Intention to treat or per protocol: which quitting question is being answered?

Assignment and attendance are different facts

A trial can assign someone to a support arrangement even if that person never attends, attends partly, or seeks other help. An assignment-focused analysis keeps people in their assigned groups. An adherence-focused question concerns following the specified intervention. Cochrane distinguishes these effects and cautions that a simple per-protocol exclusion can introduce bias.

This distinction is useful when a headline suggests that a program works if people simply commit. Look for what the actual analysis set out to estimate. A label in a table cannot turn attendance into a character judgment, and an analysis of attendance does not automatically establish the effect of choosing that program in ordinary life.

Cochrane: Cochrane Handbook, Chapter 8: Assessing risk of bias in a randomized trial

Draw a fictional assignment ledger

Imagine a fictional project assigning adults to either a scheduled conversation or a planning note. In the conversation group, one adult attends, another misses the meeting because of work, and a third receives separate help. Write two columns: assigned group and what happened afterward. Do not move the third adult into the note group merely because the conversation was missed.

Now write two questions above the sheet: what followed offering the assigned arrangement, and what would follow adhering to that arrangement? The ledger shows why the questions differ. It supplies no outcome numbers, real trial, or quitting estimate. You also cannot answer the second question just by deleting the missed-attendance rows.

Keep missing outcomes on a separate page

Not attending an intervention and not providing an outcome report are separate events. A participant might miss sessions yet answer follow-up; another might attend every session but have an unavailable outcome. Check how the report handled both situations rather than assuming they are interchangeable exclusions.

If a paper says modified intention to treat, find its actual inclusion rule. Preserve the authors’ explanation and any uncertainty. Do not substitute a homemade rule or choose the analysis showing the most appealing number. An adherence-effect estimate may require methods beyond the simple introductory distinction explained here.

Carry the effect question into your summary

A useful reading note names the intended effect, assigned groups, actual analysis method, and unresolved missing-outcome details. For example: the headline concerns assignment, but the quoted analysis excluded some participants after assignment, and I have not verified how that affects the claim. That is a focused question to bring to the full methods.

This guide does not assess a named study or tell you which treatment to use. Qualified interpretation matters when an individual care decision depends on the analysis. Your own attendance and vaping observations can both be recorded honestly; they do not create the unobserved randomized comparison that a trial is designed to investigate.