Published on

May 12, 2026

Last updated on

May 12, 2026

Taiwan FDA Introduces ICH E9(R1) Estimand Framework into Clinical Trial Review

A brightly lit laboratory scene shows two scientists conducting research. In the foreground, a woman wearing clear safety goggles, blue nitrile gloves, and a white lab coat peers into a microscope with focused concentration.

On April 15, 2026, the Taiwan Food and Drug Administration (TFDA), under the Ministry of Health and Welfare, published an announcement dated April 14, 2026 (TFDA document No. 1151403103 [FDA藥字第1151403103號] adopting “ICH E9 (R1): Estimands and Sensitivity Analysis in Clinical Trials — Addendum to the Guideline on Statistical Principles for Clinical Trials” as guidance for drug clinical trial management in Taiwan.

The update further aligns Taiwan’s clinical trial framework with internationally harmonized regulatory expectations and reinforces a broader shift already visible across major regulatory markets: greater scrutiny of whether submitted trial evidence actually answers the clinical question a study was designed to address.

The Strategic Shift Behind ICH E9(R1)

ICH E9(R1) is often discussed as a statistical guidance update. In practice, however, its significance is broader.

The framework changes how sponsors are expected to define treatment effect, interpret endpoints, manage intercurrent events, and connect protocol design with downstream regulatory interpretation.

Historically, many development programs focused primarily on endpoint selection, statistical significance, and missing data methodology.

ICH E9(R1 reframes the discussion more fundamentally: What precise clinical question is the study intended to answer, and does the trial design consistently support that objective?

That distinction matters because the same endpoint can support very different interpretations depending on how intercurrent events are handled.

For example, a study may seek to evaluate:

Treatment effect regardless of rescue medication
Treatment effect assuming rescue medication had not been used
Whether rescue medication itself should be interpreted as treatment failure

Each approach reflects a different clinical objective and can materially affect interpretation of efficacy and benefit-risk.

For sponsors, the practical implication is that estimands increasingly influence:

Protocol strategy
Operational follow-up
Endpoint interpretation
Statistical analysis planning
Regulatory positioning from early development onward

Internal Consistency Across the Evidence Chain

One of the core concepts introduced by ICH E9(R1) is alignment between:

Trial objective
Estimand
Primary estimator
Primary estimate
Sensitivity analyses

The regulatory significance is not the terminology itself, but the expectation that these elements remain internally consistent throughout the development program.

A protocol that list endpoints and statistical test without clearly defining the estimand may leave unresolved questions around:

What treatment effect is actually being estimated
How intercurrent events affect interpretation
Whether the primary analysis meaningfully answers the intended clinical question

Increasingly, regulators are assessing not only whether statistical methodology is technically valid, but whether the evidence package as a whole remains logically coherent from objective through interpretation.

Intercurrent Events Are Becoming a Central Regulatory Focus

The addendum also formalizes the distinction between intercurrent events and missing data; a point with important operational implications.

Intercurrent events occur after treatment initiation and affect either interpretation of the measurement or existence of the measurement itself.

Examples may include:

Treatment discontinuation
Rescue medication
Treatment switching
Death

Missing data, by contrast, refers to information relevant to the estimand that was not collected.

Operationally, this distinction matters because discontinuation or switching should not automatically be treated as statistical missing-data problems. In many cases, they are meaningful events that require explicit handling within the estimand strategy itself.

As a result, sponsors are increasingly expected to prospectively define:

which intercurrent events are anticipated,
how they affect the clinical question,
and whether continued follow-up remains necessary after the event occurs.

This has direct implications for protocol design, data collection strategy, follow up procedures, and operational trial execution.

The Five Estimand Strategies

ICH E9(R1) outlines five strategies for handling intercurrent events, with selection expected to reflect the underlying clinical objective.

Treatment Policy Strategy: Evaluates the treatment effect regardless of whether events occur and may reflect treatment performance under real-world conditions.
Hypothetical Strategy: Estimates outcomes under an assumed scenario in which the intercurrent event did not occur.
Composite Strategy: Incorporates the intercurrent event into the endpoint itself, such as counting discontinuation due to toxicity as treatment failure.
While-on-Treatment Strategy: Focuses on outcomes before the intercurrent event occurs.
‍Principal Stratum Strategy: Evaluates treatment effects within subgroups defined by whether certain events would occur under specified treatment conditions.

Importantly, the framework emphasizes that strategy selection should be driven by clinical relevance and scientific rationale rather than analytical convenience.

Sensitivity Analysis Is Increasingly Tied to Regulatory Credibility

The guidance also elevates the role of sensitivity analysis within regulatory decision-making. Sensitivity analyses are intended to evaluate whether conclusions remain robust when assumptions underlying the primary analysis are challenged.

This becomes especially relevant where assumptions related to:

rescue medication,
discontinuation,
treatment switching,
post-discontinuation follow-up, and
uncollected but clinically meaningful data

could materially influence the interpretation of treatment effect.

From a regulatory perspective, the expectation is increasingly not simply that alternative analyses are presented, but that sponsors can clearly explain:

Which assumptions underpin the primary analysis
Why those assumptions are appropriate
How dependent the conclusions are on those assumptions remaining true

The framework also distinguishes sensitivity analysis from supplementary analysis. Sensitivity analyses test robustness for the same estimand, whereas supplementary analyses provide additional contextual insight beyond the primary interpretive framework.

Implications for Clinical Development Programs

The update is especially relevant in confirmatory clinical trials supporting approval, labeling, or benefit-risk assessment.

At the same time, the implications extend well beyond Taiwan-specific submissions. Many global development programs rely on standardized protocol and analysis templates across regions. Those frameworks may require reassessment if estimands, intercurrent-event handling, and sensitivity-analysis assumptions are not clearly aligned throughout the evidence package.

In practice, sponsors should review whether the following elements remain internally consistent:

Trial objectives
Estimands
Endpoint interpretation
Intercurrent-event handling
Estimator assumptions
Sensitivity analyses
Operational follow-up procedures.

Increasingly, these considerations affect not only statistical analysis planning, but also:

Protocol design
Case report forms
Operational conduct
Data-collection strategy
Downstream regulatory positioning

Considerations for Ongoing and New Studies

For ongoing studies, sponsors should assess whether existing documentation adequately supports the estimand framework and whether intercurrent events and sensitivity analyses are described with sufficient clarity for regulatory review.

The TFDA materials do not specify a formal transition mechanism for ongoing studies. Nevertheless, targeted review may help identify inconsistency between:

Trial objectives
Estimands
Operational procedures
Statistical assumptions
And final interpretation

For new studies earlier alignment between clinical, regulatory, operational, and biostatistics teams becomes increasingly important.

Operationally, this becomes particularly significant where protocols or case report form fails to capture data necessary to support the intended estimand. Once those data are unavailable, resulting evidence gaps may be difficult to resolve during later-stage submission preparation or regulatory review.

Final Thoughts

TFDA’s adoption of “ICH E9(R1)” reflects a broader regulatory movement toward more explicitly defined, clinically interpretable, and decision-oriented trial evidence. The framework raises expectations around the consistency of the full evidence chain; from trial objective and estimand definition through operational conduct, statistical analysis, sensitivity testing, and final regulatory interpretation.

Increasingly, regulatory robustness will depend not only on statistical validity, but also on whether submitted evidence consistently and transparently answers the clinical question the study was designed to address.

For pharmaceutical companies operating across regional and global development programs, this may require closer integration between clinical development, biostatistics, regulatory strategy, and operational execution earlier in the trial-design process.

Companies navigating Taiwan FDA clinical development and submission planning can work with Cisema for regulatory consulting support across Asian development programs and regional regulatory strategy.