NYMC Faculty Publications

Assessing the Benefit-Risk Ratio of Approved Treatments for Bipolar Depression Using Likelihood to Be Helped or Harmed (LHH) Analyses

Author Type(s)


Journal Title

CNS Spectrums

First Page


Last Page


Document Type


Publication Date



Psychiatry and Behavioral Sciences



Four medications are FDA approved for bipolar depression: lurasidone (LUR), cariprazine (CAR), quetiapine IR & XR (QUE), and olanzapine-fluoxetine combination (OFC). Indirect comparisons for efficacy using Number Needed to Treat (NNT) and for tolerability using Number Needed to Harm (NNH) can be useful clinical benchmarks to aid treatment decisions. Benefit and risk may also be examined using the Likelihood to be Helped or Harmed (LHH). In this post-hoc analysis, we examined the benefit-risk ratio of the four treatments using LHH. Method

Individual and pooled monotherapy data from short-term clinical registration trials of patients with bipolar depression were assessed for LUR, CAR, pooled QUE (300 and 600 mg), and pooled OFC (considered as monotherapy for this study at fixed doses of 6/25, 6/50, 12/50 mg) data. NNT estimates were calculated using the proportions of MADRS responders (defined as ≥ 50% improvement at study endpoint) and MADRS remitters (defined as a score of ≤ 10 [for LUR and CAR] and ≤ 12 [for QUE and OFC]) at study endpoint. NNH data were calculated for the proportions of patients who discontinued due to an adverse event (AE) and for individual AEs commonly associated with each treatment. LHH was calculated as the ratio of NNH/NNT to determine the benefit-risk ratio. Results

The NNT estimates for response vs. placebo were: 5 for both LUR 20–60 mg and 80–120 mg; 10 for both CAR 1.5 mg and 3.0 mg; 6 for QUE; and 4 for OFC. The NNTs for remission vs placebo were: 7 for LUR 20–60 mg and 9 for LUR 80–120 mg; 10 for CAR 1.5 mg and 13 for CAR 3.0 mg; 6 for QUE; and 5 for OFC. The NNH estimates for discontinuations due to AEs were: 642 for LUR 20–60 mg and −151 for LUR 80–120 mg; 298 for CAR 1.5 mg and 31 for CAR 3.0 mg; 10 for QUE; and −37 for OFC. NNH values that were negative were assigned a value of 1000 to permit LHH to be calculated. The LHHs for response vs discontinuation due to an AE were: 128.4 for LUR 20–60 mg and 200 for LUR 80–120 mg; 29.8 for CAR 1.5 mg and 3.1 for CAR 3.0 mg; 1.7 for QUE; and 250 for OFC. The LHHs for response vs akathisia were: 3.6 for LUR 20–60 mg and 2.4 for LUR 80–120 mg; 3.6 for CAR 1.5 mg and 1.3 for CAR 3.0 mg; 34 for QUE; and not available (NA) for OFC. The LHHs for response vs EPS were: 8 for LUR 20–60 mg and 3.2 for LUR 80–120 mg; 5 for CAR 1.5 mg and 2.5 for CAR 3.0 mg; NA for QUE; and NA for OFC. The LHH for response vs weight gain was 5.8 for LUR 20–60 mg and 1110 for LUR 80–120 mg; 5 for both doses of CAR; 2.7 for QUE; and 1.5 for OFC. Conclusions

LHH can illustrate the trade-offs regarding potential benefits versus potential harms. Across a variety of measures, the lower-dose groups for both LUR and CAR generally evidenced a better benefit-risk profile than the higher-dose groups. While quetiapine and OFC demonstrated robust efficacy, their reduced tolerability resulted in a more marginal benefit-risk ratio for some of the outcomes.