1. Research Question to Protocol: Objectives, Estimands, and SAP
After this lesson you can
- Write a trial question with population, intervention, comparison, endpoint and time
- Turn a question into one primary objective and ranked secondaries
- Write an estimand with a strategy for each intercurrent event
- List the main sections of a protocol and how amendments are handled
- Explain what the statistical analysis plan adds and why it is finalized before unblinding
Video coming soon
Lessons publish daily from November 17, 2026.
This lesson turns a vague question into a five-part trial question and then into one primary objective, ranked secondary objectives and a hypothesis pair. It defines the estimand with its five attributes and strategies for intercurrent events, and explains how the protocol and the statistical analysis plan fix the question before the data arrive.
Questions and answers
Questions a learner would ask about this lesson. Think through your own answer before you open each one.
- What are the five parts of a trial question?
Show answer
Population, intervention, comparison, endpoint and time. In the fictional Oakhaven trial they are adults with the condition and a control value of 60 or above at screening, the program added to usual care, usual care alone, poor control, and twelve months. A vague part weakens the whole trial, because the question is a chain of five links.
- Why should a trial have only one primary objective?
Show answer
The sample size is computed for one endpoint, the conclusion is read from one result, and the chance of a false positive is controlled for one test. Several primaries means no primary. Secondary objectives are ranked in the protocol and reported in that order.
- What is an estimand?
Show answer
The estimand is the precise definition of the effect the trial estimates. It has five attributes: the population (on whom), the treatment (what versus what), the endpoint (measured how), intercurrent events (counting what) and the summary measure (summarized how). It is written before the first participant.
- What is an intercurrent event, and how is one handled?
Show answer
It is an event after randomization that affects how the endpoint is interpreted, such as stopping the program, switching to similar support, or death before twelve months. Each one gets a strategy fixed in advance. Oakhaven uses treatment policy for stopping and switching and a composite strategy for death, counted as poor control.
- How does the statistical analysis plan differ from the protocol?
Show answer
The protocol is the complete plan for the whole trial, fixed before the first participant. The statistical analysis plan is a separate, dated document that specifies every analysis in advance and is finalized before unblinding. The protocol says what will be estimated, and the analysis plan says exactly how.
Practice quiz
- Question 1In the Oakhaven question, the "comparison" part is:
- Question 2The lesson calls the question a chain of five links. Which weak link means "any result can be claimed"?
- Question 3How many primary objectives does a well-written trial have?
- Question 4Under the Oakhaven hypothesis, usual care has 40 percent poor control and the program 20 percent. What is the expected risk ratio?
- Question 5A participant stops attending the program at month four but is still measured at month twelve. This is:
- Question 6Which strategy for intercurrent events counts the endpoint as it is, whatever happened?
- Question 7In the Oakhaven estimand, how is death before twelve months handled?
- Question 8The statistical analysis plan must be finalized:
- Question 9Oakhaven's analysis plan splits alpha of point zero five across three secondary endpoints. About what alpha does each secondary get?
- Question 10Which of these is an example of changing the question after the data arrive?
Saved in your browser only: no account, no server.
Design of Clinical Studies is education only. Every lesson follows the fictional Oakhaven Adherence Trial with invented numbers; the course uses no real data and cites no real studies.