Metformin for Overweight and Obese Children With Bipolar Spectrum Disorders Treated With Second-Generation Antipsychotics (MOBILITY): Protocol and Methodological Considerations for a Large Pragmatic Randomized Clinical Trial

Abstract

Youth with bipolar spectrum disorders may experience improved mood stability when treated with second generation antipsychotics (SGAs); however, SGAs are associated with unhealthy weight gain and adverse metabolic effects. Metformin may mitigate this weight gain but is rarely prescribed by community mental health practitioners. Its long-term efficacy, safety, and acceptability in usual care, and factors that might moderate these effects, are unknown. The Metformin for Overweight and Obese Children and Adolescents with Bipolar Spectrum Disorders Treated with Second Generation Antipsychotics (MOBILITY) trial aims to fill these gaps. We present the design and analytic plan of this multi-site, open-label, randomized trial. Patients will be randomized to either metformin plus brief healthy diet and exercise education (MET+LIFE) or to LIFE alone. Up to 1637 patients will be followed for up to 2 years at 64 community and academic mental health treatment facilities. Patients may switch between treatment arms during follow-up. Pragmatic trials place few burdens and constraints on participating patients, families, and clinicians. This flexibility will allow MOBILITY to obtain long-term follow-up in a large, diverse sample, but produces analytic challenges. MOBILITY's flexible design has the potential to generate several novel methodological issues that we address. Some patients randomized to LIFE will fail to lose weight, and therefore metformin initiation contrary to the randomization may result from weight gain. Adherence to medications, SGAs, and lifestyle recommendations as well as satiety are potential time-varying mediators, moderators, or confounders of the effect of metformin. Adherence to metformin and SGAs may be positively correlated; therefore, a beneficial effect of metformin on weight could be obscured by the known SGA adverse effect on body weight. However, such correlation could facilitate causal inference by providing indirect information about unknown metformin adherence among patients who did not receive it. A perceived protective effect of metformin could potentially lead to risk compensation, with poorer diet and activity among those receiving metformin. We discuss limitations of traditional statistical approaches and summarize an advanced methodology ("Targeted Learning") that addresses some of these limitations. Metformin for Overweight & OBese ChILdren and Adolescents With BDS Treated With SGAs (MOBILITY); https://clinicaltrials.gov/; NCT02515773.