Changing prescription patterns for lithium and valproic acid in old age: shifting practice without evidence

BMJ 2003; 326 doi: (Published 03 May 2003) Cite this as: BMJ 2003;326:960
  1. Kenneth I Shulman (ken.shulman{at}, professora,
  2. Paula Rochon, assistant directorc,
  3. Kathy Sykora, biostatisticianb,
  4. Geoffrey Anderson, senior adjunct scientistb,
  5. Muhammad Mamdani, scientistb,
  6. Susan Bronskill, scientistb,
  7. Chau T T Tran, doctoral candidateb
  1. a Department of Psychiatry, Sunnybrook and Women's College Health Sciences Centre, Faculty of Medicine, University of Toronto, Toronto, ON, Canada M4N 3M5
  2. b Institute for Clinical Evaluative Sciences (ICES), Sunnybrook and Women's College Health Sciences Centre
  3. c Kunin-Lunenfeld Applied Research Unit, Baycrest Centre for Geriatric Care, Toronto, ON, Canada M6S 2E1
  1. Correspondence to: K I Shulman
  • Accepted 13 January 2003

Over the past decade, valproic acid (prescribed as divalproex in North America) has been marketed as an alternative to lithium for treating bipolar disorders. For elderly patients, however, there is no clear evidence that valproic acid is more beneficial than lithium. Moreover, the evidence for the superiority of valproic acid in treating bipolar disorders—mixed episodes and rapid cycling—has been challenged in a recent Cochrane review.1 Valproic acid has not benefited patients with manic and psychiatric symptoms in dementia, despite the growing use of the drug in the management of these conditions.2 Recently, the relatively rapid shift in prescription patterns has been questioned.3 We describe trends in the use of lithium and valproic acid in a large population of people over 65.

Methods and results

We obtained information on drug use from the Ontario Drug Benefit Program, which provides comprehensive drug benefits to all residents aged 65 or older in Ontario, Canada. We identified all patients who had been taking lithium or valproic acid between 1993 and 2001 (prevalent users) and we further identified those patients who had not previously taken lithium or valproic acid (new users). We restricted our study to patients aged 66 or more to enable us to examine their previous drug use for a minimum of one year. Using unique encrypted health card numbers, we linked data on this cohort to two other large datasets—the Canadian Institute for Health Information dataset of all hospital separations and the Ontario Health Insurance Plan dataset of all claims for physician services. These datasets provided information on hospitalisations and visits to physicians that helped us distinguish between psychiatric and anticonvulsant uses of lithium and valproic acid and between bipolar disorder and other indications such as dementia.


New elderly users of valproic acid and lithium who did not have a diagnosis of dementia before the incident use, Ontario, 1993-2000

Among patients who had no previous history of convulsive disorders, we identified 3902 patients who had started taking lithium and 5341 patients who had started taking valproic acid between 1993 and 2001. New valproic acid users were slightly older than lithium users (75.4 years v 73.5 years). More than three quarters of new lithium users and fewer than two thirds of valproic acid users had contact with a psychiatrist. Almost one quarter of the new lithium patients and 41% of new valproic acid users had had a diagnosis of dementia.

The number of new lithium users per year fell from 653 older adults in 1993 to 281 in 2001, whereas the number of valproic acid users rose from 183 in 1993 to 1090 in 2001. The trend was similar when we eliminated patients who had had a diagnosis of dementia (figure). The number of new valproic acid users surpassed new lithium users in 1997, with a steady decline in new lithium users and a steady increase in new valproic acid users between 1993 and 2000.


Prescription patterns have shifted in favour of valproic acid over lithium, for elderly patients with bipolar disorder. This shift is occurring in the absence of evidence based data. Lithium carbonate has been a mainstay for the management of bipolar disorders for all age groups, but elderly patients pose a special concern because of the potential for toxicity.4 Before the use of lithium is abandoned for these patients, however, adequate evidence that valproic acid offers a comparable or superior efficacy, effectiveness, and safety profile is needed. Indeed, the concerns about lithium toxicity may reflect a failure to adjust for pharmacokinetic and pharmacodynamic factors that occur in old age.4 For example, renal clearance changes notably with age, and hence also lithium clearance. Better guidelines for dosage and serum concentrations of lithium carbonate for elderly people are needed before clinicians switch to prescribing new agents, especially with the associated risk of relapse that follows the discontinuation of lithium.5 Given the complexity, morbidity, and mortality associated with bipolar disorders and dementia in elderly patients, the use of mood stabilisers and other psychotropic agents requires ongoing systematic evaluation.


Contributors: KIS conceived the study, reviewed the literature, and wrote the initial draft. PR helped to conceive the study, revised the initial and subsequent drafts, and was overseer of the research network. KS analysed the data and developed the figure. GA, MM, SB, and CT contributed to the study design, data analysis, and revisions of the manuscript. KIS is the guarantor.


  • Funding This work was supported by the Canadian Institutes of Health Research Chronic Disease New Emerging Team programme (NET 54010).

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract