Re: NOT

1

This is too much like work.


Posted by: Moby Hick | Link to this comment | 01-26-15 3:42 PM
horizontal rule
2

So if we all start taking Terbinafine, that increases the chances that Moby's nasty toenails will clear up. Is that how it works?


Posted by: Jesus McQueen | Link to this comment | 01-26-15 3:45 PM
horizontal rule
3

Childhood depression- infinite. That's pretty depressing.


Posted by: SP | Link to this comment | 01-26-15 3:55 PM
horizontal rule
4

Almost as depressing as the fact that I've been standing on a subway train for 30 minutes and it hasn't left the station.


Posted by: SP | Link to this comment | 01-26-15 3:56 PM
horizontal rule
5

I think the assumption is it's patients diagnosed with the indicated condition (although presumably that includes misdiagnoses.)
So you both need to take off your socks and play footsie then Moby will get better.


Posted by: SP | Link to this comment | 01-26-15 3:59 PM
horizontal rule
6

5: Haven't read the links yet, but NNT is frequently used to determine whether a preventive intervention (like statins for heart attack or stroke) is helpful. AFAIK, it's not really used for determining whether a novel intervention beats standard of care.


Posted by: ydnew | Link to this comment | 01-26-15 4:03 PM
horizontal rule
7

5: But it's indicative of effectiveness, right?


Posted by: biohazard | Link to this comment | 01-26-15 4:21 PM
horizontal rule
8

It's used for other things, but that's one of the most common places. For conditions that commonly get better by themselves (there were a couple on the list), I don't think it makes much sense to use NNT unless you have an outcome that is setting a very high threshold. The first example (ear aches or Acute otitis media) seems a good illustration of when it doesn't make sense. The NNT is reduced because some people get better on their own after between 7 to 14 days. But I'd bet that somebody with sufficient ear pain would want to know if maybe the antibiotic isn't faster.


Posted by: Moby Hick | Link to this comment | 01-26-15 4:23 PM
horizontal rule
9

3 - Duration of treatment is going to have a massive impact on those numbers, so maybe not as depressing? But it's still not a great result.


Posted by: MHPH | Link to this comment | 01-26-15 4:24 PM
horizontal rule
10

8 to 6, mostly. At least the first sentence.


Posted by: Moby Hick | Link to this comment | 01-26-15 4:29 PM
horizontal rule
11

8: I'm not necessarily pro antidepressant (psychotherapy is useful and can be better etc.), but I can't stand it when people say that untreated depression usually resolves within however many months. If you are moderately depressed (poor concentration, poor sleep/excessive sleep), difficulty maintaining self-care/hygiene, you are pretty fucking miserable and could lose your job while waiting for the depression to resolve on its own.


Posted by: Bostoniangirl | Link to this comment | 01-26-15 6:53 PM
horizontal rule
12

Yes, plus it's often cyclical so resolving isn't permanent.


Posted by: Moby Hick | Link to this comment | 01-26-15 7:12 PM
horizontal rule
13

12: I was trying to be as charitable as possible to the argument and assuming, for the sake of argument, that it's a one-off episode of major depression.


Posted by: | Link to this comment | 01-26-15 7:26 PM
horizontal rule
14

13 was I.


Posted by: Bostniangirl | Link to this comment | 01-26-15 7:26 PM
horizontal rule
15

Psychiatric conditions in general are probably not places where 8 applies easily (though they are absolutely ones where people try to apply them). The danger is that with a lot of conditions, like earaches, if it doesn't go away but keeps getting worse it's not hard to tell. With a lot of mental health disorders it's almost the opposite since they tend to directly attack the stuff you need in order to accurately assess how bad things are, and how you should respond to them being that way.


Posted by: MHPH | Link to this comment | 01-26-15 7:37 PM
horizontal rule
16

The issue is that improvements on low probabilities are hard, isn't it? If you lower the probability of a bad outcome from 3 percent to 2 percent, that's a 33 percent reduction but a NNT of 100 people.

Personalized medicine is not likely to be too much better, there is lots of randomness in the world.


Posted by: Disingenuous Bastard | Link to this comment | 01-26-15 8:00 PM
horizontal rule