Read the confidence interval beside a quit-vaping study estimate

Locate the interval for the result you are reading

A paper may report several outcomes and several analyses. Find the table or paragraph containing the result behind the headline, then copy the measure, estimate, interval, and confidence level into a private reading note. Do not attach an interval from a different outcome just because it appears nearby.

Cochrane explains that confidence intervals communicate uncertainty and precision around estimated effects. A relatively narrower interval indicates greater statistical precision for the same kind of measure. That concept does not tell you whether the study asked the right question or whether its participants resemble you.

Cochrane: Cochrane Handbook, Chapter 15: Interpreting results and drawing conclusions

Keep the units and comparison attached

Ask whether the estimate is an absolute difference, a risk ratio, an odds ratio, or another measure. The interval uses that measure’s scale. Values cannot be read as interchangeable percentages just because the headline discusses quitting.

Our original reading card says: “Outcome measured; group comparison; measure used; estimate and interval; question I still have.” Fill it with the actual report, not invented values. If the article does not identify the measure clearly enough for you to understand it, mark that part unresolved and look for the full explanation or an appropriate expert interpretation.

Cochrane: Cochrane Handbook, Chapter 6: Choosing effect measures

Avoid converting uncertainty into a personal promise

An interval around a group comparison does not divide readers into people who will and will not quit. Nor does a confidence level mean that each participant has that probability of success. Keep the study’s estimation question separate from your own next practical decision.

In a hypothetical example, an advertisement repeats one estimate but omits the interval that the paper reports. A careful reading note would retain both and identify the study population and follow-up. This example does not supply a real treatment result or an app success rate; it shows what a single number leaves out.

Preserve the limits outside the interval

Precision is one question; missing observations, outcome definitions, and methodological bias are others. A neat, narrow interval does not certify those features. You can record a result precisely while still needing help understanding whether it supports the claim being made.

This guide does not calculate an interval, judge the significance of a named study, or recommend a treatment. When the report remains unclear, avoid repeating a precise headline as a settled answer. The useful output is a faithful note of the reported uncertainty and the questions that require further reading, rather than a numerical forecast for your own attempt.

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