Valproate for acute mania

These commentaries are based on Dr Gillman’s peer reviewed scientific papers, see Publications


This commentary reviews valproate dosage and addresses the question of whether it is under-dosed and under-used in acute mania and hypomania. Even by the conservative standard used as advice in guidelines, the audits of actual clinical practice demonstrate that the typical dosages used are below those recommended. And ‘recommended’ dosages are themselves low. There is a strong suggestion that regimes using higher dosages (40-60 mg/kg/day) show much increased efficacy over typical regimes of only 10-20 mg/kg/day. Higher dosages show a comparatively low level of side effects and low potential for toxicity. Patient acceptability, withdrawal due to side effects, and medium-term problems (e.g. weight gain, metabolic syndrome, extra-pyramidal SEs), are all more favourable with valproate than for anti-psychotic drugs.


Is valproate under-dosed, in acute mania/hypomania? Good evidence indicates that is the case; I am not alone in that view {Vasudev, 2012 #16933}: I will proffer the evidence and my experience to explain why this is so. One factor to remember is that we are talking about short-term use in acute illness: with adequate dosages the duration of treatment is usually short, which will influence the overall impact of the burden of side-effects, as well as the risk of severe adverse effects, which are rare.

Monotherapy with valproate or an antipsychotic agent is often recommended as first-line treatment for acute mania: de novo, lithium may be slower to act and be less effective in more severe cases {Goodwin, 2009 #9453;Goodwin, 2016 #20340}, but a proportion of cases presenting will already be on lithium, so checking and adjusting levels and adding a second drug will often occur. Many patients with previous experience of anti-psychotic drugs would prefer to be given valproate.

It is worth noting that states of arousal and excitement, whether precipitated by drugs (stimulants, ST, or tyramine-induced hypertension), or psychosis, are usually safely and effectively ameliorated with high dose BDZs. They are a valuable adjunct to treatment and are much preferred to raising the dose of a dopamine antagonist, which is sometimes done mistakenly to achieve sedation. Horses for courses.

When I was in practice I was an early adopter of high-dosage valproate. The valproate dosage regime I used was informed by my friend Professor Mervyn Eadie, a neurologist who was doing research on valproate around that time {Eadie, 1976 #20617;Eadie, 1983 #20613;Eadie, 1988 #20619;Dickinson, 1989 #20620;Eadie, 1990 #20621;Eadie, 1991 #20614;Eadie, 1992 #858}, and it was aimed at achieving the maximum tolerable dosages within a few days of admission to hospital. I therefore started with the objective of using 20-60 mg/kg per day, aiming at an initial dosage in the middle of that range of 40-50 mg/kg, which is a daily dosage of 2,500 mg for smaller persons (50 kg body wt.), and 5,000 mg for larger persons (100 kg body wt.).

I quickly settled on an initial dose of 1 g on admission (in the non-slow-release form in which it was then available, ‘Epilim’ tabs), followed by a further 1 g 4 to 6 hours later, usually escalating to 3 or 4 g the following day — most patients were able to tolerate 3 to 4 g daily without difficulty.

I am not suggesting that I necessarily always agree with the FDA (in this instance I do), but it may be noted that the approved indications are: acute mania, and mixed episodes, but not maintenance treatment.

Frequency of usage

The UK study by Hayes et al. {Hayes, 2011 #20350} found that by 2009 only 20% of BPD patients were given valproate. Incidentally, they also proffered what seems to me to be an odd comment ‘by 2009 one third of women of childbearing age who were taking medication for bipolar disorder were taking valproate. This final finding is worrying as guidelines are now very clear that valproate should be avoided because of its teratogenic potential.’ Members of these guideline-groups do appear to become a little intoxicated by the buzz of being ‘elevated’ to the perceived star-status of ‘expert committee membership’ and when assembled as committees often come up with odd notions: the imperative ‘should be avoided’ is in this context inappropriately directive and commanding phraseology (by what authority do these individuals think they act?). Careful consideration of the ‘pros and cons’, yes; but should be avoided is completely out of order as a directive and will be interpreted in ways possibly not intended by persons such as administrators and lawyers.

Since the main use is for acute treatment of a serious illness with significant morbidity (and mortality), often in hospital, to ‘avoid because of its teratogenic potential’ is ridiculous.

Druschky’s usage figures from Germany found (1993-2013) that ~50% of manic patients were treated with valproate, a much higher proportion than other countries {Druschky, 2017 #20649}, well done Germany.

The Swedish data from Song et al. {Song, 2017 #21062} show 15% of a 50,000 cohort were on Valproate (but that is ever, during the study period, so it may be a rather smaller % had it as acute treatment.

The frequency of usage clearly varies greatly, but is probably often in combination, with the so-called’ ‘atypical’ APs, and with lithium. Reliable figures from North America seem unavailable. The power of the dollar influencing the prescribing of more expensive new so-called atypical antipsychotics, at the expense of cheaper drugs like lithium and valproate, is likely a relevant driving force over prescribing habits.


The peak plasma concentration (Tmax), with the standard non-slow-release preparation, is ~2-4 hrs, T1/2 11-20 hrs (steady-state 3-4 days).

It is highly, but variably, protein bound (to albumin) and only the free drug is active and able to cross the BBB {Gibbs, 2015 #21079;Lindberger, 2001 #21080}. CSF levels equilibrate rapidly (~4 hrs) with plasma free valproate and are nearly the same concentration. Plasma free levels rise in a non-linear fashion once binding sites with albumin are close to saturation, which means that the therapeutic effect can rise quite rapidly once a certain level is achieved. This explains why low-dosage valproate has minimal effect.

Excretion, following conjugation in the liver, is mostly via the kidney (but CYP2C9 plays a significant role in children).

The usually stated range for trough total valproate serum concentration has generally been 50-100 mg/ml (350-700 mmol/L), but as with most TDM evidence is ‘low-grade’, so recommendations vary (see above re protein-binding).

Experience, and research, show that most ‘TDM’ serum level requests are inappropriate, because they are not taken at the correct time — at least eight hours after the last dose, and before a morning dose {Perucca, 2002 #2491}.

Using blood levels to guide dosage is of uncertain value — ‘good servants but bad masters’.

The (active) free-fraction FF (not protein-bound) is generally around 5-10%, but rises non-linearly, to ~20-30% when total levels are >60 μg/ml {Nasreddine, 2018 #21063}. FF also varies a lot more in specific circumstances (like low albumen, other highly protein-bound drugs, uraemia etc. and may be hard to predict in ICU patients with multiple drugs and morbidities.

However, in usual practice fewer factors cause problems: the common drugs the affect protein binding and increase FF are; aspirin, clevidipine and ibuprofen.

A recent (2018) study adds to our understanding: Nasreddine et al. {Nasreddine, 2018 #21063} found that VPA exhibits a concentration-dependent protein binding and increasing total VPA concentration >60 μg/ml results in non-linear rise in FF VPA concentration (presumably because binding is saturated). They recommend the total daily dosage of VPA should be increased in smaller increments once total VPA >60 μg/ml. For drug monitoring they recommend assaying the FF VPA, but that is not available in many laboratories (NB FF VPA is a good reflection of CNS level with rapid equilibration).

In a recent review from the epilepsy field Patsalos et al. {Patsalos, 2008 #20637} highlight the value of ‘individual therapeutic concentration’, as opposed to reference ranges and concluded that the evidence validating a useful predictable dose-effect relationship is poor (see also ‘toxicology’ below), and that due to individual variation many patients require concentrations outside the reference ranges. At least for epilepsy management is best guided by determination of the ‘individual therapeutic concentration’ (the concentration needed for seizure freedom in a particular individual — which may vary with the type of seizure).

These issues make it even more relevant to assess the levels at which valproate is effective in mania separately, and not to rely only on pre-existing epilepsy data — see comment re Allen {Allen, 2006 #16941} below.

Twice daily administration of the standard preparation is a reasonable way to start therapy, and if blood tests are felt to be indicated they should be done before the morning dose.

Note that some Valproate SEs are not concentration-related (see toxicology below).

High-dosage Valproate

Until I undertook the research for this commentary I was insufficiently aware that the vast majority of trials, and other published accounts, of valproate usage in acute mania had utilised such low dosages (viz. a maximum of 10-20 mg/kg/day). Despite such low dosage they demonstrated a degree of efficacy equivalent to APs in most studies. The Cochrane review tabulates the details of the trials that were included {Macritchie, 2003 #2596}, of which none used in excess of 20 mg/kg except Freeman {Freeman, 1992 #20632}, who used up to 3,000 mg, but only in a small number of patients, and Tohen {Tohen, 2000 #20633} who used a max of 2,500 and used blood levels in a target range of 50 to 125 µg/ml.

In Druschky’s more recent European data {Druschky, 2017 #20649} from a sample of 50,000 cases the mean dosage was 1,500 mg daily. And Paton et al. in a large UK audit {Paton, 2018 #21055} of 6705 patients reported the mean dosage of valproate was in 1200 mg women and 1400 mg for men — that is less than 20 mg/kg for most men.

A review by Vasudev et al. underlines the under-dosing issue {Vasudev, 2012 #16933}. They found the mean valproate dosage being given after the first week was 1,000 mg. The maximum, after one whole month, was only 1,500 mg. They concluded, understandably, that inadequate initial dosage, and inadequate dosage increments, indicated that current practice is not evidence-based.

A small digression is relevant here, because this is an excellent example of how guidelines are a powerful negative self-fulfilling and circular exercise. Trials are often designed to show equivalent effectiveness to existing treatment, but with less side effects. The trials with the lowest dosages demonstrating equi-effectiveness get published. Others get lost. Since no trials have utilised higher dosages the guidelines will inevitably conclude that there is no evidence that higher dosages are better — and they will therefore entrench and perpetuate that unfounded misconception, especially because more and more doctors now are unwilling to use non-guideline drugs and dosages. This is exemplified by the UK NICE guidelines which are unhelpful, in more ways than one.

Therefore, few doctors gain the experience that would contradict that, because they feel constrained to follow the guidelines. Furthermore, so few psychiatrists have knowledge of the pharmacology and toxicology of valproate, that they do not have the confidence to try higher dosages.

This is despite scientific evidence that under dosing is occurring, and that higher dosages are safe, and more effective (see below).

The more one considers it, the more one sees examples of the powerful negative influence, on clinical practice, of guidelines.

Vasudev’s paper leaves me profoundly perplexed regarding the kind of difficulty in thinking and decision-making that allows a clinician to take a month to do something so simple, like to adjust the dosage of the drug above the minimum effective dose. This is especially so because it is being given for a serious acute condition which carries significant morbidity.

This frequent failure of ‘good-clinical-management’ of treatment, in relation to dosage-escalation, duration, and response, is a theme which permeates most of psychiatric therapeutics, and is given insufficient attention in guidelines, if it is mentioned at all (see the comments in my AD algorithm). This seems to relate to several factors: the generally poor critical-thinking skills of psychiatrists, the inattention to teaching competent decision-making during their training, and poor confidence, resulting largely from poor pharmacological knowledge.

Rapid control of severe mania

Valproate is relatively unusual in that the intravenous and oral doses are equivalent (i.e. bio-availability is 100%): therefore, blood levels and effectiveness are similar when equivalent IV/oral doses are administered {Ghaleiha, 2014 #20579}. Generally speaking, in manic patients, it will be easier to give drugs orally, than attempt to manage intravenous infusions, which would in any case have no advantages.

Valproate has now been in use for over 50 years, and it has become clear that dosages of up to 140 mg/kg/day are safe. Trials in epilepsy indicate that dosages of 40-60 mg/kg/day are well-tolerated and many patients have little by way of side-effects or significant adverse effects, even when doses are above 100 mg/kg/day (see also ‘toxicology’).

Typical guidelines for the treatment of epilepsy advise dosages up to a maximum of 60 mg/kg/day.

The above information helps to put the low dosages of around 10-20 mg/kg into perspective.

A conservative dosage range was simply adopted from pre-existing common practice in epilepsy patients. It is unlike most other psychiatric drugs, which gained approval specifically for a treatment indication in psychiatry: those were subjected to dose-response studies, prior to the clinical-trial phase of experimentation.

No formal dose-ranging study has ever been done concerning the use of valproate in acute mania, or any other psychiatric condition — but see Allen {Allen, 2006 #16941} regarding evidence that higher levels are better. They produced evidence of a pronounced and clear improvement in benefit showing a linear trend that did not plateau-out even at serum levels of >100 μg/ml.

The recent study from Ghaleiha {Ghaleiha, 2014 #20579} indicates that rapid normalization of manic symptoms may be expected with even moderate loading-dose strategies (20mg/kg aiming at Se levels of 100 mg/ml — their loading dosage produced Se level of 100 μg/ml within 24 hrs, and a reduction in the YMRS score from 35>10 in 7 days. That accords with my clinical experience.

Carlat sums it up as ‘very effective for acute mania, and rapidly so, usually quelling manic symptoms within a week’. ‘Trials’, and hence guidelines, suggest that even at low dose it is equally effective as lithium and APs; I take Carlat’s above comment to infer his opinion that it is much better. I concur.

Incidentally, it appears lamotrigine is used surprisingly frequently in acute mania despite complete lack of evidence for its effectiveness.

Relative safety vs AEDs and APs

Druschky’s review {Druschky, 2017 #20649} — these figures are from Germany — utilizing data from half a million patients over 20 years, indicates, as previous reports and data suggest, that valproate is the safest of all the anti-epileptic drugs (AEDs), and it is safer and much more ‘patient-friendly’ than APs. It has the lowest rate of adverse drug reactions rated as severe, compared to other AEDs {Druschky, 2017 #20649}. Since the turn of the millennium about 50% of manic patients (in Germany) were treated with VPA. The median dose of 1500 mg is much higher than the other figures given above.

Severe adverse drug reactions (ADRs) with single AED usage (monotherapy) were, per 1,000 — Valproate 2.5, Carbamazepine 7, Oxcarbazepine 9.4, Lamotrigine 4.4, pregabalin 4.3.

The probability of developing a significant reduction in thrombocyte count was substantially increased at trough VPA levels above 100 microg/ml in women and above 130 microg/ml in men; a significant negative correlation was found between VPA levels and platelet counts. Women were more likely to develop reduced platelet counts {Nasreddine, 2008 #21065}. Clinically significant thrombocytopenia did not occur.

It has low day-to-day side-effects compared to alternatives, and the lowest rate of treatment cessation from all causes, including SEs (a contrast to quetiapine where in some studies up to 50% of recipients cease it within 4 weeks).

Dose-related SEs most frequently experienced appear to be: nausea, upper abdominal cramps, diarrhoea, vomiting, anorexia, asthenia. But even in patients on high doses, these rarely require treatment discontinuation {Beydoun, 1997 #20672}. Note, not all SEs are dose-related.

It has a wide margin of safety as the toxicology data demonstrates.

Recent papers relating to higher dosages of valproate

Trinka {Trinka, 2014 #20353}: Efficacy and safety of intravenous valproate for status epilepticus: a systematic review. … 15 and 45 mg/kg in bolus (6 mg/kg/min) followed by 1-3 mg/kg/h infusion. Safety studies of intravenous VPA administration in patients with status epilepticus showed a low incidence of adverse events overall (<10%), mainly dizziness, thrombocytopenia, and mild hypotension, which was independent of infusion rates. Of note, good cardiovascular and respiratory tolerability was observed in these studies, even at high dosages and fast infusion rates (up to 30 mg/kg at 10 mg/kg/min), despite multiple morbidities or other antiepileptic drugs.

Georgoff {Georgoff, 2017 #20370}: The maximum tolerated dose of intravenous valproate in healthy subjects was 140 mg/kg; significantly higher than the previously established maximum tolerated dose of 60-75 mg/kg.  Adverse events were mild and no drug-related abnormalities were seen in clinical laboratory tests, ECG, and cognitive testing.


Case reports of interactions and toxicity can be highly misleading, as they are in my field of serotonin toxicity. The same, unsurprisingly, applies to valproate, which has, as a result of misleading case-reports, quite unjustifiably acquired a reputation for high toxicity and a narrow therapeutic index.

More comprehensive data from large series of carefully observed overdoses from toxicology units strongly contradicts that impression {Shadnia, 2015 #20671;Isbister, 2003 #20670}. In a nutshell, valproate is relatively benign in overdose: mostly producing only drowsiness, tachycardia, and gastrointestinal effects (vomiting). In Isbister’s series ‘in 8 of the 15 valproate-alone poisonings, greater than 200 mg kg-1 (14 g) was ingested, none of these had any major abnormalities except drowsiness in two cases ingesting 20 g and 25 g.’ Just as there is little correlation between SEs and serum levels at therapeutic dosages, so Isbister found there is no correlation in ODs between serum levels and severity: they concluded, ‘more than half of the valproate-alone overdoses ingested greater than 200 mg kg-1 with little effect. Drowsiness was seen in two patients ingesting 300– 400 mg kg-1.

These data are from Prof Whyte’s prospective toxicology database, that I have a connection with, because that is where most of the ST data have come from. It is high-quality data from a top clinical research group. Lastly, they also noted how much more toxic was carbamazepine.

Long-term usage

Only limited evidence, of poor quality, supports the efficacy of valproate in the long-term treatment of bipolar disorder. As Cipriani et al. state {Cipriani, 2013 #20522}: ‘Clinicians and patients should consider acceptability and tolerability profile when choosing between lithium and valproate, their combination, or other agents, as long-term treatment for bipolar disorder’.

The FDA approved indications are: acute mania, and mixed episodes, but not maintenance treatment.

More splendid irony: where it is undoubtedly highly effective in acute mania it is underused, and where it has dubious efficacy in the long-term prophylaxis of illness-episodes, it is widely used.

Nobody can accuse psychiatrists of being logical.


Valproate is under-used and greatly under-dosed in acute mania. Mono-therapy with high-dosage valproate is the first-choice preference by a wide margin. The higher doses of 20-60 mg/kg/day were, in my experience, required for a week or two at most: therefore, possible dosage-related risks (e.g. thrombocytopenia) are unlikely to be significantly elevated.

It has less side effects and a wider margin of safety than any of the other relevant drugs.

If adequately dosed, is almost certainly much more effective, despite the modest evidence for this from RCTs. That simply speaks to the fact that RCTs are frequently unhelpful, and as Parker has said {Parker, 2003 #2999}, they ‘produce clinically meaningless results’.

It is therefore important for clinicians to develop and trust their clinical judgement, and stop deferring too readily and too often to the sometimes mis-informed conclusions promulgated by the committees that produce guidelines — for a discussion of the extensive problems with guidelines see my recent commentary.

There has been little useful clinical research on valproate in mania for some years now — all the money is in new APs! But there is more data on pharmacology, dosing and toxicology, showing valproate has a substantially better profile than other options.

No dose-ranging study has been done concerning the use of valproate in acute mania: a conservative dose-range was adopted uncritically from epilepsy practice. The re-examination of those data supports the safety, theoretical rationale, and effectiveness, for administering much larger dosages.

Accordingly, I would urge clinicians to consider the use of higher dosages of valproate in acute mania (20 – 60 mg/kg/day, or more): that means, even for a little (50 kg) patient a minimum dosage of 1,500 mg per day (30 mg/kg) and for a larger person (100 kg) a minimum dosage of 3,000 mg per day.

It is probable that the immense amount of money put into the marketing of new so-called ‘atypical’ antipsychotic drugs has obscured the usefulness of valproate. As if it were needed, this is yet another example of the distorting effect on good clinical practice, produced by the huge dollar influence in sales and marketing from big Pharma, acting on the rather half-hearted and hesitant profession of psychiatry.

Consider Donating to PsychoTropical

PsychoTropical is funded solely through generous donations, which has enabled extensive development and improvement of all associated activities. Many people who follow the advice on the website will save enormously on doctors, treatment costs, hospitalization, etc. which in some cases will amount to many thousands of dollars, even tens of thousands — never mind all the reduction in suffering and the resultant destruction of family, work, social, and leisure capability. A donation of $100, or $500, is little compared to those savings. Some less-advantaged people feel that the little they can give is so small it won’t make a difference – but five dollars monthly helps: so, do not think that a little donation is not useful.

– Dr Ken Gillman

Dr Ken Gillman