Skip navigation

NNTs for Preventative Interventions

Bandolier presents three examples of numbers-needed-to-treat for preventative interventions, antibiotics in acute otitis media (NNT 7), antibiotics after dog bites (NNT 16) and compression stockings for postoperative DVT (NNT 9).

Interpreting NNTs for prophylactic interventions requires thought. How serious is the event you are trying to prevent. Could it be treated effectively as and when it arises? Does the prophylaxis have any adverse effects? And then there is cost.

Two examples may help, from opposite ends of the spectrum. From the ISIS2 trial, using streptokinase and aspirin after myocardial infarction had an NNT of 20 to prevent 1 death at 5 weeks compared with doing nothing. Here is prophylaxis against a serious event, by definition not treatable if it occurs. The NNT for the adverse effect of haemorrhagic stroke with streptokinase was 1000.

From the less dramatic end of the spectrum is prophylaxis after dog bite. If no prophylactic antibiotics are given any subsequent infection (10%) should be treatable. The NNT for antibiotic prophylaxis was 16, so that only one patient out of 16 receiving prophylaxis actually avoids infection as a result. The other 15 have antibiotics that they did not need, and which may have adverse effects. And then there is cost.

The NNTs for prophylactic interventions not only require thought but reinforce the message that NNTs are by definition relative. There is no rule that says that an NNT of 2 is always good and one of 100 is always bad: it depends on the context. A NNT of 20 in a fatal condition could be terrific - one person in 20 could be saved. The same NNT would not be terrific for prevention of postoperative vomiting if accompanied by significant adverse effects.

next story in this issue