13. Interim Analysis and Stopping Rules: The Monitoring Committee
After this lesson you can
- Give the three valid reasons to look at accumulating data
- Explain how repeated looks inflate the type one error and how alpha is spent
- Apply a conservative early boundary to an interim result
- Explain futility with conditional power
- Describe the monitoring committee's membership, charter and advisory role
Video coming soon
Lessons publish daily from November 17, 2026.
This lesson explains why and when to look at accumulating data, why repeated looks cost alpha, and how a conservative early boundary works, applied to the fictional Oakhaven interim. It covers futility by conditional power, harm rules, the data monitoring committee and the costs of stopping early.
Questions and answers
Questions a learner would ask about this lesson. Think through your own answer before you open each one.
- Why would a trial look at its data before the end?
Show answer
There are three reasons: harm (participants are being hurt), overwhelming benefit (the answer is already clear beyond reasonable doubt), and futility (the trial can no longer answer its question). Curiosity, reassuring a funder or a conference abstract is never a reason, because every look spends alpha and risks the blind.
- Why does looking early cost alpha?
Show answer
Each look is another chance for noise alone to cross the threshold. Testing at point zero five at every look pushes the overall type one error well above five percent. A spending rule treats point zero five as a budget and spends a slice at each planned look.
- What is Oakhaven's stopping boundary?
Show answer
A conservative early boundary: stop for benefit at the interim only if p is below point zero zero one, and declare benefit at the end if p is below point zero four nine. Almost all of the alpha is kept for the final analysis. The look is triggered by 125 twelve-month outcomes, not a calendar date, and futility is considered if conditional power falls below ten percent.
- What did the Oakhaven interim show, and why did the trial continue?
Show answer
Program 8 of 62 versus usual care 20 of 63 with poor control, about 13 versus 32 percent, a gap of about 19 points. An ordinary point zero five test would call that significant, but it fell far short of the point zero zero one boundary. Conditional power was well above ten percent and serious adverse events were two and two, so the committee recommended continuing.
- What does the data monitoring committee do, and who decides?
Show answer
It is an independent group, typically a clinician, a statistician and a chair, supported by an independent statistician and with no investigators. It reviews unblinded data by arm label in a closed session and recommends continue, change or stop. The sponsor or steering group decides, and a rejected recommendation must be explained in writing.
Practice quiz
- Question 1Which is NOT a valid reason to look at accumulating data?
- Question 2Repeated looks at accumulating data, each tested at point zero five without a spending rule:
- Question 3The Oakhaven interim boundary for benefit is:
- Question 4With point zero zero one spent at the interim, the final Oakhaven threshold is point zero four nine. Why not point zero five?
- Question 5At the interim, 8 of 62 program participants and 20 of 63 usual-care participants had poor control. How many of the 125 had poor control in total?
- Question 6The interim result of 8 of 62 versus 20 of 63 is:
- Question 7Conditional power is:
- Question 8A trial is stopped for futility. How should it be reported?
- Question 9Why is the harm boundary looser than the benefit boundary?
- Question 10Why can stopping early for benefit overestimate the effect?
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.