Skip to main content
v2026.11,986 entries · CC-BY 4.0

How to Report Confidence Intervals in a Manuscript

The correct format for reporting a confidence interval in a manuscript: APA bracket notation, AMA/biomedical parenthetical notation, worked examples for means, proportions, and ratios, and a checklist of the mistakes reviewers flag most.

Written and maintained by CASRAI Editorial Board

Published · Last updated

On this page

To report a confidence interval, give the point estimate, then the confidence level and both bounds. In APA style write M = 4.6, 95% CI [2.31, 6.89]; in AMA and most biomedical journals write OR, 1.8 (95% CI, 1.1-2.9). State the level once, match the decimal places of the estimate, and use the same notation in every sentence and table.

A confidence interval (CI) communicates the precision of an estimate, not just whether it cleared a significance threshold. Most current reporting standards — including CONSORT 2025 (which supersedes CONSORT 2010) and the ICMJE recommendations — expect an effect size to be reported with a measure of its precision, such as a CI. This page covers exactly how to write one: the notation, the decimal precision, the differences between style guides, and the mistakes that get flagged most often in peer review.

If you need a refresher on what a confidence interval actually means or how to calculate one, see Confidence Interval Explained: How to Calculate and Interpret One first. This page assumes you already have the interval and need to know how to write it into a Results section. If the interval came from resampling rather than a formula, bootstrapping in statistics explains how those bounds are built.

Quick reference: CI notation by style guide

Style guide Notation Worked example Notes
APA (7th edition) 95% CI [LL, UL] M = 4.6, 95% CI [2.31, 6.89] Square brackets, comma between bounds, no repeated percent sign, same decimal precision as the point estimate.
AMA / most biomedical & clinical journals (95% CI, LL-UL) or (95% CI: LL-UL) OR, 1.8 (95% CI, 1.1-2.9) Parentheses, hyphen or en dash between bounds. The exact separator (comma vs. colon) varies by journal — check the target journal’s instructions for authors before submission.
CONSORT / ICMJE (style-agnostic requirement) Estimated effect size with its precision (such as a 95% CI) Mean difference −3.2 kg (95% CI −5.1 to −1.3) CONSORT 2025 item 26 and the ICMJE statistics guidance are reporting-completeness requirements, not notation rules — the actual formatting still follows whichever style guide (APA, AMA, Vancouver) the journal has adopted.
No house style specified 95% CI [LL, UL] 95% CI [2.3, 6.7] The APA bracket format is the safest default when a journal gives no explicit statistics style guidance.

Why the CI has to be there, not just the p-value

Three overlapping sources of guidance push authors toward reporting confidence intervals as standard practice rather than an optional extra:

  • CONSORT 2025, the current reporting guideline for randomized trials (it replaces the 2010 version), has a 30-item checklist. Item 26, “Numbers analysed, outcomes, and estimation”, asks for “the estimated effect size and its precision (such as 95% confidence interval)” and, for binary outcomes, both the absolute and the relative effect size. See the CONSORT Statement guide for the checklist as a whole.
  • ICMJE’s recommendations for manuscript preparation instruct authors, in the statistics section, to “avoid relying solely on statistical hypothesis testing, such as P values, which fail to convey important information about effect size and precision of estimates,” and to “quantify findings and present them with appropriate indicators of measurement error or uncertainty (such as confidence intervals)” whenever possible.
  • APA style recommends reporting effect sizes and confidence intervals alongside significance tests, the same principle behind CASRAI’s companion guide, How to Report P Values.

The practical upshot: a Results sentence that reports only “p = .03” without an accompanying effect size and interval leaves out the precision information that ICMJE and CONSORT 2025 ask for.

Writing the CI in text: APA style

In APA style, the confidence interval follows the point estimate, in square brackets, with the lower and upper limits separated by a comma. The percent sign and “CI” are stated once, not repeated for each bound:

Correct: “Participants in the intervention group lost significantly more weight than controls, M = 3.2 kg, 95% CI [1.1, 5.3], p = .004.”

Incorrect: “… 95% CI (1.1% – 5.3%) …” — parentheses and a percent sign on each bound are AMA/biomedical conventions, not APA’s.

Keep the decimal precision of the CI bounds consistent with the point estimate itself — if the mean is reported to one decimal place, the CI bounds should generally match rather than appearing more or less precise than the estimate they bracket.

Writing the CI in text: AMA and biomedical-journal style

AMA style and most biomedical and clinical journals place the CI in parentheses, typically after a comma or colon following “95% CI,” with the bounds separated by a hyphen or en dash rather than a comma:

Example: “Patients receiving the intervention had lower 30-day readmission rates (OR, 0.62; 95% CI, 0.41-0.94; P = .02).”

Because the exact punctuation (comma vs. colon after “CI,” hyphen vs. en dash between bounds) varies by journal even within AMA-adjacent house styles, always check the specific journal’s instructions for authors rather than assuming one biomedical journal’s convention applies to another.

Writing the CI in a table

In results tables, the CI is usually given its own column (or a bracketed sub-line under the estimate) rather than folded into a single results sentence. A typical table column header reads “95% CI” with the two bounds in one cell, separated by the same convention (comma-in-brackets for APA, hyphen-in-parentheses for AMA) used in the text. Whichever format is used, keep it identical across every row of the table — mixing “[LL, UL]” in some rows with “LL-UL” in others is one of the more common copyediting flags in review.

Reporting CIs for different types of estimates

Estimate type What the CI describes Example
Mean or mean difference Plausible range for the true population mean or difference between groups Mean difference = 3.2 kg, 95% CI [1.1, 5.3]
Proportion or percentage Plausible range for the true population proportion 62%, 95% CI [54%, 70%]
Odds ratio (OR) / risk ratio (RR) / hazard ratio (HR) Plausible range for the true ratio; a CI that spans 1.0 indicates the ratio is not statistically distinguishable from no effect at that confidence level HR, 0.71 (95% CI, 0.52-0.97)
Correlation coefficient Plausible range for the true population correlation r = .34, 95% CI [.18, .49]

Note the ratio-specific rule: for odds ratios, risk ratios, and hazard ratios, the “no effect” value is 1.0, not 0 — a CI that includes 1.0 (e.g., 95% CI, 0.89-1.42) indicates the result is not statistically significant at that confidence level, even if the point estimate itself looks meaningfully different from 1. For mean differences and correlations, the corresponding “no effect” value is 0. For what the interval adds beyond a significance test, see p-value vs confidence interval.

Binary outcomes: report absolute and relative effects with their CIs

For a yes/no outcome such as event or no event, CONSORT 2025 item 26 asks for both the absolute and the relative effect size, each with its precision. The relative measure (risk ratio, odds ratio) shows the proportional difference; the absolute measure (risk difference) shows how many more or fewer people are affected. Giving only the relative effect can make a small absolute difference look large, and giving only the absolute effect hides the proportional change. The ICMJE Results guidance points the same way: give numeric results not only as derived values such as percentages but also as the absolute numbers they came from.

Illustrative example (invented numbers, not from any study): suppose 30 of 200 participants in the intervention arm and 50 of 200 in the control arm had the event. A complete sentence reads: “The event occurred in 30 of 200 participants (15.0%) in the intervention group and 50 of 200 (25.0%) in the control group; the risk difference was −10.0 percentage points (95% CI, −18.0 to −2.0) and the risk ratio was 0.60 (95% CI, 0.40-0.90).” Note that the counts, both group percentages, one absolute and one relative effect, and a CI on each effect all appear.

What to report alongside the interval

  • The point estimate the interval belongs to, named (mean difference, odds ratio, proportion).
  • The confidence level, stated explicitly, even when it is 95%.
  • Both bounds, in one consistent notation, with the units of the estimate.
  • The numbers analysed in each group, so readers can see what the estimate rests on (CONSORT 2025 item 26 is titled “Numbers analysed, outcomes, and estimation”).
  • The P value, if you report one, as an addition to the interval rather than a replacement for it: ICMJE advises authors to avoid relying solely on P values because they “fail to convey important information about effect size and precision of estimates”.

The data table below the guide sets out the specific wording of the ICMJE and CONSORT 2025 statements that support this checklist.

Common mistakes to check before submission

  • Inconsistent notation within the same manuscript. Pick one convention (APA brackets or AMA parentheses) and use it in every table and every results sentence — don’t switch formats between the text and a table, or between sections.
  • Mismatched decimal precision. A CI bound reported to more decimal places than the point estimate it belongs to reads as spurious precision.
  • Contradicting the CI in the prose. Describing a result as a meaningful effect while the CI for a ratio spans 1.0 (or a CI for a difference spans 0) is an internal contradiction reviewers routinely flag.
  • Omitting the confidence level. If the interval is not a 95% CI — a 90% or 99% CI, for instance — say so explicitly; “95%” is a convention, not a default that can be assumed.
  • Reporting the CI for the estimate but not stating what estimate it belongs to. “95% CI [1.1, 5.3]” alone is not interpretable — it must be attached to a named point estimate (a mean difference, an odds ratio, a proportion).
  • Ambiguous negative-bound formatting. A hyphen between two negative numbers (e.g., “-5.1–1.3”) is genuinely ambiguous in print. Use “to” (“−5.1 to −1.3”) or an unambiguous en dash with spacing, and check how your target journal handles this case specifically.

Frequently asked questions

What is the correct confidence interval reporting format?

There is no single universal format — it depends on the style guide the target journal follows. APA style uses square brackets with a comma (“95% CI [1.1, 5.3]”); AMA and most biomedical journals use parentheses with a hyphen (“95% CI, 1.1-5.3”). CONSORT 2025 asks for the effect size and its precision (such as a 95% CI), and ICMJE asks for appropriate indicators of uncertainty such as confidence intervals; both leave the exact notation to the journal’s chosen style guide.

How do I write confidence intervals in a Results section?

State the point estimate first, then the confidence interval immediately after it, in the notation your target journal’s style guide specifies, followed by the p-value if one is reported: for example, “M = 3.2 kg, 95% CI [1.1, 5.3], p = .004” in APA style, or “OR, 0.62; 95% CI, 0.41-0.94; P = .02” in AMA style. Keep the format identical across every outcome reported in the same section.

Is CI reporting in APA style different from AMA style?

Yes. APA style (7th edition) uses square brackets and a comma between the lower and upper bound, with the percent sign stated once. AMA style and most biomedical/clinical journals use parentheses, typically with a hyphen or en dash between bounds and a comma or colon after “CI.” Always confirm against the specific target journal’s instructions for authors, since biomedical journals do not apply AMA conventions with perfect uniformity.

How do you report a 95% confidence interval in a sentence?

Name the estimate, give its value, then the level and both bounds: “The mean difference was 3.2 kg (95% CI, 1.1-5.3)” in a biomedical style, or “M = 3.2 kg, 95% CI [1.1, 5.3]” in APA style. Use the same pattern for every outcome.

How do you report a confidence interval for a difference or odds ratio?

Attach the interval to the named effect measure: “OR, 1.8 (95% CI, 1.1-2.9)” or “mean difference −3.2 kg (95% CI −5.1 to −1.3)”. For binary outcomes, CONSORT 2025 item 26 asks for both an absolute and a relative effect size, each with its precision.

Should I report a confidence interval instead of a p-value?

ICMJE advises against relying solely on P values, because they do not convey effect size and precision; it recommends presenting findings with indicators of uncertainty such as confidence intervals. Reporting the interval with the effect size, and a P value if you wish, covers both.

Which CONSORT version is current?

CONSORT 2025 (BMJ 2025;389:e081123, doi 10.1136/bmj-2024-081123) is the current version; the 2001 and 2010 statements are its predecessors. Its checklist has 30 numbered items.

Can I put confidence intervals in a table?

Yes. Give the interval its own column or a clearly marked sub-line and keep the notation identical in every row and consistent with the text.

Do I still need to report a p-value if I report a confidence interval?

Most journals and reporting guidelines expect both, not one instead of the other — the CI conveys precision and effect size, the p-value conveys the significance-test result. See How to Report P Values for the companion formatting rules.

What confidence level should I use if the journal doesn’t specify one?

95% is the near-universal default in biomedical and social-science reporting. If a different level is used (90% or 99%, for instance), state it explicitly rather than leaving the reader to assume 95%.

Sources

Sources:

  • ICMJE, Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals (updated January 2026), Preparing for Submission: icmje.org
  • CONSORT 2025 checklist, EQUATOR Network: resources.equator-network.org; CONSORT reporting guideline page: equator-network.org

Always confirm the specific notation against your target journal’s current instructions for authors, since house style can vary even within one style-guide family.

Data: Confidence-interval reporting statements in ICMJE (January 2026) and CONSORT 2025

Compares what the ICMJE Recommendations and the CONSORT 2025 checklist say about reporting confidence intervals and precision; compiled from the wording of the two documents.

Period covered: 2025/2026.

Confidence-interval reporting statements in ICMJE (January 2026) and CONSORT 2025
StatementSourceRequirement wording
Quantify findings with uncertaintyICMJE StatisticsWhen possible, quantify findings and present them with appropriate indicators of measurement error or uncertainty (such as confidence intervals).
Do not rely only on P valuesICMJE StatisticsAvoid relying solely on statistical hypothesis testing, such as P values, which fail to convey important information about effect size and precision of estimates.
Report absolute numbersICMJE ResultsGive numeric results not only as derivatives (percentages) but also as the absolute numbers from which they were derived.
Effect size and precisionCONSORT 2025 item 26The estimated effect size and its precision (such as 95% confidence interval).
Binary outcomesCONSORT 2025 item 26For binary outcomes, presentation of both absolute and relative effect size.

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about How to Report Confidence Intervals in a Manuscript

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →