Clinical Trials · Evidence Appraisal
A 12-Month Benefit and a 30-Day Harm Are Not One Net-Benefit Number
October 21, 2026 · Aqrab editorial team
A trial abstract says treatment prevented two disease events per 100 people and caused one serious adverse event per 100. It is tempting to report “one net event prevented.” Before doing that arithmetic, check the clocks. What if the benefit was measured through 12 months but adverse events were tabulated only during the first 30 days?
A fictional timeline, not a treatment claim
Imagine 1,000 people in each randomized group. The report gives the following illustrative counts. They are not drawn from a real intervention.
| Outcome | Treatment | Control | Observation window |
|---|---|---|---|
| Disease event | 80 / 1,000 | 100 / 1,000 | Randomization to 12 months |
| Serious adverse event | 15 / 1,000 | 5 / 1,000 | Randomization to day 30 only |
The first row is a difference of two fewer disease events per 100 by month 12. The second is one more serious adverse event per 100 by day 30. The arithmetic is correct within each row. Subtracting one from the other and calling it a 12-month net benefit is not justified: the later safety experience is unknown from this table, the outcomes may differ greatly in importance, and the same person could appear in both rows.
Five questions before comparing the rows
- When does each clock start and stop? Follow-up from randomization, first dose, and “while on treatment” can answer different questions.
- Who remains observable? Report the number with outcome data at each time point. Early withdrawal or death may change what later event counts can mean.
- What is counted? People with at least one event, total events, and time-to-first-event estimates are not interchangeable denominators.
- How are post-discontinuation events handled? A safety analysis restricted to exposure time is different from one that follows everyone assigned to treatment.
- What decision is intended? Benefit and harm need an explicit common decision horizon and patient-relevant weighting before they can be combined into a net-benefit claim.
Different windows are not automatically an error: an acute toxicity may warrant a short window, while efficacy needs longer follow-up. The problem is silently treating the two estimates as if they describe the same period and decision.
The practical takeaway
Put the start date, end date, denominator, and outcome definition beside each benefit and harm estimate. If those differ, describe the findings separately and request the missing follow-up or a justified decision analysis before claiming a single net effect.
Where Aqrab fits
Use Aqrab to structure a critique of the report: extract the benefit and harm windows, denominators, treatment-discontinuation rules, and wording of the conclusion. That review can expose a time-horizon mismatch; it cannot estimate unreported later harms or decide how a patient values different outcomes.
Source anchors
The ICH E9(R1) estimands guideline explains why the treatment effect of interest, outcome variable, intercurrent events, and observation period need explicit alignment, for both efficacy and safety. The CONSORT 2025 trial-reporting statement calls for outcome results with analyzed and available-data counts, effect estimates and precision, harm reporting, and interpretation that balances benefits and harms. The numerical example and the five-question audit are editorial teaching devices, not source findings.
Keep reading
Don't stop at one method.
Good methods judgment comes from contrast. Read the neighboring guides, see where the assumptions diverge, and avoid treating every observational problem like it needs the same hammer.
The Trial Ended. Who Made It Into the Open-Label Extension?
A denominator-first audit of long-term extension findings: trace who enrolled, who stayed, and what an uncontrolled extension can actually establish.
Twenty Patients, Thirty Events: Read the Harms Denominator
A worked trial example separates the proportion of patients harmed from recurrent events per person-time, then shows why unequal follow-up needs more than a rate correction.
Clinical Importance: When a Significant Result Is Still Too Small to Matter
A practical guide to reading confidence intervals against a prespecified clinical-importance threshold. Separate evidence against the null from evidence that an effect is large enough to matter.