A population-level read of the evidence, not personal advice. “Non-sugar sweeteners” (NSS) covers aspartame, sucralose, saccharin, acesulfame-K, stevia and the rest — a chemically diverse class the evidence mostly treats as one, so a finding on saccharin is not automatically a finding on stevia. The single fact that organises everything below: the answer depends entirely on what the sweetener is replacing.
The bottom line
The WHO’s 2023 guideline suggests not using sweeteners to control weight or lower disease risk — but read the wording exactly: it is a conditional recommendation on low-certainty evidence (Non-Sugar Sweeteners). That means “the expected payoff of using them for weight control isn’t positive on current evidence,” not “they are proven to harm you.” Two separate things drive it: the short-term benefit is real but trivial and fades, and the long-term data are genuinely unresolved.
Two things you have to hold at once
1. The comparator is the whole game. Sweeteners are never eaten in a vacuum — judge them against the realistic alternative:
- Versus sugar: a small, short-lived weight benefit — pooled across trials, about 0.7 kg lost.
- Versus water: nothing — no effect on weight, a hair’s-worth increase in BMI.
- As an actual sugar swap (habitual sugar users told to switch to diet versions): the benefit shrinks to 0.6 kg and loses significance. The weight loss comes from cutting the sugar calories, not from anything the sweetener molecule does.
2. The trials and the long-term studies disagree — and WHO leaves it open. Short randomised trials show that small weight benefit; long-term observational cohorts show sweetener use tracking higher rates of hard disease:
| Outcome (cohorts, up to ~13 years) | Association | Certainty |
|---|---|---|
| Obesity | HR 1.76 (a 76% higher rate) | Low |
| Type 2 diabetes | HR 1.23-1.34 | Low |
| Cardiovascular disease | HR 1.32 | Low |
| Stroke | HR 1.19 | Low |
| All-cause mortality | HR 1.12 | Very low |
| Cancer (overall) | HR 1.02 — null | Very low |
The obvious escape hatch is reverse causation: people already heavy or heading toward diabetes switch to sweeteners because of that, making the sweetener look guilty for the condition that prompted it. But this is the honest, load-bearing point — WHO ran that check and refused to dismiss the cohorts: after adjusting for BMI, dropping early follow-up, and the rest, the associations “cannot be dismissed as being solely a result of reverse causation or residual confounding.” So the defensible statement is the uncomfortable middle: the long-term signal is unexplained — neither proven causal nor waved away — and that unresolved arrow is exactly why the recommendation is only conditional.
Your specific questions
- Do they cause cancer? In the long-term outcome data WHO reviewed, overall cancer is null (HR 1.02); the one exception is an old saccharin-bladder signal at very low certainty. Separately, in 2023 the IARC labelled aspartame “possibly carcinogenic” — and this is where hazard versus risk matters: a hazard label asks whether a thing could cause cancer at some dose, not whether it does at the amounts people actually consume (the dose-and-real-world-risk question the food-safety bodies handle). Those specific hazard/dose evaluations are not in our held evidence — a named gap — so on what we hold: no demonstrated cancer risk at normal intake, and a hazard headline that is routinely misread.
- Do they spike insulin? No — in the trials, fasting glucose, insulin, and HbA1c are flat. (This is exactly why a continuous glucose monitor sees nothing: there’s no glucose to see — Continuous Glucose Monitoring as a Health Intervention.) The live maybe is the gut microbiome: some sweeteners shift it and nudge glucose tolerance in small human studies — mechanistically real, importance unsettled. Not held here; flagged, not asserted.
- Do they increase appetite or perpetuate sweet cravings? Plausible stories exist (sweetness without calories may blunt learned satiety; keeping the “sweet setpoint” high) — but the human evidence is thin and short. This sits at insufficient evidence, not a finding.
- Do they help you lose weight? Versus sugar, marginally and briefly. Long-term, or versus water, no — and WHO’s point is that a few months of minor weight loss on a scale isn’t a health outcome (Surrogate Outcomes).
So what does this mean for a real choice?
- Sweeteners have no nutritional value, so finding “no benefit” costs you nothing by skipping them.
- WHO’s preferred swap is explicit: replace sugary drinks with water or unsweetened options (or whole fruit), not with diet versions — swapping sugar for a sweetener leaves overall diet quality roughly where it was.
- But the realistic-alternative rule cuts the other way for an individual: if your actual choice is a sugary soda versus a diet one, and water is not on the table, the diet one is still the better of those two. The guideline is answering a population-policy question (don’t build health strategy on sweeteners), not refereeing your next drink.
Two groups where the read differs
- Pregnancy: higher sweetener use is associated with preterm birth (OR 1.25) at low certainty — enough to warrant more caution than the general adult picture.
- People with existing diabetes were excluded from the reviewed evidence, so this guidance simply doesn’t speak to them.
What we cannot yet say (routed for later)
- The aspartame cancer hazard (IARC 2023) and safe-dose (JECFA 2023) evaluations are not held here — the guideline covers only the long-term outcome side and defers the hazard-and-dose question to those bodies. Closing the cancer sub-question properly needs them.
- Appetite, cravings, and the gut-microbiome route are named mechanisms, not findings — they need a dedicated trial.
The honest limits
- This grades the evidence, not the outcome. Whether sweeteners actually move a hard health outcome over decades is a fact no study here can see — you can’t randomise a lifetime of diet-soda drinking.
- Built on one guideline (WHO 2023), and the reverse-causation call rests on the review authors’ own sensitivity analyses, not an independent re-analysis.
- It appraises, it does not prescribe. The right choice depends on your realistic alternative and your goals.
- Class-level, with real within-class differences: aspartame, sucralose, stevia and saccharin differ chemically and may differ physiologically; a single-sweetener result should not be stretched to all.