Taking aspirin daily has a 1-2000 chance of preventing your heart attack
January 28, 2015 3:41 AM Subscribe
This fundamental lesson is conveyed by a metric known as the number needed to treat, or N.N.T. Developed in the 1980s, the N.N.T. tells us how many people must be treated for one person to derive benefit. An N.N.T. of one would mean every person treated improves and every person not treated fails to, which is how we tend to think most therapies work.So it turns out that e.g. you need 2000 People to take a daily aspirin for two years to prevent one heart attack.
Currently the number needed for treatment is rarely given out, but there is a site for it, TheNNT.com. Sarah Fallon in Wired magazine explains what it does:
It’s unfortunate, then, that the NNT is not a statistic that’s routinely conveyed to either doctors or patients. But you can look it up on a site that you’ve probably never heard of: TheNNT.com. Started by David Newman, a director of clinical research at Icahn School of Medicine at Mount Sinai hospital, the site’s dozens of contributors analyze the available studies, crunch the numbers on benefits and harms, and then post the results. While a low NNT is generally “good” and a high NNT is “bad,” you also have to consider the severity of both the illness and the drug’s side effects. Which is why the team added a color-coding system: Green for when a treatment makes sense, yellow for when more study is needed, red for when the harms and the benefits cancel each other out, and black when the harms outweigh the benefits.For those in a hurry, The Bandolier, "an independent journal about evidence-based healthcare", has a handy little table with NNTs for common afflictions, while Wikipedia has a nice example table explaining what it all means.
Fairmettle, unless I'm misunderstanding your question, there are prominent harm statistics on the top of every page.
posted by Bugbread at 4:06 AM on January 28, 2015 [2 favorites]
posted by Bugbread at 4:06 AM on January 28, 2015 [2 favorites]
This is a great find. Thanks for posting.
posted by wittgenstein at 4:25 AM on January 28, 2015
posted by wittgenstein at 4:25 AM on January 28, 2015
The high NNT has been known to physicians for a while now, which is why use of Aspirin for primary prevention is controversial. The cohort used is middle aged adults with at least one cardiovascular risk factor.
However, if you take a cohort of patients who have just had a heart attack, the NNT drops to 26 in preventing death, stroke or a repeat heart attack.
posted by ianK at 4:26 AM on January 28, 2015 [40 favorites]
However, if you take a cohort of patients who have just had a heart attack, the NNT drops to 26 in preventing death, stroke or a repeat heart attack.
posted by ianK at 4:26 AM on January 28, 2015 [40 favorites]
The high NNT has been known to physicians for a while now, which is why use of Aspirin for primary prevention is controversial. The cohort used is middle aged adults with at least one cardiovascular risk factor.
I've been reading a lot of basic biostatistics material at work recently (I care about noninferiority and guess who else cares about noninferiority?) and aspirin is pretty much the go to example for why you ought to think about NNT.
(Amusingly, an NNT of 2000 does not mean "Taking aspirin daily has a 1-2000 chance of preventing your heart attack". Don't ask me to rewrite it before 7am, though.)
posted by hoyland at 4:51 AM on January 28, 2015 [1 favorite]
I've been reading a lot of basic biostatistics material at work recently (I care about noninferiority and guess who else cares about noninferiority?) and aspirin is pretty much the go to example for why you ought to think about NNT.
(Amusingly, an NNT of 2000 does not mean "Taking aspirin daily has a 1-2000 chance of preventing your heart attack". Don't ask me to rewrite it before 7am, though.)
posted by hoyland at 4:51 AM on January 28, 2015 [1 favorite]
The flu vaccine's NNT is 23.
Can I read that as saying that I'll need to get 23 years of flu vaccines to prevent one flu? Because that actually feels like it might not be worth it.
posted by anotherpanacea at 4:57 AM on January 28, 2015 [2 favorites]
Can I read that as saying that I'll need to get 23 years of flu vaccines to prevent one flu? Because that actually feels like it might not be worth it.
posted by anotherpanacea at 4:57 AM on January 28, 2015 [2 favorites]
I read that as saying get 23 flu shots per year. ; )
Can you imagine if the average American understood statistical evidence-based assessment (in general) well enough to think clearly about risk in all categories? Different society, it would be.
That's why they don't really teach kids math in schools. Keeps 'em guessing and going by their gut feelings, like real Americans.
posted by spitbull at 5:04 AM on January 28, 2015 [8 favorites]
Can you imagine if the average American understood statistical evidence-based assessment (in general) well enough to think clearly about risk in all categories? Different society, it would be.
That's why they don't really teach kids math in schools. Keeps 'em guessing and going by their gut feelings, like real Americans.
posted by spitbull at 5:04 AM on January 28, 2015 [8 favorites]
Not exactly, no. It means that for every 23 flu vaccines administered in the population, we can expect one less case of flu, but your specific risk will vary.
posted by Nothing at 5:06 AM on January 28, 2015 [14 favorites]
posted by Nothing at 5:06 AM on January 28, 2015 [14 favorites]
I suspect that if you limited the cohort to all people with hypertension and middle-aged and older men you'd get a much more attractive NNT. Anyhow having gotten a spiffy new stent for my 50th birthday last year (fortunately I noticed the blood pressure spike instead of waiting for an actual heart attack) I'll keep taking the aspirin, thanks.
posted by localroger at 5:13 AM on January 28, 2015 [4 favorites]
posted by localroger at 5:13 AM on January 28, 2015 [4 favorites]
Let me get this out of the way up front: NNT is a good number to look at. When you are talking about lowering heart attacks from 2.7 to 2.6% with a drug that causes major side effects in 5% of people, there is that trade off of 0.1% reduction versus 5% increase in toxicity. In terms of NNT you treated 1000 to get one benefit while getting 50 experiencing toxicity.
However, this story is misleading, obfuscating the fact that low dose aspirin is very good. The NYT article says this:
... it is possible to calculate the chance that a person will have a first heart attack in the next 10 years. The American Heart Association recommends that people who have more than a 10 percent chance take a daily aspirin to avoid that heart attack.
How effective is aspirin for that aim? According to clinical trials, if about 2,000 people follow these guidelines over a two-year period, one additional first heart attack will be prevented.
That doesn’t mean the 1,999 other people have heart attacks. The fact is, on average about 3.6 of them would have a first heart attack regardless of whether they took the aspirin.
Okay, first of all, you are talking about people with a 10% chance or greater possibility over these 10 years. For 2000 people, that's 200 heart attacks at least. They only focus on the two years and point out that 3.6 people out of 2000 would have heart attacks. That a rate of 0.18%. The mini-aspirin reduces that by one person to 2.6 people. That's a reduction down to 0.13% or a 27.7% reduction in the number of people. If aspirin decreases heart attacks by 27.7% over the ten years, that's a reduction of heart attacks in 54 people out of 200.
I'm not sure how they jibe the 0.18% heart attack rate with the over 10% they mention in the beginning. Do the remaining 9.82% (or greater) happen between 2 years and 10 years? (Really, this is very strange. 10% chance of a heart attack and then only 0.18% in the first two years?)
Furthermore, mini-aspirins also reduce strokes and colon cancer.
posted by dances_with_sneetches at 5:15 AM on January 28, 2015 [30 favorites]
However, this story is misleading, obfuscating the fact that low dose aspirin is very good. The NYT article says this:
... it is possible to calculate the chance that a person will have a first heart attack in the next 10 years. The American Heart Association recommends that people who have more than a 10 percent chance take a daily aspirin to avoid that heart attack.
How effective is aspirin for that aim? According to clinical trials, if about 2,000 people follow these guidelines over a two-year period, one additional first heart attack will be prevented.
That doesn’t mean the 1,999 other people have heart attacks. The fact is, on average about 3.6 of them would have a first heart attack regardless of whether they took the aspirin.
Okay, first of all, you are talking about people with a 10% chance or greater possibility over these 10 years. For 2000 people, that's 200 heart attacks at least. They only focus on the two years and point out that 3.6 people out of 2000 would have heart attacks. That a rate of 0.18%. The mini-aspirin reduces that by one person to 2.6 people. That's a reduction down to 0.13% or a 27.7% reduction in the number of people. If aspirin decreases heart attacks by 27.7% over the ten years, that's a reduction of heart attacks in 54 people out of 200.
I'm not sure how they jibe the 0.18% heart attack rate with the over 10% they mention in the beginning. Do the remaining 9.82% (or greater) happen between 2 years and 10 years? (Really, this is very strange. 10% chance of a heart attack and then only 0.18% in the first two years?)
Furthermore, mini-aspirins also reduce strokes and colon cancer.
posted by dances_with_sneetches at 5:15 AM on January 28, 2015 [30 favorites]
This is super fascinating stuff. I want MOAR NNTs! Also, I want to see more of the PA and PB information. It's obvious that one NNT can mean very different things based on what the outcomes without treatment would be. That's the danger of reducing things down to one number. Yes, an NNT of 2000 sounds pathetic, but as pointed out above, "one out of five heart attacks can be prevented with this treatment" is pretty staggering, and also true based on the data I'm seeing. (And I am also curious about how this "group of patients with a 10% risk of heart attack over 10 years" ends up with this low of a heart attack rate, even over two years.)
I just woke up and the coffee's still working, so it's possible I'm Just Not Getting It yet...
posted by obfuscation at 5:34 AM on January 28, 2015 [2 favorites]
I just woke up and the coffee's still working, so it's possible I'm Just Not Getting It yet...
posted by obfuscation at 5:34 AM on January 28, 2015 [2 favorites]
the NNT drops to 26 in preventing death, stroke or a repeat heart attack
Oh, that'll never work as click bait.
posted by DarkForest at 5:39 AM on January 28, 2015 [1 favorite]
Oh, that'll never work as click bait.
posted by DarkForest at 5:39 AM on January 28, 2015 [1 favorite]
The second thing is that asprin is *dirt cheap* to implement. Patient can self-dose, no patent, made by the decatonne, and the side effect profile (bleeding in stomachs) is well known - the statistics are in and appear not to mess with mortality or other treatment necessities; the cost to treat per patient per year should be about $10, TOPS!
If this were an on-patent antibody at $50k per treatment for a non-lethal lifestyle ailment , then 2k NNT would be absurd. But these are heart attacks!
Anyone with cardiovascular risk profile; certainly anyone who has had a cardiovascular event, NOT taking Asprins is doing themselves a great disservice.
* Also the analogy with vaccination should stop; no herd immunity, heart attacks are not transmissible diseases**, no wild reservoir of mutating virii....
** outside of receiving medical bills in the USA, that is!
posted by lalochezia at 5:41 AM on January 28, 2015 [8 favorites]
If this were an on-patent antibody at $50k per treatment for a non-lethal lifestyle ailment , then 2k NNT would be absurd. But these are heart attacks!
Anyone with cardiovascular risk profile; certainly anyone who has had a cardiovascular event, NOT taking Asprins is doing themselves a great disservice.
* Also the analogy with vaccination should stop; no herd immunity, heart attacks are not transmissible diseases**, no wild reservoir of mutating virii....
** outside of receiving medical bills in the USA, that is!
posted by lalochezia at 5:41 AM on January 28, 2015 [8 favorites]
Why don't medicines work on everyone?
(This is not doubting the analysis above, but a genuine question about different pharmacological interactions with our bodies/internal chemistry.)
posted by eyeofthetiger at 5:53 AM on January 28, 2015
(This is not doubting the analysis above, but a genuine question about different pharmacological interactions with our bodies/internal chemistry.)
posted by eyeofthetiger at 5:53 AM on January 28, 2015
I'm glad you posted this – I wanted to, but I've known Dr. David Newman since he was in short pants. If you like exploring the data behind common medical interventions you should also check out his podcast, SmartEM. And note also that the developer and technical lead of TheNNT.com is MetaFilter's Own gramcracker!
posted by nicwolff at 6:05 AM on January 28, 2015 [10 favorites]
posted by nicwolff at 6:05 AM on January 28, 2015 [10 favorites]
Tight Glycemic Control for Type 2 Diabetes (Over Five Years)
Well, I feel better about my A1C numbers now.
posted by Foosnark at 6:13 AM on January 28, 2015 [2 favorites]
Well, I feel better about my A1C numbers now.
posted by Foosnark at 6:13 AM on January 28, 2015 [2 favorites]
Yeah, the A1C thing totally jumped out at me too (probably because, being a dietitian, it's one of the few that actually applies to my practice), and it was pretty shocking. It definitely doesn't match what I've seen (dozens of poorly managed diabetics with amputations, versus very, very few hypoglycemia admissions), but my experience isn't a randomized controlled study either. But again, it does point out the danger in giving one number too much credit. I was once commending a patient on his good A1C (<7), and his wife was like, "yeah, well, I wake him up with a blood sugar of 40 around 3am every morning, so..."
posted by obfuscation at 6:21 AM on January 28, 2015
posted by obfuscation at 6:21 AM on January 28, 2015
Can you imagine if the average American understood statistical evidence-based assessment (in general) well enough to think clearly about risk in all categories? Different society, it would be.
Not to derail this (very cool) conversation I just wanted to chime in and say that pretty much any developed nation would be 100% better off if we taught stats in early high school, even if we replaced calculus with stats.
posted by digitalprimate at 6:30 AM on January 28, 2015 [16 favorites]
Not to derail this (very cool) conversation I just wanted to chime in and say that pretty much any developed nation would be 100% better off if we taught stats in early high school, even if we replaced calculus with stats.
posted by digitalprimate at 6:30 AM on January 28, 2015 [16 favorites]
This is very interesting, thanks for posting!
On switching to a Mediterranean diet: Maybe it prevents only 1 heart attack in 30 people, but I think that particular statistic disregards that there may be other health benefits.
Not to mention the benefit that a Mediterranean diet tastes much better than what most people shove into their mouths...
posted by sour cream at 6:34 AM on January 28, 2015 [2 favorites]
On switching to a Mediterranean diet: Maybe it prevents only 1 heart attack in 30 people, but I think that particular statistic disregards that there may be other health benefits.
Not to mention the benefit that a Mediterranean diet tastes much better than what most people shove into their mouths...
posted by sour cream at 6:34 AM on January 28, 2015 [2 favorites]
I'm glad you posted this. When it was mentioned in Wired a while back, heavy traffic crashed the site (or made it impossibly slow) so I couldn't check it out.
posted by klausman at 6:36 AM on January 28, 2015
posted by klausman at 6:36 AM on January 28, 2015
Math is hard.
posted by allthinky at 9:34 AM on January 28 [+] [!]
Eponysterical?
posted by ssmug at 6:39 AM on January 28, 2015
posted by allthinky at 9:34 AM on January 28 [+] [!]
Eponysterical?
posted by ssmug at 6:39 AM on January 28, 2015
the cost to treat per patient per year should be about $10, TOPS!
So, $20 for two years, which prevents one heart attack in 2000 people. 20x2000=$40,000.
I suspect this is much cheaper than the cost to treat one heart attack.
posted by eriko at 6:40 AM on January 28, 2015 [3 favorites]
So, $20 for two years, which prevents one heart attack in 2000 people. 20x2000=$40,000.
I suspect this is much cheaper than the cost to treat one heart attack.
posted by eriko at 6:40 AM on January 28, 2015 [3 favorites]
This is not directly relevant, but... I am a 51 year-old-male. The only risk factor for heart attack that I have is high cholesterol, for which I've taken a statin for several years--and have had problems with mild aches and pains as a side effect. Otherwise, there's no history of heart attack in my family, I exercise vigorously (running) regularly, eat relatively well, am not obese, don't smoke.
This past year at my annual physical exam, my physician suggested that I get a heart CT exam to see if I have any plaque in my arteries. The results came back with the lowest possible score, meaning no plaque at all! So, after the results were sent to my doctor, I called her office to see what she had to say. I assumed she would tell me I could stop taking statins. But she told me to continue! I decided to stop the statins anyway; I have not seen my doctor again since then but plan to discuss this with her next time I see her.
posted by tippiedog at 6:57 AM on January 28, 2015
This past year at my annual physical exam, my physician suggested that I get a heart CT exam to see if I have any plaque in my arteries. The results came back with the lowest possible score, meaning no plaque at all! So, after the results were sent to my doctor, I called her office to see what she had to say. I assumed she would tell me I could stop taking statins. But she told me to continue! I decided to stop the statins anyway; I have not seen my doctor again since then but plan to discuss this with her next time I see her.
posted by tippiedog at 6:57 AM on January 28, 2015
By the way, I've also taken a lose-dose aspirin daily for several years now. I'll probably continue. It may not help me, but the risks of it harming me seem to be extremely low as well, and the cost is negligible to me.
posted by tippiedog at 6:58 AM on January 28, 2015
posted by tippiedog at 6:58 AM on January 28, 2015
The NNT for statins is pretty bad, especially in patients with no prior history of heart disease. The failure of patients (and far too many doctors) to understand how ineffective they are for most people has led to wild overuse. It made pharma companies a lot of money, though. This quote pretty well sums up the math:
paid gave the American Heart Association (and a lot of doctors on certain key panels and committees) a lot of money. Not long after that, the AHA recommended that t-PA be used in stroke cases. Following the AHA's recommendation, the American College of Emergency Physicians did the same. This made it the de facto standard of care, making doctors who didn't use it vulnerable to malpractice cases. Even though science is on their side, it's an expensive thing to fight.
posted by Bobby Rijndael at 7:06 AM on January 28, 2015 [4 favorites]
Several recent scientific papers peg the NNT for statins at 250 and up for lower-risk patients, even if they take it for five years or more. "What if you put 250 people in a room and told them they would each pay $1,000 a year for a drug they would have to take every day, that many would get diarrhea and muscle pain, and that 249 would have no benefit? And that they could do just as well by exercising? How many would take that?" asks drug industry critic Dr. Jerome R. Hoffman, professor of clinical medicine at the University of California at Los Angeles.Another tragic case is thrombolytics (clot-dissolving) drugs for strokes. There is very solid data (especially once you factor out manufacturer-supported studies) that it does no good and quite a lot of harm. Genentech, the manufacturer of the main thrombolytic drug in question (t-PA),
posted by Bobby Rijndael at 7:06 AM on January 28, 2015 [4 favorites]
I take aspirin for cancer prevention, and I try to keep in mind the NNH (number needed to harm) mentioned upthread.
The aspirin puts me at some higher risk of having a "major bleeding event". Hard to know what the NNH is in patients like me. thennt.com suggests that "1 in 3333 were harmed (major bleeding event requiring hospital admission and transfusion)". GI hemorrhage would be bad, but the major bleeding event I am most concerned about is hemorrhagic stroke.
I never do contact sports, and I rarely climb ladders, but I make sure to stop my aspirin a couple of weeks before any planned activity where I might bump my head.
We had a conference today where the presenter talked about NSAIDS for postoperative pain and mentioned the NNH for other bad postoperative outcomes for these drugs. Good to see the idea getting a little more exposure.
posted by etherist at 7:07 AM on January 28, 2015
The aspirin puts me at some higher risk of having a "major bleeding event". Hard to know what the NNH is in patients like me. thennt.com suggests that "1 in 3333 were harmed (major bleeding event requiring hospital admission and transfusion)". GI hemorrhage would be bad, but the major bleeding event I am most concerned about is hemorrhagic stroke.
I never do contact sports, and I rarely climb ladders, but I make sure to stop my aspirin a couple of weeks before any planned activity where I might bump my head.
We had a conference today where the presenter talked about NSAIDS for postoperative pain and mentioned the NNH for other bad postoperative outcomes for these drugs. Good to see the idea getting a little more exposure.
posted by etherist at 7:07 AM on January 28, 2015
Also from thennt.com, something that I don't think cardiac surgeons explain well to patients: the increase in longevity from coronary bypass operations isn't very good, and there are a lot of bad outcomes. Some people are never the same cognitively after major surgery, especially open-heart surgery.
The good
1 in 25 were helped (prevented death)
1 in 10-14 were helped (prevented non-fatal heart attack)
(not mentioned: does seem to increase quality of life by decreasing frequency/severity of disabling chest pain)
The unintended consequences
1 in 83 were harmed (death)
1 in 100 were harmed (stroke)
1 in 43 were harmed (kidney failure)
1 in 28 were harmed (re-operation)
1 in 14 were harmed (extended life support)
1 in 3-5 were harmed (cognitive decline)
posted by etherist at 7:11 AM on January 28, 2015 [1 favorite]
The good
1 in 25 were helped (prevented death)
1 in 10-14 were helped (prevented non-fatal heart attack)
(not mentioned: does seem to increase quality of life by decreasing frequency/severity of disabling chest pain)
The unintended consequences
1 in 83 were harmed (death)
1 in 100 were harmed (stroke)
1 in 43 were harmed (kidney failure)
1 in 28 were harmed (re-operation)
1 in 14 were harmed (extended life support)
1 in 3-5 were harmed (cognitive decline)
posted by etherist at 7:11 AM on January 28, 2015 [1 favorite]
Bobby Rijndael: The NNT for statins is pretty bad, especially in patients with no prior history of heart disease.
That was my vague understanding as well, which is why I was surprised when my doctor told me to continue on them. At least this discussion helps me to have the proper terminology when I do discuss with my physician.
posted by tippiedog at 7:20 AM on January 28, 2015
That was my vague understanding as well, which is why I was surprised when my doctor told me to continue on them. At least this discussion helps me to have the proper terminology when I do discuss with my physician.
posted by tippiedog at 7:20 AM on January 28, 2015
etherist: Also from thennt.com, something that I don't think cardiac surgeons explain well to patients: the increase in longevity from coronary bypass operations isn't very good, and there are a lot of bad outcomes. Some people are never the same cognitively after major surgery, especially open-heart surgery.
My father-in-law had a heart attack in his late 30s in the late 1970s. At the time, open heart surgery was just gaining widespread publicity. But my FiL, to his credit, saw the stats and therefore chose to limp along with nitro for another 10 years before he absolutely had to have bypass surgery. He lived another 10 years and died of cancer.
posted by tippiedog at 7:24 AM on January 28, 2015
My father-in-law had a heart attack in his late 30s in the late 1970s. At the time, open heart surgery was just gaining widespread publicity. But my FiL, to his credit, saw the stats and therefore chose to limp along with nitro for another 10 years before he absolutely had to have bypass surgery. He lived another 10 years and died of cancer.
posted by tippiedog at 7:24 AM on January 28, 2015
"Turns out."
posted by entropicamericana at 7:35 AM on January 28, 2015
posted by entropicamericana at 7:35 AM on January 28, 2015
Can you imagine if the average American understood statistical evidence-based assessment (in general) well enough to think clearly about risk in all categories? Different society, it would be.
Not to derail this (very cool) conversation I just wanted to chime in and say that pretty much any developed nation would be 100% better off if we taught stats in early high school, even if we replaced calculus with stats.
I'm pretty sure probability is part of algebra II, which is pre pre calculus. It's just not emphasized the way we are thinking. My algebra II teacher made every problem about horse races...that's a good history lesson, but medical stats would be more relevant to kids, i think.
If I were homeschooling, though, I could use this site in a lesson. cool site!
posted by eustatic at 7:53 AM on January 28, 2015 [1 favorite]
Not to derail this (very cool) conversation I just wanted to chime in and say that pretty much any developed nation would be 100% better off if we taught stats in early high school, even if we replaced calculus with stats.
I'm pretty sure probability is part of algebra II, which is pre pre calculus. It's just not emphasized the way we are thinking. My algebra II teacher made every problem about horse races...that's a good history lesson, but medical stats would be more relevant to kids, i think.
If I were homeschooling, though, I could use this site in a lesson. cool site!
posted by eustatic at 7:53 AM on January 28, 2015 [1 favorite]
I'm pretty sure probability is part of algebra II
I sure never had any probability in Algebra II. A2 was basically "intro to derivative calculus". I think that is pretty typical.
The only people who I suspect get much exposure to probability at all are people who take a college stats class, which is to say people studying the social sciences or bio, maybe chem. (Stats wasn't required when I was in undergrad as a physics major, which even now I think is absurd given how much modern physics relies on statistics, but there you go.)
Honestly, I think probability and statistics is far more useful than calculus; I think if we have to choose between the two, I'd rather make sure everyone has a grounding in stats and then teach calc to the students who really need it (physics and some other hard sciences, largely) than the other way around. And I like calculus; it's beautiful and elegant and simple once you get the hang of it... but as a life skill it's nothing like being able to understand probabilities and meaningfully interpret statistics.
posted by Kadin2048 at 8:55 AM on January 28, 2015 [3 favorites]
I sure never had any probability in Algebra II. A2 was basically "intro to derivative calculus". I think that is pretty typical.
The only people who I suspect get much exposure to probability at all are people who take a college stats class, which is to say people studying the social sciences or bio, maybe chem. (Stats wasn't required when I was in undergrad as a physics major, which even now I think is absurd given how much modern physics relies on statistics, but there you go.)
Honestly, I think probability and statistics is far more useful than calculus; I think if we have to choose between the two, I'd rather make sure everyone has a grounding in stats and then teach calc to the students who really need it (physics and some other hard sciences, largely) than the other way around. And I like calculus; it's beautiful and elegant and simple once you get the hang of it... but as a life skill it's nothing like being able to understand probabilities and meaningfully interpret statistics.
posted by Kadin2048 at 8:55 AM on January 28, 2015 [3 favorites]
The unintended consequences
1 in 83 were harmed (death)
1 in 100 were harmed (stroke)
1 in 43 were harmed (kidney failure)
1 in 28 were harmed (re-operation)
1 in 14 were harmed (extended life support)
1 in 3-5 were harmed (cognitive decline)
As someone who had quadruple bypass 10 years ago, this doesn't make me feel all warm and fuzzy.
posted by briank at 9:47 AM on January 28, 2015
1 in 83 were harmed (death)
1 in 100 were harmed (stroke)
1 in 43 were harmed (kidney failure)
1 in 28 were harmed (re-operation)
1 in 14 were harmed (extended life support)
1 in 3-5 were harmed (cognitive decline)
As someone who had quadruple bypass 10 years ago, this doesn't make me feel all warm and fuzzy.
posted by briank at 9:47 AM on January 28, 2015
"Teach stats" is the wrong message to take from this. Most doctors take stats classes and absolutely do not internalize them.
Conditional probability and risk assessment are really quite hard things to do naturally. I would guess that no human is ever going to be able to do it.
posted by TypographicalError at 9:49 AM on January 28, 2015
Conditional probability and risk assessment are really quite hard things to do naturally. I would guess that no human is ever going to be able to do it.
posted by TypographicalError at 9:49 AM on January 28, 2015
Form implementation intentions: "If I am considering a treatment, ask about its efficacy and NNT."
posted by anotherpanacea at 9:51 AM on January 28, 2015
posted by anotherpanacea at 9:51 AM on January 28, 2015
Reducing the outcome of a study to a single number is dangerous, and doubly so when it comes to informing a patient without the research or clinical background necessary to understand the context of that number. The methods of a given study are hugely important for determining its validity and relevance.
For example, the results indicating low effects on mortality with statins in patients without preexisting cardiovascular disease are derived from meta-analyses encompassing a large number of separate clinical trials, many of which had conflicting results. So while the consensus has been slowly moving towards a lack of mortality benefit in low risk patients, it has vacillated considerably over the past ten years and is still not definitive. The picture is further complicated by the fact that statins have clear benefits in reducing known risk factors for cardiovascular disease (e.g., LDL levels, plaque instability, vascular inflammation, etc)—which may or may not translate to direct effects on mortality, but are nonetheless promising from a clinical standpoint.
posted by dephlogisticated at 9:55 AM on January 28, 2015 [5 favorites]
For example, the results indicating low effects on mortality with statins in patients without preexisting cardiovascular disease are derived from meta-analyses encompassing a large number of separate clinical trials, many of which had conflicting results. So while the consensus has been slowly moving towards a lack of mortality benefit in low risk patients, it has vacillated considerably over the past ten years and is still not definitive. The picture is further complicated by the fact that statins have clear benefits in reducing known risk factors for cardiovascular disease (e.g., LDL levels, plaque instability, vascular inflammation, etc)—which may or may not translate to direct effects on mortality, but are nonetheless promising from a clinical standpoint.
posted by dephlogisticated at 9:55 AM on January 28, 2015 [5 favorites]
I am terrible with math and this stuff unsettles me, but suffice it to say that I know many folks who take baby aspirin not just as a blood thinner--that is, to prevent heart attacks--but also because it reduces inflamation, a cause, it seems, of other health problems
posted by Postroad at 10:58 AM on January 28, 2015
posted by Postroad at 10:58 AM on January 28, 2015
Reducing the outcome of a study to a single number is dangerous, and doubly so when it comes to informing a patient without the research or clinical background necessary to understand the context of that number. The methods of a given study are hugely important for determining its validity and relevance.
For example, the results indicating low effects on mortality with statins in patients without preexisting cardiovascular disease are derived from meta-analyses encompassing a large number of separate clinical trials, many of which had conflicting results. So while the consensus has been slowly moving towards a lack of mortality benefit in low risk patients, it has vacillated considerably over the past ten years and is still not definitive. The picture is further complicated by the fact that statins have clear benefits in reducing known risk factors for cardiovascular disease (e.g., LDL levels, plaque instability, vascular inflammation, etc)—which may or may not translate to direct effects on mortality, but are nonetheless promising from a clinical standpoint. (my emphasis bolded)
I agree with paragraph 1, but NOT with paragraph 2.
No one cares about clinical biomarkers unless they lead to i) mortality reduction or ii) quality of life improvement.
e.g. The obsession with LDL (an example of a 'paragraph 1 error' as practiced by doctors) as the be-all and end-all has led to a vast overprescription of statins, costing us large amounts of money and no little misery.
If I lower LDL in 100k people saving 20 heart attacks - 10 of which might be fatal in 10 years, but the drug causes 19 people to die 10 years earlier of diabetes, then NO ONE CARES and we've spent a lot of money. People usually don't care if they die early of diabetes or a heart attack.
Show me with studies designed correctly that these biomarkers' reduction evinces mortality/QoL promise or stop using the biomarkers as sole justification for vast overprescription orgies bankrolled by big pharma.
I say this as someone who is supportive of the drug industry!
posted by lalochezia at 11:32 AM on January 28, 2015 [1 favorite]
GAAAAAAAH I so want to reformat that Bandolier HTML table to make it more readable. GAAAAAAAAAH
posted by potsmokinghippieoverlord at 11:36 AM on January 28, 2015 [1 favorite]
posted by potsmokinghippieoverlord at 11:36 AM on January 28, 2015 [1 favorite]
I wish the NNT site addressed psychiatric medications, which are some of the most profitable.
posted by OmieWise at 12:04 PM on January 28, 2015 [2 favorites]
posted by OmieWise at 12:04 PM on January 28, 2015 [2 favorites]
No one cares about clinical biomarkers unless they lead to i) mortality reduction or ii) quality of life improvement.
That pretty much goes without saying. The point is that the relationships between disease pathology, biomarkers, and mortality are not at all straightforward to determine. Clinical studies routinely produce results that contradict each other. Any one study has the risk of being biased, flawed, or otherwise invalid. To help mitigate the risk of drawing incorrect conclusions, you look for studies which approach the problem from different angles and produce mutually-supportive conclusions. So ideally you want to see independently-verified relationships between heart disease and LDL, between LDL and mortality, between statin therapy and decreased LDL, and between statin treatment and mortality. It will inevitably be a mix of positive and negative results. You take all of that evidence together and try to piece together an evolving consensus of how the disease works and how to treat it, knowing that any one of those relationships may eventually prove to be false.
posted by dephlogisticated at 1:52 PM on January 28, 2015
That pretty much goes without saying. The point is that the relationships between disease pathology, biomarkers, and mortality are not at all straightforward to determine. Clinical studies routinely produce results that contradict each other. Any one study has the risk of being biased, flawed, or otherwise invalid. To help mitigate the risk of drawing incorrect conclusions, you look for studies which approach the problem from different angles and produce mutually-supportive conclusions. So ideally you want to see independently-verified relationships between heart disease and LDL, between LDL and mortality, between statin therapy and decreased LDL, and between statin treatment and mortality. It will inevitably be a mix of positive and negative results. You take all of that evidence together and try to piece together an evolving consensus of how the disease works and how to treat it, knowing that any one of those relationships may eventually prove to be false.
posted by dephlogisticated at 1:52 PM on January 28, 2015
So what I'm getting is that it's really complicated to make things simple.
posted by Sebmojo at 5:06 PM on January 28, 2015
posted by Sebmojo at 5:06 PM on January 28, 2015
anotherpanacea: The flu vaccine's NNT is 23.
Are you referring to "Neuraminidase Inhibitors Given for Influenza"? (I don't see the number 23 there, but I also don't see any other articles about influenza or flu across the entire site...)
posted by snap, crackle and pop at 5:13 PM on January 28, 2015
Are you referring to "Neuraminidase Inhibitors Given for Influenza"? (I don't see the number 23 there, but I also don't see any other articles about influenza or flu across the entire site...)
posted by snap, crackle and pop at 5:13 PM on January 28, 2015
The flu vax number is on the Bandolier site, which is linked toward the end of the FPP.
The problem with probability like this is that it's impersonal. It's good as a public health tool, but I may still want to bet on the vax or test (psa) for my own well-being.
posted by OmieWise at 5:45 PM on January 28, 2015
The problem with probability like this is that it's impersonal. It's good as a public health tool, but I may still want to bet on the vax or test (psa) for my own well-being.
posted by OmieWise at 5:45 PM on January 28, 2015
Ahhh, thanks for that.
posted by snap, crackle and pop at 5:46 PM on January 28, 2015
posted by snap, crackle and pop at 5:46 PM on January 28, 2015
it's really complicated to make things simple
Well it's only complicated when the things you want to make simple are actually complicated.
posted by localroger at 5:48 PM on January 28, 2015
Well it's only complicated when the things you want to make simple are actually complicated.
posted by localroger at 5:48 PM on January 28, 2015
Also the analogy with vaccination should stop; no herd immunity, heart attacks are not transmissible diseases...
Well, not yet anyway. Mwahahahaha... Uh, er, I mean How about those red birds?
posted by Kid Charlemagne at 7:03 PM on January 28, 2015
Well, not yet anyway. Mwahahahaha... Uh, er, I mean How about those red birds?
posted by Kid Charlemagne at 7:03 PM on January 28, 2015
laloachezia:
The obsession with LDL (an example of a 'paragraph 1 error' as practiced by doctors) as the be-all and end-all has led to a vast overprescription of statins...
There's also this:
Do statins have a role in primary prevention? An update.
and...
This Letter examines proven and associated harms with statin use. Proven harms are those that have been established in systematic reviews or randomized controlled trials (RCTs); associated harms are those identified from observational studies, case series and case reports.
...now with podcast!
posted by mandolin conspiracy at 10:22 PM on January 28, 2015
The obsession with LDL (an example of a 'paragraph 1 error' as practiced by doctors) as the be-all and end-all has led to a vast overprescription of statins...
There's also this:
Do statins have a role in primary prevention? An update.
and...
This Letter examines proven and associated harms with statin use. Proven harms are those that have been established in systematic reviews or randomized controlled trials (RCTs); associated harms are those identified from observational studies, case series and case reports.
...now with podcast!
posted by mandolin conspiracy at 10:22 PM on January 28, 2015
I wonder how you effectively use the NNT numbers to also consider the benefits of partial payoffs. If 1 heart attack is prevented are others simply lessened in severity? Bodies aren't binary, after all.
posted by phearlez at 1:46 PM on January 29, 2015
posted by phearlez at 1:46 PM on January 29, 2015
OmieWise: "I wish the NNT site addressed psychiatric medications, which are some of the most profitable."
You can get more NNTs about many aspects of antidepressants from meta-analyses like this, but in general, NNTs usually range between 5.5 (instrumented response) and 8-10 (remission). There's usually a 15-25% diff between placebo and active, but usually only if you screen out mod/mild MDD and below. The main issue I see as a psychiatrist is so many people without MDD mod-to-severe are prescribed antidepressants as a kind of knee-jerk response, and there's basically no evidence base showing efficacy for this (NNT is very high, NNH is unacceptably low). There are also big differences in NNTS for age and gender/hormonal cohorts.
The relatively low NNTs for many psych meds, combined with the high prevalence of disorders, actually makes them reasonably cost effective on a DALY basis (MH costs exceeding cardiovascular, cancer, diabetes and respiratory), and exceptionally so if you use only older, proven generics vs newer, experimental branded drugs.
posted by meehawl at 2:46 PM on January 29, 2015 [2 favorites]
You can get more NNTs about many aspects of antidepressants from meta-analyses like this, but in general, NNTs usually range between 5.5 (instrumented response) and 8-10 (remission). There's usually a 15-25% diff between placebo and active, but usually only if you screen out mod/mild MDD and below. The main issue I see as a psychiatrist is so many people without MDD mod-to-severe are prescribed antidepressants as a kind of knee-jerk response, and there's basically no evidence base showing efficacy for this (NNT is very high, NNH is unacceptably low). There are also big differences in NNTS for age and gender/hormonal cohorts.
The relatively low NNTs for many psych meds, combined with the high prevalence of disorders, actually makes them reasonably cost effective on a DALY basis (MH costs exceeding cardiovascular, cancer, diabetes and respiratory), and exceptionally so if you use only older, proven generics vs newer, experimental branded drugs.
posted by meehawl at 2:46 PM on January 29, 2015 [2 favorites]
Perhaps it's inappropriate to offer my anecdote apropos of a stats post, but basically for the last year or so I've been working on my father-in-law to re-examine his statin use. He's in his mid-eighties, and has been suffering from low energy, leg aches, and other issues for years, in a steadily increasing trajectory. Because of his age, he and the whole family basically just assumed it was part of the long slow slide from old age unto, eventually, death. It was really depressing, in part because he was apparently both strong and mentally with-it, just unable to do more than an hour's activity without hours of lying in bed afterwards.
Since it was so slow in coming on, for me, at a couple degrees remove, I didn't see it as a problem to tackle until for whatever reason a year ago I got fed up (at fucking-death-and-disease, that is) and started doing some research. Which by various paths led me to studying statins along with all the other drugs he was on. I didn't hit on the NNT stuff until late though -- to get there I first read all the various statin papers, the odds ratios, the (few) side-effects papers, etc., and only after I did all the math myself did I realize that though those odds ratios were strongly significant and even substantively significant (a factor of 2, eg), the absolute change was tiny. And the side effects were huge! I was so angry I could hardly see straight. God knows whether it had anything to do with his case, but how the hell can they be so cavalierly prescribing these things to almost everyone when the improvements (in lower-risk groups) were so small and the side effects both prevalent and really quite serious??
And then, once I had read up on it all (and finally fount the NNT stuff to back me up), it was a bit scary to recommend to my spouse's father that he stop taking something that could seriously affect his chance of heart attack, on the off chance that his just-getting-old symptoms happen to match a few documented side effects. I'm still pretty afraid of that even now, 5 months after he stopped the statins and for the first time in nearly a decade can actually spend an afternoon in the garden, take a walk around the neighborhood, drive out to visit one of his kids for lunch, and maybe even take a trip somewhere again. Every time we see him he can't stop singing the praises of getting off the statins (which he thinks was his idea; I have no problem with that!), and you can see why: it's not just that he's feeling better, it's that the terrifying inevitable-decline-unto-death has been reprieved, however briefly. He's said that if he gets a heart attack in the future it will still have been worth it, though god knows none of us can make such predictions about future attitudes. But in any case, my own take-away is still fury at the medical/pharmaceutical industry for deliberately obfuscating these basic statistics in the name of selling really powerful and dangerous drugs. And sure, maybe there's some placebo effect (though god knows a dozen other things were tried first), and maybe this is just a one-off anecdote. But I'm a Bayesian -- even if there's just a 10% chance it really was the statins, I'm still pretty damn mad.
posted by chortly at 5:58 PM on January 29, 2015 [4 favorites]
Since it was so slow in coming on, for me, at a couple degrees remove, I didn't see it as a problem to tackle until for whatever reason a year ago I got fed up (at fucking-death-and-disease, that is) and started doing some research. Which by various paths led me to studying statins along with all the other drugs he was on. I didn't hit on the NNT stuff until late though -- to get there I first read all the various statin papers, the odds ratios, the (few) side-effects papers, etc., and only after I did all the math myself did I realize that though those odds ratios were strongly significant and even substantively significant (a factor of 2, eg), the absolute change was tiny. And the side effects were huge! I was so angry I could hardly see straight. God knows whether it had anything to do with his case, but how the hell can they be so cavalierly prescribing these things to almost everyone when the improvements (in lower-risk groups) were so small and the side effects both prevalent and really quite serious??
And then, once I had read up on it all (and finally fount the NNT stuff to back me up), it was a bit scary to recommend to my spouse's father that he stop taking something that could seriously affect his chance of heart attack, on the off chance that his just-getting-old symptoms happen to match a few documented side effects. I'm still pretty afraid of that even now, 5 months after he stopped the statins and for the first time in nearly a decade can actually spend an afternoon in the garden, take a walk around the neighborhood, drive out to visit one of his kids for lunch, and maybe even take a trip somewhere again. Every time we see him he can't stop singing the praises of getting off the statins (which he thinks was his idea; I have no problem with that!), and you can see why: it's not just that he's feeling better, it's that the terrifying inevitable-decline-unto-death has been reprieved, however briefly. He's said that if he gets a heart attack in the future it will still have been worth it, though god knows none of us can make such predictions about future attitudes. But in any case, my own take-away is still fury at the medical/pharmaceutical industry for deliberately obfuscating these basic statistics in the name of selling really powerful and dangerous drugs. And sure, maybe there's some placebo effect (though god knows a dozen other things were tried first), and maybe this is just a one-off anecdote. But I'm a Bayesian -- even if there's just a 10% chance it really was the statins, I'm still pretty damn mad.
posted by chortly at 5:58 PM on January 29, 2015 [4 favorites]
Fascinating stuff, especially as I am currently working through disentangling the side effects from the blood pressure and cholesterol drugs I have been taking since a clot in my coronary artery a decade ago (which required a stent) , and the chemo treatment for cancer a year ago. In the last year, after an initial surge after my chemo ended, I've suffered a number of nagging complaints (e.g. fatigue, muscle pain, pre-diabetic symptoms despite considerable weight loss) that were initially inexplicable but which are now looking more and more like they are linked to the beta blockers and statins I've been taking for so long. Perhaps the most useful thing that I have been taking is also the cheapest, and the only one that doesn't need a prescription - 100mg aspirin per day.
Then again, perhaps it's all due to the cat's habit of using me as furniture.
An excellent and thought-provoking FPP.
posted by Autumn Leaf at 10:57 PM on January 30, 2015
Then again, perhaps it's all due to the cat's habit of using me as furniture.
An excellent and thought-provoking FPP.
posted by Autumn Leaf at 10:57 PM on January 30, 2015
Anticoagulation for venus thromboembolism: None were helped. What?!
posted by wintersweet at 12:10 PM on February 1, 2015
posted by wintersweet at 12:10 PM on February 1, 2015
Even if we take the OP's analysis at face value... Walgreens will sell you 500 83mg aspirin tablets for USD$9.99. So treating 2000 people for 2 years requires 1,460,000 tablets or 2,920 bottles at a cost of around $30,000. I don't know if you've gone to the hospital in the US recently but having a heart attack costs a lot more than $30,000, along with the risks of permanent impairment and death, so even in this worst-case analysis the numbers support using the therapy.
posted by localroger at 1:21 PM on February 1, 2015 [1 favorite]
posted by localroger at 1:21 PM on February 1, 2015 [1 favorite]
« Older The Dream of the '90s? Jon Chait on P.C. 2.0 | Mormon church softens (a little bit) on gay rights Newer »
This thread has been archived and is closed to new comments
posted by fairmettle at 4:04 AM on January 28, 2015 [1 favorite]