realm of real-world evidence (RWE) and observational studies, the use of external controls and synthetic arms has gained significant traction. These methodologies offer a viable alternative to traditional controlled trials, particularly in situations where randomization is not feasible. However, the introduction of these approaches does not come without challenges, primarily in the form of biases.

This tutorial guide provides a comprehensive exploration of common biases encountered in the use of external controls and synthetic arms, alongside strategies for mitigating these biases. Our focus will primarily be on the US, UK, and EU regulatory environments, as these regions have prominent guidelines and frameworks for addressing these methodological challenges.

Understanding External Controls and Synthetic Arms

External controls refer to the use of historical data derived from previous studies or real-world evidence to establish a comparative baseline for evaluating the efficacy and safety of a new intervention. In contrast, synthetic arms involve creating a control group using data from existing databases or registries, which mirror the characteristics of a traditional clinical trial arm. This approach is particularly relevant when traditional patient populations are difficult to assemble due to ethical considerations or logistical constraints.

Research professionals, including principal investigators in clinical trials, must understand the nuances of these approaches to ensure their studies meet regulatory scrutiny and scientific rigor.

Challenges Posed by Biases in External Controls and Synthetic Arms

The implementation of external controls and synthetic arms is fraught with potential biases that can compromise the integrity of clinical trial outcomes. Major biases include:

• Selection Bias: This occurs when the comparative groups are not adequately balanced in terms of prognosis or characteristics. For example, if a historical cohort used as an external control differs significantly from the current trial participants, the results may not be generalizable.
• Confounding Bias: This type of bias arises when external factors influence both the treatment and outcome, potentially leading to misleading associations. The lack of randomization in synthetic arms exacerbates this issue.
• Attrition Bias: Involves systematic differences between those who complete the study and those who drop out. This can alter the perceived efficacy of the treatment under investigation.
• Measurement Bias: Occurs when the data collection methods differ between the external control and the experimental group, impacting the validity of outcomes.

Step 1: Identify Potential Biases Early

The first line of defense against biases is identification. Prior to study initiation, researchers should conduct a thorough risk assessment to identify potential sources of bias associated with the use of external controls or synthetic arms. This includes evaluating:

• The comparability of cohorts: Use statistical methods to compare baseline characteristics.
• Data sources: Assess the reliability and validity of historical data or existing databases used for creating synthetic arms.
• Analytical methods: Plan out the statistical techniques that will be employed to account for confounding factors.

Utilizing platforms like ClinicalTrials.gov can provide researchers with invaluable insights into historical data and efficacy outcomes from similar trials, aiding in the mitigation of biases from the outset.

Step 2: Employ Advanced Statistical Techniques

To correct for biases, researchers should implement advanced statistical techniques in their study design and analysis. Techniques may include:

• Propensity Score Matching: This method helps reduce selection bias by matching participants in the treatment group with similar participants in the external control group based on observed characteristics.
• Inverse Probability Weighting: Weighting outcomes based on the probability of receiving treatment can help to balance the groups regarding confounders.
• Bayesian Methods: These are particularly useful for synthesizing data from external sources and accounting for uncertainty in predictions.
• Multivariable Regression Analysis: This analysis allows for adjustment of multiple confounding variables, enhancing comparison between external controls and treated groups.

These techniques will not only increase the robustness of the data but also align with the stringent standards set forth by regulatory bodies in the US (e.g., FDA) and Europe (e.g., EMA, MHRA).

Step 3: Continuous Monitoring and Interim Analysis

Incorporating interim analysis into the study design is critical, particularly in platform clinical trials, where multiple interventions are assessed concurrently. Continuous monitoring allows for early detection of potential biases that may arise as additional data is collected. Key considerations during this phase include:

• Data Monitoring Committees (DMCs): Establish independent committees to oversee trial progress, identify biases, and ensure data integrity.
• Preplanned Analyses: Establish interim analyses to assess whether pre-specified thresholds are met for efficacy or safety, allowing for modifications to the trial if biases are identified.
• Transparency: Maintaining transparency with stakeholders regarding findings at each stage fosters trust and adherence to regulatory expectations.

Step 4: Documenting and Reporting Methods for Bias Correction

Robust documentation and transparent reporting of methodologies employed to address bias are essential for the credibility of the trial. According to ICH-GCP guidelines, clinical investigators should detail:

• The rationale for using external controls or synthetic arms, including their relevance to the research question.
• Specific statistical techniques used for bias correction and the reasons for their selection.
• Any limitations of the methods employed in addressing biases, ensuring that findings are interpreted accurately.

These practices not only facilitate regulatory review but also enhance the trustworthiness of the findings published in scientific literature.

Conclusion

Correcting biases in external controls and synthetic arms is crucial for the validity of clinical trials. A thorough understanding of potential biases, coupled with the implementation of advanced statistical techniques, interim analyses, and strict documentation practices, can enhance the reliability of results. As clinical operations, regulatory affairs, and medical affairs professionals engage in trials like astellas clinical trials, awareness of these biases will contribute to more responsible use of RWE and observational studies.

Continued education on these methodologies alongside regulatory developments will promote better research practices and outcomes in the future. For further information on regulatory expectations regarding external control and synthetic arms, consult the relevant resources from FDA, EMA, and MHRA.