number needed to treat

**Definition:** NNT is the number of patients you need to treat over a specified time period to prevent one additional adverse outcome (or achieve one additional good outcome) compared with a control.

**Formula:** - NNT = 1 / ARR - ARR = control event rate (CER) − experimental event rate (EER) - Always round UP to the next whole number.

**Worked example:** Drug reduces MI rate from 10% (control) to 6% (treatment). ARR = 0.10 − 0.06 = 0.04. NNT = 1 / 0.04 = 25. You treat 25 patients to prevent 1 MI.

**Related measures:** - **Relative risk (RR)** = EER / CER - **Relative risk reduction (RRR)** = 1 − RR = ARR / CER. In the example, RRR = 0.04 / 0.10 = 40%. - **Absolute risk reduction (ARR):** the difference in event rates. It is the denominator of NNT. - **Number needed to harm (NNH)** = 1 / attributable risk (absolute risk increase). It is the number of patients treated for one additional patient to experience an adverse effect. A higher NNH means a safer drug. - **NNT and NNH are opposites in interpretation:** you want a LOW NNT and a HIGH NNH.

**Key points for boards:** - NNT depends on the baseline risk. The same drug (same RRR) has a lower NNT in high-risk patients and a higher NNT in low-risk patients. - It is only meaningful with a stated time frame (e.g., 5 years) and a defined outcome. - RRR can make an effect look large even when ARR is small. NNT, derived from ARR, gives a more realistic sense of clinical benefit. - NNT is calculated from data in randomized controlled trials and cohort studies. It is not calculated from case-control studies, because they cannot give true incidence or risk. - If ARR = 0 (no effect), NNT is infinite. If the treatment increases risk, the value becomes NNH. - Report a confidence interval when possible, since NNT CIs can be awkward when the ARR CI crosses zero.

**Clinical significance:** NNT helps compare interventions, weigh benefit against harm (likelihood of being helped vs. harmed, LHH = (1/NNT) / (1/NNH)), and guide cost-effectiveness and shared decision-making. Examples: statins for primary prevention have a high NNT (low-risk population). Antibiotics for severe infections have a low NNT.

**Exam tip:** The question usually gives event rates in two groups. Compute ARR first, take the reciprocal, and round up.