9 How much toothpaste, and does fluorosis matter?
Among the components of DBOH’s toothbrushing recommendation, this is the one that most openly trades caries prevention against a harm, and it does it in a single bullet, without saying that is what it is doing.
The advice
- using only a smear of toothpaste (children under 3)
- using a pea-sized amount of the toothpaste (children 3 to 6)
Strength of recommendation: Strong
— Delivering Better Oral Health, chapter 2, tables 1a and 1b (1)
What the guideline says its evidence is
Advice to use a smear only based on possible fluorosis risk (inconclusive evidence).
— Delivering Better Oral Health, chapter 13, table 1 (1)
and for the pea:
Advice to use pea-sized amount based on possible fluorosis risk.
— Delivering Better Oral Health, chapter 13, table 2 (1)
Read those twice, because they are doing something none of the other bullets in the same recommendation do.
This component is not about tooth decay at all. Everything else in that Strong recommendation is aimed at preventing caries. This one is aimed at preventing something else, dental fluorosis, and it does so by limiting the intervention that prevents caries. It points in the opposite direction from its neighbors.
And the guideline’s own word for the evidence is inconclusive.
What fluorosis is, and how worried to be
Fluorosis is a disturbance of enamel formation caused by swallowing too much fluoride while the permanent teeth are still forming under the gum, roughly before age six. In its mild forms, which are the forms this argument is about, it appears as faint white flecks or lines on the enamel. It is not painful, it does not weaken the tooth, and in most cases nobody but a dentist notices it.
The Cochrane review DBOH cites here is about topical fluoride as a cause of fluorosis (2). Its relevant findings, as SIGN 138 summarizes them, are that starting fluoride toothpaste before 12 months of age may be associated with an increased risk of mild fluorosis, and, notably, that there was no significant association with the frequency of toothbrushing (brushing less than twice daily compared with twice daily or more: OR 0.88, 95% CI 0.71 to 1.08) (3).
That is the 2010 version of the review, and it is the version DBOH cites.
Cochrane updated this review in June 2024 (4). Delivering Better Oral Health was published on 10 September 2025 and cites the 2010 version.
I only caught this because a reviewer checked the citation rather than taking my word for it. An earlier draft of this chapter stated that no update existed and that “the evidence on fluoride toothpaste and fluorosis has not been systematically updated in fifteen years.” That was simply wrong, and it was the same failure this book is about: I asserted an absence without running the search that would have found the thing.
The update matters in both directions.
It weakens the guideline’s stated basis. Wong and colleagues now report that starting fluoride toothpaste at or before 12 months, against after 12 months, is associated with fluorosis at RR 0.98 (95% CI 0.81 to 1.18), from two cohort studies in 260 children, at very low certainty. That is a null, on almost no data. The 12-month threshold in the guidance chain is weaker in 2024 than it looked in 2014.
It strengthens the harm side of a different component. The one thing the update rates at moderate certainty is not the amount of toothpaste but its concentration: two randomized trials in 1,968 children found that lower concentrations reduce fluorosis, at RR 0.75 (0.57 to 0.99) for 550ppm against 1,000ppm, and RR 0.72 (0.58 to 0.89) for 440ppm against 1,450ppm. So the benefit-harm trade-off does have a moderate-certainty arm. It just is not the arm this bullet is about.
SIGN’s own assessment of the trade-off is worth quoting because it is the clearest statement of it anywhere in this chain:
While the review provides some evidence that brushing a child’s teeth with fluoride toothpaste before the age of 12 months may be associated with an increased risk of developing mild fluorosis, for children assessed by their dentist to be at high risk of tooth decay the benefit of reducing caries probably outweighs the risk of mild fluorosis.
— SIGN 138 (3)
SIGN also flags a methodological point I would have made myself: fluorosis was treated as binary, present or absent, when it is in reality a continuum, and all the evidence in the review concerns mild levels judged on appearance.
The problem with a smear
Here is what nobody has established: that a smear delivers enough fluoride.
The evidence for fluoride toothpaste, set out in Chapter 8, comes from trials in which participants used a normal amount of toothpaste. The concentration was varied and randomized; the quantity was not. I did not find a trial that allocated children to different amounts of paste and measured caries, and I want to be careful about how much that sentence can carry: it means I did not find one in the search recorded below, not that none exists.
On the harm side, quantity has at least been looked at. Wong and colleagues’ 2024 update pools it as their third comparison, and finds nothing decisive: using less than half a brush of toothpaste against half or more gives OR 0.77 (95% CI 0.41 to 1.46) from two case-control studies in 258 children, and OR 0.92 (0.66 to 1.28) from three cross-sectional surveys in 2,037 children, both at very low certainty (4). So the amount of toothpaste has been studied observationally, and the answer is that nobody can tell.
So the smear recommendation asks parents to use less of an intervention than the caries trials used, on the basis of very low certainty evidence about a different outcome, inside a recommendation labeled Strong for reasons that have nothing to do with either.
I want to be careful here, because this is the point at which a reader could take the wrong lesson. I am not saying use more toothpaste. A smear or a pea is almost certainly fine, and the amount of fluoride in a pea of 1,000ppm paste is not trivial. The limit is intended to reduce fluorosis risk in a group who cannot reliably spit, which is a sensible thing to intend; whether these particular amounts achieve it is exactly what remains uncertain. What I am saying is that this is a genuine benefit-harm trade-off, made on inconclusive evidence, presented to parents as a Strong instruction with no indication that a trade-off was involved.
GRADE’s evidence-to-decision framework exists precisely for cases like this. When a recommendation balances a benefit against a harm, the panel is supposed to consider the magnitude of each, how patients value them, and the certainty of both.
Some of that work has clearly been done: the smear-versus-pea distinction by age tracks the period of enamel formation and the child’s ability to spit, which is sensible. What is not published is the reasoning. How much caries prevention is being given up? How much fluorosis is being avoided? At what certainty, on each side? Chapter 13 gives one clause, “based on possible fluorosis risk (inconclusive evidence)”, for what is a two-sided judgment.
I cannot reconstruct the calculation, and neither can a parent.
PubMed, 21 August 2026, in two passes.
The first pass searched for an update to CD007693 by review ID and found nothing, which I reported as meaning the 2010 version was current. That was a retrieval failure on my part, not an absence: the 2024 update (DOI 10.1002/14651858.CD007693.pub3, PMID 38899538) is indexed and a title search returns it immediately. The second pass, run after that was pointed out, retrieved it. Both passes are recorded, the failed one included, because a search that missed something is more informative than one that did not.
The 2024 review’s own search ran to 28 July 2022 and it includes 43 studies: three randomized trials, four cohort studies, ten case-control studies and 26 cross-sectional surveys.
I have the full abstract and summary of findings for the 2024 update, and SIGN 138’s summary of the 2010 version, but not the full text of either.
I did not find a trial randomizing the quantity of toothpaste against a caries outcome. One database, one screener: read that as the limit of what I looked at.
Full record in appraisals/searches/ch09-toothpaste-amount.md.
Verdict
- Certainty of evidence
-
No estimate at all for the quantity of toothpaste as a caries-relevant decision. I am deliberately not writing “very low” here: a certainty rating is a property of an effect estimate, and rating an unasked question would be the same category error this book opens with (Chapter 2). Nothing I found has randomized the amount of paste against caries.
For the fluorosis side, very low and inconclusive, on the 2024 update rather than the 2010 review the guideline cites: OR 0.77 (0.41 to 1.46) and OR 0.92 (0.66 to 1.28) for less versus more paste (4). The guideline’s own word for its version of this evidence is inconclusive, and that word survives the update.
- Directness to the advice as worded
- Poor, and in an unusual way. The evidence concerns fluorosis; the recommendation sits inside a caries-prevention bundle. It is not that the evidence fails to match the advice, but that the advice has quietly changed outcome without saying so.
- Is the strength label defensible?
- Not as it stands. On DBOH’s own categories this looks like a Good practice point: a sensible precaution resting on clinical consensus and inconclusive research. It may well be the right precaution. Putting it under the same Strong label as five caries bullets stops a reader of the summary table from seeing that these components rest on different certainty and on different benefit-harm reasoning.
- What would change my mind
- A trial randomizing smear against pea against a full brush-length in young children, measuring both caries increment and fluorosis at age eight. That trial has ethical complications but is not impossible; Wong and colleagues argue that new randomized trials of fluorosis are unethical and that further evidence will have to come from observational work, which if right means this question stays open indefinitely. Failing that: publication of the trade-off reasoning the panel actually used, and a guideline that cites the 2024 review rather than the 2010 one.
What this means for you
Use a smear for a child under three and a pea for a child of three to six. It is sensible, the fluorosis it aims at is real if minor, and no one is suggesting otherwise.
What you should not conclude, if you have ever squeezed out slightly too much, is that you have done something dangerous. The amount is a precaution against faint white marks on adult teeth, taken on evidence the guideline itself calls inconclusive and which the 2024 update leaves inconclusive. The part of the instruction that has actually been tested, and that matters far more, is that the toothpaste contains fluoride at all.