Number Needed to Harm in Practice
Number needed to harm, or NNH, translates extra risk into a simple count. It tells how many people must receive an exposure before one additional harmful event occurs. The value is useful because percentages can feel abstract. A small absolute risk increase may still matter when many patients, workers, or users are exposed.
Why Absolute Risk Matters
NNH depends on absolute risk increase, not relative risk alone. A treatment can double a rare side effect and still create a high NNH. Another exposure can raise a common event by a few percentage points and create a low NNH. That lower value signals a more frequent added harm.
Using Counts or Rates
This calculator accepts event counts and group totals. It can also accept ready event rates. Counts are best because they support risk difference uncertainty. When group sizes are available, the tool estimates a confidence interval for the risk difference. That interval helps show whether the apparent harm is precise, weak, or crossing no difference.
Interpreting the Result
A positive exposed minus control risk gives an absolute risk increase. The NNH is one divided by that increase. For example, an increase of 0.02 gives an NNH of 50. That means one extra harm is expected for every fifty exposed people, compared with the control group. A smaller NNH means harm appears more frequent.
Limits and Care
NNH is not a complete safety decision. It should be compared with benefit size, event severity, study quality, follow-up time, and patient preference. Mild nausea and major bleeding should not be treated as equal outcomes. Always keep the event definition clear. Use the same follow-up period for both groups. Avoid mixing rates from studies with different populations.
Better Reporting
Report the exposed risk, control risk, absolute risk increase, NNH, confidence interval, and assumptions. If the confidence interval crosses zero, the NNH interval may cross infinity. That does not make the calculation useless. It means uncertainty is high. Use the result as a structured safety summary, not as a standalone verdict.
Good inputs improve trust. Check event counts before reporting. Keep denominators visible. Save exports with study notes, dates, exposure names, outcome definitions, and reviewer initials after review.