Get access

Reporting outcomes in clinical trials for bipolar disorder: a commentary and suggestions for change


  • AstraZeneca provided funding to Parexel, a medical communications company, for drafting an earlier version of this article; however, this new, twice revised and substantially changed version is written by the authors themselves with no monetary compensation. EV and KNRC have received research grant support and have been consultants to AstraZeneca. JM and BP are employees of AstraZeneca. AM-A and SG have no conflicts of interest to disclose in connection with this article. KNRC and SG, as co-editors of Bipolar Disorders, did not participate in any editorial decisions to accept or reject this paper or in choosing any referees, which was overseen by a senior editorial board member who had no conflicts of interest in adjudicating the manuscript.

KN Roy Chengappa, MD, Professor of Psychiatry, Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, 3811 O’Hara Street, Pittsburgh, PA 15213-2593, USA. Fax: +1 412 246 5007; e-mail:


Objective:  Newer outcome measures and statistical reporting that better translate efficacy data to evidence-based psychiatric care are needed when evaluating clinical trials for bipolar disorder. Using efficacy studies as illustrations, the authors review and recommend changes in the reporting of traditional clinical outcomes both in the acute and maintenance phases of bipolar disorder.

Methods:  Definitions of response, remission, relapse, recovery, and recurrence are reviewed and recommendations for change are made. These suggestions include reporting the numbers needed to treat or harm (NNT or NNH), and a ratio of the two, likelihood of help or harm (LHH), as an important element of the effect size (ES). Moreover, models of prediction that conduct sensitivity or specificity analyses and utilize decision trees to help predict positive and negative outcomes of interest (for instance, excessive weight gain, or time to remission) using positive or negative predictive values (PPV or NPV) are reviewed for potential value to clinicians. Finally, functional and cognitive assessments are recommended for maintenance studies of bipolar disorder.

Results:  The examples provided in this manuscript underscore that reporting the NNT or NNH, or alternative effect sizes, or using PPV or NPV may be of particular value to clinicians. Such reports are likely to help translate efficacy-driven clinical data to information that will more readily guide clinicians on the benefits and risks of specific interventions in bipolar disorder.

Conclusions:  The authors opine that reporting these newer outcomes, such as NNT or NNH, area under the receiver operating curve (AUC), or PPV or NPV will help translate the results of clinical trials into a language that is more readily understood by clinicians. Moreover, assessing and evaluating functional and cognitive outcomes will not only inform clinicians about potential differences among therapeutic options, but likely will make it easier to communicate such differences to persons with bipolar illness or to their families. Finally, we hope such scientific and research efforts will translate to optimism for recovery-based outcomes in persons with bipolar disorder.