How Rare Disease Trials Compare to External Studies: Rolling Baseline, Re-Baseline, and Visit Window Mapping

 

Rare disease trials often cannot align perfectly with external studies or natural history data.
As a result, sponsors use pragmatic alignment techniques — but FDA expects them to be prespecified, justified, and sensitivity-tested.

---

1️⃣ Rolling Baseline

What it is (plain English)

A rolling baseline allows each patient’s “baseline” to be defined relative to their own treatment start, rather than at a single fixed calendar time.

Baseline is anchored to treatment initiation, not a universal study day.

---

Why it’s used in rare disease

Rare disease realities:

• Slow recruitment

• Staggered enrollment

• Variable disease progression

• No single synchronized baseline date

Rolling baseline allows:

• Each patient to contribute data even if enrolled months apart

• Alignment with external controls who start follow-up at different times

---

Example

Single-arm rare disease trial

• Patient A starts treatment in Jan

• Patient B starts treatment in June

Baseline definition:

• Baseline = last assessment within X days before first dose

• Not “Day 0 = Jan 1 for everyone”

External natural history study

• Patients enter registry at different disease stages

• Rolling baseline lets you align:

  • “Time since treatment start” vs

  • “Time since matched disease milestone”

---

Regulatory risk

⚠️ Rolling baseline can introduce bias if:

• Disease is rapidly progressive

• Baseline assessments are far apart in time

• External controls deteriorate faster before alignment

👉 FDA expects:

• Tight baseline windows

• Clear clinical justification

• Sensitivity analyses

---

2️⃣ Re-Baseline

What it is

Re-baseline means resetting baseline at a new clinically meaningful time point, instead of the original study baseline.

You deliberately ignore the original baseline and define a new one.

---

Why it’s used in rare disease

Common scenarios:

• Open-label extensions (OLE)

• Delayed treatment start

• Crossover from placebo to active

• Comparing treated patients to external natural history cohorts

Re-baseline aligns:

• Treated patients’ “start of exposure”

• External controls’ “start of observation”

---

Example

Rare neuromuscular disease

• Patients observed untreated for 1 year

• Then all receive gene therapy

To compare with natural history:

• Re-baseline at first dose

• Ignore pre-treatment trajectory

• Analyze post-treatment change only

External control

• Natural history patients indexed at same disease milestone (e.g., loss of ambulation risk)

---

Regulatory risk

⚠️ Re-baseline can:

• Discard informative pre-treatment data

• Favor treatment if disease worsens rapidly pre-baseline

👉 FDA expects:

• Clear estimand definition

• Justification for discarding earlier data

• Sensitivity analyses using original baseline

---

3️⃣ Visit Window Mapping

What it is

Visit window mapping aligns non-identical visit schedules between studies.

You map external control visits to protocol-defined windows.

---

Why it’s unavoidable with external controls

Rare disease comparisons often involve:

• Trial visits at fixed weeks (e.g., Week 24, 48, 96)

• External data collected irregularly (registries, EHRs)

Visit window mapping:

• Assigns external observations to predefined windows

• Allows comparison at “approximately the same time”

---

Example

Trial

• Week 48 ± 4 weeks

External registry

• Visits at months 10 and 14

Mapping rule:

• Month 10 → Week 48 window

• Month 14 excluded or mapped to later window

---

Regulatory risk

⚠️ Risks include:

• Outcome-dependent visit timing

• Informative visit frequency (sicker patients seen more often)

• Asymmetric window widths

👉 FDA expects:

• Prespecified windows

• Same rules applied to both arms

• Sensitivity analyses with alternative windows

---

4️⃣ How They Compare (Quick Table for Teaching)

Concept	What it Aligns	Why Used	Key Risk
Rolling baseline	Baseline timing	Staggered enrollment	Baseline drift
Re-baseline	Start of follow-up	Delayed treatment	Data loss, bias
Visit window mapping	Assessment timing	Irregular visits	Informative timing

---

5️⃣ Why FDA Tolerates These in Rare Disease

FDA understands that:

• Randomization may be infeasible

• External controls are sometimes the only option

• Perfect alignment is impossible

BUT FDA requires:

• Prespecification

• Clinical justification

• Transparent assumptions

• Sensitivity analyses

• Conservative interpretation

These techniques are risk-mitigation tools, not bias eliminators.

References:

FDA (2023) Externally Controlled Trials Guidance;
FDA (2019) Rare Diseases: Natural History Studies;
ICH E9(R1); ICH E10;
Suissa (2008); Hernán & Robins (2020)