Polypharmacy – Is It Evidence-Based?

Polypharmacy essentially means taking too many pills. It’s a real problem, especially in the elderly.

A family doctor gives an elderly patient one pill for diabetes, another for high blood pressure, and another to lower cholesterol. The patient sees a rheumatologist for his arthritis and gets arthritis pills. Then he sees a psychiatrist for depression and gets an antidepressant. He takes a sleeping pill. He takes a laxative. He buys some over-the-counter cold medicine and Tylenol. Then he goes to his local GNC store and buys a smorgasbord of vitamins, minerals, supplements and herbal products. It would be surprising if some of these didn’t interact with each other to cause some problems.

One doctor may not know what the other doctors have prescribed. The patient may not think to tell his doctors about the non-prescription products he’s taking. Or he may not want to admit it for fear the doctors will disapprove.

In another common scenario, the patient gets drug A which causes a side effect. He is given drug B to counteract that side effect. Drug B causes a side effect of its own, so he’s given drug C to counteract that one. If drug A is needed to save life or limb, such an approach may be justified; but sometimes the doctor doesn’t recognize that the symptoms are drug side effects and he ends up just chasing his tail.

Any effective drug can have side effects. When patients are on multiple drugs, the risk of reactions and drug interactions increases.

We try to practice evidence-based medicine, but there may not be any evidence for a specific combination of drugs. We have evidence that drugs A, B, C, D and E are each safe, but we don’t have evidence from controlled studies to show that drugs A, B, C, D, and E can safely be used together.

I recently received an inquiry about polypharmacy. The writer stated that 100,000 people die from adverse drug reactions every year and we don’t know how many die from polypharmacy. He repeated a commonly cited estimate that only 30% of medical practice is evidence-based. He asked whether the use of unstudied polypharmacy can really be considered evidence-based, “scientific” medicine.

That’s an excellent question. The answer is yes and no.

It is certainly true that patients are often put on a number of different drugs that have not been specifically tested in combination. It would be nice if every possible combination could be properly tested, but that simply isn’t possible. There are too many combinations and permutations. I’m no statistician, but I’m pretty sure the number of studies needed would far exceed the number of people in the world.

The statistics for adverse drug reactions are misleading. A drug that kills 5 people might also have saved 500 people. Some drugs are prescribed as risky “last resort” treatments for patients who would have died without any treatment. Sometimes drugs with potential interactions are prescribed together because there is no better alternative. Nevertheless, it’s a reasonable guess that polypharmacy does considerable harm. By one estimate, it may be responsible for up to 28% of hospital admissions. How can we avoid that harm when we don’t have comprehensive studies? How can we practice evidence-based medicine when there’s no evidence?

The Wikipedia article on polypharmacy makes a distinction between “thoughtful” and “thoughtless” polypharmacy. We know a lot about drug interactions and can try to avoid them. We know that the more medications, the greater risk of side effects. On the other hand, sometimes we deliberately use multiple medications so that the dose of each can be reduced and the risk of side effects minimized. Multiple drugs are commonly used for infectious diseases like TB and AIDS to reduce the development of drug resistant strains. The different drugs in these regimens have different mechanisms of action so they are less likely to have additive toxicities. Combinations of that sort usually have been adequately tested.

When patients see more than one doctor or fill their prescriptions at more than one pharmacy, thoughtful polypharmacy may be sabotaged. And patients may be to blame for unforeseen or foreseeable interactions when they take over-the-counter medications, herbal medicines, or dietary supplements without the knowledge of their doctors.

Ideally, you would have to test every combination of prescription drugs with every combination of OTCs, herbs, and dietary supplements. Not to speak of combining them with different foods. Grapefruit juice is a notorious offender: it affects drug metabolism by blocking the enzyme CYP3A4. Once we know what it does, do we really need to separately test grapefruit with every drug that is metabolized by that pathway?

The medical literature recommends frequent re-assessment to see if a patient really needs to continue all the drugs he’s on, and it recommends asking the patient to put all his meds and dietary supplements in a paper bag and bring them to each appointment – the “brown bag” review. Guidelines have been published for reducing polypharmacy, for instance this CME article with nine key questions for the provider to ask.

The claim that only 30% of medical practice is evidence-based is misleading. There are some things we can never study with RCTs. You couldn’t ethically do a study of appendicitis with a “no-treatment” arm. You couldn’t ethically compare cancer surgery to placebo surgery. You couldn’t try a placebo instead of casting a fracture. RCTs are the gold standard, but they’re not the only standard. Sometimes we have no recourse but to accept a less perfect kind of evidence.

There are some things we may never know. Does a single chest x-ray increase the risk of cancer? If there is a risk it is very small; the risk from natural background radiation is much greater. Factors like living at different altitudes and living in brick vs. wood houses cause wide variations in individual radiation exposure that would mask any effect from a single x-ray.

Medicine is not an exact science like physics, and the nature of clinical trials means that even in the best of cases we must use common sense to apply imperfect knowledge to a unique individual.

I think there is widespread awareness of the polypharmacy problem, and “thoughtful” polypharmacy IS evidence-based. It’s based on knowledge of the individual drugs, their mechanisms of action and their known interactions. What’s really promising is our increasing knowledge of the genome: we will eventually be able to predict which patients will respond to which drug and which combinations of drugs will be more dangerous for one individual than for another. Calling for RCTs to study every possible combination of drugs in the general population would not be a good use of scarce research money even if it were possible. If certain drugs are frequently prescribed in combination, then those combinations would warrant study, especially if adverse reactions are reported.

Even when we don’t have good evidence for specific combinations of pills, we do have evidence to guide us. We have evidence about the mechanism of action and metabolism of individual drugs, and we can use that evidence to predict some drug interactions. We can do the “brown bag” review and look for evidence that the patient still needs each drug. When a patient develops a new symptom we can review the known side effects for each of his meds to check for a possible drug reaction.

We don’t have all the facts, but we can apply all the knowledge we do have. We can follow general principles like “keep the number of drugs to a minimum.” We can avoid thoughtless polypharmacy – and most good clinicians do. Science-based medicine doesn’t mean requiring an impossible standard of perfection; it means using all the evidence we do have and using common sense to apply it to clinical decisions.

Posted in: Herbs & Supplements, Pharmaceuticals, Science and Medicine

Leave a Comment (12) ↓

12 thoughts on “Polypharmacy – Is It Evidence-Based?

  1. mckenzievmd says:

    Excellent points! I work in a field (verterinary medicine) where the quantity and quality of evidence is considerably less than that available to MDs, and I am constantly faced with the question of how to practice rational, evidence-based medicine in the absence of ideal, or sometimes any, evidence. I am also faced with the challenge you discuss in which proponents of CAM argue that since so much of conventional medicine is not based on ideal evidence it is really no different from CAM. As you point out, the practice of relying on the best available evidence where possible, and plausability and other rational means of imperfect assessment when RCT or better quality evidence is not available, is still superior to accepting that anecdote and myth is the best we can do and giving up trying to do better.

  2. Harriet – a thoughtful entry as usual. I would add two points:

    1 – the 30% figure for how much modern medicine is evidence-based is not only misleading, it’s wrong. Figures from 15-30% are often given to impugn modern medicine – but these figures are nothing more than myths.

    I tell the full story here:

    The bottom line is that reviews give a median result of 78% for the number of modern medical practices that are reasonably evidence-based.

    2 – Another source of evidence for the safety of polypharmacy comes from RCT’s themselves. When a new investigational drug is tested in phase III trials, the subjects are real patients with the disease in question, and they are on a representative assortment of other medications. New drugs are not tested only in people on no other medication. So the trials are measuring the net effect of adding the new drug to the polypharmacy that is likely to be encountered in practice. In fact the degree to which a study population matches a likely treatment population is often assessed in such trials, and polypharmacy is one such factor taken into consideration.

  3. Fifi says:

    Good article and thanks for the clarifications Dr Novella.

  4. daedalus2u says:

    Since essentially every natural product is a complex mixture, essentially every natural product given alone is an example of polypharmacy. An exception being homeopathy which is actually just plain water but which is claimed to be exceedingly complex.

  5. Jules says:

    When patients see more than one doctor or fill their prescriptions at more than one pharmacy

    I find this interesting because this (kinda-sorta, I know it’s not your intention) blames the patient for shortchanging himself. But what else are you supposed to do if you get your health care from a WalMart clinic or the pharamcist you use most regularly won’t give you birth control pills?

    Mathematical models for antagonism and synergism are getting pretty complicated. Isoboles are pretty simple to understand, but now the math is showing complex waveforms in 3-dimensional space–and that’s just with two drugs! I’m still waiting for someone to come out with a three-drug interaction model. The point of all this? Trial and error–the way pediatric cancer treatments were worked out–is really the only way to go.

    I’d be curious to know whether epigentics is a major factor in drug-drug interactions, actually, because that would REALLY complicate matters. We know that many mental illnesses are in part epigenetic in origin; I’d be curious to know how many somatic illnesses, are, too–probably a good reason to stay away from power lines, right? ;-)

    As an aside: I also find this quote interesting “Medicine is not an exact science like physics” because if you’ve read Brian Green’s books you’ll realize that physics isn’t really a science, either :-D !

  6. daedalus2u says:

    Jules, the only difference between a liver cell and a nerve cell and between any somatic cells is just epigenetics.

    Every illness has to be in part epigenetics because without epigenetics all you have are undifferentiated pluripotent stem cells.

    Everything in physiology is really complicated. The only exceptions are the parts that we don’t know enough about and so we think they are simple.

    Essentially all interactions in physiology are coupled and non-linear. Beyond a few coupled non-linear parameters, everything becomes chaotic and completely intractable to model mathematically. Even if the system is completely 100% known in complete 100% detail with complete 100% accuracy. This is a fundamental property of such systems.

    It is not possible to predict the course of the unperturbed state (normal physiology), the perturbed state (the state plus the drug) is yet more complicated.

  7. pec says:

    “published results show an average of 37.02% of interventions are supported by RCT (median = 38%). They show an average of 76% of interventions are supported by some form of compelling evidence (median = 78%).”

    “Three quarters is a completely different world than “as low as 15%.” Also, we must consider that the other 25% is based on some evidence, even if not compelling, and typically will have a high degree of plausibility. You also have to take into consideration the context of treatment. If there is no evidence-based treatment available for a patient (or what is available has failed or cannot be used) then it is reasonable to try plausible if unproven therapies.”

    This is how Novella gets his 78% figure. Actually only 38% are based on RCT. The rest of the 78% is “compelling,” but we are left wondering how Novella defines “compelling.”

    Well probably the same way he defines “plausible,” as agreeing with his predetermined faith system.

    So even a hard-line materialist like Novella has conceded that almost two thirds of mainstream medical decisions are not based on experimental evidence. And for decisions concerning combinations of drugs the rate is probably much lower.

    As Harriet point out, sometimes drugs are prescribed to treat side effects of other drugs. I know of cases that would almost be amusing, if they weren’t tragic, with chains of drugs, each treating a side effect caused by another, eventually forming a complete circle. Would a sane MD prescribe a drug that causes hypertension to a patient whose original complaint was hypertension? Or a drug that causes depression to a patient whose original complaint was depression? Well, it happens all the time.

  8. qetzal says:

    It’s amazing how certain people seem so unwilling/unable to read. One might even think they were deliberately misrepresenting things (i.e. lying).

    It’s such a simple matter to click on the link Dr. Novella provided above:

    From there, one can click on the link provided to the original analysis by Dr. Imrie here:

    And there, one can see all of the citations of the original scientific literature offered in support of the claim that on average, 76% of interventions are supported by some form of compelling evidence.

    Now, an honest critic might attempt to dispute those primary literature citations or their interpretations. A dishonest one (e.g a liar) would claim that it’s Dr. Novella’s figure, and imply that it’s influenced by his bias. A dishonest one (e.g. a liar) would also equivocate between “[only] 37.02% of interventions are supported by RCT” and “almost two thirds of mainstream medical decisions are not based on experimental evidence.”

    As if RCTs are the only form of experimental evidence.

    I suppose an uneducated person might not understand that there are other forms of experimental evidence besides the RCT. But someone with a PhD would never be so foolish, right?

    So what are we left to think about a person who claims to have a PhD but simultaneously says such things? One might feel forced to question such a person’s honesty.

  9. pec says:

    Harriet was talking about drug interactions, most of which we know have not been tested experimentally. Novella decided to include well-known and accepted things like surgery for appendicitis. And everything he considers plausible is thrown in with evidence-based medicine. Well no, it is not evidence-based just because you think it’s true. You would not accept that reasoning from CAM.

  10. Harriet wrote: “He repeated a commonly cited estimate that only 30% of medical practice is evidence-based. ”

    “Medical practice” not drug therapy. I quoted the literature on medical practice.

    The figure of 78% was not based upon my or anyone else’s assessment of plausibility, but rather on evidence for efficacy. “Evidence-based” is not black or white – there are degrees of evidence. Reasonably evidence based was the standard used in these studies – you can read them to see their methods.

  11. Numenaster says:

    Wait, I want to know more about this brick vs wood houses differential radiation exposure. Any pointers to more info on that?

  12. Harriet Hall says:


    Background Radiation (Millirems Per Year)
    Cosmic Rays 45
    Air 4
    The Earth 15
    Food 25
    Building Materials – Living in a Brick House 45
    Living in a Stone House 50
    Living in a Wood House 35

    Man-Made Radiation (Millirems Per Year)
    Dental X-Rays – Bitewing Series 40
    Panoramic 500-1000
    Coast to Coast Airline Flight 4
    Color Television 1
    Living within One Mile of a Nuclear Power Plant 1/10


Comments are closed.