Algorithm Finds Thousands of Unknown Drug Interaction Side Effects 121
ananyo writes "An algorithm designed by U.S. scientists to trawl through a plethora of drug interactions has yielded thousands of previously unknown side effects caused by taking drugs in combination (abstract). The work provides a way to sort through the hundreds of thousands of 'adverse events' reported to the U.S. Food and Drug Administration each year. The researchers developed an algorithm that would match data from each drug-exposed patient to a nonexposed control patient with the same condition. The approach automatically corrected for several known sources of bias, including those linked to gender, age and disease. The team then used this method to compile a database of 1,332 drugs and possible side effects that were not listed on the labels for those drugs. The algorithm came up with an average of 329 previously unknown adverse events for each drug — far surpassing the average of 69 side effects listed on most drug labels."
Re: (Score:3)
Now instead of 60 warnings inside the bow we'll have 329 ? Sounds pretty futile to me. Nobody's going to read that.
Why not just writing: In addition to cure you, this drug might kill you or otherwise trigger another disease of allergy. Be warned.
Re:not surprising (Score:4, Insightful)
Presumably so doctors can better select functionally similar drugs to minimise these interactions...
For example, TFA says that the high-blood-pressure medication class thiazides and SSRIs can interact. Neither of these is available without prescription therefore a doctor could use such data to make better treatment decisions...
Re:not surprising (Score:5, Interesting)
Re:not surprising (Score:5, Interesting)
There are databases and search applications that can be made more accurate with the new data. For example, Denmark has an online system where citizens can enter the name of two drugs and get a list of possible side effects and warnings. There are also big US and European databases of this kind, although less open to the public (I believe).
Re: (Score:2, Redundant)
You can gget a drug interaction app for android and I assume others. You can also download the full blown deal on your PC... or go to a bookstore (maybe more so one on a campus) and purchase the "official" book off of the shelf....
Re: (Score:2)
Epocrates [epocrates.com] will do that. It's free. BUT it has the same problem that every other adverse drug effect database has - it is sensitive, but not specific. Most entries basically imply that you will explode, dissolve or get turned into a Newt if you take the drug.
What's needed are accurate statistics on how many people get Newted and whether you are more likely to get into trouble if you take other medications or have other conditions. This paper says you can do some data mining to get at new insights to the
Re: (Score:2)
A newt or a Newt? The former, I can live with.
Re: (Score:2)
Re:not surprising (Score:5, Insightful)
Re: (Score:1)
ERs/pharmacies/drugstores/monitoring\ stations/schools/etc
Re: (Score:2)
I love Slashdot.
Re: (Score:1)
Not to burst your bubble but that is the reason pharmacists exist. Read all the comments. People think doctors know the interactions. They don't. Pharmacists have a doctorate in pharmacy, yes just as many years of training(school:intern/residency) as a "doctor" but specialize in drug interactions not treatment(using your data). (Meta: Pharmacists earn a Pharm.D degree, a doctorate of pharmacy, they are "doctors" but the uppity M.D.s think they have a stranglehold on "Dr." unless it's in another field
Re: (Score:2)
No-one is going to read 329 warnings, but no-one is going to read sine tables either. Biomedical Informatics - and indeed any form of information clearing - is useful to the extent that we can avoid information saturation and filter down to what is actually important in some specific case.
There, of course, is the crunch. Services like PubMed are highly restricted, so the number of people with the skills to write data digestion software AND who have access to the data AND who have an interest (even a contrac
Re: (Score:2)
Sherlock Holmes always recommended a 7% solution.
Re: (Score:1)
Now instead of 60 warnings inside the bow we'll have 329 ? Sounds pretty futile to me. Nobody's going to read that.
Medical doctors are going to read that, it's their job.
Re: (Score:3)
While I go to my doctor to get treatment and any required prescriptions, I *always* double-check with my pharmacist to ensure there are no likely conflicts between the drugs I have been prescribed, my allergies to certain drugs etc. I trust my pharmacist will have read this stuff in detail, even if my doctor missed something.
Re: (Score:2)
I believe Walgreens automatically checks.
They also have an online interaction checker on the web.
Re:not surprising (Score:4, Insightful)
>> Medical doctors are going to read that, it's their job.
I think you mean "Medical doctors SHOULD read that...", or under the best cases "Medical doctors are going to TRY to read that..."
Realistically? They won't have the time to do it properly. Doctors are massively overworked, trying to see far too many patients and dealing with a field that is too broad and grows way too rapidly to keep up with even if they *didn't* have the inconvenience of actually applying their knowledge. I mean, this study alone claims to have discovered 438,228 new drug interactions and side effects. (329 side effects per drug x 1332 drugs) You try to do a thorough read-through and analysis of that kind of data without taking any time off from work; and work quick, you probably only have a week at most until something new you need to learn comes along....
they do the same thing the pharmacists do: (Score:1)
Actually, think of the upsides (Score:3)
This algorithm may be able to idenify sideeffects when combining medicines.
However, sideeffects are by definition only negative. Once "results" rather than side effects are put through the same algorithm, we may be able to identify better and cheaper ways to combat disease. And we may even be able to find cures for known diseases with combinations of drugs that have never been tested before.
It's all about the data.
Re: (Score:3)
However, sideeffects are by definition only negative.
Your geek card, please.
Re: (Score:2)
Why? He's right. He meant "de facto" and not "by definition," but he's right.
Re: (Score:2)
Whap!! Drink your coffee as he's completely correct. The Medical defination of Side-Effect is negative only as in detrimental. Otherwise it's classified as synergesic if beneficial.
Re: (Score:2)
Re: (Score:2)
Lawyers will love this. It's additional things they can use against doctors if a patient has an adverse side effect.
Re: (Score:2)
Surely you jest? That's like asking why somebody would bother to write a phonebook when nobody will read the whole thing. The big table of drug interactions is easily remembered by a computer. You enter a set of medications and it tells you what conflicts exist (and with what probability, hopefully) in that set.
I was wondered about something (Score:3, Interesting)
Re:I was wondered about something (Score:5, Informative)
Re: (Score:2)
From the fine (real article) [sciencemag.org]
Our new method (i) accomplishes the goals of stratification, dampening or removing the effect of covariates, without the need to divide drug-exposed reports into strata; (ii) is both adaptive (it removes different covariates for different drugs) and appropriate for systematic application and routine analysis; and (iii) is designed to complement modern signal detection approaches and thus extends the applicability and power of existing methods. Our model is inspired by the case-control approach to cohort selection in observational clinical studies. Each drug-exposed patient is matched to one (or more) nonexposed patients (controls). The nonexposed patients are selected on the basis of how well they match an exposed patient on a set of predefined covariates. Propensity score matching (PSM)—a statistical method designed to yield an unbiased estimate of treatment effects—has emerged as the preferred method of matching exposed and nonexposed patients in observational cohort studies and has yielded similar estimates of effects when compared to the results of randomized control trials (9–11). However, like other confounder controlling methods, PSM requires the covariates to be both known and measured; neither parameter is guaranteed to be present in spontaneous reporting systems. Instead, to match patients, we adapted PSM to use only the co-reported drugs and co-reported indications. We hypothesize that many confounders correlate with these key variables and do not need to be modeled.
As usual with these sorts of studies, my head asplode.
Re:I was wondered about something (Score:5, Informative)
As usual, Science&Nature only provide high-level info, so you'll have to dig deeper than the article ( http://stm.sciencemag.org/content/4/125/125ra31.full [sciencemag.org] )
On the authors website, http://www.tatonetti.com/cv.html [tatonetti.com] there is a paper that describes the machine-learning algorithms used:
Tatonetti, N.P., Fernald, G.H. & Altman, R.B. A novel signal detection algorithm for identifying hidden drug-drug interactions in adverse event reports. J Am Med Inform Assoc (2011) DOI:10.1136/amiajnl-2011-000214 [tatonetti.com]
Re: (Score:2)
They call it novel so they can justify patenting it.
Like if I wanted to re-invent the wheel:
"A novel transportation facilitation device consisting of a round object on an axle to reduce friction and optionally provide a driving force to move the object the axle is attached to."
You'd need the word "novel" to get it past the patent office.
Re: (Score:2)
I can imagine the commercials already.. (Score:2)
Re:I can imagine the commercials already.. (Score:4, Insightful)
A better solution would be to just ban the placement of ads for prescription drugs anywhere other than medical literature.
Re: (Score:3)
Not with this Supreme Court.
Re: (Score:2)
They'd only overturn the ban for companies donating to political campaigns, surely.
Re: (Score:3)
Instead of consult your doctor they'll say consult our computer.
Re: (Score:1)
Yeah, something like that... http://en.wikipedia.org/wiki/Clinical_decision_support_system [wikipedia.org]
Re: (Score:2)
Yeah, something like that... http://en.wikipedia.org/wiki/Clinical_decision_support_system [wikipedia.org]
Obviously you missed my use of the word "our."
Re: (Score:1)
oh, sorry :)
Re: (Score:2)
No, instead of reciting 69 side effects in ten seconds at the end of the commercial they'll recite 359 side effects in ten seconds.
Re: (Score:2)
Commercial break?
Chucko, they're going to take up the entire program.
Say! (Score:5, Funny)
Say! Are there any new prescription drugs out there that I'm not taking, but should be? Those seem pretty safe.
Perhaps they'll soon come out with glossy color catalogs for the new ones each season. They'll be full of loads of bikini-clad women draped over cars, popping pills.
Re: (Score:2)
Just say NO to drugs!
Multiple testing problem? (Score:5, Interesting)
I am a statistician.
I've only done a light skim of the paper, but it seems to me that the OP (but not the paper itself) is being way too positive here. Their methodology seems to be very vulnerable to false positives - with a massive database of drugs and potential adverse effects, you'd expect a *lot* of apparent side effects occuring solely by chance. For example:
"We constructed a database of 438,801 off-label side effects for 1332 drugs and 10,097 adverse events."
Supposing you are doing a hypothesis test at the standard 0.05 significance level, for each of the 1332*10097 drug-side effect combinations. Then, with naive assumptions, on a null hypothesis, you'd be picking up an average of 666k+ 'side effects' anyway, purely by chance. With the drug interactions case, this multiple testing problem gets even worse.
Now, there are ways to correct for multiple testing, but for things as large and complicated as this problem, I'm not sure the standard methods are going to cut it. At best, this study should be considered more a *filter* on the set of potential side effects, than really an enumeration of effects that are actually there. This is ignoring other issues like the placebo effect.
Re: (Score:3)
Re:Multiple testing problem? (Score:4, Insightful)
I both agree and disagree. In principle people really should stop using frequentist methods. In practice, though... There's still substantial disadvantages with Bayesian methodologies. For example, off the top of my head:
1. Relative slowness of computational algorithms for large datasets
2. Difficulty of presenting results to people with different prior beliefs. (Strictly speaking, in Bayesian terms, the answer you give must always be relative to *someone*.)
3. Ease of 'cheating', even unintentionally, by choosing priors to favour a certain result.
4. Proliferation of methods that pretend to be Bayesian but are in fact probably not. (e.g. Empirical bayes methods)
I'm saying this because this always comes up, but people don't realise the bayesian approach is necessarily a magic bullet either.
Re: (Score:2)
Why not just report likelihoods instead and let the reader multiply it with any prior they want? In many cases, the prior won't make much of a difference anyway, I suppose.
Sure, do away with empirical Bayes. Anyway, I don't think "When using Bayesian methods, you run the risk of using non-Bayesian methods." is an argument for not using Bayesian methods.
I'm saying this because this always comes up, but people don't realise the bayesian approach is necessarily a magic bullet either.
Regardless of what practical obstacles there might be for using Bayesian inference, using something else would be wrong, leading to results that make you tak
Re:Multiple testing problem? (Score:5, Informative)
"Why not just report likelihoods instead and let the reader multiply it with any prior they want? In many cases, the prior won't make much of a difference anyway, I suppose."
Reporting likelihoods (or rather summary statistics of likelihoods, because generally likelihoods are functions, not single spot values) is precisely what frequentist statistics is. The use of significance tests, p values and so on can be simply considered a standardised representation of the likelihood.
In the case of multiplying with any prior they want, in many (even most...) cases, the reader simply does not know what their prior is. And often, they do not even know, without a lot of work and some experience in statistical theory, what a 'sensible prior' is. For example, setting a flat prior for x (all values are equally likely a priori) can be actually *very* informative if later calculations make use of 1/x.
In most interesting cases, I'm afraid that the choice of prior *does* make a great difference. This is even more complicated if the prior is in the form of hyperparameters, in which case your prior choice can have a dramatic and non-linear effect on your result.
"Sure, do away with empirical Bayes. Anyway, I don't think "When using Bayesian methods, you run the risk of using non-Bayesian methods." is an argument for not using Bayesian methods."
But Empirical Bayes represents a huge chunk of how 'bayesian stats' is done in practice. The point here is that speakers don't define or understand clearly what is, or is not bayesian in the way that is proposed.
"Regardless of what practical obstacles there might be for using Bayesian inference, using something else would be wrong, leading to results that make you take the wrong actions!"
It depends on how the results are represented. Frequentist results are not 'wrong'. They represent a real value of the data. They may simply be irrelevant. And on the flip side, merely switching to Bayesian inference *does not prevent you from taking wrong actions*. What Bayesian inference accomplishes is that it shifts and makes explicit the mistake you are making.
For example, the false positive results I stated can be *exactly duplicated* by using a prior that is excessively permissive of finding positives, and that prior can look extremely reasonable. The real solution to these problems is to *understand what you are doing*, and this can be done in both a bayesian and frequentist way.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
'Factoring out your prior belief' in Bayesian statistics is like asking for water without the wetness. The best you can do is a prior sensitivity analysis that tells you what you have to believe in a priori to conclude a certain way, but the risk here is that you don't know if you are disbelieving in something because you are being reasonably skeptical, or just being stubborn.
Re: (Score:3, Informative)
Re: (Score:2)
GWAS done in the traditional hypothesis testing way have a *massive* multiple testing problem, which is only partially solved by imposing *extremely* stringent requirements on significance. That's why things like the Group Lasso and so on are being devised to work on them. It's definitely a non-trivial problem.
I agree that my calculations are very naive. But in my reading of the paper, I'm not really satisfied that they've done enough to deal with the multiple testing issue. (There is a mention of correctio
Re: (Score:1)
>> vulnerable to false positives
I don't know crap about the two tailed t-test, but I do know that if I ask for side effects in my drugs I better damn well get them!
Re: (Score:3, Insightful)
"Their methodology seems to be very vulnerable to false positives..."
I would agree, and go on to suggest that this is intentional. Even after applying "corrective" measures, one has to pick a preference: false negative or false positive, and then show your work (just like in math class). When it comes to drugs, the control methods are never *really* enough. If you're doing an in silico screen, depending on the algorithms used, you may want false positives, because you're just going to throw everything into a high-throughput screen and let the robots do the rest of the work.
B
Re: (Score:1)
IANAStatistician.
Is your computation of 1332*10097 correct? I think it's wrong.
Isn't the combinatorial function without repetition ((n+1)!) / (r!(n-r)! )?
Re: (Score:3)
No, that would be the right calculation for arbitary combinations of a certain length. We're looking for the number of pairs (drug, side effect), of which there are #drugs*#sideeffects total.
Re: (Score:2)
There's a placebo (strictly speaking, nocebo) effect in that people may report (and to be fair, often genuinely feel) side effects solely because of the fact they've been given medication, instead of any chemical or biological reason.
Are these really the result of drug interactions? (Score:5, Interesting)
Re: (Score:2)
Re: (Score:2)
Forty-two.
Just kidding. There is no single percentage. You need to know what percentage of people with the same disorders would exhibit that particular symptom while taking one or zero of those drugs, and compare the various percentages using a T-test or similar.
The minimum statistically significant difference in those percentages further depends on the sample size, and w
Looks like... (Score:3, Funny)
Simple (Score:1, Interesting)
All drugs have side-effects. In some those side-effects can be serious and even deadly, but it's pretty unpredictable what the side-effects will be and/or their severity in a particular patient, let alone one that takes other drugs. In others, the side-effects will never appear.
For instance, I'm one of those annoying people who doesn't take drugs unless absolutely necessary - not because I distrust the medical establishment (because I don't) but because if I don't need a drug, I won't take it - and even t
Re:Simple (Score:4, Insightful)
People keep telling me to take headache tablets, cold/flu "remedies", painkillers, etc. etc. etc. and I avoid them like the plague. The people who use them use them CONSTANTLY and still get headaches, flu and pain worse than I ever have. If you have a pack of pills in your bag "just in case" of headache, cold, etc. then you should be made to throw them away - they are purely placebo.
Look, somebody should hit your head with a hammer to make sure you know what you're talking about.
You ignore the fact that we are all different from each other. Headaches are a good example: I practically never get headaches.Other people I'm close to get absolutely terrible headaches from time to time that are so bad that they keep them awake and only the strongest Paracetamol can give some remedy for a short time. You either lack empathy (working in management?) or have really no idea how bad headaches or migraine can be.
Re: (Score:3)
Migraines are a different matter that I covered separately in that same post BECAUSE they are nothing to do with headaches and because they have unique drugs that can combat them quite effectively if taken at the onset.
Anybody that confuses or merges migraine and headaches is lacking in understanding themselves. My current and previous partners both suffer severe migraine (up to and including visual effects such as not being able to cross a road because they see cars on the road and/or "seeing" people as h
Re: (Score:2)
I understand what you're getting at - I myself won't take painkillers for mild headaches (drinking water usually does the trick - whether it's dehydration or placebo). However, in my early 20s I suffered from quite bad migraines (blindness aura, skin crawling insect feeling, crushing pain, an
Re: (Score:2)
If you have a headache that a headache tablet gets rid of, or helps, you don't have a need for the tablet, it's just convenience.
WTF does that mean? If a drug helps, then you didn't really need the drug? Yeah, and if you don't feel hungry after eating, then you don't really need food. And if you can see after turning on a light, then it wasn't really dark.
As for "it's just convenience," are there any recorded cases of terminal migraines or headaches? Anything anyone does to alleviate headache pain is a convenience.
If you have a headache that *isn't* affected enough by paracetamol, you need to get your doctor to give you something stronger.
Well doctor, I get headaches that paracetamol has no effect on, but will go away if I take ibuprofen early enough. T
Re: (Score:2)
"Headache" is a vague symptom. Migraine is one of many conditions which produces a (very distinct kind of) headache as a symptom.Yes, it has distinct drugs which are effective at treating the condition or mitigating the symptom.
Fevers also cause headaches (which are very different than those
Re: (Score:3)
Have the people you're close to tried the other over-the counters, Aspirin or ibuprofen, etc?
I've personally found paracetamol to be pretty ineffective in just about any dosage, for pretty much every type of pain I've ever had (however, I've known people who found it effective for their own pain). This sucked growing up when I used to get terrible headaches and my parents were afraid of reye's syndrome and assumed all NSAIDS were aspirin for some reason (tylenol marketing dept, I assume...), so that's all
Re: (Score:1)
Have the people you're close to tried the other over-the counters, Aspirin or ibuprofen, etc? I've personally found paracetamol to be pretty ineffective in just about any dosage
Paracetamol (aka acetaminophen, aka Tylenol) is a troubling drug. It's therapeutic dose and toxic dose are very close. In some individuals, there's no safe effective does. There are times it's called for, but ibuprofen and aspirin are safer. That said, the anti-over-the-counter drug rant was just crazy. Those things are amazing when use correctly, sadly, most people take the wrong drugs and never read the labels.
Re: (Score:2)
My personal experience is that they're pretty variable in pain relief. I assume that different causes of pain are more susceptible to different drugs, and sometimes I happen to pick the right one, but I'm not aware of any way to tell which kind of pain I'm having or how it matches with which kind of OTC pain reliever. Is there such a resource for determining it? I don't think the labels are very helpful in this regard.
Re: (Score:2)
But, literally ANYTHING I pop into my mouth that I haven't had before could kill me the instant it touches my stomach. You have no way to know.
You've convinced me. I am switching to a zero-food diet.
Need an app to prioritize (Score:1)
Re: (Score:3)
So the algorithm found 329 adverse effects which were not known previously. Has someone spent the time in checking the validity of the algorithm by doing studies to check these claims? If not, this is mostly paranoia.
RTFS. They've looked at electronic medical records and confirmed the prediction in the case of thiazides and SSRIs. They're planning more tests - eg a clinical trial.
The Infancy of Medicine (Score:3)
OMG. I really do hope medicine outgrows its infancy during my lifetime.
It will not. (Score:2)
Doctors and nurses are still resisting checklists! SIMPLE CHECKLISTS!!? a proven method to drastically cut down errors to which there is no decent counter argument but here we are STILL - no checklists.
You think the human ego will not drastically hold back progress more than the other humans contribute towards progress? I don't.
Health for profit is not ultimately beneficial; it could be removed and replaced with something better but the culture is so brainwashed we'll continue down in the wrong direction --
A video comment on drugs. (Score:1)
A nice song from Consumers Union about drug side effects to enrich your day.
http://youtu.be/mYodDH4qZQo [youtu.be]
Side effects (Score:3)
Ive had some pretty nasty side effects from a drug that weren't on the label. It really sucked. I'm still dealing with them.
Let me tell you that if you goto a doctor and say that you think that side effects are related to the drug he put you on and those side effects aren't yet on the label they will treat you like you are crazy. It was really eye openning to see how doctors can talk down to patients. The drug I was taking for example can cause tendons to snap. That was known. The doctor flat out didn't believe that incrediblely painful tendinitis could be related to taking the drug.
It really sucks. I got lucky though because a month later the FDA released a new list of side effects that for this drug that included my ones. Then I got a doctor to pay attention and help me out.
I knew the side effects were related to the drug thanks to a quick google search that showed thousands of other people had experienced the same things that I had. There were literally 100,000 posts in a forum discussing what I had experienced in the previous weeks.
I can't even imagine the complexity when combining two drugs and what havoc that could cause.
You know what this means? (Score:1)
Where is the database? (Score:2)
I'd like to have a go at the database myself. Is it included in the article (which I don't have access to)? (It should be, IMHO, because how do you otherwise replicate their results?) Can it be found elsewhere?
Re: (Score:2)
Thank you. I realise I shall need the article itself as well to be sure I understand what some of the numbers in the database mean.
This is not directed at side effects! (Score:1)
cool results from data mining (Score:2)
But it's not science yet. The more important findings need to be verified in controlled experiments.
Live as clean and natural as possible (Score:2)
I think it goes without saying that we should all live as clean and natural as possible. I'm not saying we shouldn't use drugs to improve our health, to fight disease or to manage pain and discomfort. We should. It's very useful to do so. But I am saying we should actively seek to AVOID using drugs and to seek out the most healthy foods and to live the most healthy lifestyles available to us.
But you know what? I'm fat. I'm definitely overweight. I have been better than I am today but I ate too much o
Why not (Score:2)
This is going to make those drug commercials... (Score:2)
a *lot* longer. Maybe, we can just turn them into the whole darn show. I see sitcoms developing around viagra. (Stay tuned for "What's up, Doc?!)
More Side Effects? (Score:2)
My favorite side effect: (Score:2)
"May result in increased desire for gambling or sexual activity."
I almost swore it was a joke the first time I saw the commercial, especially given that it was to treat restless leg syndrome.