Number Needed to Treat (NNT) and Number Needed to Harm (NNH)

Number Needed to Treat questions are the most common scenario in the USMLE, followed by Number Needed to Harm. The Number Needed to Diagnose (NND), Number Needed to Expose (NNE) and Number Needed to Screen (NNS) and, Confidence Intervals (CIs), are used in research but ~not~ in the in USMLE.

Core concepts of Number Needed to Treat (NNT) and Number Needed to Harm (NNH)

NNT and NNH are clinical assessments between two groups:
new (experimental) treatment versus placebo (or, standard treatment).

NNT and NNH are the number of patients — it's NEVER a percentage.

As NNT ↑, effectiveness ↓ — the optimal Number Needed to Treat is, 1.
As NNH ↑, effectiveness ↑ — the optimal Number Needed to Harm is as far away from 1 as possible (e.g., 1,000,000,000).

Calculating the Number Needed to Treat (NNT) and Number Needed to Harm (NNH)

There are three steps to computing the number needed:
  1) In each of the two groups, establish the ratio of [those suffering] / [total people in the group];
  2) Subtract the smaller number from the larger number (number needed can never be negative); &
  3) Invert (flip) the answer above & calculate (e.g., 0.20 inverted = 1/0.20; 05 inverted = 1/5; and 3/4 inverted = 4/3).


The USMLEbiostatistics workbook provides more details, and the tricks and commonly-presented word problems with thorough explanations, and more. Buy now! More content will be added!!


Back to Biostatistics






















Double hump

Renamed to "double chump," I guarantee that this seemingly troublesome test question will become completely intuitive and absolutely NO trouble.

Read more about the tools...

All these IMGs and medical students are saying, “Biostatistics is EASY!!!”


Simply yet thorough – read about the med students & IMGs’ experiences with the USMLEbiostat workbook of tools!



Links just for med students & IMGs.