Drug-Nutrient Interactions:  Solving Three Practical Problems  

Understanding drug-nutrient interactions is critical in the practice of integrative medicine.  Despite an abundance of online resources, reliable information is hard to find.

Entering a drug and a herb into an online drug-supplement interaction checker is likely to retrieve a long list of “possible” or “potential” interactions.  It’s hard to know which interactions really matter, and which ones are merely educated guesses.   

For example, entering “melatonin” into the Medscape interaction checker retrieves a list of the top prescribed medications as “possible interactions:”

• Anticoagulants
• Antiplatelet drugs
• Anticonvulants
• Antidiabetes drugs
• CNS depressants
• Immunosuppressants
• CYP1A2 substrates
• CYP2C19 substrates
• Oral contraceptives

If you add up the statistics and usage data, this list adds up to about 2200 FDA-approved prescription and OTC drug entries.   The majority of American adults who take medications are taking one of these drugs.   Should they all avoid melatonin?

Absolutely not.   

Rule of thumb: Most of the interactions you read about are derived from leaps and liberties in scientific “interpretation,” that would never be acceptable elsewhere. Most interaction warnings are further magnified by an overabundance of caution and an agenda to appear scientifically “comprehensive.”   Beware that this “comprehensiveness” increases the likelihood of clinical errors on your part.

You may already experience “overalerting” fatigue, where real interactions are easy to overlook because of so many warnings on your screen. Part of your success depends on knowing what to ignore.

Let’s define 3 practical problems in this area, their underlying causes, and solutions.

• There are far too many warnings.   The scientific evidence behind drug-supplement interactions is perhaps one of the least mature areas of medical knowledge.  Nevertheless, intearaction checkers generally provide too many warnings, many of which are merely hypothetical because of a sparse evidence base.  This “overalerting” or “crying wolf” phenomenon leads to precaution fatigue, which could lead to errors when real warnings present themselves.  All plants (herbs, fruits, and vegetables) contain p450 enzyme inhibitors that block drug metabolism conduits in cell culture dishes, but they typically fail to produce this effect in clinical trials.  These studies generally don’t translate to the clinic, but interaction checkers use this logic liberally. This is the only area of medicine where undisciplined use of in vitro data is accepted, and it needs to stop.

• You don’t have time to become a pharmacologist.  I spent many extra years of education preparing to work in this area, and it still takes me several hours to review relevant publications appropriately to make a good recommendation on a possible interaction to a physician.    To discern real warnings from guesses, you need time to do the reading, along with expertise in interpreting this type of research data.

• Inconsistent and conflicting information is endemic in the world of interaction checkers.  Coverage of dietary ingredients varies widely and different tools describe the nature of interactions in vastly different ways. Contrasting information is everywhere. The only way to “vet” a checker is to have a Zoom call with the scientific curators of the tool, and spend several days examining the continuity between evidence and output.

The solution to the foregoing challenges is quite simple.  Checkers should display only what matters, based on rational, responsible scientific data interpretation. If the science experts can’t weed out irrelevant data, then how will you? This diligence should reduce the number of warnings you see, not increase them.

• Deliver clinically relevant information.

There isn’t much direct clinical research on interactions between drugs and dietary supplements.  Clinical trials seldom investigate drug-supplement combinations in a way that informs us about possible interactions. These trials are expensive, with limited external validity because of dose and formulation discrepancies.  Consequently, online checkers and resources are left with limited information to work with.  They must resort to lower quality evidence from anecdotal adverse event reports and preclinical experiments.  Case reports can seldom establish an interaction because of too many confounding variables, and preclinical (in vitro and animal studies) generally cannot be translated to the clinic.    In vitro experiments can be useful if the concentration of drug and nutrient can be achieved clinically (we’ll come back to this). 

• Communicate the evidence and its limitations

Many interactions that you read about are merely hypothetical (educated guesses) based on limited evidence from cell culture, case reports and animal studies.  A good interaction checker should provide the reference(s) that you can review when making your decision.   Proven interactions that are covered in the Prescriber’s Desk Reference (PDR) and the drug package insert should be included.  

• Filter for clinical signifiance

I reviewed 949 interactions indexed in a leading drug-nutrient interaction resource and found that 66% of them failed to meet the following criteria for clinical significance:

1. Evidence.  The current body of evidence should support or suggest a possible interaction.   
2. Effect size.  The interaction produces a discernable effect on the body.
3. Extrapolation.  The interpretation of the original research is correct.    
4. Exposure.  All interactions are dose-dependent, so dosage and bioavailability are major determinants of clinical significance. Both factors are usually ignored in checkers.

I revisited the warnings for melatonin, and by filtering for these “4E” criteria, I reduced the number of potential interactions from 2200 to 575.  That’s a 74% reduction in meaningless “noise.”

Less is more

A checker that provides a large, unfiltered list interactions may give the impression of being scientifically diligent, complete and comprehensive.  But the scientific reliability of a checker isn’t about the volume of information – it’s about the quality of information.  It should give you only what you need, based on thorough scientific evaluation, to minimize integrative dispensing errors.

Stay tuned for upcoming learning resources on this topic.