Putting these statistics in words

To make all these numbers useful to decision makers, we have to be able to express them in words. Using the example from the previous section, here are some suggestions of how to express your results.

The RR of 'still being infected' on antibiotics relative to no antibiotics was 0.12. We can express this as

• The risk of still being infected on antibiotics was about 12% of the risk on control
• Treatment reduced the risk to 12% of what it would have been
• Treatment reduced the risk by 88% of what it was in the control group

We could reasonably exchange the word 'risk' for 'chance' or 'probability', as they are commonly used to mean the same thing.

Odds ratios are harder. The OR in this example is 0.02. You could express this as:

• Antibiotics reduced the odds of still being infected to about 2% of what they would have been
• Treatment reduced the odds by 98% of what they were in the control group

Note that we have to use the word odds and must not use words like chance, risk or probability.

The risk difference in this example was -0.76. This is best expressed as:

• Antibiotics reduced the risk of still being infected by 76 percentage points.

It is important to be clear about how you express the reduction. If we said the reduction was 76%, it is difficult to know if this reduction is 76% of the risk without treatment (i.e. the control group risk), which in this case was 0.86, or it is a reduction of 76 percentage points. In most cases RD should be expressed as percentage points, as these are the units we need to calculate number needed to treat.

Most people find risk ratio and risk difference (and NNT) easier to interpret than odds ratios. However, communication of the results of your review is only one factor to consider when choosing the best statistic to use.