[HN Gopher] The NNT (Number Needed to Treat), explained (2013)
___________________________________________________________________
The NNT (Number Needed to Treat), explained (2013)
Author : yamrzou
Score : 20 points
Date : 2022-12-25 11:10 UTC (11 hours ago)
(HTM) web link (www.thennt.com)
(TXT) w3m dump (www.thennt.com)
| jphoward wrote:
| NNT is an interesting metric that people were very sanctimonious
| about when I was at university, but has rightly lost favour.
|
| This is because NNT is actually a terrible metric for any
| treatment that reduces risk over a long period, including the
| most important outcomes like mortality.
|
| Imagine a drug that halves your risk of dying of heart disease.
| Heart disease kills 1 in 4 people in developed countries, and so
| realistically it would be a no brainer that everyone middle aged
| or over should take it.
|
| If you did this in a randomised controlled trial, you might
| recruit 1000 patients, and randomise 500 to this new drug and 500
| to placebo. If you ran the trial for a year, you might find 1% of
| the placebo arm died, and only 0.5% in the control arm. This
| demonstrates a risk reduction of 0.5%, or a NNT of 200.
|
| An NNT, by conventional teaching, is underwhelming.
|
| Is this the right way to think about it? ABSOLUTELY NOT! The high
| NNT is a consequence of a low event rate, NOT an ineffective
| therapy. In preventative medicine, where everyone will EVENTUALLY
| die of something (and 25% the thing you are looking at), it is
| the RELATIVE risk reduction (50% here, not 0.5%) that matters.
|
| It's different if you have a single one-off therapy, where all
| the benefit will have been delivered by a certain timepoint, at
| which you measure the NNT. But for a lot of medicine it is not
| that simple, and the NNT sucks.
| ggm wrote:
| But how about if you project it into health economics? If you
| use it to calculate something like QALY cost?
| nkurz wrote:
| You've probably thought about this more than I have, but your
| "take down" of NNT doesn't work for me. Let's use your example.
| We treat 500 people and observe another 500. 5 of the untreated
| die, and 2.5 of the treated die (I don't know how either). We
| observe that the NNT is 200.
|
| And we this is great! If a drug is inexpensive, has minor but
| not severe side effects, and saves 1/200 people from dying
| within a year you should absolutely be giving it to everybody.
| It's a home run. The problem isn't the NNT, it's the
| expectation of what constitutes a a high or low number.
|
| By contrast, consider another study. We treat 500,000,000
| people, and observe another 500,000,000. We observe a 50%
| relative risk reduction among the treated. The drug is
| inexpensive, and has minor but not severe side effects. Should
| we create everyone?
|
| I have no idea. It has to depend on how much risk we started
| with. If the mortality rate is 1% per year among the untreated,
| then almost definitely yes. But if we observed 2 deaths in the
| untreated and only 1 from the treated, probably there is a
| better area to spend our efforts, even if we somehow knew the
| life we saved just noise.
|
| So yes, the reduction in risk matters, but only in concert with
| a concept of the baseline risk. By itself, it's not very
| helpful, but in concert with other data it's great. NNT is a
| way of expressing this in a single number. There is no apriori
| reason to believe that 200 is high or low. It may not be
| perfect, but it's better than just a bare relative risk
| percentage.
| oldgradstudent wrote:
| > If you did this in a randomised controlled trial, you might
| recruit 1000 patients, and randomise 500 to this new drug and
| 500 to placebo. If you ran the trial for a year, you might find
| 1% of the placebo arm died, and only 0.5% in the control arm.
| This demonstrates a risk reduction of 0.5%, or a NNT of 200.
|
| That's because the only thing you've learned during the above
| trial is that it halves the death rate for one year. You have
| no idea what happens after that year. It could get better,
| worse, or the adverse effects would be worse than the cure.
|
| Your assumption that that drug cuts the death rate beyond that
| single year is an assumption, not a fact.
|
| > An NNT, by conventional teaching, is underwhelming.
|
| As it should be. That's the only thing you can conclude from
| the trial.
|
| > Is this the right way to think about it? ABSOLUTELY NOT! The
| high NNT is a consequence of a low event rate, NOT an
| ineffective therapy.
|
| Yes. The absolute _proven_ benefit is miniscule.
|
| > In preventative medicine, where everyone will EVENTUALLY die
| of something (and 25% the thing you are looking at), it is the
| RELATIVE risk reduction (50% here, not 0.5%) that matters.
|
| But you've only had a 1 year trial. Drugs often have long term
| adverse effects that dwarf the benefit.
|
| As an example, when steroids were first discovered they were
| viewed as miracle drugs and given in massive doses. It didn't
| take long to understand the adverse effects were far worse than
| the symptoms it was treating. Today we are far more careful
| with steroids.
|
| Similarly with Aspirin, or any number of very useful drugs.
|
| So the question is what gives you the confidence to make any
| claims beyond what the trial showed?
| yawboakye wrote:
| bullshit. what about those killed by stopattack? there should at
| least be 3 distinct effects related to the medication: no effect,
| positive effect, negative effect. somehow the potential negative
| effect of the medication is gladly overlooked.
| nickprins wrote:
| There is a corresponding measure - number needed to harm (NNH).
|
| https://en.m.wikipedia.org/wiki/Number_needed_to_harm
|
| TheNNT also highlights treatments where the evidence for
| benefit isn't clear (or controversial), e.g. thrombolytics in
| stroke.
|
| https://www.thennt.com/nnt/thrombolytics-for-stroke/
| yawboakye wrote:
| it shouldn't be a separate quantity. we should they should be
| discussed cumulatively at all times.
___________________________________________________________________
(page generated 2022-12-25 23:01 UTC)