Orbiter of the fluoride cluster (nucleus: Fluoride and Health). This page holds the
benefit arm — the reason fluoride is added to water at all — cashed from the 2024 Cochrane update
(supersedes the 2015 pub2). The decision object is community water fluoridation (CWF, ~0.7 mg/L) ->
child dental caries. Caries (decayed/missing/filled teeth) is a patient-important outcome, not a
surrogate — it transmits directly to pain, tooth loss and quality of life -> Surrogate Outcomes.
The headline: the benefit is real but SMALL and LOW-CERTAINTY in the contemporary era
The 2024 update’s whole move is to stratify by era — before vs after ~1975, when fluoride toothpaste became widespread. All 157 included studies are non-randomised (RCTs judged «logistically unfeasible for this topic»), so «we downgraded the certainty of all evidence for these risks» (Iheozor-Ejiofor et al., 2024).
Contemporary (after-1975) initiation of CWF, primary-teeth caries (the best-certainty benefit estimate the review holds):
- «the initiation of CWF may lead to a slightly greater change in dmf over time (mean difference (MD) 0.24, 95% confidence interval (CI) -0.03 to 0.52; P = 0.09; 2 studies, 2908 children; low-certainty evidence). This equates to a difference in dmf of approximately one-quarter of a tooth in favour of CWF; this effect estimate includes the possibility of benefit and no benefit.» (Iheozor-Ejiofor et al., 2024)
So the contemporary benefit is ~0.24 of a tooth, low certainty, and the CI crosses the no-effect line — «benefit and no benefit» both remain in play. The permanent-teeth arm is weaker still: DMFT MD 0.27 (95% CI -0.11 to 0.66; 4 studies; very low, heterogeneity + imprecision — CI crosses 0) (Iheozor-Ejiofor et al., 2024). The caries-free proportion moves ~3-4 percentage points toward CWF (primary MD -0.04, 95% CI -0.09 to 0.01, low; permanent -0.03, -0.07 to 0.01, low — both CIs cross 0) (Iheozor-Ejiofor et al., 2024). Only children were studied — «We found no eligible studies that report caries outcomes in adults» (Iheozor-Ejiofor et al., 2024).
The pooled-all-eras figure OVERSTATES the contemporary benefit — the era-stratum is the decision
The large, familiar fluoridation benefit is a pre-1975 number, and quoting it for a contemporary population is the benefit-side analogue of quoting a bare «fluoride lowers IQ» without the >1.5 mg/L bound (the harm-side stratum error on Fluoride and Child IQ). Pre-1975 initiation gave a dmf MD of «2.10 (1.71 to 2.49)» (5 studies, 5709; CI excludes 0) (Iheozor-Ejiofor et al., 2024) — roughly nine times the contemporary point estimate, and unlike it, statistically clear. The review downgrades all pre-1975 evidence for indirectness (inapplicability to contemporary settings). The authors’ conclusion states the attenuation plainly: «Contemporary studies indicate that initiation of CWF may lead to a slightly greater reduction in dmf and may lead to a slightly greater increase in the proportion of caries-free children, but with smaller effect sizes than pre-1975 studies» (Iheozor-Ejiofor et al., 2024).
A contemporary UK single-time-point series (collected informally, outside the formal review) lands the same small magnitude — «the most recent single time point studies showing a mean difference of 0.16 to 0.21 dmf between fluoridated and low-fluoridated/non- fluoridated areas» (Iheozor-Ejiofor et al., 2024) — a cross-design corroboration of the ~0.24 estimate, though the review flags that the comparison «is not strictly appropriate» (design + outcome-measure mismatch).
Why the benefit shrank: substitution, not a failed intervention (the compensation mechanism)
The attenuation is a net-effect / realistic-alternative point, not evidence that fluoride stopped working. The counterfactual (no-CWF) population now gets fluoride from other routes, so the marginal benefit of adding it to water is smaller: «In countries where the widespread use of fluoride toothpastes has increased from the mid to late 1970s, along with increased access to other caries-preventive strategies of proven effectiveness, such as fluoride varnishes … and dental sealants …, the benefit of water fluoridation may be diluted» (Iheozor-Ejiofor et al., 2024). This is the Layer-3 substitution rule made concrete — judge CWF against the realistic alternative (a population already using fluoride toothpaste), not against no-fluoride-at-all.
The pre-1975 evidence is not worthless — it may still transport to settings the modern era does not resemble: «they may still be relevant to countries in which tooth decay is very high and people don’t have easy access to fluoride toothpaste» (Iheozor-Ejiofor et al., 2024). The benefit is baseline-risk-dependent (route (a)): larger where caries burden is high and other fluoride sources are scarce.
The paired harm at the SAME dose: dental fluorosis (established, but mild)
At the fluoridation dose itself, one harm is established (unlike the IQ and skeletal arms, which need >1.5 mg/L). The fluorosis estimates are carried unchanged from the 2015 review (current to Feb 2015, not re-searched): at 0.7 ppm, «the percentage of partici- pants with dental fluorosis of aesthetic concern was esti- mated to be 12% (95% CI 8% to 17%)» (40 studies, 59,630), and any level ~40% (95% CI 35% to 44%; 90 studies) (Iheozor-Ejiofor et al., 2024), low certainty. Dental fluorosis is a cosmetic / enamel outcome — a real unwanted effect at the fluoridation dose, but not on the patient-important tier of the child-IQ or fracture arms. It is the one harm that sits squarely at 0.7 mg/L rather than above 1.5.
Certainty, and why it stays low
- Design floor: every estimate is non-randomised (ROBINS-I). Contemporary caries evidence is low or very low; the two contemporary low-RoB studies (downgraded one level only) are Blinkhorn 2015 and Goodwin 2022 (Iheozor-Ejiofor et al., 2024). Pre-1975 evidence is very low (RoB + indirectness).
- Imprecision is the binding downgrade contemporary-side: the benefit CIs «include the possibility of benefit and no benefit».
- Cessation is insufficient-evidence, not a null — one after-1975 study, very low; the effect of stopping CWF cannot be determined -> The Insufficient-Evidence Statement.
- Delivered dose is not the target dose: English artificially fluoridated supplies were «within the optimal range of 0.7-1.0 mg F/L in 27.7%-77.8% of samples» (Moore 2019, quoted in the review) (Iheozor-Ejiofor et al., 2024) — real-world exposure is more variable than the nominal 0.7.
Non-health axes named, not priced (one-axis rule): the review runs a brief economic commentary (CWF «appears to offer good value for money due to its low per capita intervention delivery costs», context-dependent) (Iheozor-Ejiofor et al., 2024) and an environmental life-cycle analysis, and frames CWF as an equity intervention — but only one study reported disparities data («no evidence that deprivation influences the relationship») (Iheozor-Ejiofor et al., 2024). Cost, environment and equity are separate axes: the wiki records that they exist and stops.
Why this is a decision-change, not a restatement
The public framing is fluoridation prevents cavities, so it is good. The fabric’s move is the era-stratum: for a contemporary population already using fluoride toothpaste, the marginal caries benefit of CWF is about a quarter of a tooth, low-certainty, with a CI that includes no benefit — while a familiar large number describes a 1950s-60s population that no longer exists. The benefit is genuine and baseline-risk-dependent (larger where caries is high and other fluoride is scarce), and it sits at the same 0.7 mg/L dose where the serious harm arms are insufficient-evidence and only mild dental fluorosis is established. The dose- and era-stratified balance is synthesised on Fluoride and Health.