NNT*

NNT*

by Eduardo Rodriguez Almaraz -

I am having a hard time wrapping my head around the NNT*. I understand the mathematical calculation to get the this number but, If I understand correctly the Number needed to treat (without star) is the number of interventions that we need to do in order to avoid one outcome in a given population. But the NNT* is calculated with the prior probability of a specific patient. Therefore, the NNT* will vary from patient to patient and this is the concept that I am having issues with. 

In the lecture an example of a girl with  a prior probability of 45% for flu yields an NNT* of 20 for prophylaxis with oseltamivir was given. What is this number exactly telling us? Does this mean that we need to treat 20 close contacts in order to avoid her getting flu for that specific patient?


Thank you

Re: NNT*

by Michael Kohn -

NNT is calculated assuming that the patient has the condition being treated.  NNT* is calculated when there is uncertainty about whether the patient has the condition.  You can interpret NNT* as the number of patients *just like this one* who I have to treat to prevent one bad outcome.  The flu prevention in household contacts is a little more complicated.  The NNT* there refers to how many *household contacts* you need to treat to prevent one case of the flu.


Remember that NNT is just 1/ARR.  You might find it easier to think in terms of ARR.  If the ARR of treating is when I'm sure the patient has the disease, then if the patient has probability P of having the disease, ARR* = PxARR and NNT* = 1/ARR*.