On this page
What should a REMS assessment methodology specify?
Connect each program objective to a measurable question, population, data source, instrument and analysis. Define the counting unit, denominator, timing, missing-data treatment and interpretation before reviewing results. Document feasibility, limitations and the actual review path. The 2019 assessment draft offers planning recommendations; it does not create one universal survey, passing percentage or submission deadline for every REMS.
Before you begin
Methodology for an actual REMS assessment. Product-specific decisions and correspondence control applicability; the February 2019 landing-page guidance is a January 2019 draft, not a final universal method.
What you will prepare: A versioned question-to-method plan with data definitions, limitations, analysis rules and a documented review path.
Begin with the program objective and decision
Identify the relevant goal, participant behavior or operational question. State what decision the result is intended to support. Distinguish reach, knowledge, implementation and clinical outcomes; they require different evidence. A survey of knowledge cannot by itself measure every aspect of safe use. Map each question to the operative assessment plan and identify where agency agreement or clarification is still needed.
Make the method reproducible
| Design item | Specify before analysis | Missing-input action |
|---|---|---|
| Population | Eligibility, sampling frame and relevant exclusions | Resolve who the result represents |
| Measure | Operational definition, instrument and timing | Do not substitute an undefined proxy |
| Data | Source, linkage, duplicates and missingness | Document feasibility and access limits |
| Analysis | Denominators, planned comparisons and uncertainty | Have qualified analysts define the method |
| Interpretation | Decision criteria and limitations | Do not invent a passing threshold after results |
Keep the protocol, instrument and analysis specification version-aligned. Describe how deviations and amendments will be recorded.
Separate draft recommendations from actual obligations
The 2019 draft discusses submitting methodology/protocol detail and updating the supporting document. Use it as clearly labeled planning context. Establish actual submission timing and requested content from the approved program and agency interaction; do not treat an illustrative schedule as a binding rule. Link the method to the supporting document and preserve the prior version when proposing a change.
Fictional example: a biased sampling frame
A survey recruits only participants who completed training successfully. That may answer a narrow question about trained respondents, but it cannot automatically represent all intended participants. Document the limitation and obtain an appropriate design decision. If the sampling frame is unknown, resolve it before promising representative results or selecting a sample size.
Define the measure before choosing a sample size
An assessment can be precise and still answer the wrong question. Start by stating the intended inference: for example, what a defined participant population knows at a specified point, or whether a particular safe-use action occurred. A completed questionnaire and a verified action record measure different things.
| Definition | Specify | Ambiguity to resolve |
|---|---|---|
| Objective and decision | Actual REMS objective and how results will be used | Is the measure outreach, knowledge, behavior or outcome? |
| Target population | Eligibility, setting, geography and assessment timing | Does the available sampling frame cover that population? |
| Counting unit | Person, site, dispensing event or other defined unit | Can repeated records make the sample look larger than it is? |
| Measure and instrument | Question wording, response rules and operational definition | Does the instrument test the intended concept without teaching the answer? |
| Data handling | Linkage, duplicate handling, missing fields and exclusions | Are absent records distinguishable from unmet requirements? |
| Analysis | Denominator, uncertainty, comparisons and sensitivity analyses | Is the method capable of supporting the intended inference? |
| Interpretation | Justified criterion or alternative evaluation approach | Would the same rule be applied if results were unfavorable? |
Only after those decisions should the qualified methodological team justify sampling and sample size. State assumptions about recruitment, expected response, subgroup coverage and precision. A larger sample from the wrong population does not repair the sampling frame, and repeated responses by the same person do not automatically provide independent information.
Fictional timing exercise: the objective concerns retained knowledge several months after training, but the proposed survey is administered immediately after the teaching module while the answers remain visible. That design may assess a narrower immediate task; it does not establish the intended later knowledge. Resolve the assessment timing and instrument with the responsible owners before collection, and record the version and decision. Relabeling the result afterward cannot make the earlier method prespecified.
The January 2019 draft recommends justifying performance thresholds and discusses alternatives when prespecifying a threshold is infeasible. It does not establish a universal 80%, 90% or 100% pass rate. Document the proposed interpretation, its rationale and any actual agency agreement. If the method changes after data are seen, disclose the change and distinguish exploratory analysis from the planned assessment.
Plan the assessment of access and operational burden where relevant. Define which delays or difficulties are attributable to the REMS process and what comparison or contextual evidence is needed; avoid treating every access problem as program-caused. Complementary data sources can help, but their definitions and linkage limitations must be understood.
Maintain one version chain across the protocol, questionnaire, analysis specification and supporting document. The draft discusses advance methodology review, including a 90-day recommendation in VI.A; it remains a draft recommendation, not a universal statutory clock. Establish the actual submission expectations through the approved program and agency interaction. Use the assessment-report guide to carry those definitions and deviations into results.
Your preparation checklist
0/3 checkedUse this to track your review in this visit. Checks are not saved and do not establish regulatory compliance.
Frequently asked questions
Does FDA prescribe one passing knowledge percentage for every REMS survey?
The cited assessment draft does not establish one universal percentage. It recommends justified performance thresholds and discusses alternatives when prespecification is infeasible. Connect the interpretation to the actual objective, study design and agency interaction. Do not choose a convenient threshold after seeing the results.
Can a larger sample compensate for recruiting the wrong participants?
Not by itself. Define the target population and assess whether the sampling frame and recruitment process reach it. More responses from only highly engaged or successfully trained participants may leave the same selection problem. Explain representativeness, nonresponse and the limits of the intended inference.
Is the draft’s 90-day methodology recommendation a universal binding deadline?
No. The January 2019 document remains draft guidance and its advance-submission recommendation should be labeled accordingly. Establish the actual timing and content expectations from the applicable REMS and FDA interaction. Do not treat an illustrative planning recommendation as a substitute for the product’s operative requirements.
Can an analysis selected after results are known be called prespecified?
No. Preserve the dated protocol and analysis versions and disclose later decisions. A new analysis may be useful, but it should be distinguished from the planned assessment, with its rationale and limitations explained. Do not backdate the method or silently replace an unfavorable planned result with a favorable exploratory one.
Sources and revisions
Requirements, source recommendations and editorial preparation advice have different roles. Review the scope and revision of the source you use.
Draft guidance
FDA draft: REMS Assessment: Planning and Reporting ↗January 2019 PDF cover; February 2019 FDA landing page. Still draft/not for implementation when checked September 22, 2026. III–VI provide proposed assessment organization, not a replacement for the approved REMS.
Guidance
FDA: Format and Content of a REMS Document ↗January 2023 final guidance; III.A–F and IV. Distinguishes REMS document, materials and supporting document. Checked September 22, 2026.
Technical specification · placement only
FDA eCTD v4.0 comprehensive hierarchy ↗Version 2.2, February 2025. Section 1.16.2.4. A heading identifies placement, not mandatory applicability.