How to decide which supplements to drop (subtraction as discipline)

By Marcus Renshaw ·

My supplement stack only ever grew. A podcast, a good month, a compelling study, and something new went in. Almost nothing ever came out, because dropping a supplement means admitting you wasted money on it and betting it was not the one quietly doing the work. The stack you end up with is expensive, unmeasurable, and mostly faith. The skill nobody sells you is subtraction.

Adding a supplement is easy; subtracting is the skill. Every item in a stack should have a specific job and a way to tell if it is doing it. To test one, drop it for a window long enough for the marker it targets to move, and watch. If nothing changes that you can feel or measure, it was not earning its place.

Why stacks only grow

Adding is frictionless and feels productive. Dropping fights three biases at once. Sunk cost: you already paid for the bottle. Loss aversion: the fear that the thing you cut was the one thing working. And identity: a big stack feels like taking your health seriously. None of those is evidence. Together they produce a pile of capsules that no one, including you, can account for.

Why a supplement seems to work when it does not

Before you defend anything in your stack, account for the two ways it can fool you.

The first is regression to the mean. You tend to start something new when you feel worst, and from your worst you are likely to drift back toward your average anyway. The supplement gets the credit that time alone earned. This statistical artifact, not the placebo effect, is the main reason ineffective things appear to work, and placebo on top of it is real but usually small.

The second is confounding. If you started three things the same month, changed your sleep, and got a lighter workload, you cannot assign the result to any one of them. Self-experimentation with no control group cannot separate a real effect from natural improvement unless you change one variable at a time and watch.

The subtraction test

This is the method, and it is deliberately boring.

First, give every item a job. Write next to each supplement the specific reason it is there: a marker it should move, a symptom it should touch, and the tier of evidence that says it can do either. Anything you cannot finish that sentence for is already a candidate to drop.

Second, drop one and watch. Stop a single item for a defined window, long enough for whatever it targets to actually change. That window is the same one from retest cadence: weeks for a hormone or a blood marker, not days. Change one thing at a time, or you learn nothing.

Third, read the result honestly. If a number you track moves, or a symptom returns, it had a job and did it, so it goes back. If nothing you can feel or measure changes across the window, that is your answer: it was not doing anything you can detect, and detection is the only thing you are paying for. This is the same rule as the low-effort changes: keep what moves the number, drop what does not.

A real one from my own stack: I ran berberine for glucose control and dropped it when it wrecked my gut. That subtraction was easy, because by then a GLP-1 was doing the glucose work far more powerfully, so berberine had no job left to justify the cramping. Nothing to replace, and one fewer variable confounding every future draw. The item I dropped taught me more than most of the ones I kept.

The same logic cuts complexity, not just items. I moved my zinc to the morning, not because it absorbs better there, it does not, but because I kept skipping the extra step at night. A supplement you do not actually take has no job by definition, so compliance beats the optimal timing you never follow.

The hidden tax: your stack corrupts your own labs

Subtraction is not only about money. A bloated stack degrades the exact data you use to make decisions.

The cleanest example is biotin. It is in most hair, skin, and nails formulas at 5 to 20 mg, hundreds of times the roughly 0.03 mg you need, and at those doses the FDA warns it interferes with common immunoassays. It can push thyroid results (TSH, FT4, FT3), hormones, and even troponin either falsely high or falsely low, and there are documented cases of misdiagnosis from it. So a supplement you barely think about can be the reason your thyroid panel reads wrong, which then feeds a bad decision about a real drug. The more you take, the more your reference ranges mean something other than what you think. Every item you drop is one fewer thing confounding your bloodwork.

Keep a Tried and Dropped list

The list of what you dropped is more useful than the list of what you take, because it is your actual n-of-1 evidence. Record each item you cut, the dose, how long you ran it, what you watched, and what happened when it left. That record does two things: it stops you re-adding the same supplement the next time a podcast makes it sound new, and it turns your stack from a collection of hopes into a set of decisions you can defend.

This is the whole reason the supplement and compound journal in the bundle I built has a column for the drop, not just the add: every item with its job, its dose, its start date, and what changed when it went. The add is the easy half. The drop is where the learning is.

Take your current stack and write one job beside each item. Anything you cannot fill in is your first subtraction test. Drop it for a window, watch, and start the list.

FAQ

How do I know if a supplement is actually working?

Drop it, one at a time, for a window long enough for whatever it targets to move, and watch a number or symptom you track. If nothing changes that you can feel or measure, it was not doing anything you can detect. Improvement while you were taking it is not proof, because you may have started it at your worst and simply regressed toward your average.

How long should I stop a supplement to test it?

Match the window to the biology of what it targets, the same as retesting bloodwork. A blood marker or hormone needs weeks, not days, to settle at a new level, so a meaningful test is usually four to eight weeks, not a long weekend.

Isn’t it risky to stop a supplement?

For ordinary supplements, stopping to test is low-risk and informative. Prescription medication is different: do not stop anything a clinician manages without talking to them. The subtraction test is for the shelf of optional capsules, not your prescriptions.

Why do supplements seem to work at first?

Mostly regression to the mean. You start them when you feel worst, and you were likely to improve toward your average regardless. That statistical artifact, plus a usually small placebo effect, credits the supplement for recovery that time alone produced.

Can supplements really mess up my blood tests?

Yes. High-dose biotin, common in hair and nails formulas, interferes with many immunoassays and can make thyroid, hormone, and cardiac results read falsely high or low, per an FDA warning. A stack you forgot about can be the reason a panel looks wrong.

What should I write in a Tried and Dropped list?

For each item: the dose, how long you ran it, the job it was supposed to do, what you measured, and what happened when you stopped. That record is your own evidence, and it is what stops you buying the same thing twice.

How many supplements is too many?

The moment you can no longer say what each one is for is the moment you have too many. Count is not the problem. Unaccountable count is.

Sources