Reappraising a Preference for Valproate

October 29, 2020
David N. Osser, MD

David Osser, MD, comments on the low priority of valproate for treating patients with bipolar disorder.


I am pleased to be offered the opportunity to respond to the commentary by Bernardo Dell’Osso, Monica Macellaro, and Terence A. Ketter. I appreciate that these distinguished clinicians and scientists have taken an interest in this piece, especially because they largely agree with most of the points made, adding some validity to the perspectives presented.

They begin by proposing, without citing evidence, that valproate does not deserve a low priority in some subgroups of bipolar patients. They affirm that it can be reasonable for male and female patients not of child-bearing potential, especially the “very young” (presumably younger than the children and adolescents in the Wagner et al study1, in which valproate did not do better than placebo), and elderly females. They did agree that valproate had inferior results compared with most other competing antimanic medicines in the 3 meta-analyses cited. I can add the more recently published GERI-BD study involving 224 elderly inpatients (with equal numbers of men and women) that found, consistent with the meta-analyses, lithium to be superior to valproate for acute mania over 9 weeks and comparably tolerated.2

The commentators then proposed that I failed to note contraceptives could mitigate the teratogenicity and related risks to offspring exposed to valproate. However, they fail to note that 50% of pregnancies are unplanned. Thus, it seems difficult to be confident in contraceptive adherence (especially should a hypomania or mania occur). Combining that consideration with the marginal efficacy in the evidence base and the very high risk of fetal harm, valproate still seems to be nearly the lowest priority option for women of child-bearing potential. Use of valproate in this population is considered to be a major public health problem in Europe.3

The commentators next raise the possibility of “methodological limitations,” explaining the negative controlled studies of valproate in acute mania without describing them. The authors of the negative Hirschfeld et al study4 of valproate proposed that failure could have been due to their use of what they termed a “moderate” mean dose of 2200 mg daily. In the previous, only modestly positive study5, participants received a mean dose of 3350 mg daily. However, there were more side effects of somnolence, nausea, vomiting, and dizziness in that study. Few clinicians use such a high dose. Also, the authors noted that the protocol of the negative study encouraged earlier hospital discharge, which may have made it more difficult to find a difference from placebo. The negative study also allowed longer use of adjunctive lorazepam, which may have added support to the placebo group. However, shorter hospital stays are the standard today, and experts encourage liberal use of adjunctive benzodiazepines in treating mania.6 Thus, the conditions that may have resulted in valproate not being found efficacious in its most recent study are likely to be present in usual practice today.

The commentary discussed the lack of FDA-approval for valproate as a maintenance treatment. The cited maintenance study that failed to show efficacy indeed had faults—and one that is often mentioned is that the lithium group had more suicidal issues than the placebo group, contrary to most other evidence, suggesting that there was something peculiar about the patient sample. The commentators encouraged not making too much of FDA-approval as a basis for decision-making. They noted that the FDA has approved adding medications like second generation antipsychotics (SGA) to mood stabilizers including valproate and lithium. However, these SGAs are also efficacious and approved as monotherapies for mania, and in the design of all these enhancement studies, the patient must first have been on the valproate or lithium for 2 weeks and not responded before the SGA or placebo is added. So, these studies do not contribute to the evidence base for the efficacy of valproate: the improvement could have been entirely from the SGA. Other studies including the BALANCE 2-year trial of lithium versus valproate versus the combination found valproate far less effective than lithium.7,8 Effectiveness as a maintenance treatment should be an important consideration in choosing a medication for an acute episode. 

After some remarks in agreement with the article, the commentary discusses the relative weight gain from valproate versus olanzapine. The authors assert that even if valproate causes more weight gain than olanzapine in head-to-head trials, olanzapine is more widely avoided. This is likely true. Probably both should be avoided in favor of other agents for mania at least for first-line use, as asserted in important national and international guidelines.6

Next, there is mention of the relative benefit or harm from valproate versus lithium in patients with suicidal tendencies. There is nothing of importance in the literature to suggest that valproate has any benefit for this common problem in patients with bipolar disorder, even if, as the commentary authors assert, the FDA’s suicide risk warning on valproate was not based on strong evidence.

On the other hand, I agree with the importance of mentioning that carbamazepine, which may have a larger effect size for efficacy in acute mania, has almost as many problems for women of child-bearing potential. It also carries the same suicide risk warning that the FDA required for anticonvulsants.

In the final remarks, the commentary reiterated and added evidence that valproate is being very widely used especially in the United States, but that agreed that lithium, SGAs, lamotrigine, and carbamazepine should figure prominently in selecting medication for bipolar patients. The purpose of my article was, indeed, to urge clinicians to reappraise their preference for valproate. 

Dr Osser is associate professor of psychiatry, Harvard Medical School, and co-lead psychiatrist at the US Department of Veterans Affairs, National Telemental Health Center, Bipolar Disorders Telehealth Program, Brockton, MA. He is also on the Editorial Board of Psychiatric Times.


1. Wagner KD, Redden L, Kowatch RA, et al. A double-blind, randomized, placebo-controlled trial of divalproex extended-release in the treatment of bipolar disorder in children and adolescents. J Am Acad Child Adolesc Psychiatry. 2009;48:519-32.

2. Young RC, Mulsant BH, Sajatovic M, et al. GERI-BD: A Randomized Double-Blind Controlled Trial of Lithium and Divalproex in the Treatment of Mania in Older Patients With Bipolar Disorder. Am J Psychiatry. 2017;174:1086-93.

3. Sher J, Frank JW, Doi L, de Caestecker. Failures in reproductive health policy; overcoming the consequences and causes of inaction. J Public Health. 2018;41:e209-e15.

4. Hirschfeld RM, Bowden CL, Vigna NV, Wozniak P, Collins M. A randomized, placebo-controlled, multicenter study of divalproex sodium extended-release in the acute treatment of mania. J Clin Psychiatry. 2010;71:426-32.

5. Bowden CL, Swann AC, Calabrese JR, et al. A randomized, placebo-controlled, multicenter study of divalproex sodium extended release in the treatment of acute mania. J Clin Psychiatry. 2006;67:1501-10.

6. Mohammad O, Osser DN. The psychopharmacology algorithm project at the Harvard South Shore Program: an algorithm for acute mania. Harv Rev Psychiatry. 2014;22:274-94.

7. Geddes JR, Goodwin GM, Rendell J, et al. Lithium plus valproate combination therapy versus monotherapy for relapse prevention in bipolar I disorder (BALANCE): a randomised open-label trial. Lancet. 2010;375:385-95.

8. Kessing LV, Hellmund G, Geddes JR, Goodwin GM, Andersen PK. Valproate v. lithium in the treatment of bipolar disorder in clinical practice: observational nationwide register-based cohort study. Br J Psychiatry. 2011;199:57-63.