Target trial emulation is a design and analysis approach for observational data that first specifies the research question as a hypothetical randomized trial. Only once inclusion criteria, treatment strategies, allocation time, follow-up, endpoint, causal contrast, and analysis plan are established, is it described how these components are replicated using routine data. This approach reveals which parts of the desired comparison are actually observable in the data and where assumptions become necessary.
The Study Protocol Precedes Data Analysis
The starting point is not an existing dataset with numerous variables, but rather a protocol for the target trial. This protocol defines which individuals would be eligible at what baseline time point, which treatment regimens are compared, and which event counts within what risk period. A uniform time zero for both strategies is particularly important: individuals who can only start the investigational therapy later must not be considered treated from the time of cohort entry.
The specification also names the covariates relevant for allocation and outcome, the censoring rules, and the target measure, such as a difference in risk at a defined time point. For repeated treatment opportunities, emulation can generate several consecutive “copies” of suitable individuals and assign them to the strategies possible at that time. This separates the question of “start now or not” from a retrospective classification based on a later prescription.
Translation into Real-World Data
For each protocol component, an operational equivalent must be defined. Diagnoses, prescriptions, laboratory values, and treatment discontinuations are not treated as self-explanatory but are defined with codes, capture windows, and priority rules. For example, a prescription documented in a record may mean something different from a therapy actually dispensed or taken; the chosen definition must align with the intended treatment strategy.
The subsequent evaluation addresses the lack of random allocation in non-randomized data. Measured baseline covariates are considered for comparability, and deviations from a strategy or loss to follow-up are handled according to pre-defined rules. However, target trial emulation does not eliminate unobserved confounding structures and does not replace plausible temporal order or sufficient measurement of relevant clinical factors.
Furthermore, multiple permissible treatment starts require a rule against selective choice of the most favorable time point. In a sequence of monthly decision points, the same eligible individual is therefore mapped for the strategies still possible at each point, while the analysis correctly accounts for the resulting reuse. This construction transforms a vague question about “treated patients” into a comparison of clearly dated treatment decisions.
Distinction: Not a Separate Study Design
Target trial emulation is not a separate study design and does not transform a database analysis into a randomized clinical trial. It is an analysis strategy for real-world data that aligns the planning of an observational study with an explicit counterfactual design. Thus, it complements the existing terms real-world data, real-world evidence, observational study, and non-interventional study, without being synonymous with them.
Its concrete benefit lies in identifying avoidable design flaws early. For example, if the treated group is only classified as treated after an event-free interval, the target trial protocol points out the resulting immortal time bias. The solution is not a general “data cleaning” but a consistent definition of eligibility, index date, exposure, and risk period for each comparative strategy.
Relevance for clinical trials
Therefore, in studies on healthcare practice, the protocol must reflect the clinical decision-making situation and not just an available data variable. This concerns, for example, the permissible pre-treatment status, the start of follow-up for each regimen, the assignment of competing therapies, and the recording of changes. For a non-interventional study to be acceptable to regulatory authorities, it must be comprehensible what information was available at the index date and which variables only became known through the subsequent course.
Full-service CROs like Mediconomics support target trial emulations by developing an RWE protocol, defining cohort algorithms and endpoints, coordinating the variable catalog and statistical analysis plan, and verifying the temporal data logic. Biostatistics and data management can thus document rules for index dates, exposure windows, censoring, and sensitivity analyses before the start of outcome analysis.
Frequently Asked Questions (FAQ)
What information must a target trial protocol contain at a minimum?
Required elements include eligibility criteria, the compared treatment strategies, the allocation mechanism, time zero, follow-up, the endpoint, the causal contrast, and the analysis plan. For each point, a concrete implementation within the available data must then be specified.
Can target trial emulation replace randomization?
No. It can structure an observed comparison such that its assumptions are transparent and measured confounders are specifically accounted for. Unobserved factors that influence both treatment choice and outcome remain a potential source of biased effect estimates.
Why is time zero so crucial?
It determines when individuals are exposed to risk and what information may be used for assignment to a strategy. If the start of exposure and the start of follow-up are defined differently, a strategy may be attributed event-free time that it has not earned.
Regulatory References
- FDA, Real-World Evidence: Considerations Regarding Non-Interventional Studies for Drug and Biological Products – names index date, risk period, covariates, and bias strategies as protocol elements.
- ICH E9(R1), Addendum on Estimands and Sensitivity Analysis in Clinical Trials – classifies the precise definition of the clinical question and the estimand.
- ICH E10, Choice of Control Group and Related Issues in Clinical Trials – describes the role of a clearly defined comparative regimen.