It Got Better After the Treatment — That Alone Proves Nothing

On day three of a cold you feel wretched, so you take something — a supplement, a syrup, a remedy a friend swears by. By day five you are clearly better, and the conclusion seems to write itself. But a cold clears up on its own in about a week, sometimes two, so almost anything taken during those days will look like it worked. Getting better after a treatment is not, by itself, evidence that the treatment did anything.

What is happening

Three ordinary explanations sit underneath almost every "it helped me," and from the inside all three feel exactly the same.

The illness was going to end anyway. CDC (US) describes the common cold as usually lasting less than a week; the NHS (UK) says people generally improve within one to two weeks, and young children take longer. A cough can hang on for up to three weeks on its own — longer than that is worth checking with a doctor. Whatever you took landed inside that window, and the window closed on schedule.

Regression to the mean. You do not reach for a remedy on a random Tuesday. You reach for it on the worst day, when symptoms are at their peak — and from a peak, the likely direction is down, remedy or no remedy. Hartman describes this as the engine that keeps ineffective treatments in business: by their nature they get taken exactly when things are about to improve anyway. A 2020 paper in Social Science & Medicine put the same point plainly — conditions that resolve spontaneously can increase the belief that ineffective treatments work. A self-limiting illness is perfect soil for a false cure. The general shape of this pattern is on behind extreme outcomes lies a return to the mean; this page is its medicine-cabinet version.

Placebo. It is real, but smaller than its reputation. A Cochrane review of placebo interventions across all clinical conditions found no clinically important effects on average, with a possible exception for subjective outcomes such as pain. So "it was just placebo" is not a universal explanation either — and you cannot tell placebo from natural recovery by how you feel. Only a control group, people who did not get the treatment, separates them.

Notice what none of this claims. It does not say your remedy failed. It says your experience does not contain the information needed to tell.

What to do

• Ask when you started. At the peak, or on the way up? Anything begun on the worst day is almost guaranteed to be followed by improvement.
• Ask what would have happened with nothing at all. That missing comparison, not your memory, is the entire question.
• Look for people, not dishes. Evidence worth the name is a trial in people comparing the treatment against placebo or against standard care. "It kills the virus in a test tube" is a fact about a test tube.
• Read "it helped me" and "it went away on its own" as the same sentence until something separates them. A single vivid case separates nothing — see one example is not a pattern.
• Watch the story slide from sequence to cause. Two things in a row is correlation, not causation.
• Check any numbers that come with the claim. The question in what to do when you see a scary statistic works just as well on hopeful ones.
• Ask what would change your mind. If nothing could — if recovery proves the remedy and a bad course proves you started too late — you are not looking at evidence.

When this is not the case

Not every condition resolves by itself. This reasoning has force precisely because a cold ends on its own. Where an illness does not, improvement after treatment carries far more information. The argument here is about self-limiting illness, not about medicine in general.

Relief is real even when the course is unchanged. If something makes you feel better for an hour, that is a reason to use it for relief — as long as it is safe with whatever else you take, and without concluding that it treats anything. Whether it shortened the illness is a separate question, and that one needs a control group.

The logic runs both ways. Feeling worse after taking something does not prove it harmed you either. The same week, the same peak, the same missing comparison.

This is a way to read evidence, not a reason to wait. Which warning signs mean an emergency number and which mean a doctor today is a medical question, not a question of study design. What needs a doctor today and what can wait is set out on what to do at the first signs of a cold; which cold remedies have actually been tested is on cold remedies and what the evidence says.

Sources

• Hartman, 2009 — The therapeutic effects of natural course and regression to the mean
• Social Science & Medicine, 2020 — Diseases that resolve spontaneously can increase the belief that ineffective treatments work
• Cochrane — Placebo interventions for all clinical conditions
• CDC — About the common cold
• NHS — Common cold

Checked: September 6, 2026