A Simple Way to Find Number Needed to Treat (NNT)

The findings from clinical research can be difficult to translate into everyday care.

The Number Needed to Treat (NNT) provides an easy to understand alternative for evaluating the relevance of findings.

Here is a very easy way to find NNT from a study

A SimpleWay to Find Number Needed to Treat (NNT)

The findings from clinical research can be difficult to translate into everyday care. Most often the Relative Risk is presented, it is the fraction between the prevalence of the studied event in a treatment group and the prevalence in the comparison group of that event. The comparison group is typically another treatment or placebo. Typically with statements like “the risk of stroke was reduced by 50% by treatment X”. This sounds good and great…who would not half the risk for stroke, but it does not provide much useful information if any at all. By this very common type of statements, you do not know if it was one person in one group and two individuals in the other study group haveing a stroke in a total study population of 2000, or if it was 1000 in one group and 500 in the other. It does indeed make a difference.

What you want to know is the change in Absolute Risk (AR). AR is simply the fraction of persons in each group haveing the studied event (here stroke). If the treatment group(t) had 560 individuals and 22 had a stroke during the study period, and the comparison(c) group had 572 persons and 45 persons, then you can find the AR as AR(t)= 22/560 = 0.039 (0r 3.9%), and AR(c)=45/572= 0.78 (or 7.8%). In this example the change in AR (ARD) is ARD= 0.078-0.039 = 0.039 (or 7.8%-3.9% = 3.9%). Note that in this example the RR is a 50% reduction.

IMG_2130So in this example, the RR was 50% but we “saved” 3.9%. This change in AR can be used to calculate the “Number Needed to Treat” (NNT), which is a calculated (hypothetical) number used to assess the effectiveness of a healthcare intervention, typically a treatment with medication. The NNT is the average number of patients who need to be treated for one to have or prevent a clear outcome (i.e. the number of patients that need to be treated for one to benefit compared with a control in a clinical trial). NNT is only meaningful when evaluated in the context from which it is calculated, i.e. time frame and a comparator.
NNT is calculated based on measured changes the Absolute Risks, the ARD from above, as NNT = 1/ARD. Int eh example above NNT= 1/0.039 =25.6, round off to 26 persons as a fraction is without meaning concerning persons.

In other words, we need (theoretically) to treat 26 persons at the same time as the study period, for one person to benefit from the treatment. It is the same result as “the risk reduce by 50%”, but more relevant at the point of care.

Note that NNT

  • is an easy way to estimate clinical relevance of a treatment
  • is only relevant concerning a period
  • cannot be compared across studies, as the study population and design will be different
  • heavily depend on the study design and method, which must be clinically relevant for NNT to be useful.

Zignifica_Easy_NNT_Finder The graph provides a simple way to find NNT. Just locate the ARD in the horizontal line, and use the graph to find the corresponding NNT.Zignifica_Easy_NNT_Finder


Download it and save it on your smartphone for easy access when you need it.