MMRM
Also written as: MMRM — Mixed Model for Repeated Measures
A statistical analysis method for longitudinal data that accounts for missing observations without imputation.
Full Definition
Mixed Model for Repeated Measures (MMRM) is a statistical approach for analyzing longitudinal clinical trial data that uses all available observations without requiring imputation for missing data points. MMRM models assume that missing data are missing at random (MAR) and use maximum likelihood estimation to include all participants with at least one post-baseline assessment. This method has become the preferred approach in psychiatric clinical trials, replacing older techniques like Last Observation Carried Forward (LOCF), because it provides more accurate estimates and appropriate handling of uncertainty due to missing data. MMRM is particularly valuable in psychiatry where dropout rates can be substantial and missing data patterns complex.
Usage
Usage note: Spell out on first use, then use MMRM. Emphasize advantages over LOCF when appropriate.
In Context
- "The primary analysis employed MMRM with an unstructured covariance matrix to handle missing data." — statistical analysis plan
- "MMRM results were consistent with the LOCF analysis but showed more conservative treatment effects." — results interpretation