What Is Confidence Interval? Definition for Peptide Research

Understanding Confidence Intervals

The Basic Concept

A confidence interval represents uncertainty in a measurement:

Component	Meaning
Point estimate	Best single guess (e.g., 15% weight loss)
Lower bound	Lowest plausible value
Upper bound	Highest plausible value
Confidence level	How confident we are the true value is within range

How to Read a CI

Example: Weight loss of 15.0% (95% CI: 13.5% to 16.5%)

15.0%: Best estimate of the true effect
13.5%: Lowest plausible weight loss
16.5%: Highest plausible weight loss
95%: If we repeated this study 100 times, 95 intervals would contain the true value

What 95% Confidence Really Means

Common Misinterpretation	Correct Interpretation
”95% chance the true value is in this range”	Incorrect - the true value is fixed
”95% of the data falls in this range”	No - CIs are about estimates, not data
”Correct interpretation”	If we repeated the study many times, 95% of CIs would contain the true value

Confidence Intervals in Peptide Research

Trial Results Examples

STEP 1 (Semaglutide 2.4 mg):

Weight change: -14.9% (95% CI: -15.7% to -14.2%)
Interpretation: True weight loss likely between 14.2% and 15.7%
Narrow CI indicates precise estimate

SURMOUNT-1 (Tirzepatide 15 mg):

Weight change: -22.5% (95% CI: -23.6% to -21.4%)
Interpretation: True effect between 21.4% and 23.6%
Very narrow CI reflects large sample size

Comparing Treatments Using CIs

Semaglutide:    |----[====15%====]----|
                     13.5%      16.5%

Tirzepatide:               |----[====22.5%====]----|
                                21%         24%

                |_________|_________|_________|
                    10%      15%       20%      25%

Non-overlapping CIs suggest statistically significant difference.

Interpreting Confidence Intervals

What Wide vs Narrow CIs Tell You

CI Width	Indicates	Causes
Narrow	High precision	Large sample size, low variability
Wide	Low precision	Small sample size, high variability

Relationship to Statistical Significance

CI Relationship to Zero/Null	Statistical Significance
CI excludes zero (or null value)	P < 0.05 (significant)
CI includes zero	P > 0.05 (not significant)
CI barely excludes zero	P close to 0.05

Example:

Effect: 5.0% (95% CI: 1.2% to 8.8%) - Significant (excludes 0)
Effect: 2.0% (95% CI: -0.5% to 4.5%) - Not significant (includes 0)

CIs for Risk Ratios and Odds Ratios

For ratios, the null value is 1.0, not 0:

Result	Interpretation
RR = 0.70 (CI: 0.55-0.90)	Significant reduction (excludes 1)
RR = 0.85 (CI: 0.70-1.05)	Not significant (includes 1)
RR = 0.60 (CI: 0.50-0.72)	Strong significant reduction

Factors Affecting CI Width

What Makes CIs Narrower

Factor	Effect on CI	Explanation
Larger sample size	Narrower	More data = more precision
Lower variability	Narrower	Less spread in outcomes
Higher confidence level (99% vs 95%)	Wider	More certain = wider range

Sample Size Impact

n = 50:   |--------[====Effect====]--------|  (Wide CI)

n = 200:     |---[====Effect====]---|         (Moderate CI)

n = 1000:       |-[Effect]-|                  (Narrow CI)

Larger trials produce more precise estimates.

CIs vs P-Values

Complementary Information

Metric	What It Tells You
P-value	Probability result is due to chance
Confidence interval	Range of plausible effect sizes
Effect size	Magnitude of the difference

Why CIs Are Often Preferred

Advantage of CIs	Explanation
Show effect magnitude	P-values don’t indicate size
Display precision	Width shows certainty
Enable comparison	Can visually compare treatments
Not dichotomous	Avoid “significant vs not” thinking

Practical Applications

Evaluating Clinical Meaningfulness

Question: Is this effect clinically important?

Look at the entire CI, not just the point estimate:

Scenario	CI	Interpretation
All values clinically meaningful	12-18% weight loss	Clearly beneficial
Mixed values	2-8% weight loss	Uncertain clinical importance
Includes trivial effects	-1% to 5% weight loss	May not be meaningful

Decision Making with CIs

Ask: “Would I be satisfied if the true effect is at the lower bound?”

CI: 15-22% weight loss - Yes, even 15% is substantial
CI: 2-10% weight loss - Maybe, depends on context
CI: -1% to 5% weight loss - No, might have no real effect

Frequently Asked Questions

Why do some studies report 90% or 99% confidence intervals?

The confidence level is a choice. 95% is conventional, but 90% CIs are narrower (used when precision matters more) and 99% CIs are wider (used when more certainty is required). FDA bioequivalence studies often use 90% CIs. The choice should be pre-specified, not selected after seeing results.

How do confidence intervals relate to clinical decision-making?

CIs help clinicians understand uncertainty. A narrow CI around a large effect provides confidence the treatment works well. A wide CI suggests results could vary considerably. Clinicians should consider whether even the lower bound represents worthwhile benefit for their patient.

Can two treatments with overlapping CIs still be statistically different?

Yes. Visual CI overlap is a rough guide, but overlapping CIs don’t necessarily mean non-significance. Two intervals can overlap substantially yet still show a statistically significant difference when directly compared. Formal statistical tests of the difference are more reliable than visual assessment.

Confidence Interval