无糖口香糖里那个甜的东西，被发现和心血管事件有关：一万七千人，最高的一组高出百分之五十七

This episode is about a study just made public, and about a question worth more than the study itself: how do we decide that something is "healthier"?

The date is early September this year; the venue is the annual congress of the European Society of Cardiology.

A team from a large teaching hospital in Berlin presented long-term observational results on xylitol.

Let me define the subject first.

Xylitol is a sugar alcohol, a class of sweetener; it is close to sucrose in sweetness but lower in calories, and it does not raise blood glucose the way sucrose does.

Its most familiar homes are sugar-free chewing gum and toothpaste, and beyond that it is widespread across processed foods.

The logic behind its heavy use is very simple: consumers want sweetness, and they do not want the word "sugar" on the ingredient list.

Over the past twenty years, use of substitutes like this has climbed steadily alongside the public consensus on cutting sugar.

And a problem is already buried in that sentence — the thing being replaced has been studied exhaustively; the thing replacing it has not.

The design of the study itself is fairly direct.

The researchers enrolled just over seventeen thousand seven hundred participants, average age about sixty-one.

The sample came from two existing long-term cohorts: one in Canada, followed for six years, the other in Britain, followed for thirty.

The critical step is how exposure was measured: the researchers assayed xylitol levels in the participants' blood, rather than asking people to recall how much they had eaten.

That choice matters.

Self-reported dietary questionnaires carry enormous error; people's memory of what they have consumed over years is not reliable.

Using a blood marker instead of a questionnaire makes this data harder than that of comparable studies.

Then came follow-up, recording major adverse cardiovascular events — in plain terms, heart attacks and strokes.

The results are as follows.

In the Canadian cohort, the group with the highest blood xylitol had a subsequent cardiovascular event risk fifty-seven percent above the lowest group.

In the British cohort, the same comparison produced eighteen percent.

The two cohorts agree in direction and differ in magnitude.

Agreement in direction raises confidence; the difference in magnitude suggests that background population and follow-up length both affect the estimate.

The researchers also offered a plausible biological mechanism.

The normal duty of platelets is to aggregate into a plug at a break in a vessel and stop the bleeding.

In vitro work indicates that xylitol increases the tendency of platelets to aggregate.

At a wound, that capacity saves lives.

Inside an intact vessel, the same capacity is thrombotic risk.

The mechanism is coherent, and that makes the result harder to dismiss as noise.

But what really determines how this study should be used is its type.

This is an observational study.

Observational studies can establish correlation. They cannot establish causation.

The team put that sentence near the front themselves, which is the responsible thing to do.

Why is this distinction not academic nit-picking?

Because the single observation "blood xylitol is elevated" admits at least two explanatory paths running in opposite directions.

Path one: intake is high, blood level is high, platelet activity rises, vessels are damaged — on this path xylitol is the cause.

Path two: this person has an underlying metabolic disturbance, blood level is elevated as a result, and what is actually damaging the vessels is the disturbance itself — on this path xylitol is only a marker.

Both paths generate identical correlational data, and an observational design cannot separate them.

Only a randomised controlled trial can: recruit participants, randomise them, have one arm consume it and the other not, then compare endpoint events.

Trials like that are costly and slow, and on dietary components they are particularly hard — it is very difficult to hold people to a prescribed diet for decades.

For xylitol, no such trial exists yet.

So the only sound conclusion is this: a signal worth pursuing has appeared; a conclusion you could act on has not.

The researchers' advice sits at exactly that calibration: no need to panic, and no need to pretend you did not see it.

That covers the study.

I want to push one step further, because the decision structure is really the more interesting part.

For twenty years, public health guidance has pointed unambiguously at "reduce added sugar", and that direction has ample evidence behind it.

The problem is at the execution stage: the overwhelming majority of people cut sugar by substituting.

And the thing they substitute in usually satisfies exactly one condition — it is not sugar.

"Not sugar" is a negative description. It carries no positive information about safety.

Proving one thing harmful and proving another thing harmless are two independent judgements, each needing its own evidence.

In real decisions, we routinely transfer the evidence for the first to the second.

The words "sugar-free" on an ingredient list are what carry that transfer.

Legally it is only a statement of composition. Psychologically it gets read as a promise about health.

Two questions to leave you with.

The first is concrete: how many things in your house did you buy specifically because they were "healthier"?

And that judgement of "healthier" — where did you get it?

The second is bigger: when the evidence in a field only supports saying "we do not know yet", what should a person act on?