Randomized controlled trials (RCTs) remain the gold standard for assessing the efficacy or safety of a given medication or intervention, with RCT data at the top of the hierarchy of evidence. By design, the randomization process eliminates the risk of selection bias and ensures that RCTs are unconfounded, thereby avoiding important sources of bias that are particularly challenging to control for in observational studies. However, conducting an RCT might not always be feasible owing to various constraints — the study might be unethical, too expensive, or too complex to design and monitor. RCTs might also lack generalizability, which is an important barrier in nephrology because patients with reduced kidney function are often excluded from trials. In these situations, observational studies represent an alternative, making use of the increasing amount of routinely collected administrative and health-care data and benefiting from the advantage of being implemented in real-world settings, thereby increasing their generalization potential. However, observational studies are more susceptible to confounding effects and bias than RCTs.

By applying the design of an RCT to observational data, the target trial emulation (TTE) framework aims to get the best of both worlds and limit the introduction of biases that often affect observational research, such as the prevalent-user bias and immortal-time biases. This approach is being increasingly and successfully used, including in kidney research1,2. Emulating a trial from observational data essentially involves writing the protocol of a hypothetical trial that would answer the research question, explicitly specifying each one of its components (for instance, eligibility criteria, treatment strategies and modalities of treatment assignment). This protocol is useful to guide the study design and emulate each component of the trial using observational data. The key feature is the alignment of eligibility criteria, treatment assignment and start of follow-up, as would be the case in a RCT at randomization3.