Source-backedReimbursement

Real-World Evidence and Registry Design

A registry-design resource for the post-approval evidence questions that trials and economic models rarely settle on their own.

Minimum domains: 6
Advanced domains: 6+
Key limitation: Not randomization

At a glance

What to take from this page

A registry should be built around named uncertainties, not around every variable that could be collected.
Real-world evidence is strongest when questions, definitions, follow-up windows, and analysis plans are defined before routine use starts.
RWE can track durability, retreatment, safety, resource use, equity, and site variation, but it usually cannot replace a weak comparator trial.

Who this helps

Payers

Specify which uncertainties should be monitored after a conditional or staged access decision.

Developers

Prepare post-launch evidence infrastructure before assessment asks for it.

Researchers

Separate minimum registry fields from deeper observational research questions.

Step 1

Evidence

Step 2

Approval

Step 3

Reimbursement

Step 4

Delivery

Step 5

Access

Registry architecture

Collect fields in tiers, tied to named uncertainties

More fields are useful only when providers can collect them reliably and decision-makers know how they will be interpreted.

Minimum

EligibilityBaselineExposureSymptomsSafety

Decision-grade follow-up

RelapseRetreatmentResource useFunctioning

Implementation learning

FidelitySite variationEquityLinked data

Registry field tiers

Tier	Fields	Main use
Minimum	Eligibility, baseline severity, prior treatment, treatment exposure	Define who received care and whether use matches the intended population
Clinical follow-up	Symptoms, response, remission, relapse, retreatment, adverse events	Track persistence, safety, and need for further care
Resource use	Visits, staff time, hospital use, rescue medication, crisis care	Inform budget impact and economic models
Implementation	Site, staff roles, fidelity, protocol deviations, waiting time	Detect delivery variation and bottlenecks
Equity and linkage	Geography, socioeconomic proxies, referral source, claims linkage	Understand who receives access and whether outcomes generalize

Evidence questions and registry design choices

Question	Data to collect	Why it matters
Durability	Common symptom scale at baseline, acute endpoint, 3 months, 6 months, and 12 months where feasible	Trial endpoints are shorter than many HTA model horizons, so observed follow-up constrains waning assumptions.
Retreatment and relapse	Relapse definition, rescue medication, additional therapy, repeat psychedelic exposure, hospital or crisis care	Payers need to know whether an intensive treatment episode is one-off, periodic, or followed by substantial extra care.
Safety	Acute psychological distress, suicidality, cardiovascular events, persisting perceptual symptoms, medication interactions, serious adverse events	Routine care will treat a broader and less selected population than trials.
Service intensity	Preparation, administration, monitoring, integration, follow-up, no-shows, staff mix, room time, and indirect work	Economic models are highly sensitive to staff-hours and the shape of the care pathway.
Fidelity and variation	Protocol version, staff credentials, supervision, deviations, site, and escalation events	Outcomes may depend on delivery quality, not only the medicine.
Equity	Referral source, geography, language, deprivation proxy where lawful, wait time, exclusion reasons, and completion	Access can narrow upstream if referral and screening pathways disadvantage complex or underserved patients.

The registry should be small enough to run and specific enough to answer named uncertainty. More fields are only useful when they can be collected reliably.

Start with the uncertainty, not the dataset

A useful registry begins with the decisions it needs to inform: durability, relapse, retreatment, adverse events, resource use, equity, site variation, and whether the service model is being delivered as intended.

This question-first approach keeps the registry from becoming either too thin for payers or too burdensome for providers.

Minimum viable registry fields

A first-pass registry should usually capture eligibility, baseline severity, prior-treatment history, treatment exposure, symptom outcomes, adverse events, relapse or retreatment, and basic health-resource use.

Those fields do not answer everything, but they create a shared language for whether the therapy is being used in the intended population and whether outcomes persist in routine care.

Advanced fields add payer and implementation value

More mature systems can add functioning, quality of life, productivity, site-level variation, therapist fidelity, protocol deviations, equity indicators, and linkage to claims or hospital data.

These fields are often where the access argument becomes stronger, but they also raise burden, governance, privacy, and data-quality questions.

What the psychedelic literature already points toward

Psychedelic studies already show which outcomes a registry should not ignore: rapid symptom change, later waning or relapse, adverse psychological events, suicidality signals, rescue care, protocol deviations, support intensity, and participant selection. EPIsoDE, Stockholm, COMPASS follow-up, MDMA reviews, and group-format studies each highlight different pieces of that evidence map.

The registry should not only ask whether someone improved after dosing. It should ask who was treated, how much preparation and integration they received, what happened during administration, whether the person relapsed or needed retreatment, and whether the real-world pathway resembles the studied protocol.

Managed-access logic changes the registry brief

A managed-access or staged-reimbursement design should define the uncertainty first: durability, retreatment frequency, safety, resource use, site variation, subgroup response, or budget impact. The registry then collects the minimum data required to reduce that uncertainty.

This is different from a broad academic registry. A payer-linked registry needs stable definitions, auditable follow-up windows, prespecified analyses, data-quality checks, and a clear decision point where evidence is reviewed.

Other service models show the value of routine measurement

Behavioral-health services such as NHS Talking Therapies show why routine outcome capture matters for stepped care, service management, supervision, and equity monitoring. Medicines with conditional or managed access show why data collection must be tied to a decision rather than treated as passive surveillance.

Psychedelic therapy needs both habits: repeated symptom and function measurement from psychological services, and uncertainty-focused evidence generation from HTA and managed-access frameworks.