Patient centricity has become one of the most widely shared ambitions in clinical research. The industry increasingly recognises that engaging patients early in development leads to more meaningful therapies, more efficient trials, and ultimately better outcomes for all stakeholders. Translating this recognition into how trials are structured and how treatment effects are measured is an area where methodology continues to evolve, and where more innovative and structured approaches have a clear role to play.
Why Traditional Trial Design Can Fall Short
Randomised trials are typically designed around a single primary endpoint, a structure well-suited to answering one confirmatory question with statistical robustness and efficiency. Secondary endpoints add supportive evidence but, by design, do not contribute to the primary evaluation of treatment effect. This is appropriate when a single dimension adequately captures what matters clinically but may not always reflect the multidimensional nature of patient experience. In many settings, and particularly in complex or chronic conditions where benefits and risks trade-off are inherent, the decision-relevant question is inherently heterogeneous. Addressing this gap requires a methodological framework that can systematically integrate several patient-relevant outcomes into a coherent assessment of treatment effect.
Patients may value improvements in symptoms, preservation of daily functioning, or avoidance of specific adverse effects differently than clinicians or regulators. A trial that focuses narrowly on one outcome risks missing these nuances, leading to results that are statistically significant but less informative for the decisions patients themselves face.
Regulatory initiatives have reinforced the importance of incorporating patient perspectives into drug development. Efforts such as the FDA’s patient-focused drug development series and recent guidance on the use of patient preference information across the product lifecycle highlight a growing consensus that what patients value should inform decision-making throughout all clinical development. These developments create a context in which more structured approaches to integrating multiple outcomes are not only scientifically appealing but increasingly expected.
Incorporating Multiple Outcomes into a Single Assessment with Net Treatment Benefit
The Net Treatment Benefit (NTB) framework provides such an approach. It is based on Generalised Pairwise Comparisons (GPC), a statistical method applied to randomised clinical trials in which every patient in the experimental arm is compared with every patient in the control arm across a prioritised list of outcomes.2 These comparisons are performed sequentially across a prioritised list of outcomes. For each pair, the most important outcome is assessed first, and if a difference is observed, the comparison reflects whether the experimental treatment is better or worse. If no meaningful difference is observed, the comparison proceeds to the next outcome in the hierarchy. This process continues across all prioritised outcomes, allowing multiple dimensions of benefit and risk to be evaluated within a single framework. The overall result estimates the Net Treatment Benefit, a metric that reflects the net probability that a randomly selected patient on the experimental treatment did better than a randomly selected patient on control, according to the full hierarchy of outcomes.
A defining feature of this approach is that benefit and risk are assessed together rather than in isolation. Instead of reporting efficacy and safety separately and leaving their integration to post-hoc judgment, the framework integrates them based on their relative importance. This requires that outcomes are not only selected carefully but also prioritised in a way that reflects their clinical relevance. The prioritisation step is critical, as it determines how different aspects of treatment effect contribute to the overall evaluation. It also introduces a level of transparency that is often missing in traditional analyses, as stakeholders can understand how the structure of the endpoint influences the final result.
