow as the epilepsy population ages. (Source 19)

• Cohort study, Low certainty.
• Size: 26,179 patients: 8,598 aged 0 to 15, 16,476 aged 16 to 64 and 1,105 aged 65 and over.
• Who: Patients with epilepsy whose serum sodium was measured, Japan, January 2006 to December 2020.
• How long: 15 years of records.
• Result: 677 patients (2.6%) developed moderate-severe hyponatraemia, defined as serum sodium below 130 mEq/L. Incidence per 1,000 person-years was 3.1 in children, 19.8 in adults and 50.4 in older adults, and in older adults it rose from 36.8 in 2007 to 58.5 in 2020.
• Funding: Independent; see the paper's own disclosure.

Limit of this finding: This is the authors' recommendation to prescribers, not a finding, and it is quoted as theirs. The study itself is a database analysis without a comparison group of untreated patients, so it shows which drugs went with hyponatraemia, not how much of it each drug caused.

Serum sodium levels should be monitored carefully when patients are receiving first-generation antiseizure medications or antipsychotics or combinations of these drugs.

What the evidence supports

On remission outcomes, carbamazepine did beat gabapentin and sodium valproate in people with focal seizures. (Source 23)

• Meta-analysis, High certainty.
• Size: Same individual participant dataset of 14,789 participants.
• Who: Adults and children with focal onset seizures.
• How long: Time to 12-month and 6-month remission.
• Result: Carbamazepine performed better than gabapentin for 12-month remission and better than sodium valproate for 6-month remission; no differences between lamotrigine and any drug for focal seizures.
• Funding: Independent review; industry disclosure as above.

for individuals with focal seizures, carbamazepine performed better than gabapentin (12-month remission) and sodium valproate (six-month remission)

What the evidence does not support

For generalised tonic-clonic seizures carbamazepine performed worse than sodium valproate on treatment failure. (Source 23)

• Meta-analysis, Moderate certainty.
• Size: Same individual participant dataset; evidence for generalised onset seizures was described as more limited.
• Who: Adults and children with generalised onset tonic-clonic seizures, with or without other generalised seizure types.
• How long: Time to treatment failure.
• Result: Treatment failure for any reason, sodium valproate versus carbamazepine HR 1.52 (95% CI 1.18 to 1.96); versus topiramate 1.37 (1.06 to 1.77); versus phenobarbitone 2.13 (1.20 to 3.79); no difference versus lamotrigine 1.06 (0.81 to 1.37) or levetiracetam 1.13 (0.89 to 1.42)
• Funding: Independent review; industry disclosure as above.

HRs (95% CIs) for treatment failure for any reason for sodium valproate versus: lamotrigine 1.06 (0.81 to 1.37), levetiracetam 1.13 (0.89 to 1.42), gabapentin 1.13 (0.61 to 2.11), phenytoin 1.17 (0.80 to 1.73), oxcarbazepine 1.24 (0.72 to 2.14), topiramate 1.37 (1.06 to 1.77), carbamazepine 1.52 (1.18 to 1.96)

Absence seizures do not respond to carbamazepine, and it can increase generalised convulsions in mixed seizure disorders that include atypical absences. (Source 3)

• Official position, Certainty not rated.
• Size: Not quantified in the label.
• Who: People with absence (petit mal) or atypical absence seizures.
• How long: Not applicable.
• Result: No effect estimate; the label states the seizure type is not controlled.
• Funding: Not applicable.

Absence seizures (petit mal) do not appear to be controlled by carbamazepine (see PRECAUTIONS, General).

For neuropathic pain other than trigeminal neuralgia there is no good-quality evidence of benefit, and two in five extra people get a side effect. (Source 4)

• Systematic review, Very low certainty.
• Size: Ten studies (11 publications), 480 participants; adverse event data from four studies (173 on carbamazepine, 173 on placebo) and withdrawal data from eight studies (268 versus 255)
• Who: Adults with trigeminal neuralgia, diabetic neuropathy or post-stroke pain.
• How long: No trial longer than four weeks; most four weeks or less.
• Result: No first-tier or second-tier evidence for any efficacy outcome; 65% (113/173) on carbamazepine had at least one adverse event versus 27% (47/173) on placebo, which the review expresses as two extra people with an adverse event for every five treated; 3% (8/268) withdrew for adverse events on carbamazepine versus none (0/255) on placebo.
• Funding: Authors report research support from charities, government and industry at various times, none related to this review; one author has consulted for and received lecture fees from pharmaceutical companies.

In four studies 65% (113/173) of participants experienced at least one adverse event with carbamazepine, and 27% (47/173) with placebo; for every five participants treated, two experienced an adverse event who would not have done so with placebo.

In schizophrenia, carbamazepine alone did not reduce relapse and adding it to an antipsychotic did not improve mental state. (Source 6)

• Systematic review, Low certainty.
• Size: Ten randomised trials, 283 participants.
• Who: Adults with schizophrenia or schizoaffective psychosis.
• How long: Varied; one sole-treatment trial was abandoned early at three months.
• Result: Sole treatment versus placebo, relapse RR 1.07 (95% CI 0.78 to 1.45, 1 RCT, n = 31) with 26 of 31 participants relapsing by three months; augmentation versus placebo, 50% reduction in BPRS RR 0.86 (95% CI 0.67 to 1.12, 6 RCTs, n = 147, low quality evidence); leaving the study early RR 0.47 (95% CI 0.16 to 1.35, 8 RCTs, n = 182, very low quality evidence)
• Funding: Independent review; the first author reports consulting and lecture honoraria from many pharmaceutical companies and that Eli Lilly provided medication.

Limit of this finding: Two things in this Cochrane abstract are wrong in the source itself and must not be read as written. It prints a parkinsonism result as "RR 0.03 CI 0.00 to 0.043", where the upper limit 0.043 is below the point estimate's own scale and is almost certainly a misplaced decimal for 0.43; that interval is not safe to use as a precise figure. It also prints "randomised still 283.One study" with no space and "we were able use GRADE" with a word missing. Both have been kept exactly as published. None of this affects the result quoted here, which is the absence of any difference in mental state. The whole review rests on 283 people in ten small trials.

There were no differences for the mental state outcome of 50% reduction in BPRS scores (6 RCTs n = 147, RR 0.86 CI 0.67 to 1.12, low quality evidence).

Cochrane's own verdict is that carbamazepine cannot be recommended for routine use in schizophrenia. (Source 24)

• Systematic review, Low certainty.
• Size: Ten randomised trials, 283 participants.
• Who: Adults with schizophrenia or schizoaffective psychosis.
• How long: Varied.
• Result: A negative conclusion rather than an effect size.
• Funding: Independent review; author industry disclosures as above.

Based on currently available randomised trial-derived evidence, carbamazepine cannot be recommended for routine clinical use for treatment or augmentation of antipsychotic treatment of schizophrenia.

There is no evidence that carbamazepine is habit-forming or produces dependence. (Source 15)

• Official position, Low certainty.
• Size: Not quantified.
• Who: Humans taking carbamazepine.
• How long: Not stated.
• Result: A categorical statement of absence rather than a measured rate.
• Funding: Not applicable.

No evidence of abuse potential has been associated with carbamazepine, nor is there evidence of psychological or physical dependence in humans.

The label does not support carbamazepine as a general painkiller: it states the drug is not a simple analgesic and should not be used for trivial aches or pains. (Source 3)

• Official position, Certainty not rated.
• Size: Not applicable.
• Who: People with facial pain.
• How long: Not applicable.
• Result: No numbers.
• Funding: Not applicable.

This drug is not a simple analgesic and should not be used for the relief of trivial aches or pains.

The label states that the effect of race and sex on how the body handles carbamazepine has never been systematically studied. (Source 21)

• Blood level study, Certainty not rated.
• Size: No studies reported.
• Who: Not applicable; a stated absence of data.
• How long: Not applicable.
• Result: No data exist to quote. This is an explicit gap in the label's own pharmacokinetic section, and it sits alongside a genetic risk that differs sharply by ancestry.
• Funding: Not applicable.

The effects of race and gender on carbamazepine pharmacokinetics have not been systematically evaluated.

The label does not present settled human pregnancy-safety data; it directs patients who become pregnant to a registry that is still collecting that information. (Source 8)

• Official position, Certainty not rated.
• Size: Not applicable; an ongoing registry rather than a completed study.
• Who: People who become pregnant while taking carbamazepine.
• How long: Open-ended; the registry collects data as pregnancies occur.
• Result: No effect estimate. The label gives the North American Antiepileptic Drug Pregnancy Registry enrolment line, 1-888-233-2334.
• Funding: Not stated in the label.

Limit of this finding: A registry collects reports from people who enrol; it is not a controlled study, so it can describe what happens in pregnancies on the drug but cannot by itself establish what the drug caused.

Patients should be encouraged to enroll in the NAAED Pregnancy Registry if they become pregnant. This registry is collecting information about the safety of antiepileptic drugs during pregnancy.

Where the evidence is mixed

For focal onset seizures carbamazepine sits in the best-performing group of antiseizure drugs, but lamotrigine beat it on treatment failure. (Source 23)

• Meta-analysis, High certainty.
• Size: Individual participant data for 14,789 of 22,049 eligible participants (67%) from 39 of 89 eligible trials (43%)
• Who: Adults and children with focal onset seizures or generalised tonic-clonic seizures.
• How long: Time-to-event outcomes over the trials' follow-up.
• Result: Treatment failure for any reason, lamotrigine versus carbamazepine HR 1.26 (95% CI 1.10 to 1.44); versus oxcarbazepine 1.30 (1.02 to 1.66); versus phenytoin 1.44 (1.11 to 1.85); versus phenobarbitone 1.97 (1.45 to 2.67)
• Funding: Independent review; one author discloses that a consortium of GSK, EISAI and UCB Pharma funded an unrelated national audit through grants to his university.

lamotrigine performs better than most other treatments in terms of treatment failure for any reason and due to adverse events, including the other first-line treatment carbamazepine; HRs (95% CIs) for treatment failure for any reason for lamotrigine versus: levetiracetam 1.01 (0.88 to 1.20), zonisamide 1.18 (0.96 to 1.44), lacosamide 1.19 (0.90 to 1.58), carbamazepine 1.26 (1.10 to 1.44)

In trigeminal neuralgia almost everyone responds at first, but roughly a quarter cannot stay on the drug because of side effects. (Source 25)

• Cohort study, Low certainty.
• Size: 200 outpatients (100 started on carbamazepine, 100 on oxcarbazepine); 22 excluded to remove possible secondary trigeminal neuralgia.
• Who: Patients with typical classical trigeminal neuralgia at a tertiary neuropathic pain centre, all with MRI and trigeminal reflex studies.
• How long: Mean 8.6 months for the carbamazepine group, 13 months for oxcarbazepine.
• Result: Initial responders 98% with carbamazepine at a median 600 mg (range 200-1200) and 94% with oxcarbazepine at median 1200 mg; 27% of carbamazepine responders stopped or cut the dose to an unsatisfactory level because of undesired effects versus 18% on oxcarbazepine; late resistance in only 3 carbamazepine and 2 oxcarbazepine patients.
• Funding: Not stated.

Limit of this finding: This was a retrospective review of case records at one tertiary neuropathic-pain centre, with no control group, no placebo arm and no blinding, so it shows what happened to patients who were treated, not how much of that was the drug. The word "retrospective" appears only in the paper's background section, which is now included in the quoted block. Note also the denominator: 200 patients were selected but 22 were excluded, so the 98%, 27% and 3% figures rest on 178 patients, not 200. The paper does not restate that denominator.

The initial number of responders was 98% with CBZ with a median dose of 600 mg (range 200-1200), and of 94% with OXC, with a median dose of 1200 mg (range 600-1800). In a mean period of 8.6 months, 27% of responders to CBZ incurred in undesired effects to a level that caused interruption of treatment or a dosage reduction to an unsatisfactory level.

In acute bipolar mania carbamazepine beat placebo on response and symptom improvement, but was not among the better-tolerated options. (Source 5)

• Meta-analysis, Moderate certainty.
• Size: 72 double-blind randomised trials of 23 drugs plus placebo, 16,442 participants; response analysis 56 trials and 14,503 participants.
• Who: Adults with acute bipolar mania, mean age 39.55, 50.93% male.
• How long: Mean study duration 3.96 (SD 2.39) weeks.
• Result: Carbamazepine was among the drugs that outperformed placebo for response and for improvement of mania symptoms and had lower discontinuation due to inefficacy; only aripiprazole, olanzapine, quetiapine and risperidone had lower all-cause discontinuation than placebo.
• Funding: Authors declare no conflicts relating to this study; the first author reports speaker honoraria and research grants from multiple pharmaceutical companies in the past three years.

Limit of this finding: These results apply only to the trials the review let in: randomised trials of a single oral medicine taken on its own for at least 10 days, with trials that allowed a rescue antipsychotic excluded. They say nothing about carbamazepine added to another drug, which is how it is often used. Trials averaged under four weeks, so nothing here speaks to longer-term treatment. The source also prints "In conclusions", its own grammatical slip, which has been kept. PubMed dates this paper 2022 and Crossref dates the same DOI 2021 (online first); 2022 is the issue date.

In conclusions, these antipsychotics, carbamazepine, lithium, tamoxifen, and valproate were effective for acute mania. However, only aripiprazole, olanzapine, quetiapine, and risperidone had better acceptability than the placebo.

The label's own instruction is that the risk of suicidal thoughts has to be weighed against the risk of the untreated illness, because epilepsy itself carries an increased risk of suicidal thoughts and behaviour. (Source 20)

• Official position, Moderate certainty.
• Size: Not applicable; the label's instruction to prescribers.
• Who: Anyone being considered for carbamazepine or any other antiepileptic drug.
• How long: Not applicable.
• Result: No effect estimate; the label states that the illnesses antiepileptic drugs are prescribed for are themselves associated with morbidity, mortality and an increased risk of suicidal thoughts and behaviour.
• Funding: Not applicable.

Anyone considering prescribing carbamazepine or any other AED must balance the risk of suicidal thoughts or behavior with the risk of untreated illness. Epilepsy and many other illnesses for which AEDs are prescribed are themselves associated with morbidity and mortality and an increased risk of suicidal thoughts and behavior.

Where the research disagrees

Whether carbamazepine should still be a first-line choice for newly diagnosed focal epilepsy

• The UK National Institute for Health and Care Excellence, as reported in Nevitt 2022, A clinical guideline, that is, a position held by a body, not an experiment: current guidelines from the National Institute for Health and Care Excellence (NICE) in the United Kingdom for adults and children recommend carbamazepine or lamotrigine as first-line treatment for focal onset seizures and sodium valproate for generalised onset seizures (Source 26)
• Nevitt and colleagues, Cochrane 2022, Network meta-analysis of individual participant data from 39 trials and 14,789 participants, which puts carbamazepine in the best group overall but still behind lamotrigine on treatment failure: High-certainty evidence demonstrates that for people with focal onset seizures, current first-line treatment options carbamazepine and lamotrigine, as well as newer drug levetiracetam, show the best profile in terms of treatment failure and seizure control as first-line treatments. (Source 27)

How much of the benefit claimed for carbamazepine in neuropathic pain the evidence can actually carry

• Wiffen and colleagues, Cochrane 2014, Systematic review applying three tiers of evidence quality to ten small, mostly cross-over trials: Carbamazepine is probably effective in some people with chronic neuropathic pain, but with caveats. No trial was longer than four weeks, had good reporting quality, nor used outcomes equivalent to substantial clinical benefit. (Source 4)
• Di Stefano and colleagues, Journal of Headache and Pain 2014, Retrospective cohort of 200 outpatients in one tertiary centre, specifically in classical trigeminal neuralgia rather than neuropathic pain generally: CBZ and OXC were confirmed to be efficacious in a large majority of patients, but the side effects caused withdrawal from treatment in an important percentage of patients. (Source 25)

How much

• Reference intake: There is no reference intake for a prescription medicine; dosing is set by the prescriber and titrated against blood levels and response. The label's position, as SPL version 29 effective 16 September 2026, is that treatment starts low and rises gradually, is taken with meals, and is then cut back to the minimum effective level; for epilepsy in adults and children over 12 that is 200 mg twice a day initially with maintenance usually 800 to 1200 mg daily, and for trigeminal neuralgia 100 mg twice a day initially with maintenance usually 400 to 800 mg daily. (Source 14)
• Upper limit: No nutrition body sets an upper limit. The label's maximum daily doses, as a regulatory position, are 1000 mg in children 6 to 12 and in those 12 to 15 years of age, 1200 mg above 15 years, with up to 1600 mg used in adults in rare instances; under 6 years no recommendation can be made above 35 mg/kg/24 hours; and for trigeminal neuralgia the dose should not exceed 1200 mg daily. The quoted block now runs to the end of the label's section, so it contains the Dosage Information table that its own opening words, 'DOSAGE AND ADMINISTRATION (SEE TABLE BELOW)', point at; every maximum in that table is already stated in the prose above it. These are the prescriber's limits, not a dose for anyone to take. (Source 14)
• Studied: In the trigeminal neuralgia cohort the median effective dose was 600 mg a day with a range of 200 to 1200 mg. (Source 25)
• Studied: Usual adult therapeutic plasma levels are between 4 and 12 mcg/mL, and measured plasma levels range from 0.5 to 25 mcg/mL with no apparent relationship to the daily dose taken. (Source 21)
• Studied: The neuropathic pain trials pooled by Cochrane used any dose by any route for at least two weeks, in ten studies of 480 participants, almost all lasting four weeks or less. (Source 4)

A common belief, and what the research shows

The belief: Having an antiseizure drug for epilepsy means the seizures are controlled, and the genetic skin-rash test only matters for people from East Asia.

What the research shows: Two things in the literature cut against that. First, carbamazepine is an old drug that is no longer the clear best option even where it is licensed: "HRs (95% CIs) for treatment failure for any reason for lamotrigine versus: levetiracetam 1.01 (0.88 to 1.20), zonisamide 1.18 (0.96 to 1.44), lacosamide 1.19 (0.90 to 1.58), carbamazepine 1.26 (1.10 to 1.44)". Second, the genetic risk is not confined to one ancestry. HLA-B1502 is the East Asian allele, but in Europeans the relevant one is HLA-A3101, and "The presence of the allele increased the risk from 5.0% to 26.0%, whereas its absence reduced the risk from 5.0% to 3.8%." The label adds that "Many HLA-B1502-positive and HLA-A3101-positive patients treated with carbamazepine will not develop SJS/TEN or other hypersensitivity reactions", so a positive test is a risk marker, not a diagnosis.

Questions and answers

What is it?

Carbamazepine is an old prescription anticonvulsant, taken by mouth as tablets, chewable tablets, a suspension or extended-release forms. Chemically it is a dibenzazepine, related to the tricyclic antidepressants rather than to other epilepsy drugs. It is licensed both as an anticonvulsant and as a specific painkiller for trigeminal neuralgia. It is a manufactured medicine and is not found in food. (Source 28)

What does it do in the body?

It calms over-excitable nerve signalling. In animal studies it reduces polysynaptic responses, blocks post-tetanic potentiation and abolishes pain triggered by stimulating the facial nerve involved in trigeminal neuralgia. In people the precise mechanism is still unknown, which the label says plainly. A second, quite separate action matters just as much in practice: it strongly induces liver enzymes, so it speeds up the destruction of many other drugs and eventually of itself. (Source 1)

Is it good or bad for you?

It depends entirely on which condition and which person. For focal epilepsy and for trigeminal neuralgia the evidence supports it, though lamotrigine outperformed it on treatment failure and nearly a third of trigeminal neuralgia patients could not tolerate it. For absence seizures it does not work, and for schizophrenia the trials are negative. Its risks are the serious ones: a boxed warning for Stevens-Johnson syndrome and toxic epidermal necrolysis at 1 to 6 per 10,000 new users (about ten times that in some Asian countries), and for aplastic anaemia and agranulocytosis at 5 to 8 times the background rate; plus hyponatraemia, cardiac conduction problems, liver injury, and a class-wide near-doubling of suicidal thoughts. (Source 2)

How do you get more of it?

Carbamazepine is prescription-only and there is no dietary route to it. The label's position is that a low initial dose is raised gradually and then trimmed back to the minimum that works, and that it should be taken with meals. Doses studied in trigeminal neuralgia clustered around a median 600 mg a day; usual adult therapeutic blood levels are 4 to 12 mcg/mL. Because of autoinduction, the same dose produces lower levels after three to five weeks than it did on day one. None of this is advice for any reader. (Source 14)

If it is harmful, what reduces it?

When carbamazepine is doing harm the documented response is to stop or reduce it, under medical supervision, never abruptly if it is controlling seizures. The label says to discontinue at the first sign of a rash unless the rash is clearly not drug related, and that if signs suggest Stevens-Johnson syndrome, toxic epidermal necrolysis or a bullous fixed drug eruption the drug should not be restarted at all. There is no supplement or food that clears it. Levels fall faster once liver enzymes are induced, which is why the half-life shortens from 25 to 65 hours to 12 to 17 hours. (Source 18)

Why might someone be low in it or missing it?

The question does not apply as a deficiency: the body does not make carbamazepine. Someone may not be on it because it is not indicated, because a genetic test rules it out, or because something else wrecks it. The label says people with ancestry in populations where HLA-B*1502 may be present should be screened first and those testing positive should not be treated unless the benefit clearly outweighs the risk. Levels can also fall far below the useful range when it is combined with enzyme-inducing drugs such as rifampin, phenytoin, phenobarbital or primidone. (Source 7)

Which whole foods contain it or feed it?

No whole food contains carbamazepine. Food matters in two other ways. The label says to take it with meals. And grapefruit juice blocks the CYP3A4 enzyme that clears it, so the label lists grapefruit juice among the agents expected to push carbamazepine blood levels up, alongside the supplement form of vitamin B3, niacinamide (nicotinamide). (Source 7)

What happens if you do not have it?

Nothing happens from lacking the drug; the risk is losing its effect once it has been started. The label warns that abrupt discontinuation of any anticonvulsant in a responsive patient may lead to seizures or even status epilepticus, which is life-threatening, and that is why withdrawal is gradual. For someone never treated, what remains is the untreated condition: in classical trigeminal neuralgia 98% of patients responded to carbamazepine, so for most people the alternative to treatment is continuing attacks of facial pain. (Source 11)

How can you test for it?

Several tests apply, and they are unusually well specified for an old drug. Before starting, high-resolution HLA-B*1502 typing is recommended for genetically at-risk patients; the test is positive if one or two alleles are found. A full blood count including platelets is taken as a baseline, with liver function, urinalysis, BUN and eye examinations repeated periodically. Blood levels of the drug itself can be measured, and the label says monitoring has increased the efficacy and safety of anticonvulsants; the usual adult therapeutic range is 4 to 12 mcg/mL, but levels correlate poorly with dose, especially in children. Serum sodium should be watched because hyponatraemia is common and dose-related. (Source 29)

References

1. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, CLINICAL PHARMACOLOGY / Mechanism of Action (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
2. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets / Carbamazepine Chewable Tablets, Boxed Warning section (LOINC 34066-1) (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
3. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, INDICATIONS AND USAGE (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
4. Cochrane Database of Systematic Reviews. Carbamazepine for chronic neuropathic pain and fibromyalgia in adults. 2014. PMID 24719027, DOI 10.1002/14651858.CD005451.pub3. Read the source
5. Molecular Psychiatry. Pharmacological treatment for bipolar mania: a systematic review and network meta-analysis of double-blind randomized controlled trials. 2022. PMID 34642461, DOI 10.1038/s41380-021-01334-4. Read the source
6. Cochrane Database of Systematic Reviews. Carbamazepine for schizophrenia. 2014. PMID 24789267, DOI 10.1002/14651858.CD001258.pub3. Read the source
7. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, PRECAUTIONS / Drug Interactions / Agents That May Affect Carbamazepine Plasma Levels (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
8. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, PRECAUTIONS / Information for Patients (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
9. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, PRECAUTIONS / Effect of Carbamazepine on Plasma Levels of Concomitant Agents (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
10. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, untitled section closing the printed boxed warning (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
11. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, ADVERSE REACTIONS (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
12. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, WARNINGS / General (hyponatraemia, gradual withdrawal, porphyria) (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
13. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, MEDICATION GUIDE (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
14. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, DOSAGE AND ADMINISTRATION (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
15. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, DRUG ABUSE AND DEPENDENCE (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
16. The New England Journal of Medicine. Carbamazepine-induced toxic effects and HLA-B*1502 screening in Taiwan. 2011. PMID 21428768, DOI 10.1056/NEJMoa1009717. Read the source
17. The New England Journal of Medicine. HLA-A*3101 and carbamazepine-induced hypersensitivity reactions in Europeans. 2011. PMID 21428769, DOI 10.1056/NEJMoa1013297. Read the source
18. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, WARNINGS / SJS-TEN and HLA-B1502 Allele and Hypersensitivity Reactions and HLA-A3101 Allele (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
19. Heliyon. Incidence trends and risk factors for hyponatremia in epilepsy patients: A large-scale real-world data study. 2023. PMID 37554799, DOI 10.1016/j.heliyon.2023.e18721. Read the source
20. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, WARNINGS / Suicidal Behavior and Ideation (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
21. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, CLINICAL PHARMACOLOGY / Pharmacokinetics (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
22. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, ADVERSE REACTIONS, reactions by body system (Hemopoietic System through Metabolism) and the closing paragraphs on lupus erythematosus-like syndrome, plasma lipids and aseptic meningitis (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
23. Cochrane Database of Systematic Reviews. Antiepileptic drug monotherapy for epilepsy: a network meta-analysis of individual participant data. 2022. PMID 35363878, DOI 10.1002/14651858.CD011412.pub4. Read the source
24. Cochrane Database of Systematic Reviews. Carbamazepine for schizophrenia (Authors’ conclusions). 2014. PMID 24789267, DOI 10.1002/14651858.CD001258.pub3. Read the source
25. The Journal of Headache and Pain. Natural history and outcome of 200 outpatients with classical trigeminal neuralgia treated with carbamazepine or oxcarbazepine in a tertiary centre for neuropathic pain. 2014. PMID 24912658, DOI 10.1186/1129-2377-15-34. Read the source
26. Cochrane Database of Systematic Reviews. Antiepileptic drug monotherapy for epilepsy: a network meta-analysis of individual participant data (Background). 2022. PMID 35363878, DOI 10.1002/14651858.CD011412.pub4. Read the source
27. Cochrane Database of Systematic Reviews. Antiepileptic drug monotherapy for epilepsy: a network meta-analysis of individual participant data (Authors’ conclusions). 2022. PMID 35363878, DOI 10.1002/14651858.CD011412.pub4. Read the source
28. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, DESCRIPTION (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source
29. DailyMed (U.S. National Library of Medicine), Structured Product Label. Carbamazepine Tablets, PRECAUTIONS / Laboratory Tests (Major Pharmaceuticals). SPL version 29, effective 2026-09-16. Read the source