10 Fluoride varnish twice a year
This is the first professional intervention Part II audits. Everything before it is something you do at home; fluoride varnish is applied in a chair, by somebody else, and it is the reason a great many children are booked in twice a year. Chapter 11 covers the other one.
It is also, on the evidence, one of the better-supported things in the guideline, with one caveat that has grown more serious every year since 2013.
The advice
Apply fluoride varnish (2.26% NaF) to teeth 2 times a year
Strength of recommendation: Strong
— Delivering Better Oral Health, chapter 2, table 1b (1)
That is table 1b’s word order. Table 1d, for children aged 7 to 18, moves the concentration to the end: “Apply fluoride varnish to teeth 2 times a year (2.26% NaF)”. The adult row is table 1g, not 1f, and is worded as 1d is. For children at higher risk, table 1e says “2 or more times a year.” An earlier draft of this chapter attributed the 1b wording to tables 1b, 1d and 1f; table 1f carries no varnish recommendation at all.
One note on the chemistry, since this book quotes the guideline verbatim and the guideline’s notation here is loose. The product used across these trials is a 5% sodium fluoride varnish, which delivers approximately 2.26% fluoride ion (22,600 ppm). DBOH writes “2.26% NaF”, which reads as 2.26% sodium fluoride and is not the same statement. I have kept the quotation as printed, because changing words inside quotation marks is the thing this book most often complains about, but the figure to carry away is 5% NaF, 2.26% F.
What the guideline says its evidence is
Recommendation based on moderate certainty evidence.
— Delivering Better Oral Health, chapter 13, tables 2, 4 and 7 (1)
That is the whole of it for the twice-yearly rows. A longer statement exists, but it belongs to the higher-risk “2 or more times a year” row:
Recommendation based on moderate certainty evidence. Most studies used 2 applications per year. For guidance: manufacturers recommend application every 6 months, or a maximum of every 3 months.
— Delivering Better Oral Health, chapter 13, table 5 (1)
An earlier draft of this chapter glued the two together and presented the longer one as the evidence for the twice-yearly recommendation.
The citation is the Cochrane review of fluoride varnishes (2).
Following the citation
The citation is on point, which after Part II’s earlier chapters is worth saying plainly. Marinho and colleagues asked exactly the question the recommendation makes, in the right population, with caries as the outcome.
Twenty-two trials, published between 1975 and 2012, allocating 12,455 participants, of whom 9,595 contributed to the analyses. Not all were randomized: the review’s eligibility criteria are “randomised or quasi-randomised controlled trials,” and several included studies used quasi-random allocation. I called them twenty-two randomized trials in an earlier draft.
One population note that matters later: the review includes children and adolescents aged 16 or under. The adult recommendation in table 1g is therefore an extrapolation, and the guideline says so, rating it “moderate certainty evidence from children and adolescents.” The primary measure is the prevented fraction: the difference in caries increment between varnish and control, as a percentage of the control increment.
| Dentition | Prevented fraction | 95% CI | Trials | Certainty |
|---|---|---|---|---|
| Permanent teeth (D(M)FS) | 43% | 30% to 57% | 13 | moderate |
| Primary teeth (d(e/m)fs) | 37% | 24% to 51% | 10 | moderate |
A 43% reduction is a large effect. I am not going to rank it against the SMD in Chapter 8 or the odds ratio in Chapter 5, as an earlier draft did: a prevented fraction, a standardized mean difference and an odds ratio sit on three different scales, from three different populations and comparisons, and lining their numbers up produces something that looks quantitative and is not.
The review’s own conclusion contains the caveats, and they matter.
The review suggests a substantial caries-inhibiting effect of fluoride varnish in both permanent and primary teeth, however the quality of the evidence was assessed as moderate, as it included mainly high risk of bias studies, with considerable heterogeneity.
Risk of bias. The review states its included trials are mainly at high risk of bias.
And an upgrade, which I had left out. The route to moderate is not simply “randomized evidence, downgraded once.” The summary-of-findings footnotes record that the reviewers downgraded for heterogeneity and study limitations and then upgraded on the basis of “a consistent, large clinically important effect,” landing at moderate. That is worth flagging twice over: it is a fuller account than the one I gave, and upgrading is a move GRADE ordinarily reserves for observational evidence, so applying it to a randomized body is a judgment a reader should be able to see rather than one buried in a footnote.
Heterogeneity. Substantial and statistically confirmed (I² = 75%, P < 0.0001) for the permanent-teeth analysis. A pooled estimate of 43% across trials that disagree that much is a summary of a wide range, not a precise prediction for any one child.
Nothing explained the variation. This is the most interesting part and the least discussed. The review ran meta-regressions looking for what modifies the effect, and found no significant association with baseline caries severity, background fluoride exposure, whether a prophylaxis was done first, the concentration of fluoride, or the frequency of application. Nor with length of follow-up, use of a placebo versus no treatment, or individual versus cluster randomization.
That last one looks like the chapter’s punchline, and in an earlier draft I made it one: I wrote that no comparison of application frequencies supports the twice-yearly schedule, and that such a trial “has never been done.”
That is wrong, and the trials are inside this very review.
Weintraub and colleagues randomized 376 young children to counseling alone, counseling plus varnish once a year, or counseling plus varnish twice a year, and reported a dose-response effect on the number of applications actually received (p < 0.01) (3).
I then reported their odds ratios backwards, which is worth correcting carefully because the direction is the entire content of the result. The published comparisons take counseling alone as the exposure: children receiving counseling only had higher caries odds than those assigned annual varnish, OR 2.20 (95% CI 1.19 to 4.08), and than those assigned twice-yearly varnish, OR 3.77 (1.88 to 7.58). My earlier draft attached those numbers to the varnish arms, which reads as varnish causing more decay.
Turned round so that varnish is the exposure, by taking the reciprocal of the estimate and of each interval bound, the same results are approximately 0.45 (0.24 to 0.84) for annual varnish and 0.27 (0.13 to 0.53) for twice-yearly. Those are re-expressions of the published comparisons, not new estimates, and the reciprocal of an interval reverses which bound is which.
Sköld and colleagues ran a four-arm trial comparing twice, three times and eight times a year against no treatment.
So frequency has been randomized. What has not been done is a pooled direct estimate of once against twice: the review combined active arms for its main analysis, and its frequency meta-regression is a trial-level comparison of whole studies, underpowered by the review’s own warning that “all the meta-regressions have low power and the findings should not be interpreted as no effect.”
The honest version is therefore narrower than the one I wanted, and more useful. Twice a year is the schedule most of the evidence was generated under, 17 of the 22 trials used it, and at least one randomized trial points to twice being better than once. What nobody has produced is a precise estimate of how much better, or of whether three or four times would be better still.
Adverse effects. “There was little information concerning possible adverse effects or acceptability of treatment.” Twenty-two trials, and the harm side is close to blank.
The caveat that grows every year
The newest trial in this review was published in 2012, and the search closed on 13 May 2013. That is thirteen years ago as I write.
Why that matters more here than elsewhere: the benefit of a topical fluoride depends on how much fluoride the population is already getting. When most of these trials ran, fluoride toothpaste was less universal than it is now. The review looked for exactly this effect, and did not find background fluoride exposure to be a significant modifier, which is reassuring. But that null result comes from the same heterogeneous body of evidence, and the modifier analyses rest on subsets of seven to twelve trials rather than all twenty-two, with the I² of 75% belonging to the thirteen-trial permanent-surface meta-analysis. Underpowered trial-level regressions cannot confidently exclude effect modification by background fluoride exposure.
This is the same structural problem that turns up much more dramatically in Chapter 24, where splitting the water fluoridation evidence at 1975 changed the answer substantially.
PubMed, 21 August 2026, for an update to CD002279. None found: the 2013 version (.pub2) remains current. The review’s own search ran to 13 May 2013 across CENTRAL, MEDLINE, EMBASE, CINAHL, LILACS, BBO, ProQuest Dissertations and Web of Science Conference Proceedings, with no language restriction.
I have the full text of this review, supplied during writing, including the characteristics of included studies. Statements above about the meta-regression findings and the adverse-effects gap are from the review body.
Full record in appraisals/searches/ch10-fluoride-varnish.md.
Verdict
- Certainty of evidence
-
Moderate that fluoride varnish substantially reduces caries increment in children and adolescents. I agree with DBOH’s rating. The effect is large and consistent in direction across 22 trials, and downgraded for risk of bias and heterogeneity rather than for anything about relevance.
Separately, for the “2 times a year” schedule: low. Frequency has been randomized, in at least two included trials, and one of them found a dose-response on applications actually received. What the review does not provide is a pooled direct estimate of once against twice, because it combined active arms. So the schedule is not unevidenced, as I first wrote; it is imprecisely evidenced.
- Directness to the advice as worded
- Good on the intervention, poor on the schedule, and aging. Right population, right comparator, caries measured directly over years. But the newest trial is from 2012, in a population with different background fluoride exposure from today’s, and nothing in the review tests the recommended frequency.
- Is the strength label defensible?
-
Yes for applying varnish. This is the second-best-supported recommendation in the book after fluoride toothpaste, and a Strong label is right.
Arguable for “2 times a year” as a specific number. DBOH reports openly that most included studies used two applications a year, and separately cites manufacturer guidance; it does not claim the number came from a head-to-head comparison. Twice a year is the schedule the evidence was generated under and the one a randomized trial found better than once. What has not been shown is that it is optimal, and a reader would be better served if the recommendation said which of those two things it meant.
- What would change my mind
- An updated review including post-2012 trials, which would either confirm the effect in a modern fluoride-exposed population or show it shrinking, as happened with water fluoridation. And an analysis extracting the arm-level frequency contrasts that already sit inside Sköld and Weintraub, which would turn “twice is what was studied” into an estimate of how much twice beats once.
What this means for you
If a dentist offers fluoride varnish for your child, it is among the best-evidenced things in this book: a large effect, measured on actual tooth decay, across twenty-two randomized or quasi-randomized trials.
Two honest qualifications. Nobody has produced a precise estimate of how much better twice a year is than once, although the trials that randomized it point that way; and the evidence is now over a decade old, gathered in populations whose background fluoride exposure was probably lower than today’s. Neither is a reason to decline an application a clinician has judged appropriate. Both are reasons someone should be redoing this work.
One practical note this chapter owes you, since it is recommending a treatment. Varnish is not for everyone: the guideline lists contraindications including ulcerative gingivitis and stomatitis, and cautions around colophony allergy and severe asthma. That is a conversation with the person applying it, not something a book can settle.