Tirzepatide in Hue: Relative Risk, Absolute Risk, and How to Read Safety Tables in Clinical Trials
Featured Answer: How Should You Read Tirzepatide Safety Tables?
To interpret a tirzepatide safety table correctly, look beyond the percentage in each treatment group. Identify the number of participants who experienced the event, the total number evaluated, the comparator, the observation period, and the event definition. Then compare both the relative risk (RR) and the absolute risk difference (ARD).
Relative risk describes how many times as common an event was in one group compared with another. Absolute risk describes how often the event occurred in each group. Absolute risk difference shows the percentage-point difference between the groups. These measures answer different questions and should ideally be considered together.
For example, if nausea occurs in 20% of one group and 10% of another, the relative risk is 2.0, but the absolute risk difference is 10 percentage points. If a rare event occurs in 2 people per 1,000 versus 1 per 1,000, the relative risk is also 2.0, but the absolute difference is only 1 per 1,000. Neither number should be interpreted without considering the event’s severity, uncertainty, study duration, and the population studied. The CONSORT Harms guidance recommends presenting both relative and absolute effects for binary safety outcomes when appropriate. [1]
Key Takeaways
- Relative risk is a ratio. It compares the probability of an event between two groups.
- Absolute risk is the observed event frequency. It is meaningful only when the denominator, event definition, and observation period are understood.
- Absolute risk difference uses percentage points. A change from 10% to 12% is a 2-percentage-point increase, not a 2% relative increase.
- Relative and absolute effects can tell different stories. A large relative increase can correspond to a small absolute increase when the event is rare.
- Safety percentages are not all interchangeable. The percentage of participants with at least one event differs from the number of events per person-time.
- A serious adverse event is not necessarily caused by the medicine. Seriousness, severity, and causality are distinct concepts.
- “Not statistically significant” does not mean “no risk.” The confidence interval and number of events help show how uncertain the estimate is.
- Trial context matters. Dose, comparator, concomitant medicines, participant characteristics, follow-up, and study design all influence interpretation.
1. Why Relative and Absolute Risk Matter in Tirzepatide Research
Tirzepatide is a dual glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1) receptor agonist. Randomized clinical trials have evaluated its effects on glycemic control, body weight, and adverse events in different populations.
Safety tables from these studies often report percentages for nausea, diarrhea, vomiting, hypoglycemia, treatment discontinuation, and serious adverse events. These numbers help researchers characterize tolerability and identify potential safety signals. However, a percentage alone does not tell the whole story.
Consider two headlines:
- “The event was twice as common with treatment.”
- “The event occurred in 2% of the treatment group and 1% of the comparator group.”
These statements can describe exactly the same result. The first emphasizes the relative difference; the second communicates the absolute frequency. A reader who sees only the first statement may overestimate the practical size of the difference. A reader who sees only the second may overlook a potentially important proportional increase.
Good safety interpretation requires both perspectives. It also requires knowing whether the difference is precise, whether it reflects a causal treatment effect, and whether the event is clinically important.
2. What Is Absolute Risk?
Absolute risk is the proportion of participants in a defined group who experience a specified event over a specified period.
For a simple binary outcome, such as whether a participant experienced nausea at least once during a trial, the calculation is:
Absolute risk = Participants who experienced the event ÷ Participants evaluated
Absolute risk (%) = Absolute risk × 100
Suppose 120 of 1,000 participants experience an event during follow-up. The observed absolute risk is 12%.
Absolute risk is useful because it describes the event frequency in a specific group rather than expressing the difference as a ratio. It can help readers understand the approximate number of people affected under the study conditions.
However, absolute risk is not a universal property of a medicine. It depends on the underlying population, how the event is defined, how long participants are observed, the comparator, and the methods used to collect adverse events.
A 12% risk over 40 weeks cannot automatically be compared with a 12% risk over two years. Nor should the event frequency in adults with type 2 diabetes receiving concomitant glucose-lowering medicines automatically be applied to a different population.
3. What Is Relative Risk?
Relative risk (RR), also called the risk ratio, compares the absolute risk in an intervention group with the absolute risk in a comparator group.
Relative Risk (RR) = Risk in the treatment group ÷ Risk in the comparator group
- RR = 1: The observed risks are equal.
- RR greater than 1: The event is more common in the treatment group.
- RR less than 1: The event is less common in the treatment group.
If nausea occurs in 20% of participants receiving treatment and 10% receiving the comparator, RR = 0.20 ÷ 0.10 = 2.0. The event was twice as common in the treatment group during the observation period.
Relative risk is useful for comparing proportional differences. But it does not reveal the baseline frequency on its own. An RR of 2.0 could describe an increase from 1% to 2%, or from 20% to 40%. Both are a doubling, but the absolute differences are very different.
Relative risk should also not be confused with an odds ratio. An odds ratio compares the odds of an event rather than the probabilities themselves. When events are common, the odds ratio can be numerically farther from 1 than the corresponding risk ratio. A table reporting an odds ratio should not be described as though it were a risk ratio.
4. Absolute Risk Difference: The Percentage-Point Change
Absolute risk difference (ARD) is the difference between the risk in the treatment group and the risk in the comparator group.
ARD = Risk in the treatment group − Risk in the comparator group
If an adverse event occurs in 12% of the treatment group and 9% of the comparator group, the absolute risk difference is 3 percentage points.
That is not the same as a 3% relative increase. The relative risk is 12% ÷ 9% = 1.33, corresponding to a relative increase of approximately 33%.
These statements describe different aspects of the same data:
- Absolute risk: 12% in the treatment group and 9% in the comparator group.
- Absolute risk difference: 3 percentage points higher in the treatment group.
- Relative risk: 1.33, or approximately 33% higher relative risk.
For adverse events, a positive risk difference generally indicates more events in the treatment group; a negative risk difference indicates fewer. Whether either result is important depends on the event’s clinical consequences and the uncertainty around the estimate.
5. A Simple Example: How the Same Relative Risk Can Mean Different Things
The following examples are hypothetical and are included to explain the mathematics. They are not observed estimates of tirzepatide’s safety.
| Scenario | Treatment Risk | Comparator Risk | Relative Risk | Absolute Difference |
|---|---|---|---|---|
| A: Common event | 20% | 10% | 2.0 | +10 percentage points |
| B: Less common event | 2% | 1% | 2.0 | +1 percentage point |
| C: Rare event | 0.2% | 0.1% | 2.0 | +0.1 percentage point |
All three scenarios have the same relative risk: 2.0. But the absolute differences range from 10 percentage points to 0.1 percentage point.
This is why safety headlines based solely on relative risk can be misleading. Conversely, a small absolute difference should not automatically be dismissed: the event may be serious, the estimate may be uncertain, or the data may be insufficient to rule out a larger effect.
A relative risk has no useful interpretation without its comparator. The same event rate can look different when compared with placebo, an active medicine, or the background rate in a different population. Check the control group, the study duration, and the clinical context before drawing conclusions.
6. Applying These Concepts to SURPASS-2
SURPASS-2 was a randomized, open-label, 40-week Phase 3 trial involving 1,879 adults with type 2 diabetes. Participants received tirzepatide 5 mg, 10 mg, or 15 mg once weekly, or semaglutide 1 mg once weekly. The trial’s primary focus was glycemic control, with safety outcomes reported alongside efficacy results. [2]
Gastrointestinal events were among the most commonly reported adverse events. The following table uses the published percentages to demonstrate how relative and absolute comparisons can be interpreted. These are observed trial frequencies, not predictions for every patient or every tirzepatide formulation.
| Adverse Event | Tirzepatide 5 mg | Tirzepatide 10 mg | Tirzepatide 15 mg | Semaglutide 1 mg |
|---|---|---|---|---|
| Nausea | 17.4% | 19.2% | 22.1% | 17.9% |
| Diarrhea | 13.2% | 16.4% | 13.8% | 11.5% |
| Vomiting | 5.7% | 8.5% | 9.8% | 8.3% |
| Decreased appetite | 7.4% | 7.2% | 8.9% | 5.3% |
Source: Frías et al., SURPASS-2, New England Journal of Medicine, 2021. Published trial.
The table shows that gastrointestinal event frequencies varied by event and dose. It does not mean that every adverse event increased consistently with each dose. Nor does it establish that any observed difference was statistically significant unless the relevant analysis and uncertainty estimates support that conclusion.
7. Worked Example: Nausea in SURPASS-2
Using the published percentages, nausea occurred in 17.4% of participants receiving tirzepatide 5 mg and 17.9% receiving semaglutide 1 mg.
- Absolute risk in the tirzepatide 5 mg group: 17.4%.
- Absolute risk in the semaglutide group: 17.9%.
- Absolute risk difference: 17.4% − 17.9% = −0.5 percentage points.
- Approximate relative risk: 17.4 ÷ 17.9 ≈ 0.97.
In this comparison, the observed nausea frequency was approximately 0.5 percentage points lower with tirzepatide 5 mg than with semaglutide 1 mg. The relative risk is approximately 0.97, meaning the observed risk was about 3% lower on a relative scale.
This calculation is descriptive. Without a confidence interval or an appropriate inferential comparison for this specific event, it should not be interpreted as proof that tirzepatide 5 mg reduces nausea compared with semaglutide. A small numerical difference may reflect random variation.
Now consider the 15 mg tirzepatide group: nausea occurred in 22.1%, compared with 17.9% for semaglutide.
- Absolute risk difference: 22.1% − 17.9% = +4.2 percentage points.
- Approximate relative risk: 22.1 ÷ 17.9 ≈ 1.23.
This means the observed nausea frequency was 4.2 percentage points higher, or approximately 23% higher on a relative scale, in the 15 mg group compared with the semaglutide group. The clinical meaning still depends on severity, duration, treatment discontinuation, uncertainty, and the wider benefit-risk profile.
8. What Is Number Needed to Harm (NNH)?
Number needed to harm (NNH) is an estimate of how many people would need to receive an intervention, rather than a comparator, for one additional person to experience a specified adverse outcome over a defined period. It is derived from the absolute risk increase when the treatment is associated with more events.
NNH = 1 ÷ Absolute Risk Increase
Use the absolute risk difference as a proportion, not as a whole-number percentage.
For the SURPASS-2 nausea example comparing tirzepatide 15 mg with semaglutide 1 mg, the observed risk difference was 4.2 percentage points, or 0.042 as a proportion.
The simple reciprocal is 1 ÷ 0.042 ≈ 23.8, which rounds to approximately 24 participants for one additional participant reporting nausea over the trial’s observation period, if the observed difference were a reliable estimate of the underlying treatment effect.
This is an illustrative calculation, not a definitive clinical NNH estimate. A formal NNH should be accompanied by uncertainty estimates derived from the risk difference’s confidence interval. If that interval crosses zero, the corresponding NNH interval may be unbounded or may include both benefit and harm. In addition, NNH applies to a particular outcome, comparator, population, and duration; it should not be treated as a general measure of a medicine’s overall safety.
9. How to Read Serious Adverse Events and Severe Events
Clinical-trial tables often report “serious adverse events,” “severe adverse events,” and “adverse events leading to treatment discontinuation.” These terms are not interchangeable.
- Adverse event (AE): An unfavorable medical occurrence during a study, which does not necessarily have a causal relationship with treatment.
- Serious adverse event (SAE): An event meeting regulatory seriousness criteria, such as death, a life-threatening event, hospitalization or prolongation of hospitalization, significant disability, or another medically important event.
- Severe event: An event with high intensity or severity. A severe symptom is not automatically a serious adverse event under regulatory definitions.
- Adverse event leading to discontinuation: An event that causes a participant to stop the assigned treatment. This is a distinct outcome from simply experiencing the event.
For example, a participant may report severe nausea that leads to treatment discontinuation without meeting the formal definition of a serious adverse event. Conversely, an event can be serious because it requires hospitalization even if the symptom intensity is not described as severe.
Another important distinction is causality. An adverse event occurring during treatment is not automatically caused by the medicine. Randomized comparisons can help estimate whether events occur more frequently in one group, but individual causal attribution may remain uncertain.
When reviewing tirzepatide safety data, read the table footnotes and definitions. A table of serious adverse events should not be interpreted as a table of the most intense symptoms, and a table of common side effects should not be assumed to capture every medically important risk. The CONSORT Harms guidance emphasizes transparent definitions and reporting of harms. [1]
10. Why the Denominator and Observation Period Matter
Every safety percentage has a denominator. It may represent all randomized participants, participants who received at least one dose, participants who completed follow-up, or participants with data available for a specific analysis. Those denominators can differ.
For example, 20 participants experiencing an event among 200 participants corresponds to 10%. The same 20 events among 100 participants corresponds to 20%. The event count is identical, but the observed risk differs because the denominator differs.
The observation period matters just as much. If one group is followed longer, participants have more opportunity to experience an event. For a binary outcome, investigators often report the proportion of participants experiencing at least one event during a defined period. For recurrent events, they may report the total number of events or a rate per person-time.
These measures answer different questions:
- Participants with at least one event: How many people experienced the event at least once?
- Total event count: How many occurrences were recorded, including repeated events in the same participant?
- Incidence rate: How frequently did events occur relative to the accumulated time participants were observed?
- Time-to-event analysis: How did the timing of the first event differ between groups, accounting for follow-up and censoring under the analysis method?
A table reporting 100 episodes of nausea does not necessarily mean 100 participants experienced nausea. Some participants may have reported multiple episodes. Always check the table heading and footnotes before interpreting counts as risks.
11. Confidence Intervals: How Certain Is the Estimate?
A point estimate, such as RR = 1.5, summarizes the observed comparison. A confidence interval (CI) communicates statistical uncertainty around that estimate under the analysis assumptions.
For a risk ratio, a confidence interval that includes 1.0 is generally compatible with no relative difference at the corresponding confidence level. For an absolute risk difference, a confidence interval that includes zero is generally compatible with no absolute difference.
That does not prove the groups have identical risks. A wide interval may include both a meaningful increase and a meaningful decrease, indicating that the data are insufficiently precise to distinguish those possibilities.
Rare adverse events are especially difficult to estimate. If only a few events occur, the point estimate can be unstable and the confidence interval wide. If no events occur in one group, a simple risk ratio may be undefined or difficult to estimate. Zero observed events should not be interpreted as proof that the true risk is zero.
Interpret the confidence interval alongside the event’s clinical importance, the number of participants, follow-up duration, and the overall body of evidence. Statistical significance alone is not a complete measure of safety.
12. What Does the Official Tirzepatide Label Tell Us?
The U.S. prescribing information for Mounjaro reports common adverse reactions from pooled placebo-controlled trials in adults with type 2 diabetes. In the cited pooled table, the percentage of participants reporting nausea was 4% with placebo, 12% with tirzepatide 5 mg, 15% with 10 mg, and 18% with 15 mg. Diarrhea was reported in 9%, 12%, 13%, and 17%, respectively. These figures come from the label’s pooled placebo-controlled trial context and should not be mixed directly with the SURPASS-2 percentages without acknowledging the different study populations and pooling methods. [3]
The label also presents hypoglycemia data by treatment setting. Hypoglycemia risk can differ when tirzepatide is used alone compared with when it is used alongside insulin or other glucose-lowering medicines. This illustrates why a safety estimate should not be separated from the clinical context in which it was measured.
The label is a valuable source for approved indications, warnings, precautions, and trial safety summaries. However, it does not replace the original trial publication when a reader needs detailed study design, event definitions, comparator characteristics, statistical methods, or outcome-specific uncertainty estimates.
13. Common Mistakes When Interpreting Safety Tables
- Reporting only relative risk: “Twice the risk” is incomplete without baseline risk and the absolute difference.
- Confusing percent with percentage points: An increase from 10% to 12% is +2 percentage points, or a 20% relative increase.
- Comparing different trial populations as if they were identical: Background risks and event collection methods can differ.
- Ignoring the follow-up period: A frequency over 40 weeks is not automatically comparable with one over a longer or shorter period.
- Assuming every adverse event was caused by treatment: Adverse events are recorded during a study; causality requires additional assessment.
- Equating no observed events with no risk: Small samples and short follow-up can miss rare or delayed harms.
- Ignoring the denominator: A percentage is only interpretable when the population included in the calculation is clear.
- Overinterpreting nonsignificant findings: A nonsignificant result may reflect imprecision rather than equivalence.
- Confusing event counts with participant counts: One participant can experience several episodes.
- Using NNH without uncertainty: A reciprocal of a point estimate is not a complete safety assessment.
14. A Practical Checklist for Reading a Tirzepatide Safety Table
Before drawing a conclusion from a published table, ask these questions:
- What exact adverse event is being measured, and how was it defined?
- Does the table report participants with at least one event or total event counts?
- What is the denominator in each group?
- What are the treatment and comparator groups, and are they clinically comparable?
- What doses and concomitant medicines were used?
- How long were participants followed, and was follow-up similar across groups?
- What are the absolute risks in each group?
- What is the absolute risk difference in percentage points?
- What is the relative risk or other relative effect measure?
- Are confidence intervals available, and how wide are they?
- Were events mild, moderate, severe, serious, or associated with treatment discontinuation?
- Was the adverse event actively and systematically assessed?
- Does the study have enough participants and follow-up to assess rare or delayed harms?
- Do the trial findings agree with the wider evidence and current prescribing information?
This approach makes it easier to separate a numerical difference from a clinically meaningful conclusion. It also helps identify when a claim goes beyond the evidence shown in the table.
15. Statistics Summary: The Measures at a Glance
| Measure | Definition | How to Interpret It |
|---|---|---|
| Absolute risk | Participants with an event divided by the group evaluated | How frequently the event occurred in that group during the defined period |
| Relative risk (RR) | Treatment-group risk divided by comparator-group risk | The proportional relationship between the two risks |
| Absolute risk difference (ARD) | Treatment risk minus comparator risk | The difference in percentage points or absolute probability |
| Relative risk increase | (RR − 1) × 100%, when RR is greater than 1 | How much higher the risk is on a relative scale |
| Number needed to harm (NNH) | Reciprocal of the absolute risk increase, when that increase is positive and interpretable | Approximate number treated for one additional event over a specified period; uncertainty must be considered |
| Confidence interval | An interval expressing statistical uncertainty around an estimate | How precise the estimate is under the specified analysis assumptions |
16. Frequently Asked Questions
Absolute risk is the frequency of an event in a group. Relative risk compares that frequency with the frequency in another group. Relative risk expresses a ratio, while absolute risk communicates how common the event was under the study conditions.
It means the observed event risk in the treatment group was twice the risk in the comparator group. It does not tell you whether the absolute risk was 1% versus 2% or 20% versus 40%, so the absolute frequencies must also be checked.
It means the event frequency increased from one group to the other by 0.02 in absolute probability. For example, a change from 10% to 12% is a 2-percentage-point increase and a 20% relative increase.
They provide complementary information. Relative risk describes the proportional change, while absolute risk difference describes how much the event frequency differs. Reporting both helps prevent exaggerated or understated interpretations.
NNH estimates the number of people who would need to receive an intervention rather than a comparator for one additional person to experience a specified adverse outcome over a defined period. It is based on the absolute risk increase and should be interpreted with uncertainty estimates.
No. A nonsignificant result means the analysis did not establish a statistically significant difference at the chosen threshold. The confidence interval may still include clinically important increases or decreases in risk.
No. Seriousness is based on regulatory criteria such as death, hospitalization, life-threatening events, or significant disability. Severity describes intensity. A severe symptom may not meet the definition of a serious adverse event.
Not always. Trials can differ in population, dose, comparator, follow-up, concomitant medicines, event definitions, and how adverse events were collected. These differences must be considered before comparing percentages.
No. Zero observed events do not prove zero underlying risk. The study may be too small or too short to detect a rare event, and the estimate may be highly uncertain.
Start with the event definition, denominator, treatment and comparator groups, and observation period. Then calculate or identify the absolute risks, absolute risk difference, relative effect, and confidence intervals before interpreting the clinical importance.
No. A commercial listing does not establish regulatory approval, product identity, purity, clinical efficacy, or safety. Published trials and official prescribing information refer to specific study products and authorized formulations in their respective contexts.
17. Related Articles and Research Resources
- Vietnam Peptides Knowledge Hub — explore educational resources on peptide science, clinical trials, and evidence interpretation.
- Peptide FAQ: Research, Storage, and Usage Questions — review general questions about research compounds and terminology.
- Peptide Products Directory — browse the site’s catalog. Product pages are commercial references, not clinical evidence.
18. Related Research Products
Tirzepatide 20mg
This catalog listing is included as a commercial reference related to the article’s subject. The clinical-trial evidence discussed above should not be assumed to establish the quality, safety, efficacy, or regulatory status of a separately listed product.
Retatrutide 20mg
Retatrutide is a related research-area reference involving metabolic pharmacology. It is a different compound with a distinct receptor profile and evidence base. Safety findings for tirzepatide should not be transferred to retatrutide, and a catalog listing does not establish clinical safety or regulatory approval.
Related Educational Resource
The Peptide Plans Resource offers broader educational context around peptide-related goals. It is not a substitute for evidence-based medical care or product-specific prescribing information.
19. Scientific References
- Junqueira DR, Zorzela LM, Golder S, et al. CONSORT Harms 2022 statement, explanation, and elaboration: updated guideline for the reporting of harms in randomised trials. BMJ. 2023;381:e073725. Full article.
- Frías JP, Davies MJ, Rosenstock J, et al. Tirzepatide versus semaglutide once weekly in patients with type 2 diabetes. New England Journal of Medicine. 2021;385:503–515. doi:10.1056/NEJMoa2107519.
- U.S. National Library of Medicine. MOUNJARO (tirzepatide) injection: Prescribing Information. DailyMed. Common adverse reactions, hypoglycemia, warnings, and precautions. Official label.
- U.S. Food and Drug Administration. Drug Trials Snapshots: MOUNJARO. Trial and adverse-reaction information. FDA resource.
- Ioannidis JPA, Evans SJW, Gøtzsche PC, et al. Better reporting of harms in randomized trials: an extension of the CONSORT statement. Annals of Internal Medicine. 2004;141(10):781–788. doi:10.7326/0003-4819-141-10-200411160-00009.
- Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. doi:10.1136/bmj.c332.
- Hernán MA, Robins JM. Causal Inference: What If. Chapman & Hall/CRC; 2020. Background on causal interpretation, comparisons, and sources of bias. Open-access book.
Conclusion
Relative risk, absolute risk, and absolute risk difference are complementary tools for interpreting tirzepatide safety data. Relative risk describes the proportional relationship between groups; absolute risk shows how often an event occurred; and absolute risk difference quantifies the difference in percentage points. Number needed to harm can provide an additional perspective, but it depends on the reliability of the risk difference and the specified observation period.
Published tirzepatide trials show why these distinctions matter. Gastrointestinal event frequencies vary by event, dose, comparator, and trial context. A numerical difference in a safety table is a starting point for interpretation, not a complete conclusion about causality or clinical importance.
The most reliable approach is to read the event definition, denominator, follow-up duration, comparator, and confidence interval before interpreting a result. Consider both relative and absolute effects, distinguish seriousness from severity, and avoid treating a nonsignificant finding as proof of no risk. These habits make clinical-trial safety claims more transparent and scientifically defensible.