Algorithm Finds Thousands of Unknown Drug Interaction Side Effects | Hacker News Reader

Algorithm Finds Thousands of Unknown Drug Interaction Side Effects

nature.com

103 pointsvoodoochilo42 comments

This should come as no surprise as our effort at better living through chemistry is barely in its infancy. Millions of people take drugs prescribed to them by their physicians with the implicit understanding that they are safe because the USDA says they are when in fact we have precious little idea if they really are.

The potential revenue from new pharmaceuticals is huge and the pressure to get new drugs out into the market is immense. It's initially comforting to see that people are working on this problem of drug interactions since as the article mentions, we couldn't do it in the clinical trials. Still, the cynic in me thinks this is just another step in getting everyone taking more and more prescription medicines. If the pharmaceutical industry can show that two drugs have no interactive side effects, it's much easier to convince a patient to begin taking a new drug. Western medicine is already heavily dependent on fixing the symptoms instead of the cause.

It's clear that in many cases we have almost no clue why a particular drug works. It's thrilling to see risks removed from the system by finding drug interactions previously unknown. However, it's frightening how much more likely it is that physicians will rely on prescribing a pill because it's easier than convincing the patient to change.

"Western medicine"? As opposed to what?

As opposed to a more holistic approach beyond "You have high cholesterol, take this pill for the rest of your life". Some of this is our natural inclination to do things that are easy (taking a pill is easy, fixing your diet is not) but as we become more and more reliant on prescription medicines to treat symptoms and disease, we will naturally become less and less inclined in the medical profession to advocate strongly for a different patient lifestyle before putting them on the meds.

All Doctors treating the average american say to eat better, exercise more, and get more sleep. The only difference is when people don't do these things they still get better with "Western medicine" vs. blaming the patent.

That's not true on several fronts. One, doctors spend little time explaining options to patients oftentimes because they don't actually know the options. Drugs are almost always the first line of defense. Obviously, this is partially because drugs are easy. It's also because doctors don't have to worry about it since the USDA says it's fine whereas if they tell a patient to alter their diet and take supplemental iodine to treat hypothyroidism, they have no support for that treatment plan.

Secondly, many times patients do not get better. Statins are prescribed in astronomically large numbers in this country to lower cholesterol when the science behind the benefits of lower cholesterol is highly mixed. In many of the supposedly supporting studies, overall mortality is unchanged in groups taking statins.

This has nothing to do with blaming the patient. It has everything to do with the fact we have little comprehensive understanding of the long term ramifications of our increasing reliance on chemicals to help us live more healthy lives. Hopefully finding ways to identify drug interactions is a first step towards current complications but it doesn't change the fact that this will only make it easier to prescribe medications for symptoms without treating the underlying cause.

Teach the controversy!

In the meantime, the data supporting statins are among the best ever compiled. The advances contributed by groups calling themselves "holistic" are entirely absent. And the willingness of modern medicine to introspect and change in the face of data is slow but meaningful.

I asked this same question the other day and still have yet to get a meaningful response: what is the "holistic" treatment for HIV?

> what is the "holistic" treatment for HIV?

There are, unfortunately, people who believe in treating HIV / AIDS with nutrition or folk remedies. A bit of debunking before they arrive.

African Potato hinders AIDS treatment

(http://www.plusnews.org/report.aspx?ReportID=38168)

Ugandan Health Minister claims to know 3 people cured of HIV through prayer. (1.2m people living with HIV, 150,000 children; this is especially troubling because Uganda is often cited as good example)

(http://www.plusnews.org/report.aspx?ReportID=93411)

South Africa's health minister delayed ARVs for years; President doubts links between HIV and AIDS; 330,000 people die as a direct result. Even after ARVs introduced minister claims olive oil, beetroot and garlic are effective.

(http://www.plusnews.org/report.aspx?ReportID=87470)

South Africans, with the worlds highest HIV caseload, trust traditional medicine because it's wisdom passed on.

(http://www.plusnews.org/report.aspx?ReportID=73872)

South African Deputy Health Minister sacked after rolling out widely welcomed plan

(http://www.plusnews.org/report.aspx?ReportID=73679)

Swaziland government fails to tender properly; a shortage of drugs is covered up by that government promoting folk remedies. (33% of the sexually active adult population were HIV+; about 66% of the population lived on less than $2 per day)

(http://www.plusnews.org/report.aspx?ReportID=62579)

Dr Matthius Rath is evil.

(http://www.plusnews.org/report.aspx?ReportID=38577)

Just in case you think all those links are old, it still happens. Here's a link about people selling 37ml of Ubhejane for $210. (About 80% of the Ugandan population earn less than $1 per day.)

(http://www.plusnews.org/report.aspx?ReportID=94679)

It's not all bad news, all that death was useful for some businesses.

(http://www.plusnews.org/report.aspx?ReportID=62655)

(http://www.plusnews.org/report.aspx?ReportID=62604)

(http://www.plusnews.org/report.aspx?ReportID=38262)

I completely appreciate the research in your post. I should have phrased my question: "what is the successful holistic treatment for HIV?"

And I see from your comment that you agree that there appear to be none, so I'm not trying to make it look otherwise.

http://people.csail.mit.edu/seneff/ has some interesting pages against statins. Personally I'd scrutinize things as best I could before considering following a doctor's recommendation for statins.

I'm pretty ignorant about the holistic health thing, but wouldn't a fairer question be what's the holistic treatment for a given person with HIV? I thought the whole point was to consider a whole person as a complex system, not a database record with the feature "HIV+".

Thank you for sharing the link. Her articles are not peer reviewed, and I don't think they would pass muster. She seems to blend evidence with anecdote to argue for a point. For example, lines such as "Increasingly, orthopedic clinics are seeing patients whose problems turn out to be solvable by simply terminating statin therapy, as evidenced by a recent report of three cases within a single year in one clinic, all of whom had normal creatine kinase levels, the usual indicator of muscle damage monitored with statin usage, and all of whom were "cured" by simply stopping statin therapy (Shyam Kumar et al., 2008)".

One case report does not bear the same weight as a randomized, controlled trial (of which there are now dozens). Yet that's the kind of stuff that's the basis for her argument. The reference about fructose in the next paragraph twists Vila's paper 180 degrees. Etc. I would be interested to see what would happen if she submitted this as a review paper. I sincerely doubt that it would get published, not because of her conclusions, but because of the way in which this paper mistreats evidence.

Just one anecdote on that point. I'm pre-med and I recently shadowed the director of cardiac research (a cardiac surgeon) at the big hospital/medical school in this area. He was asked point-blank by a patient, "I've been reading that with diet and exercise I might be able to reverse my heart disease. What do you think?"

The doctor's answer was, "Absolutely not."

This is one data point, so it's hardly conclusive, but coming from a doctor who (a) teaches doctors and therefore plants seeds for the next generation and (b) should be an expert on new medical treatments with his position as the director of research, it's troubling.

Meanwhile, there is a growing minority of doctors who take patients through an eating "bootcamp" where they help them take out processed foods/white-flour/high-fructose corn syrup from their diet, teach them how to shop and cook plant-based foods, etc. I saw a documentary on it, "Forks Over Knives" I believe. Fascinating stuff.

In the documentary, they take a man with advanced heart disease, cut out all meat/dairy/processed foods from his diet, and take before-and-after blood readings (cholesterol, blood pressure, etc.). It was surprising how quickly the intervention reversed his numbers.

There are other pioneers working in this area. There's Dr. Terry Wahls, an MD who reversed her multiple sclerosis by eating what some would call a "paleo" diet (only vegetables, fruits, nuts, meat). Here's her TED talk if you're interested: http://www.youtube.com/watch?v=KLjgBLwH3Wc She's currently working on human clinical trials to test this diet-intervention in other patients.

To your point, "all doctors say eat better, etc." I have two thoughts.

1. I'm glad you have doctors who have mentioned that but I've never had that experience.

2. A doctor who says, "You really need to eat better," is not quite what we need. For reversible conditions like heart disease, diabetes, (MS?) etc we need doctors who say, "Your next appointment will be with a food specialist/nutritionist. He will help you completely change your eating habits, go shopping with you to get new staple-foods, and you will stick to this new lifestyle for x months. Here's a list of foods you must not eat and here are x great-tasting recipes that are easy to make. Come back in 6 weeks and we'll do another round of blood-tests to see how it's working. If you're not willing to do this I can give you [statins, etc. fill-in-the-drug] but they won't actually treat the disease and may not make you feel better. They will keep you alive though. It's totally up to you and I'll support you in either decision."

Just my 2 cents. I think medicine will eventually look more and more like this, but it will take time to change the culture and the traditions of western medicine.

Heart disease can be plaque, which, theoretically, might be removable with some diet change. It can also, however, be hardening of the arteries, which is a loss of elasticity of the artery walls. We don't know how to reverse tissue loss of elasticity do we? So when he said it could not be reversed with diet, that seems to be correct to me.

In no way do I mean this as a personal attack, I simply find this comment fascinating, and it's a good example of something I see more and more of.

You basically say heart disease is either A or B. Because we don't know how diet effects A, and doctor says diet wouldn't help, doctor is right. (Notice the fallacies [false dichotomy, jump to conclusion, circular logic] become very clear when we simplify the wording.)

What I find interesting is the way we all seem to work to show that what we believe in must be right or that what we don't believe in must be wrong. Diet can't magically cure heart disease, and [logical fallacies], so diet doesn't cure heart disease. The Earth isn't warming, and [conspiracy theory], so the Earth isn't warming.

Again, I don't mean this as a personal attack (maybe your post was just had a few typos), but it's something that I see more and more frequently, even places like HN. Instead of being more like Feynman, our technology seems to be prodding us to be more like rush-hour talking heads.

I'm still trying to digest your argument, but my point was a simpler one.

I am very interested in whether or not we can restore elasticity in tissue, such as the lens of your eye, the skin on your face, or the walls of your arteries.

When a medical student, who doubtless has more knowledge than I do implies that heart disease can be reversed in any manner, diet or not, I'm interested and wanted to move the conversation forward in order to clarify.

To quote:

> We don't know how to reverse tissue loss of elasticity do we?

Ok.

> So when he said it could not be reversed with diet, that seems to be correct to me.

Doesn't follow. If something has certain effects, that reality necessarily precedes our knowledge of it. Our discovery necessarily follows the reality.

Note, he's pre med, not in med school. Pre med doesn't give you any specific medical knowledge.

When people talk about ischemic heart disease, they mean atherosclerosis. If they mean arteriosclerosis (e.g., Monckeberg's calcific sclerosis) they will call it out explicitly.

Here's one of the placebo-controlled studies that suggests lifestyle can reverse coronary artherosclerosis (hardening of the arteries): http://www.ncbi.nlm.nih.gov/pubmed/1973470

And for something more digestible, here's Dr. Dean Ornish's TED talk discussing his research on the same:

http://www.ted.com/talks/dean_ornish_on_the_world_s_killer_d...

By the time a patient is seeing a cardiac surgeon, it's extraordinarily unlikely that it would be remotely responsible to advise a patient to try to treat their condition by using diet and exercise without medication.

I think that's mainly because of tradition and the difficulty involved in getting people to make a wholesale change in their diet, not because of the superiority of the medication in reversing the condition.

See my other comment for sources: http://news.ycombinator.com/item?id=3709426

Your link shows a study comparing diet to placebo. In contrast, your comment talks about diet vs medication. A nuanced but important difference. To the best of my knowledge, diet-vs-medication has not been tested systematically in a high quality trial.

Another study that you might like, and which is potentially more relevant by the point you're talking with a cardiac surgeon, is the Lyon Heart Study, which actually looks at secondary prevention:

Easy to digest article: http://circ.ahajournals.org/content/103/13/1823.full

Original study: http://circ.ahajournals.org/content/99/6/779.full.pdf

Just to add.

Depends on why he was seeing a cardiac surgeon (who usually sees patients that have had or need surgery), and what his heart disease is. I don't think you have enough medical knowledge to assume the cardiac surgeon was, in fact, wrong. The question could easily have been interpreted as:

"Will diet and exercise reverse my [multiple 95% blocked coronary arteries before I die from a massive heart attack]"

"Will diet and exercise reverse my [dead heart tissue from my heart attack]"

Both answers are reasonably close to "absolutely not".

You seem to equate 'heart disease' to high cholesterol levels, and blood pressure readings. It is not. And if he "just" had high cholesterol and high blood pressure, he would not be seeing a cardiac surgeon.

Meanwhile, the answer to "Will diet and exercise reverse my [high cholesterol and high blood pressure]" The answer to that is sure, definitely possible.

Great points! Wish I had time to respond.

Edit: Meeting got out early so I have a minute.

You're right. Surgery is the only option in some cases. The question to the surgeon was about heart disease in general, or hardening of the arteries.

The doctor and I had a chat about it after the patient left. He said, "There are some researchers whose studies show that you can reverse heart disease with lifestyle changes, but when you look more closely at their data you can't tell which interventions actually help. The bottom line is that most people need the drugs and surgery to help them."

To me, it seemed like the doctor had a bias against non-drug or non-surgery options, and he didn't want to explore it with an open mind. Makes sense for him. His entire practice was based on surgery and drugs. If I were in his shoes I'd want to believe that drugs and surgery were the best option.

You seem to have the opposite bias, no?

Do you believe that diet alone is better than diet + drugs? Do you believe that diet alone is better than drugs alone?

I think it's fair to say that I'm biased against drugs, yes.

To your second set of questions, it totally depends on the situation (patient's willingness to change, type of disease, science of drug-based vs non-drug-based interventions, whether the illness is acute or chronic, etc). It's really impossible to make blanket statements about which interventions are most appropriate.

"This is one data point, so it's hardly conclusive, but coming from a doctor who (a) teaches doctors and therefore plants seeds for the next generation and (b) should be an expert on new medical treatments with his position as the director of research, it's troubling."

Why should this be troubling? Why do you presume to know more as a pre-med than this director of cardiac research does?

There's another reason that this general mindset exists: think drug interactions are poorly understood? Try researching nutrition and food interactions.

People eat thousands of chemicals daily, and we still haven't figured out if something as superficially simple as, e.g., the "food pyramid" is a good thing or not. What you eat is massively more complex than the drugs you are prescribed. The biggest difference there is that (most of) the chemicals have been around and in our diet for quite a long time, so they're evidently not (very) dangerous (to most. see also Celiac / Coeliac disease for an example of how long problems can exist)

Holistic approaches may be good; they may even be the best approach. But due to the insane amount of variables, they're also about the furthest from science as it's possible to get, and the least actionable in a specific sense. Everyone knows they should "eat better". Well... define "better". And be sure it applies to everyone, or be able to know who to apply what advice to.

I cringe whenever I hear the term "Western medicine" being used to describe modern evidence-based medicine. There's little Western cultural bias to this approach; a drug, lifestyle, whatever either has significant effects on health or it doesn't.

Just as the unproven components of traditional Western medicine (leaches anyone?) are mostly bunk, so are such aspects of traditional Eastern Medicine/alternative medicine/whatever.

You mean the FDA? No offense, but just because you are well versed in one subject does not mean you really understand the idiosyncracies of another. Every time I see a drug / medicine related thread on HN I cringe. Much in the same way you would if you read a technology related post on a forum geared towards doctors and pharmacists.

As a pharmacist, I approve this comment!

All that cringing and you only call out a misnamed variable?

Just to be fair, the original title of the research article is "Data-Driven Prediction of Drug Effects and Interactions" ( http://stm.sciencemag.org/content/4/125/125ra31 ). Emphasis on 'Prediction' mine.

47 are then supported by retrospective analysis of an EMR database.

Obviously this algorithm has lots of false positives.

Why is that obvious?

A false positive costs much lower than a false negative in this instance so it should follow that FP should be more common. However, without the data it is still speculation.

Whether false positives or false negatives are more harmful is a difficult question. A false positive might stop millions from taking a life-saving drug. A false negative might cause millions to take a life-threatening drug.

By nature a false positive interaction could be re-tested more easily than a false negative though, correct? Also, that fits in with medical "first, do no harm" framework more readily.

Regarding your second point, I agree. But it must be balanced against the other tenets of medical ethics. At any rate, this seems to be an important study. I look forward to seeing the same study carried out across different EMR databases and then meta-analyzed. That will be extremely powerful.

Typically scientists try to reject the null hypothesis (in this case, that a particular pair of drugs do not interact) by showing that it's unlikely (less than a 5% chance) that an apparent interaction is caused by random chance. But if you go looking at millions of possible interactions, you're going to find a lot that seem to meet that criteria by chance. 5% of a million is 50,000. 0.1% of a million is 1,000.

That's why drug companies have to prespecify the outcomes they're looking for before starting the study. If you look through your data after the fact, you're going to find something that makes your drug look good just by chance.

An association should be treated with suspicion until it has been replicated in an independent sample. Even then, the association is based on deviation from a simple null model which is little more than a straw man, so its causal basis is obscure. The authors have tried to correct for this by identifying patients with similar characteristics, but you can do so much of that before you destroy your statistical power.

Anyone know if that dataset is publicly available?

Yes, its public.

I think the below link should have some of the data.

http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInforma...

Reply on news.ycombinator.com