clinical-research

---
name: clinical-research
description: Use when designing a prospective clinical study: selecting/classifying endpoints, estimating sample size and power, or scoring phase-gate feasibility before a protocol reaches submission.
category: Research
version: 1.0.0
tools: []
---

# Clinical Research Study Design

Prospective clinical study design: endpoint selection, sample size/power, and phase-gate feasibility. Every output is an estimate with stated assumptions, routed to a named human owner. This never states clinical conclusions as fact and never substitutes for a biostatistician or regulatory affairs.

## Purpose

R&D clinical teams, medical monitors, and biostatistics functions operate in the gap between "we have a hypothesis" and "we have a protocol ready for submission." Structure three of the hardest design decisions using the methodology below: endpoint selection and classification, sample-size/power estimation, and phase-gate feasibility scoring.

## When to Use

Use this when: choosing a primary endpoint and needing to defend it against surrogate-endpoint scrutiny; producing a defensible first sample-size estimate for a protocol synopsis; giving a study plan a feasibility read before a phase-gate review; or pressure-testing whether planned enrollment is achievable given the eligible population and sites.

Do not use this for: preparing a regulatory submission or clinical evaluation report (that is downstream regulatory/QM work — ISO 13485, EU MDR technical docs, FDA 510(k)/PMA/De Novo/QSR), finding or positioning grant funding, designing a live product A/B experiment, or replacing a biostatistician's final sample-size justification.

## Sample Size / Power Estimation

Compute closed-form power/sample-size estimates for two-arm designs:

- **Means** — Cohen's d effect size
- **Proportions** — normal approximation on the two-proportion difference
- **Survival** — Schoenfeld (1983) events-based formula

Trace the effect size, proportion difference, or hazard ratio to a published or anchor-based source — never to "the n we can afford." Inflate the raw sample size for dropout using n_adjusted = n / (1 − dropout rate). Always label the result as an ESTIMATE requiring biostatistician confirmation: these are first-pass numbers using normal approximations and a standard z-table: a biostatistician produces the final justification and may use simulation, adaptive designs, or exact methods.

## Endpoint Selection

Score each candidate endpoint across five weighted dimensions — clinical relevance, measurability, regulatory acceptance, sensitivity-to-change, and burden — then classify it as **PRIMARY**, **KEY-SECONDARY**, or **EXPLORATORY**. Apply a penalty that disqualifies unvalidated surrogate endpoints from the PRIMARY classification. Apply customary regulatory priors by development area (drug, device, biologic, diagnostic, digital-therapeutic); company- or indication-specific precedent should override the generic prior. If more than one primary endpoint emerges, plan a multiplicity control (pre-specified alpha allocation, e.g. hierarchical or Bonferroni) before proceeding.

## Phase-Gate Feasibility Scoring

Score the overall study plan 0-100 across five dimensions — recruitment feasibility, endpoint readiness, statistical power, operational complexity, and budget fit — using a cost-per-patient benchmark by development-area profile (override with the real budget when available). Return one of four verdicts: **GO**, **GO-WITH-CONDITIONS**, **REDESIGN**, or **NO-GO**, together with the specific blockers and the named owners who must sign off.

## Workflow

1. Draft a protocol synopsis covering objectives, design, population, endpoints, a statistical-plan placeholder, and the owners who must sign
2. Select and classify the endpoint(s) per the endpoint-selection methodology above; if more than one primary emerges, plan multiplicity control
3. Estimate sample size using the appropriate design (means/proportions/survival), tracing the effect to a cited source and inflating for dropout
4. Score feasibility per the phase-gate methodology, by development phase (1-4), and read the verdict, blockers, and named owners
5. Assemble the synopsis and estimates into a gate packet and route it as a recommendation for sign-off by the named biostatistician, medical monitor, and regulatory owner

## Onboarding

Before first use, capture defaults once: development-area profile, alpha, power, dropout rate, and the named biostatistician, medical monitor, and regulatory owner who will be credited on every output. These defaults keep every subsequent estimate consistently framed and correctly attributed; override any of them per-request when a specific study calls for it.

## Reference Canon

Study design: ICH E8(R1) general considerations; ICH E9 plus the E9(R1) estimand addendum; CONSORT 2010; SPIRIT 2013; FDA Multiple Endpoints guidance (2022).

Endpoints and power: Cohen, *Statistical Power Analysis*; Schoenfeld (1983) survival sample size; FDA Surrogate Endpoint Table / BEST (Biomarkers, EndpointS, and other Tools) glossary; FDA PRO guidance (2009); Chow, Shao & Wang, *Sample Size Calculations in Clinical Research*.

Trial operations: ICH E6(R2/R3) Good Clinical Practice; TransCelerate risk-based monitoring; FDA RBM guidance; CTTI recruitment best practices; site-feasibility scoring literature.

## Anti-Patterns

- **Presenting a power estimate as fact.** Every output is an estimate with a named owner who must sign.
- **Powering for a convenience effect size.** The effect must trace to a published or anchor-based minimal clinically important difference (MCID), not to the sample size that's affordable.
- **Anchoring a primary endpoint on an unvalidated surrogate.** Surrogate endpoints need validation evidence for the specific indication.
- **Ignoring multiplicity.** More than one primary endpoint requires pre-specified alpha allocation.
- **Skipping dropout inflation.** Raw n undersizes the study; inflate by 1/(1 − dropout).

## Distinct From

This designs the study, not the submission: regulatory/QM work (ISO 13485 QMS, ISO 14971 risk, EU MDR technical docs and clinical evaluation, FDA 510(k)/PMA/De Novo/QSR submission) is a separate downstream step. This is also distinct from grant-funding discovery and positioning, from live product A/B experiment design, and from R&D program budgeting — this scopes the study; it doesn't fund the program, find the grant, or run the product test.

## Forcing-Question Library

Work through these one at a time, with a recommended answer and canon citation, before finalizing a design. Don't bundle them.

1. **Is the primary endpoint a clinical outcome or a surrogate — and if a surrogate, is it on FDA's validated table?** Recommended: clinical outcome, unless the surrogate is validated for this indication. Canon: FDA Surrogate Endpoint Table; BEST glossary.
2. **What minimal clinically important difference are you powering for, and where did that number come from?** Recommended: a published or anchor-based MCID, cited — never a convenience effect size. Canon: ICH E9; Cohen, *Statistical Power Analysis*.
3. **What dropout rate are you assuming, and is the sample size inflated for it?** Recommended: inflate n by 1/(1 − dropout) using a justified rate. Canon: Chow, Shao & Wang; ICH E9(R1).
4. **Single primary endpoint or multiple — and if multiple, what's the multiplicity control?** Recommended: pre-specify alpha allocation (hierarchical or Bonferroni). Canon: FDA Multiple Endpoints guidance (2022).
5. **Who is the named biostatistician, medical monitor, and regulatory owner signing this synopsis?** Recommended: name them now — this output is a recommendation, not a protocol. Canon: ICH E6(R2) GCP roles and responsibilities.

Walk depth-first — lock questions 1-2 before opening 3-5. Once all are answered, apply endpoint selection, then sample-size estimation, then phase-gate scoring, in that order.