Beyond the mean: Rethinking how we should describe clinical outcomes

Clinical research in orthopaedics and rehabilitation often relies on mean-based statistical analyses to describe treatment effects and outcomes. However, many typical outcomes, such as pain, mobility, function, recovery time, and patient-reported outcomes, are bounded, skewed, zero-inflated, or ordinal and exhibit significant heterogeneity among patients. In these cases, a single average can hide important differences in recovery, variability, response, and the proportion of patients with meaningful improvement. This is even more important in the era of personalised medicine. This editorial argues for a shift from viewing outcomes as single numbers to viewing them as distributions better reflecting the complexity of clinical recovery. Using examples from orthopaedics and rehabilitation, the editorial shows how mean-based summaries can obscure key data features such as heterogeneous responses, ceiling and floor effects, hierarchical outcomes, and clinically relevant variability. It discusses why common statistical methods may fail when outcomes deviate from their assumptions and emphasises understanding the nature of the outcome before choosing an analytical model. To facilitate this perspective, the editorial presents a proposal for a practical framework that guides researchers through three stages of outcome analysis: understanding the distribution and clinical meaning of the outcome before modelling, selecting analytical approaches that match its statistical properties, and reporting results in ways that better support clinical decision-making. Highlighted examples of methodological approaches suited to different outcome structures include hurdle and zero-inflated models, beta regression, generalised additive models for location, scale and shape, quantile regression, mixed-effects models, compositional methods, and threshold-based reporting. Rather than advocating the replacement of traditional statistical methods, the researchers are encouraged to align analytical strategies with the characteristics of the data and the clinical questions being asked. Moving beyond the mean can improve the interpretability, clinical relevance, and translational value of research findings by better reflecting how patients actually experience recovery.