Current issues of ACP Journal Club are published in Annals of Internal Medicine


Divalproex was an effective agent for migraine prophylaxis

ACP J Club. 1995 Sept-Oct;123:46. doi:10.7326/ACPJC-1995-123-2-046

Source Citation

Mathew NT, Saper JR, Silberstein SD, et al. Migraine prophylaxis with divalproex. Arch Neurol. 1995 Mar;52: 281-6.



To investigate the effectiveness and safety of divalproex in patients with migraine headache.


Randomized, double-blind, placebo-controlled trial with a 16-week follow-up.


8 headache or neurology clinics in the United States.


107 patients aged 16 to 75 years (mean age 45 y, 78% women) with migraine episodes of ≥ 6 months' duration and ≥ 2 episodes/mo in the previous 3 months who were not receiving any prophylaxis or had failed ≤ 2 trials of established prophylactic antimigraine regimens. Exclusion criteria were only migraine episodes not associated with headache; chronic daily headaches or tension-type headaches occurring > 15 d/mo; cluster headaches; history of medical or psychiatric disorder; poor compliance; or previous valproate use. Women of childbearing potential were also excluded. Follow-up was 84%.


During the baseline phase (4 weeks), patients received 1 placebo tablet/d. For a 4-week dose-adjustment phase, patients were allocated in a 2:1 ratio schedule to divalproex sodium, 250 mg/d titrated upward in increments of 250 mg every other day to a trough plasma valproate sodium concentration of 70 to 120 mg/L (n = 70), or to a similarly adjusted dose of placebo (n = 37). During the 8-week maintenance phase, the dosage remained constant unless adverse events necessitated a reduction.

Main outcome measures

The primary outcome measure was the mean number of headaches per 4-week period during the dose-adjustment and maintenance phases (treatment phase). Secondary outcomes were reduction of ≥ 50% from baseline in 4-week headache frequency; mean duration of episodes; mean severity at peak intensity; mean severity relating to functional restriction; mean symptomatic medication use per episode; 4-week headache frequency associated with nausea, vomiting, photophobia, and phonophobia; and the number of days per 4 weeks with migraines.

Main results

Patients receiving divalproex had lower mean 4-week migraine frequency than patients receiving placebo (3.5 vs 5.7, P ≤ 0.001). More patients receiving divalproex had reductions of ≥ 50% in 4-week headache frequency than those receiving placebo (P < 0.001) (Table). Compared with patients receiving placebo, those receiving divalproex had less severity related to functional restriction (P ≤ 0.05), fewer days during which symptomatic medication was required (1.3 d vs 1.7 d, P ≤ 0.05), less vomiting and phonophobia (P ≤ 0.05 for both), and fewer days with headaches (3.9 d vs 6.2 d, P ≤ 0.01). Patients receiving divalproex had more nausea, asthenia, somnolence, vomiting, tremor, and alopecia than those receiving placebo. Symptoms were mild or moderate in severity.


Compared with placebo, divalproex was effective and safe in providing relief to patients with migraine headaches.

Source of funding: In part, Abbott Laboratories.

For article reprint: Dr. N.T. Mathew, Houston Headache Clinic, 1213 Hermann Drive, Suite 350, Houston, TX 77004, USA. FAX 713-526-6369.

Table. Divalproex vs placebo for migraine headache*

Outcome at 16 wk Divalproex Placebo RBI (95% CI) NNT (CI)
Reduction of ≥ 50% in 4-wk headache frequency 48% 14% 244% (59 to 711) 3 (2 to 6)

*Abbreviations defined in Glossary; RBI, NNT, and CI calculated from data in article.

Updated commentary

A surprisingly large number of drugs commonly used for prophylaxis of migraine are not formally approved for that purpose by the Food and Drug Administration. Interest in the development of new headache medications and approval of existing medications, however, is increasing. Divalproex, already approved as an antiepileptic agent, has benefited from this attention. Widely valued by headache specialists on the basis of impressive anecdotal evidence of efficacy and early trial results, divalproex therapy was initially reserved for patients refractory to more standard therapies. The positive results of this large-scale trial by Mathew and colleagues were instrumental in encouraging divalproex use earlier in the course of migraine treatment.

Where does divalproex fit in the clinical treatment of headache? This and other studies establish its superiority over placebo in migraine prophylaxis but provide no information on its efficacy compared with that of standard drugs. We also have little information about efficacy in chronic daily headache or cluster headache, although considerable clinical enthusiasm exists for its use in both conditions. Initial worries about the potential for divalproex to cause liver function test changes have not proved clinically relevant. Many patients benefit with doses lower than that used in this study (e.g., 250 mg twice daily). In these patients, regular monitoring of serum drug levels or liver function tests is not necessary. Divalproex is still not a first-line agent for migraine treatment because less expensive and less dangerous alternatives exist. β-blockers and low-dose tricyclic antidepressants are well-tolerated, effective treatments for migraine that do not increase, as does divalproex, the risk for neural tube defects in exposed fetuses. The latter is an important consideration in the population requiring migraine prophylaxis, which is mainly women of childbearing age. I would reserve divalproex from patients who fail treatment with β-blockers or tricyclics, but would use divalproex before using medications that are potentially more dangerous, such as methysergide.

Finally, realistic expectations of benefit are essential in treating a chronic illness such as migraine. The goal of treatment is the reduction in duration, severity, or frequency of headache episodes, not complete elimination. In this trial, 48% of patients showed a 50% reduction in migraine frequency; in clinical practice, with less compliant, more complicated patients, results will be less impressive. The most common reason for failure of a prophylactic agent, however, is still the failure to use an adequate dose of the drug for an adequate length of time (8 to 12 weeks).

Elizabeth Loder, MD
Spaulding Rehabilitation HospitalBoston, Massachusetts, USA