Slate has a good article on some good consumer critical thinking about statistics

Slate has a good “article”: on how to think about whether you should take a drug. In the confusing world of “relative risk vs. absolute risk”:, it’s really hard to know the effect of a drug.

Enter the NNT(Number Needed to Treat). The idea behind this number is the _expected_ number of people that you would have to treat so that _one_ person would realize the benefit of the treatment. For example, if the NNT is 3, then you would expect one out of every three people to benefit from the treatment.

Let’s take the Pravachol (a statin, like Lipitor) example from the article. In a 1995 study in ??NEJM(New England Journal of Medicine)??, researchers reported a 31% reduction in the risks of heart attack in men who took one Pravachol every day for five years. 7.5% in the placebo experienced a heart attack vs. 5.3% in the Pravachol group — a 31% relative reduction in risk or a 2.2% absolute reduction. The NNT (see more “here”: is 1/2.2% = 45.5. So you would expect to have to give over 45 men Pravachol once a day for five years to prevent one heart attack. Turned around, we expect that over 44 of them would not avoid a heart attack (either would not experience one any way, or would not be prevented).

I’ll leave all commentary aside about whether drug companies want you to think that way. The data coming from premarketing approval has to be made public (as a certain company just found out), and anyone with a calculator and absolute risk in hand can calculate an NNT.

Slate has a few interesting NNTs:

|cortisone|painful shoulder|3|
|amoxicillin|shorten fever for ear infection|20|
|Proscar – 4 yr|Avoid surgery for enlarged prostate|18|
|Aspirin|Avoid heart attack|208|

Think about it. Think about how much you spend each year on some of these drugs, and think about what the chance is they help.

(h/t “insider”:

Technorati Tags: ,


8 Responses

  1. Dear John,

    The NNT is a very interesting statistic that I had never seen before.

    But here is my question: what if it is found that the drug sometimes causes death. Shouldn’t that number be factored into the NNT?

    There has been a lot of controversy in the news about chemotherapy for breast cancer. On Orac’s site, he says that chemotherapy in Stage 1 breast cancer after surgery gives a 3% better chance of survival which translates to an NNT of 33.3. But if some women die due to the effects of chemo, this information changes the risk picture considerably. And, given that a certain percentage of patients in a hospital will have errors in their drug protocol, that could change the risk picture also. It would be interesting to have a number that factors in all of these things.

    Oh, and one more thing I don’t understand: is that 3% an average number? In other words, on average if you give chemo to 100 women, three women will benefit. If so, there has to be a confidence interval around it that would include zero, right? Or is that an upper bound as in: if you give 100 women chemo, at most 3 women will benefit. That is very different and really changes the risk picture. So is the NNT and average or best case number?


  2. Interested –

    Let me address you second question first (it’s a lot easier). NNT is exactly as you say – an “average” (or, more in more precise terminology, estimate) with a confidence interval. It is also subject to manipulation as are means, standard deviations, relative risks, and absolute risks. The confidence interval won’t include zero (interpretation of NNT=0: the existence of the treatment will automatically give an infinite number of potential subjects the desired outcome!) NNT=1 means that everyone who is treated will benefit (and, in this one exception, it isn’t an average – an NNT of 1.1 is an average for example and means that, on average, out out of 10 people treated _won’t benefit_).

    Now, for the first question — it really depends on how you count people who benefit from the treatment. If you count a death as a treatment failure, then NNT automatically accounts for death. But then it gets even more complicated: are the deaths from all-cause (including the poor soul who steps out in front of a Mack truck before treatment is concluded), treatment-related, disease-related, or further.

    Similar questions can be asked for other side effects. For example, if NNT is 33.3 in your example and, on average, 6% of surgery subjects end up with a serious complicating infection (I’m really pulling this number out of the air as an example), then, on average, of those 33.3 people you need to treat to get one success you get 2 serious complicating infections. Whether the surgery is worth it is up to further judgment. (It may be.)

    NNT is a very nice number, but there are more questions to ask. I hope this helps you understand — if I can be more clear let me know.

  3. I should make one more comment about this statistic. Since it is an estimate based on a population sample, it does give useful insight into more of a public health or health policy judgment rather than an individual judgment such as “should I get surgery?” The information required for individual judgment is a lot different and a lot more complicated to obtain (and, speaking as the father of two children, a lot more frustrating to obtain as well — I seem to remember that you are a parent as well?)

  4. John,

    I think you’re right, but, up until lately, I never really asked those questions.

    I intuitively knew that I didn’t have all the answers and I wasn’t getting all the answers, but I didn’t have the vocabulary to ask for what I wanted.

    As an example, my daughter’s allergist (when she still needed one) wanted to prescribe Advair. I knew that it had a low level (whatever that means) of steroids and that steroids can stunt growth. I did ask for side effects, but the PA was pretty dismissive (of the side effects). I didn’t have the vocabulary or knowledge to ask what the effectiveness of the drug is. I believe that doctors should offer all these numbers to you and explain them to you. I read later that there were several (or at least one) deaths of children associated with this drug.

    So what is the best source(s) for finding all the info you need? I have never had good luck getting a full article from the medical journals. Unless you are willing to pay for a subscription, all you ever get is abstracts.

  5. One place to start is the prescribing information:

  6. One more thing you can look at:

    This is the FDA’s drug page. Another is the FDA’s Medwatch page.

  7. This is a great article. I am new to your blog and i like what I see. I look forward to your future work.

  8. Glad you enjoyed it!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: