Chapter 09 / Understanding Trial Designs Through Real Project Work / Paid beta preview

Chapter 9: Choosing the Right Design Is Also a Project Decision

How PMs help teams choose designs that are scientifically useful, feasible, ethical, and governable. ## 9.1 The Wrong Question: Which Design Is Best? Lauren Brooks once asked a question that sounded reasonable and was, in the way of clinical trials, slightly dangerous. "Which design is best?" We were in a DIAB-220 design review. The program team was deciding how to move the metabolic therapy forward after early evidence suggested glycemic benefit but left questions about dose, tolerability, cardiovascular risk, patient burden, and long-term use. The room had almost every function represented: medical, biostatistics, regulatory, operations, safety, site feasibility, patient engagement, data management, vendors, budget, quality, supply, and governance. Everyone had a preferred answer. Claire Jiang wanted the design to protect interpretability. Samuel Reeves wanted the endpoints to be clinically meaningful and the safety plan proportionate to the population. Thomas Gallagher wanted a design that could support the intended regulatory conversation. Maggie Chen wanted something sites could actually run. Rafael Ortiz wanted enrollment assumptions that survived contact with real clinics. Aisha Nwosu wanted to know who the design would quietly exclude. Victor Stein wanted the data systems to support the endpoint and visit structure. Priya Raman wanted to know whether the vendor model fit the design. Maya Desai wanted the budget to reflect the complexity instead of pretending the complexity was free. Grace Kim wanted the rationale and decision trail to be inspectable. Lauren saw the tension and tried to simplify it. "So which design is best?" I answered, "Best for what?" That is the better question. A trial design is the planned structure for answering a clinical question. It includes who will be studied, what intervention or exposure will be evaluated, what comparison will be used, what endpoints will matter, how participants will be assigned or observed, how long they will be followed, what data will be collected, what bias protections will be used, and what decision the evidence should support. Design choice is not the project manager's solo decision. A project manager (PM) should not pretend to replace medical judgment, statistical design, regulatory strategy, ethics review, or patient input. But the PM has a crucial role: making design consequences visible early enough for the right people to choose responsibly. Chapters 5 through 8 gave us the ingredients. Chapter 5 showed that observational studies and real-world evidence can be powerful when the question, data, bias controls, governance, and limits fit. Chapter 6 showed that phases are different evidence machines. Chapter 7 showed that randomization, blinding [ICH E9], controls, and comparators are operating commitments. Chapter 8 showed that complex designs create value only when flexibility is specified, firewalled, operationalized, and documented. Chapter 9 is where those lessons become judgment. The right design is not always the most complex design, the fastest design, the cheapest design, the most statistically elegant design, or the design that sounds best in an executive slide. The right design is the one that can answer the needed question credibly, ethically, and operationally. That sentence is easy to say. It is hard to govern. ## 9.2 Start With the Decision the Study Must Support Design selection begins with the decision. Not the template. Not the vendor deck. Not the phrase everyone likes that month. The decision. For DIAB-220, the team had several possible decisions in mind, and they were not interchangeable: | Program Decision | Possible Design Families | Common PM Mistake | Governance Question | |---|---|---|---| | Select a dose | Phase II dose-ranging, adaptive dose selection, exposure-response work | Treat enrollment completion as dose readiness | What evidence supports the dose and what uncertainty remains? | | Show proof of concept | Randomized [ICH E9] Phase II, controlled exploratory trial, biomarker-enriched trial | Over-read a noisy or secondary signal | Is the signal strong enough for the next investment? | | Confirm benefit-risk | Phase III randomized [ICH E9] controlled trial, superiority or noninferiority design | Underestimate scale, endpoint, supply, or monitoring burden | Can the design support the intended claim? | | Characterize long-term safety | Extension study, Phase IV interventional study, registry, database study | Assume postapproval work is simple | Which safety question requires assignment, and which can be observed? | | Understand routine use | Observational registry, pragmatic trial, real-world data study | Confuse routine care with lower rigor | Are we observing practice or assigning an intervention? | | Compare against current therapy | Active-comparator trial, standard-of-care add-on design, pragmatic randomized [ICH E9] trial | Choose comparator by convenience | Is the comparator ethical, interpretable, feasible, and acceptable? | | Study a narrow subgroup | Enrichment design, biomarker-defined trial, external-control-supported strategy | Assume smaller means easier | Can we identify, enroll, and interpret this subgroup credibly? | A design rationale is the documented explanation of why a design was chosen for a specific question, population, evidence need, ethical context, and operating reality. It should not be written after everyone has already fallen in love with the design. The PM should help the team produce the rationale before the protocol hardens. The PM questions are simple: - What decision must this study

...