Water fluoridation for the prevention of dental caries

Zipporah Iheozor-Ejiofor; Tanya Walsh; Sharon R Lewis; Philip Riley; Dwayne Boyers; Janet E Clarkson; Helen V Worthington; Anne-Marie Glenny; Lucy O'Malley

doi:10.1002/14651858.CD010856.pub3

. 2024 Oct 4;2024(10):CD010856. doi: 10.1002/14651858.CD010856.pub3

Water fluoridation for the prevention of dental caries

Zipporah Iheozor-Ejiofor ¹, Tanya Walsh ², Sharon R Lewis ², Philip Riley ², Dwayne Boyers ³, Janet E Clarkson ^2,⁴, Helen V Worthington ², Anne-Marie Glenny ^2,^✉, Lucy O'Malley ²

Editor: Cochrane Oral Health Group

PMCID: PMC11449566 PMID: 39362658

Abstract

Background

Dental caries is a major public health problem in most industrialised countries, affecting 60% to 90% of school children. Community water fluoridation (CWF) is currently practised in about 25 countries; health authorities consider it to be a key strategy for preventing dental caries. CWF is of interest to health professionals, policymakers and the public. This is an update of a Cochrane review first published in 2015, focusing on contemporary evidence about the effects of CWF on dental caries.

Objectives

To evaluate the effects of initiation or cessation of CWF programmes for the prevention of dental caries.

To evaluate the association of water fluoridation (artificial or natural) with dental fluorosis.

Search methods

We searched CENTRAL, MEDLINE, Embase and four other databases up to 16 August 2023. We also searched two clinical trials registers and conducted backward citation searches.

Selection criteria

We included populations of all ages.

For our first objective (effects of initiation or cessation of CWF programmes on dental caries), we included prospective controlled studies comparing populations receiving fluoridated water with those receiving non‐fluoridated or naturally low‐fluoridated water. To evaluate change in caries status, studies measured caries both within three years of a change in fluoridation status and at the end of study follow‐up.

For our second objective (association of water fluoridation with dental fluorosis), we included any study design, with concurrent control, comparing populations exposed to different water fluoride concentrations. In this update, we did not search for or include new evidence for this objective.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

For our first objective, we included the following outcomes as change from baseline: decayed, missing or filled teeth (‘dmft’ for primary and ‘DMFT’ for permanent teeth); decayed, missing or filled tooth surfaces ('dmfs' for primary and 'DMFS' for permanent teeth); proportion of caries‐free participants for both primary and permanent dentition; adverse events. We stratified the results of the meta‐analyses according to whether data were collected before or after the widespread use of fluoride toothpaste in 1975.

For our second objective, we included dental fluorosis (of aesthetic concern, or any level of fluorosis), and any other adverse events reported by the included studies.

Main results

We included 157 studies. All used non‐randomised designs. Given the inherent risks of bias in these designs, particularly related to management of confounding factors and blinding of outcome assessors, we downgraded the certainty of all evidence for these risks. We downgraded some evidence for imprecision, inconsistency or both. Evidence from older studies may not be applicable to contemporary societies, and we downgraded older evidence for indirectness.

Water fluoridation initiation (21 studies)

Based on contemporary evidence (after 1975), the initiation of CWF may lead to a slightly greater change in dmft 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 dmft of approximately one‐quarter of a tooth in favour of CWF; this effect estimate includes the possibility of benefit and no benefit. Contemporary evidence (after 1975) was also available for change in DMFT (4 studies, 2856 children) and change in DMFS (1 study, 343 children); we were very uncertain of these findings.

CWF may lead to a slightly greater change over time in the proportion of caries‐free children with primary dentition (MD ‐0.04, 95% CI ‐0.09 to 0.01; P = 0.12; 2 studies, 2908 children), and permanent dentition (MD ‐0.03, 95% CI ‐0.07 to 0.01; P = 0.14; 2 studies, 2348 children). These low‐certainty findings (a 4 percentage point difference and 3 percentage point difference for primary and permanent dentition, respectively) favoured CWF. These effect estimates include the possibility of benefit and no benefit. No contemporary data were available for adverse effects.

Because of very low‐certainty evidence, we were unsure of the size of effects of CWF when using older evidence (from 1975 or earlier) on all outcomes: change in dmft (5 studies, 5709 children), change in DMFT (3 studies, 5623 children), change in proportion of caries‐free children with primary dentition (5 studies, 6278 children) or permanent dentition (4 studies, 6219 children), or adverse effects (2 studies, 7800 children).

Only one study, conducted after 1975, reported disparities according to socioeconomic status, with no evidence that deprivation influenced the relationship between water exposure and caries status.

Water fluoridation cessation (1 study)

Because of very low‐certainty evidence, we could not determine if the cessation of CWF affected DMFS (1 study conducted after 1975; 2994 children). Data were not available for other review outcomes for this comparison.

Association of water fluoridation with dental fluorosis (135 studies)

The previous version of this review found low‐certainty evidence that fluoridated water may be associated with dental fluorosis. With a fluoride level of 0.7 parts per million (ppm), approximately 12% of participants had fluorosis of aesthetic concern (95% CI 8% to 17%; 40 studies, 59,630 participants), and approximately 40% had fluorosis of any level (95% CI 35% to 44%; 90 studies, 180,530 participants). Because of very low‐certainty evidence, we were unsure of other adverse effects (including skeletal fluorosis, bone fractures and skeletal maturity; 5 studies, incomplete participant numbers).

Authors' conclusions

Contemporary studies indicate that initiation of CWF may lead to a slightly greater reduction in dmft and may lead to a slightly greater increase in the proportion of caries‐free children, but with smaller effect sizes than pre‐1975 studies. There is insufficient evidence to determine the effect of cessation of CWF on caries and whether water fluoridation results in a change in disparities in caries according to socioeconomic status. We found no eligible studies that report caries outcomes in adults.

The implementation or cessation of CWF requires careful consideration of this current evidence, in the broader context of a population's oral health, diet and consumption of tap water, movement or migration, and the availability and uptake of other caries‐prevention strategies. Acceptability, cost‐effectiveness and feasibility of the implementation and monitoring of a CWF programme should also be taken into account.

Plain language summary

Does adding fluoride to water supplies prevent tooth decay?

Key messages

‐ Adding fluoride to water supplies may lead to slightly less tooth decay in children’s baby teeth.

‐ It may also lead to slightly more children being free of tooth decay.

‐ The benefits of fluoride in water supplies may be smaller than they were before the widespread addition of fluoride to toothpaste.

Tooth decay and the use of fluoride

Tooth decay is a worldwide problem affecting most adults and children. Untreated decay may cause pain and lead to teeth having to be removed.

Fluoride is a mineral which occurs naturally in water at different concentrations. It prevents tooth decay. Since 1975, fluoride has been an ingredient in most toothpastes. It is available in some mouth‐rinses, and dentists use treatments that contain fluoride. It is possible to add fluoride to the local water supply. In this case, everyone in a community will have access to fluoride.

If young children swallow too much fluoride while their permanent teeth are forming, marks may develop on those teeth – this is called dental fluorosis. This can be very mild, with barely noticeable white lines or streaks. Rarely, some fluorosis is more noticeable, and people can dislike how their teeth look.

What did we want to find out?

We wanted to find out if water with added fluoride in the local water supply is better than water without added fluoride at:

‐ reducing the number of teeth, or tooth surfaces, with signs of decay;

‐ increasing the number of people who have no tooth decay.

We also wanted to find out about unwanted effects.

What did we do?

We searched for studies comparing communities that had fluoride added to their water supplies with communities that had no additional fluoride in their water.

The last time we published this Cochrane review, we also searched for studies that reported dental fluorosis and the concentration of fluoride in the water. Because the association of fluoridated water with dental fluorosis is widely accepted, we did not update the evidence on this occasion.

What did we find?

We found 21 studies that assessed the effects of adding fluoride to a water supply. We also found one study that assessed the effects of stopping artificially‐added fluoride in a water supply. Studies only measured tooth decay in children.

In the last version of the review – not updated on this occasion – we found 135 studies that assessed the association of fluoridated water with dental fluorosis.

Main results

Studies conducted after 1975 showed that adding fluoride to water may lead to slightly less tooth decay in children’s baby teeth. We could not be sure whether adding fluoride to water reduced tooth decay in children’s permanent teeth or decay on the surfaces of permanent teeth.

Adding fluoride to water may slightly increase the number of children who have no tooth decay in either their baby teeth or permanent teeth. However, these results also included the possibility of little or no difference in tooth decay.

Studies conducted in 1975 or earlier showed a clear and important effect on prevention of tooth decay in children. However, due to the increased availability of fluoride in toothpaste since 1975, it is unlikely that we will see this effect in all populations today.

We were unsure whether there were any effects on tooth decay when fluoride is removed from a water supply.

We were unsure if fluoride reduces differences in tooth decay between richer and poorer people.

In the last version of the review, we found that adding fluoride to water supplies increases the number of people with dental fluorosis. If water contains 0.7 mg/L of fluoride, about 12% of people may have dental fluorosis that causes them to be bothered about how their teeth look, and about 40% of people may have dental fluorosis of any level. We were unsure whether fluoride in water leads to other unwanted effects.

What are the limitations of the evidence?

Our confidence in the evidence is limited because this review included studies in which communities were deliberately selected to have changes to fluoride levels in the water supply. Although a common study approach for this topic, it can mean that there are differences between communities that might affect the results. In addition, the findings in some studies were different from others, and some results included the possibility of benefit and no benefit.

Older studies were conducted before the widespread use of fluoride toothpaste and other improvements in tooth decay prevention. This meant we could not tell if these results were applicable to current times. However, they may still be relevant to countries in which tooth decay is very high and people don't have easy access to fluoride toothpaste and other prevention strategies.

How current is this evidence?

For the effects of water fluoridation on tooth decay, this review updates our previous review and the evidence is current to August 2023.

For the association of fluoridated water with dental fluorosis, the review evidence is current to February 2015.

Summary of findings

Summary of findings 1. The initiation of community water fluoridation programmes on the prevention of dental caries.

Initiation of water fluoridation compared with low‐/non‐fluoridated water for the prevention of dental caries
Population: people of all ages included in the review (although no studies on the effect of water fluoridation in adults met the inclusion criteria) Settings: community setting Intervention: initiation of water fluoridation Comparison: low‐/non‐fluoridated water
Outcomes	Impact of initiation of water fluoridation		No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Effect in area with low‐/non‐fluoridated water	Comparative effect; mean difference (95% CI)
Change in number of decayed, missing and filled teeth in the primary dentition (dmft) Scale from 0 to 20 (greater reduction = better)^a Follow‐up: range from 3 to 12 years	Contemporary evidence (after 1975) with lower burden of disease^b
	The change in the mean dmft from baseline to follow‐up in the control group ranged from 0.44 to 0.88.	In the areas with water fluoridation, there was a greater reduction (change) in mean dmft from baseline to follow‐up of 0.24 (‐0.03 to 0.52).	2908^c (2 NRSI)	Low^d ⊕⊕⊝⊝	The mean dmft at baseline in the non‐/low‐fluoridated areas ranged from 1.18 to 2.09. One study reported data according to disparities, and found no evidence that deprivation influences the relationship between water fluoridation and the severity of caries (as measured by dmft counts).
	Data from 1975 or earlier with higher burden of disease^b
	The change in the mean dmft from baseline to follow‐up in the control group ranged from 0.3 to 1.04.	In the areas with water fluoridation, there was a greater reduction (change) in mean dmft from baseline to follow‐up of 2.10 (1.71 to 2.49).	5709^c (5 NRSI)	Very low ⊕⊝⊝⊝^e	The mean dmft at baseline in the non‐/low‐fluoridated areas ranged from 4.76 to 8.1. We downgraded the certainty of the evidence for indirectness (i.e. inapplicability of the evidence to contemporary settings). We acknowledge that in some countries, caries levels remain high and access to fluoridated toothpaste and other caries prevention strategies may remain limited.
Change in number of decayed, missing and filled teeth in the permanent dentition(DMFT) Scale from 0 to 32 (greater reduction = better)^a Follow‐up: range from 4 to 11 years	Contemporary evidence (after 1975) with lower burden of disease^b
	The change in the mean DMFT from baseline to follow‐up in the control group ranged from 0.27 to 2.83; caries increments ranged from ‐0.4 to ‐4.85.	In the areas with water fluoridation, there was a greater reduction (change) in mean DMFT from baseline to follow‐up of 0.27 (‐0.11 to 0.66).	2856^c (4 NRSI)	Very low^f ⊕⊝⊝⊝	The mean DMFT at baseline in the non‐/low‐fluoridated areas ranged from 0.99 to 8.23, where reported. One study reported data according to disparities, and found no evidence that deprivation influences the relationship between water fluoridation and the severity of caries (as measured by DMFT counts).
	Data from 1975 or earlier with higher burden of disease^b
	The change in the mean DMFT from baseline to follow‐up in the control group ranged from ‐0.73 to 0.65.	In the areas with water fluoridation, there was a greater reduction (change) in mean DMFT from baseline to follow‐up of 1.00 (0.54 to 1.47).	5623^c (3 NRSI)	Very low ⊕⊝⊝⊝^g	The mean DMFT at baseline in the non‐/low‐fluoridated areas ranged from 3.01 to 4.03. We downgraded the certainty of the evidence for inapplicability to contemporary settings. We acknowledge that in some countries, caries levels remain high and access to fluoridated toothpaste and other caries prevention strategies may remain limited.
Change in number of decayed, missing and filled tooth surfaces in the primary dentition (dmfs)	‐	‐	‐	‐	There were no data for dmfs reported in the included studies conducted before or after 1975.
Change in number of decayed, missing and filled tooth surfaces in the permanent dentition (DMFS) Scale from 0 to 128 (lower values indicate reduction in caries) Follow‐up: 4 years	Contemporary evidence (after 1975) with lower burden of disease^b
	The mean DMFS increment was 9.19.	The mean DMFS increment was 2.46 lower (1.11 lower to 3.81 lower).	343 (1 NRSI)	Very low^h ⊕⊝⊝⊝	There were no data for DMFS reported in the included studies conducted in 1975 or earlier.
Change in proportion of caries‐free participants (primary teeth) Scale: 0 to 1 (greater increase = better)ⁱ Follow‐up: range 3 to 11 years	Contemporary evidence (after 1975) with lower burden of disease^b
	The change in the proportion of caries‐free children from baseline to follow‐up in the control group ranged from ‐0.19 to ‐0.11.	In the areas with water fluoridation, there was a greater increase (change) in the proportion of caries‐free children from baseline to follow‐up of ‐0.04 (‐0.09 to 0.01).	2908^c (2 NRSI)	Low^d ⊕⊕⊝⊝	The proportion of caries‐free children at baseline in the non‐/low‐fluoridated areas ranged from 0.49 to 0.68. One study reported data according to disparities, and found no evidence that deprivation influences the relationship between water fluoridation and the presence or absence of caries in the primary dentition.
	Data from 1975 or earlier with higher burden of disease^b
	The change in the proportion of caries‐free children from baseline to follow‐up in the control group ranged from ‐0.14 to 0.02.	In the areas with water fluoridation, there was a greater increase (change) in the proportion of caries‐free children from baseline to follow‐up of ‐0.17 (‐0.20 to ‐0.13).	6278^c (5 NRSI)	Very low ⊕⊝⊝⊝^e	The proportion of caries‐free children at baseline in the non‐/low‐fluoridated areas ranged from 0.08 to 0.20. We downgraded the certainty of the evidence for inapplicability to contemporary settings. We acknowledge that in some countries, caries levels remain high and access to fluoridated toothpaste and other caries prevention strategies may remain limited.
Change in proportion of caries‐free participants (permanent teeth) Scale: 0 to 1 (greater increase = better)ⁱ Follow‐up: range 3 to 11 years	Contemporary evidence (after 1975) with lower burden of disease^b
	The change in the proportion of caries‐free children from baseline to follow‐up in the control group ranged from ‐0.78 to ‐0.05.	In the areas with water fluoridation, there was a greater increase (change) in the proportion of caries‐free children from baseline to follow‐up of ‐0.03 (‐0.07 to 0.01).	2348^c (2 NRSI)	Low^d ⊕⊕⊝⊝	The proportion of caries‐free children at baseline in the non‐/low‐fluoridated areas was 0.62, where reported. One study reported data according to disparities, and found no evidence that deprivation influences the relationship between water fluoridation and the presence or absence of caries in the primary dentition.
	Data from 1975 or earlier with higher burden of disease^b
	The change in the proportion of caries‐free children from baseline to follow‐up in the control group ranged from ‐0.07 to 0.05.	In the areas with water fluoridation, there was a greater increase (change) in the proportion of caries‐free children from baseline to follow‐up of ‐0.06 (‐0.14 to 0.02).	6219^c (4 NRSI)	Very low ⊕⊝⊝⊝^g	The proportion of caries‐free children at baseline in the non‐/low‐fluoridated areas ranged from 0.05 to to 0.12. We downgraded the certainty of the evidence for inapplicability to contemporary settings. We acknowledge that in some countries, caries levels remain high and access to fluoridated toothpaste and other caries prevention strategies may remain limited.
Adverse effects (including dental fluorosis, skeletal fluorosis, hip fractures, cancer, congenital malformations, mortality)	Contemporary evidence (after 1975) with lower burden of disease^b
	‐		‐	‐	No studies reported data for this outcome.
	Data from 1975 or earlier with higher burden of disease^b
	In one study, there was a small increase in the number of children with mild fluorosis (0.12% increase). In another study, there were no cases of "unsightly mottling".		7800 participants (2 NRSI)	Very low ⊕⊝⊝⊝^j
CI: confidence interval; NRSI: non‐randomised studies of interventions
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Population: people of all ages included in the review (although no studies on the effect of water fluoridation in adults met the inclusion criteria) Settings: community setting Intervention: cessation of water fluoridation Comparison: fluoridated water
Outcomes	Impact of cessation of water fluoridation		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Effect in area with continuously fluoridated water	Comparative effect; mean difference (95% confidence interval)
Change in number of decayed, missing and filled teeth in the primary dentition (dmft)	No evidence to determine the effect of the cessation of water fluoridation on dmft
Change in number of decayed, missing and filled teeth in the permanent dentition(DMFT)	No evidence to determine the effect of the cessation of water fluoridation on DMFT
Change in number of decayed, missing and filled tooth surfaces in the primary dentition (dmfs)	No evidence to determine the effect of the cessation of water fluoridation on dmfs
Change in number of decayed, missing and filled tooth surfaces in the permanent dentition (DMFS) Scale from: 0 to 128 (lower = better) Follow‐up: 3 years			2994^a (1 observational study)	Very low^b ⊕⊝⊝⊝	Insufficient evidence to determine the effect of the cessation of water fluoridation on caries
Change in proportion of caries‐free participants (primary teeth)	No evidence to determine the effect of the cessation of water fluoridation on proportion of caries‐free participants (primary teeth)
Change in proportion of caries‐free participants (permanent teeth)	No evidence to determine the effect of the cessation of water fluoridation on proportion of caries‐free participants (permanent teeth)
Adverse effects	No evidence to determine whether cessation of a water fluoridation programme is associated with any harms
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

The association of water fluoridation (artificial or natural) with dental fluorosis
Population: people of all ages Settings: community settings Intervention: water with any concentration of fluoride from either natural sources or artificially added Comparison: n/a
Outcomes	Impact	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Dental fluorosis of aesthetic concern^a (measured by Dean's Index, TFI, TSIF)	For a fluoride level of 0.7 ppm, the percentage of participants with dental fluorosis of aesthetic concern was estimated to be 12% (95% CI 8% to 17%). Controlling for study effects, we would expect the odds of dental fluorosis of aesthetic concern to increase by a factor of 2.90 (95% CI 2.05 to 4.10) for each one unit increase in fluoride level (1 ppm F).	59,630 (40 NRSI)	⊕⊕⊝⊝ Low^b
Dental fluorosis of any level^a ‐ including dental fluorosis that can only be detected under normal clinical conditions and other enamel defects (measured by Dean's Index, TFI, TSIF)	For a fluoride level of 0.7 ppm, the percentage of participants with any level of dental fluorosis was estimated to be 40% (95% CI 35% to 44%). Controlling for study effects, we would expect the odds of dental fluorosis of any level to increase by a factor of 3.60 (95% CI 2.86 to 4.53) for each one unit increase in fluoride level (1 ppm F)	180,530 (90 NRSI)	⊕⊕⊝⊝ Low^b
Other adverse effects	‐	596,410^c (5 NRSI)	⊕⊝⊝⊝ Very low^d	Only a small number of studies reported other types of adverse effects (including skeletal fluorosis, bone fracture and skeletal maturity). We did not analyse the data, and we were unable to draw conclusions from data reported by individual study authors.
CI: confidence interval; n/a: not applicable (because the studies assessed the association of an effect, a comparison group was not necessary for this objective); NRSI: non‐randomised studies of interventions; ppm: parts per million; TFI: Thylstrup and Fejerskov index; TSIF: Tooth Surface Index of Fluorosis
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Study ID	Preliminary questions	Risk of bias due to confounding	Risk of bias in classification of interventions	Risk of bias in selection of participants into the study (or into the analysis)	Risk of bias due to deviations from intended interventions	Risk of bias due to missing data	Risk of bias arising from measurement of the outcome	Risk of bias in selection of the reported result	Overall risk of bias^a
Adriasola 1959		Moderate	Low	Low	Low	Serious	Moderate	Low	SERIOUS
Arnold 1956	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Ast 1951	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Backer‐Dirks 1961		Serious	Low	Low	Low	Serious	Moderate	Low	SERIOUS
Beal 1971	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Beal 1981		Serious	Low	Low	Low	Serious	Moderate	Low	SERIOUS
Blinkhorn 2015		Low	Low	Low	Low	Serious	Moderate	Low	SERIOUS
Brown 1965	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
DHSS England 1969		Serious	Low	Low	Low	Serious	Moderate	Low	SERIOUS
DHSS Scotland 1969		Serious	Low	Low	Low	Serious	Moderate	Low	SERIOUS
DHSS Wales 1969		Serious	Low	Low	Low	Moderate	Moderate	Low	SERIOUS
Goodwin 2022		Low	Low	Low	Low	Low	Moderate	Low	MODERATE
Gray 2001	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Guo 1984	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Hardwick 1982		Moderate	Low	Low	Low	Serious	Low	Low	SERIOUS
Holdcroft 1999^b		‐	‐	‐	‐	Serious	‐	‐	SERIOUS
Kim 2019	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Kunzel 1997	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Loh 1996	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Maupome 2001		Moderate	Low	Moderate	Low	Serious	Moderate	Low	SERIOUS
Pot 1974	No further assessment	‐	‐	‐	‐	‐	‐	‐	CRITICAL
Tessier 1987		Moderate	Low	Low	Low	Serious	Moderate	Low	SERIOUS

Study ID	Date	Age (years)	Fluoridated area						Non‐/low‐fluoridated area
			Baseline (before/at initiation)			Follow‐up			Baseline			Follow‐up
			Mean	SD	N	Mean	SD	N	Mean	SD	N	Mean	SD	N
Blinkhorn 2015	> 1975	5 to 7	2.02	3.13	781	0.72	1.63	844	2.09	2.91	523	1.21	2.27	612
Blinkhorn 2015	> 1975	5 to 7	Mean (SD) change in dmft: 1.3 (3.49), N = 813^a						Mean (SD) change in dmft: 0.88 (3.77), N = 568^a
Goodwin 2022	> 1975	5	1.06	2.16	699	0.49	1.40	609	1.18	2.41	911	0.74	1.96	835
Goodwin 2022	> 1975	5	Mean (SD) change in dmft: 0.57 (2.61), N = 654^a						Mean (SD) change in dmft: 0.44 (3.17), N = 873^a
Adriasola 1959	≤ 1975	5	8.9	5.03	186	6.4	4.18	340	8.1	4.77	174	7.8	4.67	140
Adriasola 1959	≤ 1975	5	Mean (SD) change in dmft: 2.5 (7.04), N = 263^a						Mean (SD) change in dmft: 0.3 (6.72), N = 157^a
Arnold 1956^b	≤ 1975	4	4.19	3.30	323	2.13	2.26	168	5.05	3.66	20	4.46	3.42	63
		5	5.37	3.79	1633	2.27	2.34	853	6.82	4.33	402	5.25	3.74	351
		6	6.43	4.19	1789	2.98	2.73	750	7.17	4.46	462	5.67	3.91	294
		7	6.29	4.14	1806	4.03	3.23	423	6.66	4.28	408	5.77	3.95	223
		8	5.78	3.95	1647	4.12	3.27	470	6.06	4.06	376	5.32	3.77	275
		4 to 8	Mean (SD) change in dmft: 2.75 (4.99), N = 4931^a						Mean (SD) change in dmft: 1.18 (5.8), N = 1437^a
Beal 1971	≤ 1975	5	4.91	4.86	182	2.45	3.24	182	4.97	4.12	217	5.09	4.84	229
Beal 1971	≤ 1975	5	Mean (SD) change in dmft: 2.46 (5.8), N = 182^a						Mean (SD) change in dmft: ‐0.12 (6.27), N = 223^a
Beal 1981	≤ 1975	5	4.29	3.50	196	1.8	2.48	170	4.28	3.58	205	3.49	3.62	180
		8	5	2.89	189	3.42	2.84	167	5.36	3.06	163	4.97	3.00	186
		5 or 8	Mean (SD) change in dmft: 2.02 (4.18), N = 361^a						Mean (SD) change in dmft: 0.57 (4.6), N = 367^a
DHSS England 1969^b	≤ 1975	3	2.7	2.58	43	0.6	1.11	133	1.4	1.79	44	1.2	1.64	144
		4	3.6	3.03	66	1.3	1.71	131	2.6	2.53	47	1.8	2.06	162
		5	5.4	3.80	148	1.6	1.92	111	5	3.64	110	2.8	2.63	119
		6	5.7	3.92	182	2.5	2.47	130	5.4	3.80	127	4.1	3.26	107
		7	6.4	4.18	192	2.7	2.58	172	6	4.03	121	4.3	3.35	133
		3 to 7	Mean (SD) change in dmft: 3.09 (4.3), N = 654^a						Mean (SD) change in dmft: 1.04 (4.22), N = 557^a
DHSS Scotland 1969^b	≤ 1975	3	4.87	3.6	97	1.88	2.11	135	5.2	3.72	107	4.45	3.44	130
		4	7.12	4.43	112	2.97	2.71	171	7.16	4.47	77	6.86	4.35	132
		3 to 4	Mean (SD) change in dmft: 3.49 (4.92), N = 258^a						Mean (SD) change in dmft: 0.59 (5.64), N = 223^a
DHSS Wales 1969^b,c	≤ 1975	3	3.9	3.17	310	1.4	1.79	171	4	3.21	146	3.3	2.89	105
		4	5.54	3.86	413	2.6	2.53	267	5.8	3.96	210	4.8	3.56	122
		5	5.5	3.84	556	2.9	2.69	284	5.5	3.84	256	4.8	3.56	138
		6	6.3	4.15	603	3.1	2.79	310	6.2	4.11	331	5.9	4.00	133
		7	6.85	4.35	640	3.65	3.05	266	7.3	4.50	346	6.8	4.33	130
		3 to 7	Mean (SD) change in dmft: 2.87 (4.68), N = 1910^a						Mean (SD) change in dmft: 0.64 (5.54), N = 959^a
Guo 1984	≤ 1975	3	3	3.4	202	2.6	3.3	79	1.3	3.2	205	3.7	3.9	128
		4	4.6	4	354	4.5	4.7	164	5.6	4.6	246	7.1	4.6	164
		5	6.5	4.4	589	5.5	4.3	345	6.4	4.2	218	8.5	4.6	387
		6	6.7	4.4	695	6.2	4.8	297	5.8	4.2	309	9	4.3	354
		7	5.5	3.7	399	5.6	3.7	240	5.4	3.7	335	7.9	3.6	352
		8	4.2	3	392	4.4	2.9	279	3.5	2.7	343	6	3.1	350
		3 to 8	Mean (SD) change in dmft: 0.23 (5.39), N = 2018^a						Mean (SD) change in dmft: ‐2.47 (5.35), N = 1696^a
Kunzel 1997^b,d	≤ 1975	5	2.4	2.42	688	1.4	1.79	1306	3.3	2.89	172	2.9	2.68	597
		8	4.9	3.60	2438	2.8	2.63	3020	4.9	3.60	777	4.9	3.60	1078
		5 to 8	Mean (SD) change in dmft: 2.1 (5.01), N = 3726^a						Mean (SD) change in dmft: 0.13 (5.0), N = 1313^a

Fluoride exposure (ppm)	Probability of dental fluorosis of aesthetic concern (95% CI)
0.1	0.08 (0.05 to 0.12)
0.2	0.09 (0.06 to 0.13)
0.4	0.10 (0.06 to 0.15)
0.7	0.12 (0.08 to 0.17)
1	0.15 (0.11 to 0.21)
1.2	0.18 (0.13 to 0.24)
2	0.31 (0.23 to 0.40)
4	0.59 (0.46 to 0.71)

Fluoride exposure (ppm)	Probability of any dental fluorosis (95% CI)
0.1	0.28 (0.23 to 0.33)
0.2	0.30 (0.25 to 0.34)
0.4	0.33 (0.28 to 0.38)
0.7	0.40 (0.35 to 0.44)
1	0.47 (0.42 to 0.52)
1.2	0.52 (0.47 to 0.56)
2	0.68 (0.62 to 0.73)
4	0.83 (0.77 to 0.88)

Study ID	Type of adverse effect	Age (years)	Fluoride level (ppm)	Assigned fluoride level (ppm)	Number of participants	Proportion of participants with outcome
Chen 1993	Skeletal fluorosis	16 to 65	5.5	5.5	28	82.1
			3.1	3.1	114	71.1
			0.4	0.4	50	46
			3.1	3.1	50	86
Wang 2012^a	Skeletal fluorosis	≥16	2.2	2.2	406,298	10.8
Wang 2012^a	Skeletal fluorosis	≥16	0.5	0.5	188,400	4.8
Wenzel 1982^b	Skeletal maturity	12 to 14	2.4	2.4	122	0.59 (0.1)^c
Wenzel 1982^b	Skeletal maturity	12 to 14	< 0.2	0.1	113	0.59 (0.09)^c
Alarcon‐Herrera 2001	Bone fracture	6 to 12	< 1.5	0.75	97	5.2
			1.51 to 4.99	3.25	112	8.9
			5 to 8.49	6.75	38	2.6
			8.5 to 11.99	10.25	27	11.1
			12 to 16	14	59	8.5
		13 to 60	< 1.5	0.75	192	3.1
			1.51 to 4.99	3.25	330	7.9
			5 to 8.49	6.75	146	8.9
			8.5 to 11.99	10.25	138	7.2
			12 to 16	14	96	6.3
Jolly 1971^b	Skeletal fluorosis	Not stated	0.7	0.7	Not stated	3.6
			1.4	1.4	Not stated	2.4
			2.4	2.4	Not stated	17
			2.4	2.4	Not stated	23
			2.5	2.5	Not stated	33
			3	3	Not stated	19.6
			3	3	Not stated	42.2
			3.3	3.3	Not stated	10
			3.3	3.3	Not stated	45
			3.6	3.6	Not stated	33.1
			4.3	4.3	Not stated	19.4
			5	5	Not stated	60
			5.1	5.1	Not stated	44.5
			5.5	5.5	Not stated	31.3
			7	7	Not stated	47.4
			8.5	8.5	Not stated	58.9
			9.4	9.4	Not stated	70.1

World Health Organization (WHO) region	Prevalence of caries	Percentage change in prevalence
World Health Organization (WHO) region	2019	1990 to 2019
African Region	28.50%	‐1.66%
Eastern Mediterranean Region	32.25%	‐0.27%
European Region	33.63%	‐3.91%
Region of the Americas	28.24%	‐0.05%
South‐East Asia Region	28.69%	+0.67%
Western Pacific Region	25.41%	‐6.50%
GLOBAL	28.70%	‐2.59%

Date	Event	Description
4 October 2024	New citation required and conclusions have changed	The conclusions of the review reflect contemporary evidence of greater certainty.
4 October 2024	New search has been performed	Update of objective 1: To evaluate the initiation or cessation of community water fluoridation programmes for the prevention of dental caries. Two new studies added (average number of particpants used in analysis 4193); risk of bias updated to reflect advances in methods (ROBINS‐I used); single‐time point studies provided to provide context Objective 2 was not updated: To evaluate the association of water fluoridation (artificial or natural) with dental fluorosis. Two additional review authors (PR and SL) contributed to this review update. Four review authors (RM, RA, VW, PT) did not contribute to this review update (see Acknowledgements).
7 September 2015	Amended	Plain Language Summary amended for simplification.
19 June 2015	Amended	Minor edit to Plain Language Summary for clarification. Missing referee name added to Acknowledgements.
2 February 2015	Amended	Background updated to justify the need for the review. Change to risk of bias domains, incorporating an item on 'sampling' Change to the handling of missing data; imputation of missing standard deviations for DMFT and dmft data

Judgement	Interpretation
Low risk of bias	There is little or no concern about bias with regard to this domain.
Moderate risk of bias	There is some concern about bias with regard to this domain, although it is not clear that there is an important risk of bias.
Serious risk of bias	The study has some important problems in this domain: characteristics of the study give rise to a serious risk of bias.
Critical risk of bias	The study is very problematic in this domain: characteristics of the study give rise to a critical risk of bias, such that the result should generally be excluded from evidence synthesis.

Study ID	Date	Age (years)	Fluoridated area				Non‐/low‐fluoridated area
			Baseline (before/at initiation)		Follow‐up		Baseline		Follow‐up
			n	N	n	N	n	N	n	N
Blinkhorn 2015	> 1975	10 to 12	525	777	486	642	272	436	307	455
Goodwin 2022	> 1975	11	N/A^a	N/A^a	461	570	N/A^a	N/A^a	486	622
Guo 1984	> 1975	5	575	589	338	345	214	218	358	387
		6	616	695	266	297	284	309	249	354
		7	305	399	189	240	272	335	162	352
		8	278	392	204	279	273	343	104	350
		9	242	388	167	275	195	310	98	352
		10	215	346	161	310	199	323	84	436
		11	213	330	133	307	245	451	65	365
		12	240	468	90	208	475	841	91	493
		13	227	469	88	232	434	801	77	504
		14	161	322	69	221	455	795	73	490
		15	78	164	11	38	66	121	11	63
Adriasola 1959^b	≤ 1975	12	7	292	8	419	3	197	9	211
Beal 1981	≤ 1975	8	77	189	115	167	56	163	82	186
Beal 1981	≤ 1975	12	51	192	41	189	13	188	14	197
Brown 1965^c	≤ 1975	9 to 11	34	595	220	502	35	571	42	521
Brown 1965^c	≤ 1975	12 to 14	7	593	94	503	3	486	11	485
DHSS England 1969	≤ 1975	8	40	199	50	95	33	148	29	79
		9	25	227	57	135	20	166	20	95
		10	13	134	36	115	14	160	10	80
		11	12	145	12	200	3	126	12	122
		12	3	111	20	134	0	51	4	99
		13	3	91	9	132	2	52	8	127
		14	0	70	4	90	2	36	9	180
DHSS Wales 1969	≤ 1975	8	143	607	112	283	88	351	26	125
		9	73	553	78	260	49	325	15	134
		10	63	502	44	241	25	308	8	133
		11	30	278	15	126	35	270	0	42
		12	15	186	10	108	27	265	2	108
		13	7	178	0	93	14	274	1	105
		14	8	158	3	93	15	243	1	96
Kunzel 1997	≤ 1975	8	1021	2419	2147	3016	334	777	333	1076
		12	120	1626	801	2426	42	563	50	925
		15	118	1995	249	1897	27	744	18	756

Study ID	Age	Group	Measure	Socioeconomic status	Baseline				Final follow‐up
Study ID	Age	Group	Measure	Socioeconomic status	F level	N	% caries‐free	dmft (SD)	F level	N	% caries‐free	dmft (SD)
Beal 1971^a	5	Balsall Heath	Descriptive	Poor area	Low	115	9	5.16 (0.44)	1	132	48	1.94 (0.22)
		Northfield		Industrial area	Low	182	29	4.91 (0.36)	1	182	41	2.45 (0.24)
		Dudley		Industrial area	< 0.1	217	16	4.97 (0.28)	< 0.1	229	24	5.09 (0.32)
Gray 2000^b	5	Southeast Staffordshire	Jarman 1984 score	‐23.09	Low	3435	66	1.21 (0.59)	1	3120	75	0.64 (1.46)
		Sandwell		18.1	Low	3950	51	1.93 (2.88)	1	3598	69	0.83 (1.68)
		Walsall		1.67	Low	3120	54	1.85 (2.31)	1	363	67	0.94 (1.77)
		Dudley		‐13.68	Low	3657	58	1.6 (2.54)	1	3474	73	0.78 (1.75)
		North Birmingham		21.57	Low	1965	72	0.88 (1.97)	1	1904	74	0.71 (1.65)
		North Staffordshire		‐3.59	Low	464	47	2.24 (3.04)	Low	1947	59	1.49 (2.46)
		Herefordshire		‐13.01	Low	406	57	1.61 (2.55)	Low	305	50	1.79 (2.68)
		Shropshire		‐12.34	Low	366	61	1.29 (2.22)	Low	311	60	1.33 (2.33)
		Kidderminster		‐13.13	Low	904	58	1.74 (2.81)	Low	1053	61	1.4 (2.52)
Holdcroft 1999^b	Not stated	North Birmingham	Jarman 1984 score	‐7.85	Not stated	Not stated		2.18	High	Not stated		0.68
		Sandwell		15.03	Not stated	Not stated		2.55	High	Not stated		1.13
		North Staffordshire		‐4.07	Not stated	Not stated		2.24	Not stated	Not stated		1.48
		Shropshire		‐11.73	Not stated	Not stated		1.76	Not stated	Not stated		1.29
		Herefordshire		‐11.97	Not stated	Not stated		2.56	Not stated	Not stated		1.53

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Change in the number of decayed, missing or filled primary teeth (dmft)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 Studies conducted after 1975	2	2908	Mean Difference (IV, Random, 95% CI)	0.24 [‐0.03, 0.52]
1.1.2 Studies conducted in 1975 or earlier	5	5709	Mean Difference (IV, Random, 95% CI)	2.10 [1.71, 2.49]
1.2 Change in the number of decayed, missing or filled permanent teeth (DMFT)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.2.1 Studies conducted after 1975	4	2856	Mean Difference (IV, Random, 95% CI)	0.27 [‐0.11, 0.66]
1.2.2 Studies conducted in 1975 or earlier	3	5623	Mean Difference (IV, Random, 95% CI)	1.00 [0.54, 1.47]
1.3 Change in the number of decayed, missing or filled permanent surfaces (DMFS)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
1.4 Change in the proportion of caries‐free participants (primary teeth)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 Studies conducted after 1975	2	2908	Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.09, 0.01]
1.4.2 Studies conducted in 1975 or earlier	5	6278	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.20, ‐0.13]
1.5 Change in the proportion of caries‐free participants (permanent teeth)	6		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.5.1 Studies conducted after 1975	2	2348	Mean Difference (IV, Random, 95% CI)	‐0.03 [‐0.07, 0.01]
1.5.2 Studies conducted in 1975 or earlier	4	6219	Mean Difference (IV, Random, 95% CI)	‐0.06 [‐0.14, 0.02]
1.6 Sensitivity analysis ‐ all included studies: change in the number of decayed, missing or filled primary teeth (dmft)	11		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.6.1 Studies conducted after 1975	3	6622	Mean Difference (IV, Random, 95% CI)	1.08 [‐0.53, 2.70]
1.6.2 Studies conducted in 1975 or earlier	8	17520	Mean Difference (IV, Random, 95% CI)	1.91 [1.60, 2.23]
1.7 Sensitivity analysis ‐ all included studies: change in the number of decayed, missing or filled permanent teeth (DMFT)	12		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.7.1 Studies conducted after 1975	6	12906	Mean Difference (IV, Random, 95% CI)	0.53 [0.00, 1.06]
1.7.2 Studies conducted in 1975 or earlier	6	30334	Mean Difference (IV, Random, 95% CI)	1.35 [0.77, 1.94]
1.8 Sensitivity analysis ‐ all included studies: change in the proportion of caries‐free participants (primary teeth)	12		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.8.1 Studies conducted after 1975	4	9608	Mean Difference (IV, Random, 95% CI)	‐0.10 [‐0.19, ‐0.01]
1.8.2 Studies conducted in 1975 or earlier	8	12383	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.19, ‐0.15]
1.9 Sensitivity analysis ‐ all included studies: change in the proportion of caries‐free participants (permanent teeth)	9		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.9.1 Studies conducted after 1975	3	10502	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.33, 0.09]
1.9.2 Studies conducted in 1975 or earlier	6	17459	Mean Difference (IV, Random, 95% CI)	‐0.13 [‐0.24, ‐0.03]
1.10 Sensitivity analysis ‐ change in analytical approach: change in the number of decayed, missing or filled primary teeth (dmft)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.10.1 Studies conducted after 1975	2	2825	Mean Difference (IV, Random, 95% CI)	0.28 [0.12, 0.43]
1.11 Sensitivity analysis ‐ excluding studies with imputed standard deviations: change in the number of decayed, missing or filled primary teeth (dmft)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.11.1 Studies conducted after 1975	2	2908	Mean Difference (IV, Random, 95% CI)	0.24 [‐0.03, 0.52]
1.11.2 Studies conducted in 1975 or earlier	2	1148	Mean Difference (IV, Random, 95% CI)	1.59 [1.01, 2.16]
1.12 Sensitivity analysis ‐ excluding studies with imputed standard deviations: change in the number of decayed, missing or filled permanent teeth (DMFT)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.12.1 Studies conducted after 1975	2	1535	Mean Difference (IV, Random, 95% CI)	0.53 [‐0.45, 1.51]
1.12.2 Studies conducted in 1975 or earlier	1	736	Mean Difference (IV, Random, 95% CI)	0.62 [0.25, 0.99]

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: India Geographic location: Davangere‐Nallur, Naganur, Doddabathi, Kundawada and Holesirigere Year of study: not stated Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: school children aged 12 to 15 years; lifetime residency Exclusion criteria: absence on the day of the survey Other sources of fluoride: not stated SES: socioeconomic position was similar in all villages Ethnicity: not stated Residential history: lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 0.43 ppm Group 2: 0.72 ppm Group 3: 1.1 ppm Group 4: 1.22 ppm Group 5: 3.41 ppm
Outcomes	Dental fluorosis (Dean's Index) Age at assessment: 12 to 15 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	5 villages were selected out of a possible 90. There was insufficient detail reported in order to determine how selection took place
Confounding	High risk	Did not account for use of other fluoride sources
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	Low risk	No other apparent bias

Judgement	Interpretation	How reached
Low risk of bias except for concerns about uncontrolled confounding	There is the possibility of uncontrolled confounding that has not been controlled for (given the observational nature of the study), but otherwise little or no concern about bias in the result.	Low risk of bias except for concerns about uncontrolled confounding in Domain 1, and low risk of bias in all other domains.
Moderate risk of bias	There is some concern about bias in the result, although it is not clear that there is an important risk of bias.	At least one domain is at moderate risk of bias, but no domains are at serious or critical risk of bias.
Serious risk of bias	The study has some important problems: characteristics of the study give rise to a serious risk of bias in the result.	At least one domain is at serious risk of bias, but no domains are at critical risk of bias. OR Several domains are at moderate risk of bias, leading to an additive judgement of serious risk of bias.
Critical risk of bias	The study is very problematic: characteristics of the study give rise to a critical risk of bias in the result, such that the result should generally be excluded from evidence synthesis.	At least one domain is at critical risk of bias. OR Several domains are at serious risk of bias, leading to an additive judgement of critical risk of bias.

Including studies assessed as having a critical risk of bias
Outcome	Year of study	Effect estimate in primary analysis^a	Effect estimate in sensitivity analysis^a	Interpretation
Change in number of dmft	After 1975	MD 0.24, 95% CI ‐0.03 to 0.52; I² = 26%; 2 studies, 2908 participants; Analysis 1.1	MD 1.08, 95% CI ‐0.53 to 2.70; I² = 99%; 3 studies, 6622 participants; Analysis 1.6	The sensitivity analysis includes a larger effect size than the primary analysis, and has higher levels of statistical heterogeneity.
Change in number of dmft	1975 or earlier	MD 2.10, 95% CI 1.71 to 2.49; I² = 44%; 5 studies, 5709 participants; Analysis 1.1	MD 1.91, 95% CI 1.60 to 2.23; I² = 63%; 8 studies, 17,520 participants; Analysis 1.6	The sensitivity analysis includes a smaller effect size than the primary analysis.
Change in number of DMFT	After 1975	MD 0.27, 95% CI ‐0.11 to 0.66; I² = 83%; 4 studies, 2856 participants; Analysis 1.2	MD 0.53, 95% CI 0.00 to 1.06; I² = 98%; 6 studies, 12,906 participants; Analysis 1.7	The sensitivity analysis includes a larger effect size than the primary analysis.^b
Change in number of DMFT	1975 or earlier	MD 1.00, 95% CI 0.54 to 1.47; I² = 80%; 3 studies, 5623 participants; Analysis 1.2	MD 1.35, 95% CI 0.77 to 1.94; I² = 97%; 6 studies, 30,334 participants; Analysis 1.7	The sensitivity analysis includes a larger effect size than the primary analysis.
Change in the proportion of caries‐free participants (primary dentition)	After 1975	MD ‐0.04, 95% CI ‐0.09 to 0.01; I² = 0%; 2 studies, 2908 participants; Analysis 1.4	MD ‐0.10, 95% CI ‐0.19 to ‐0.01; I² = 90%; 4 studies, 9608 participants; Analysis 1.8	The sensitivity analysis includes a larger effect size than the primary analysis.
	1975 or earlier	MD ‐0.17, 95% CI ‐0.20 to ‐0.13; I² = 13%; 5 studies, 6278 participants; Analysis 1.4	MD ‐0.17, 95% CI ‐0.19 to ‐0.15; I² = 0%; 8 studies, 12,383 participants; Analysis 1.8	The sensitivity analysis has a very similar effect to the primary analysis.
Change in the proportion of caries‐free participants (permanent dentition)	After 1975	MD ‐0.03, 95% CI ‐0.07 to 0.01; I² = 0%; 2 studies, 2368 participants; Analysis 1.5	MD ‐0.12, 95% CI ‐0.33 to 0.09; I² = 98%; 3 studies, 10,502 participants, Analysis 1.9	The sensitivity analysis includes a larger effect size than the primary analysis.^c
	1975 or earlier	MD ‐0.06, 95% CI ‐0.14 to 0.02; I² = 93%; 4 studies, 6278 participants; Analysis 1.5	MD ‐0.13, 95% CI ‐0.24 to ‐0.03; I² = 98%; 6 studies, 17,459 participants; Analysis 1.9	The sensitivity analysis includes a larger effect size than the primary analysis.^d
Change of analytical approach^d
Outcome	Year of studies	Effect estimate in primary analysis	Effect estimate in sensitivity analysis	Interpretation
Change in number of dmft	After 1975	MD 0.24, 95% CI ‐0.03 to 0.52; I² = 26%; 2 studies, 2908 participants; Analysis 1.1	MD 0.28, 95% CI 0.12 to 0.43; I² = 0%; 2 studies, 2825 participants; Analysis 1.10	Using the caries increment from the longitudinal analysis^e resulted in a very similar pooled effect estimate to the primary analysis. However, imprecision is no longer a concern using the longitudinal analysis.
Excluding studies in which missing standard deviations were imputed
Outcome	Year of studies	Effect estimate in primary analysis	Effect estimate in sensitivity analysis	Interpretation
Change in number of dmft	1975 or earlier	MD 2.10, 95% CI 1.71 to 2.49; I² = 44%; 5 studies, 5709 participants; Analysis 1.1	MD 1.59, 95% CI 1.01 to 2.16; I² = 0%; 2 studies, 1148 participants; Analysis 1.11	The sensitivity analysis includes a smaller effect size than the primary analysis.
Change in number of DMFT	After 1975	MD 0.27, 95% CI ‐0.11 to 0.66; I² = 83%; 4 studies, 2856 participants; Analysis 1.2	MD 0.53, 95% CI ‐0.45 to 1.51; I² = 89%; 2 studies, 1535 participants; Analysis 1.12	The sensitivity analysis includes a larger effect size than the primary analysis.
Change in number of DMFT	1975 or earlier	MD 1.00, 95% CI 0.54 to 1.47; I² = 80%; 3 studies, 5623 participants; Analysis 1.2	MD 0.62, 95% CI 0.25 to 0.99; 1 study, 736 participants; Analysis 1.12	The sensitivity analysis includes a smaller effect size than the primary analysis.

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: USA Geographic location: Warren County, Georgia Year of study: not stated Year of change in fluoridation status: not stated Study design: cross‐sectional
Participants	Inclusion criteria: children attending sole elementary and middle schools in study area Exclusion criteria: children whose homes were served with well‐water Other sources of fluoride: parents completed questionnaire regarding dentifrice use, home water source and current use of systemic fluoride supplements; all children received school water fluoridated at 0.5 ppm Other sources of fluoride: not stated SES: not stated Ethnicity: not stated Residential history: not considered Other confounding factors: not stated
Interventions	Group 1: 0.5 to 1.2 ppm (both natural and artifical fluoridation) Group 2: < 0.1 ppm (natural fluoridation)
Outcomes	Dental fluorosis (Dean's Index); caries data collected but not presented in this review due to study design Age at assessment: 8 to 10 years, and 11 to13 years
Funding	NIDR Grant DE‐06113
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	Participants were children attending the sole elementary and middle/high schools in Warren county. There was insufficient detail reported in order to determine how selection took place
Confounding	High risk	SES was not accounted for
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data for over 80% of participants were reported
Selective reporting (reporting bias)	High risk	Outcome of interest reported. However, data were not presented clearly enough to be considered reliable
Other bias	High risk	Exposure to fluoride water could not be controlled for. Some children had fluoride water at school across groups. Some had non‐fluoridated well‐water at home

*Study characteristics*
Methods	CARIES STUDY Country of study: Chile Geographic location: Group 1: Curico (F); Group 2: San Fernando (non‐F). Total population sizes in each location not stated Year study started: 1953 Year study ended: 1956 Year of change in fluoridation status: 1953 Study design: CBA. A different sample of children was assessed at baseline and at end of study
Participants	Inclusion criteria: children aged 3 to 15 years; children from 2 primary schools in the study areas Exclusion criteria: none stated Sample size at baseline: Group 1: 1279 children; Group 2: 748 children Sample size at final assessment: Group 1: 3060 children; Group 2: 1680 children SES: based on author's knowledge of the demographics, culture and social economy of the intervention and control areas, it was assumed that the study areas were comparable Co‐interventions: not stated Ethnicity: not stated Gender: distribution was similar across groups Residential history: not stated
Interventions	Initiation of water fluoridation Group 1: 1 ppm (artificial fluoridation) Group 2: ppm not stated (natural fluoridation)
Outcomes	% caries‐free participants Age at baseline assessment: 3 to 8 years and 11, 12 and 15 years  Age at final assessment: 3 to 8 years, and 11, 12 and 15 years
Funding	In collaboration with members of the committee Pro‐Fluoridation
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries	See Table for ROBINS‐I assessment Confounding. Efforts were made to control for confounding through design. The groups were considered to be comparable by the author team owing to the areas being neighbouring cities. "The comparison is based on the knowledge of their demographics, culture and social economy". While data were not collected on SES as part of the study, existing data were used to provide reassurance of comparability on SES proxies including infant mortality rate and illiteracy rates. These rates were reported in the paper to allow readers to judge that the areas were comparable across these characteristics. Classification of interventions. Intervention status classified correctly Selection of participants into the study/analysis. All eligible children in the cities were invited to the study Deviations from intended interventions. No deviations from intended intervention Missing data. No missing outcome data, however, no data regarding the confounder. The study reported on existing data only at baseline rather than collecting data with regard to the confounder directly from study participants. Measurement of the outcome. Outcome assessment was conducted by unblinded assessors Selection of the reported result. Outcome of interest reported
Notes	Data extracted from Adriasola 1959 differs from that presented in McDonagh 2000 (additional data extracted) Paper translated from Spanish

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: UK Geographic location: Anglesey (F); Leeds (non‐F) Year of study: 1973 Year of change in fluoridation status: 1955 Study design: cross‐sectional
Participants	Inclusion criteria: lifetime residents of study areas; children aged 12 to 16 years Exclusion criteria: missing, fractured or crowned teeth; refusal to participate (1 school in Leeds) Other sources of fluoride: not stated SES: not stated Ethnicity: not stated Residential history: lifetime residents Other confounding factors: not stated
Interventions	Group 1: 0.9 ppm (artificial fluoridation) Group 2: < 0.01 ppm (natural fluoridation)
Outcomes	Dental fluorosis Age at assessment: 12 to 16 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes	Data extracted from Al‐Alousi 1975 differs from that presented in McDonagh 2000
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	Children were selected from schools in Leeds in a quasi‐random way whereby every nth child (n = total children in school/20) from the register was selected. Eligible children in Anglesey were selected from schools randomly
Confounding	High risk	Did not account for use of other fluoride sources or SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	A clinical investigation and double‐blinded photographic examination were conducted. However, the results reported are those of the unblinded clinical investigation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Unclear risk	Outcome of interest reported
Other bias	High risk	Diagnoses had to be 'agreed' on by the 2 examiners and there was no mention of any sort of calibration of the examiners. This may have resulted in measurement bias

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Mexico Geographic location: Durango Year of study: not stated Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: children aged 6 to 12 years who had established permanent residence in the area Exclusion criteria: not stated Other sources of fluoride: not stated SES: not stated Ethnicity: not stated Residential history: permanent residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: non‐detectable to 1.5 ppm Group 2: 1.51 to 4.99 ppm Group 3: 5.0 to 8.49 ppm Group 4: 8.5 to 11.9 ppm Group 5: > 12 ppm
Outcomes	Dental fluorosis (Dean's Index) Adverse effects (bone fracture) Age at assessment: 6 to 12 years
Funding	Project grant from the Mexican National Council of Science and Technology Conacyt‐Sivilla, Project 9502160
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	Through a polystage conglomerate random sampling, 380 families were selected and prorated into 77 to 80 families per concentration area zone. The division yielded a total of 1437 individuals from the 5 different areas
Confounding	High risk	Did not account for use of other fluoride sources or SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	Unclear risk	No information examiner calibration with regard to detection of the outcome variable

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Hungary Geographic location: Bár and Dunaszekcső Year of study: 2004 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: healthy schoolchildren, aged 6 to 18 years; lifelong residents in the communities Bár or Dunaszekcső; only permanent teeth were investigated Exclusion criteria: any systemic disease Other sources of fluoride: not stated SES: not stated Ethnicity: not stated Residential history: lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 1.7 ppm Group 2: 2 ppm
Outcomes	Dental fluorosis (Dean's Index and TSIF) Age at assessment: 6 to 18 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes	Paper translated from Hungarian
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	There was insufficient detail reported in order to determine how selection took place
Confounding	High risk	Did not account for use of other fluoride sources or SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	Low risk	No other apparent bias

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Saudi Arabia Geographic location: Riyadh Year of study: 2010 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: Saudi nationality; lifetime residence in the area Exclusion criteria: non‐Saudi nationality; absence from school on the day of dental examination Other sources of fluoride: not stated SES: both schools from urban and rural areas were included in the sample frame Ethnicity: Saudi nationals, no further details Residential history: lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 0 to 0.3 ppm Group 2: 0.31 to 0.6 ppm Group 3: 0.61 to 1 ppm Group 4: 1.01 to 1.5 ppm Group 5: 1.51 to 2 ppm Group 6: 2.01 to 2.5 ppm Group 7: ≥ 2.51 ppm
Outcomes	Dental fluorosis (TF Index) Age at assessment: 6 to 18 years
Funding	Supported by a grant from King Abdulaziz City for Science and Technology, Riyadh, Saudi Arabia
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	A list of zones was considered as the sampling frame for the schools, and municipalities were randomly chosen from each zone to represent the urban area. Additionally, rural areas in the municipality with ≥ 1 school were surveyed. However, there was insufficient detail reported in order to determine how selection of schools and children within those schools took place
Confounding	High risk	Did not account for use of other fluoride sources
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	High risk	Over 95% of the subjects sampled were examined. However, it is not clear why fluorosis was not scored in permanent teeth of the 6‐ to 7‐year‐olds
Selective reporting (reporting bias)	High risk	The study authors did not report or justify not presenting fluorosis data for the age group 15 to 18 years
Other bias	Unclear risk	Clinical examination was carried out by 2 dentists, but no information on whether the examiners were calibrated with regard to detection of the outcome variable was given

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Italy Geographic location: areas around Naples (F); Catanzaro (non‐F) Year of study: 1997 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: lifetime residents of study areas (children only); children aged 12 years; used community water supply as main sources of drinking water Exclusion criteria: partially erupted teeth; orthodontic banding Other sources of fluoride: tooth brushing habits (frequency of tooth brushing); fluoride tablets; fluoride dentifrices SES: parents' employment status Ethnicity: not stated Residential history: lifetime residents Other confounding factors: sweet consumption; climate
Interventions	All natural fluoridation Group 1: ≥ 2.5 ppm Group 2: ≤ 0.3 ppm
Outcomes	Dental fluorosis; caries data evaluated in study but not included in review due to study design Age at assessment: 12 years
Funding	Partially supported by a grant of Acquedotto Vesuviano S.p.A.
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	Schools were selected at random, as were classes with the schools. All eligible children within the selected class were recruited to the study
Confounding	High risk	There was a reported imbalance between groups in the use of fluoride supplements, toothbrushing behaviour and in SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data for the majority of participants presented
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	Unclear risk	The 2 examiners involved had previously been trained and calibrated, but details not presented

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: India Geographic location: Nagaur district Year of study: 2013 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: only villages where the mean fluoride concentration was > 1.0 mg/L were selected for the dental fluorosis survey. No other information provided for participants Exclusion criteria: not stated Other sources of fluoride: not stated SES: not stated Ethnicity: not stated Residential history: not stated Other confounding factors: not stated
Interventions	54 villages receiving water with different natural fluoride concentrations ranging from 0.9 to 5.8 ppm
Outcomes	Dental fluorosis (Dean's Index) Age at assessment: not stated
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	Only villages where the mean fluoride concentration was > 1.0 ppm were selected. There was insufficient detail reported in order to determine how selection took place.
Confounding	High risk	Did not account for use of other fluoride sources or SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information to determine whether data presented for all participants as study details were poorly reported
Selective reporting (reporting bias)	Low risk	Outcome of interest not reported in paper, but made available by study authors via email
Other bias	High risk	Fluoride concentration for the different villages overlapped making the data impossible to interpret

PERMALINK

Water fluoridation for the prevention of dental caries

Zipporah Iheozor-Ejiofor

Tanya Walsh

Sharon R Lewis

Philip Riley

Dwayne Boyers

Janet E Clarkson

Helen V Worthington

Anne-Marie Glenny

Lucy O'Malley

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Summary of findings 1. The initiation of community water fluoridation programmes on the prevention of dental caries.

Summary of findings 2. The cessation of community water fluoridation programmes on the prevention of dental caries.

Summary of findings 3. The association of water fluoridation (artificial or natural) with dental fluorosis.

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Evaluation of initiation or cessation of community water fluoridation programmes for the prevention of dental caries

Evaluation of the association of water fluoridation (artificial or natural) with dental fluorosis

Types of participants

Types of interventions

Evaluation of initiation or cessation of community water fluoridation for the prevention of dental caries

Evaluation of the association of water fluoridation (artificial or natural) with dental fluorosis

Types of outcome measures

Evaluation of initiation or cessation of community water fluoridation for the prevention of dental caries

Evaluation of the association of water fluoridation (artificial or natural) with dental fluorosis

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Initiation or cessation of community water fluoridation for the prevention of dental caries

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

1.

Included studies

Evaluation of the effects of initiation or cessation of community water fluoridation on dental caries

Initiation of water fluoridation

Cessation of water fluoridation

Evaluation of the association of water fluoridation (artificial or natural) with dental fluorosis

Excluded studies

Risk of bias in included studies

Evaluation of the effects of initiation or cessation of community water fluoridation on dental caries

1. ROBINS‐I assessment for studies evaluating the initiation or cessation of community water fluoridation programmes on the prevention of dental caries.

Risk of bias due to confounding

Risk of bias in classification of interventions

Risk of bias in selection of participants into the study (or into the analysis)

Risk of bias due to deviations from intended interventions

Risk of bias due to missing data

Risk of bias arising from measurement of the outcome

Risk of bias in selection of the reported result

Evaluation of the association of water fluoridation (artificial or natural) with dental fluorosis

*Study characteristics*
Methods	CARIES STUDY Country of study: USA Geographic location: Group 1: Grand Rapids (F); Group 2: Muskegon (non‐F). Total population sizes for each location not stated Year study started: 1944 Year study ended: 1951 (after which time the control group became fluoridated; evaluated until 1954) Year of change in fluoridation status: 1945 Study design: CBA. A different sample of children was assessed at each time point, according to age at last birthday.
Participants	Inclusion criteria: children aged 4 to 16 years; used city water supplies since birth Exclusion criteria: children who lived outside study areas for more than 3 months of any 1 year Sample size at baseline: Group 1: 19,680; Group 2: 4291 (all school children) Sample size at final assessment: Group 1: 4590; Group 2: 2192 (sample of school children in 1951) SES: not stated Co‐interventions: study author stated that there were no concerted efforts to commence special caries control programmes e.g. topical fluoride programmes, in either of the cities since the study began Ethnicity: not stated Gender: not stated Residential history: lifetime residents
Interventions	Initiation of water fluoridation Group 1: 1 ppm (artificial fluoridation) Group 2: < 0.2 ppm (natural fluoridation)
Outcomes	DMFT; dmft Age at baseline assessment: 5 to 13 years (primary dentition); 6 to 16 years (permanent dentition) Age at final assessment: 5 to 13 years (primary dentition); 6 to 16 years (permanent dentition)
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries	See Table for ROBINS‐I assessment. Confounding. Study assessed as at critical risk for confounding due to lack of details regarding SES. No further assessment Classification of interventions. Intervention status classified correctly Selection of particpants into the study/analysis. Children were selected through schools. Almost all eligible children in the areas of study were examined Deviations from intended interventions. No deviations from intended intervention reported Missing data. "samples consist of all available children in certain grades (or in sections of the grades)" Number of children examined each year presented, however, numbers varied across each age group and each year (not a continuous study sample). It is noted in the results that fluorosis observations had been made, but no details were given for the methods and data (just % increase). Also, SD not reported. Measurement of the outcome. No blinding of assessors Selection of the reported result. It is noted in the results that fluorosis observations had been made, but no details were given for the methods and data (just % increase). Also, SD not reported
Notes	Data extracted from Arnold 1956 differed from that presented in McDonagh 2000 (additional data extracted)

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Tanzania Geographic location: Arusha and Moshi Year of study: 1996 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: age 9 to 14 years; lifelong residence in respective towns or villages Exclusion criteria: not stated Other fluoride sources: toothpaste use: Arusha = 94%; Arusha Meru = 100%; Moshi = 97.1% and Kibosho = 40%; Magadi use: Arusha = 31 (47%); Arusha Meru = 1(2.9%); Moshi = 41 (58.6%); Kibosho = 83 (97.6%) SES: peasant mothers: Arusha = 1 (1.5%); Arusah Meru = NR; Moshi = 7 (10%); Kibosho = 33 (38.8%); other: Arusha = 65 (98.5%); Arusha Meru = 35 (100%); Moshi = 63 (90%); Kibosho = 52 (61.2%) Ethnicity: Arusha area (Arusha and Arusha Meru) – mainly ethnic Asians; Kilimanjaro region (Moshi and Kibosho) ‐ Africans Residential history: lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 0.2 ppm Group 2: 0.3 ppm Group 3: 3.6 ppm
Outcomes	Dental fluorosis (TF Index) Age at assessment: 9 to 14 years
Funding	Supported by the Norwegian State Educational Loan fund, NUFU project 61/96, and the committee for Research and Postgraduate Training, Faculty of Dentistry, University of Bergen, Norway
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	Schools in all villages (except in Arusha Meru) as well as participants were randomly selected. For schools where participants were not randomly selected, including the school in Arusha Meru, all the registered schoolchildren were chosen to participate
Confounding	High risk	There was a reported imbalance between groups in terms of SES and use of fluoride from other sources
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	High risk	Outcome of interest not fully reported, rather presented as a median score
Other bias	High risk	Only 1 examiner was involved; no testing for intra‐rater reliability with regard to detection of the outcome variable.

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Argentina Geographic location: Sampacho (F); Porteña (non‐F) in the Cordoba province Year of study: 1993 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: children aged 6 to 7 years (1^st grade) and 12 to 13 years (7^th grade) at primary school Exclusion criteria: none stated Other sources of fluoride: frequency of tooth brushing Group 1 (aged 6‐7): 56% brushed at least once a day (28/50) Group 1 (aged 12‐13): 74% brushed at least once a day (37/50) Group 2 (aged 6‐7): 46% brushed at least once a day (23/50) Group 2 (aged 12‐13): 50% brushed at least once a day (25/50) SES: determined by occupation and highest attained level of schooling attained by main breadwinner in family (classified as high, medium, and low social class) Group 1 (aged 6‐7): 80% low SES (40/50) Group 1 (aged 12‐13): 82% low SES (41/50) Control (aged 6‐7): 74% low SES (37/50) Control (aged 12‐13) 80% low SES (40/50) Residential history: not stated Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 9.05 ppm Group 2: 0.19 ppm
Outcomes	Dental fluorosis (Dean’s Index); caries data evaluated in study but not included in review due to study design Age at assessment: 6 to 7 years and 12 to 13 years
Funding	Part of this work was subsidised by the Ministry of Science and Technology of the National University of Córdoba, Córdoba, Argentina
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	Stratified random selection was used. Following stratification by age, gender and SES, 100 school children were randomly selected from each village
Confounding	High risk	Although SES was considered during sampling, it was not controlled for within the analysis. No details were reported on the use of fluoride from other sources
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not stated, however the two calibrated operators, as study authors, were likely to have knowledge of the study areas
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Low risk	Outcome of interest was fully reported on and balanced across both groups
Other bias	Low risk	No other apparent biases

*Study characteristics*
Methods	CARIES STUDY Country of study: Holland Geographic location: Group 1: Tiel (F); Group 2: Culemborg (non‐F). Total population sizes for each location not stated Year study started: 1952 Year study ended: 1959 Year of change in fluoridation status: 1953 Study design: CBA. A different sample of children was assessed at baseline and time points throughout the study, according to age at the time of assessment.
Participants	Inclusion criteria: children aged 11 to 15 years; lifelong residents of the study areas; used the piped water supply; 100 children of each age examined Exclusion criteria: not stated Sample size at baseline: Group 1: not specified but assumed to be 100 participants per year of age from the information in the study report (i.e. 500 children); Group 2: not specified but assumed to be 100 participants per year of age (i.e. 500 children) Sample size at final assessment: Group 1: as above, assumed to be 500 participants ; Group 2: as above, assumed to be 500 participants SES: areas similar in social class structure and proportional numbers of children selected from each school type Co‐interventions: not stated Ethnicity: not stated Gender: equally balanced Residential history: lifetime residents
Interventions	Initiation of water fluoridation Group 1: 1.1 ppm (artificial fluoridation) Group 2: 0.1 ppm (natural fluoridation)
Outcomes	Carious approximal surfaces per child; approximal surfaces with caries of the dentine; pit and fissure caries per child; pit and fissure caries with cavitation; carious lesions of smooth surfaces. No totals for all surfaces Age at baseline assessment: 11 to 15 years (permanent dentition) Age at final assessment: 11 to 15 years (permanent dentition)
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries	See Table for ROBINS‐I assessment Confounding. "Attention was given to population structure, site, size (above 15,000 inhabitants), migration and water composition, and two cities were selected which were as equal as possible". Areas similar in social class structure and proportional numbers of participants selected from each school type, although no details on how SES was measured or distributed. Study authors therefore provided some reassurance that the areas were comparable in terms of characteristics which are proxy measures of SES, but these data were not reported for the reader to make this judgement on compatibility for themselves. Classification of interventions. Intervention status classified correctly Selection of participants into the study/analysis. A proportion of children were chosen at random from different types of schools (public school, Roman Catholic, Protestant) Deviations from intended interventions. No deviations from intended intervention reported Missing data. No missing outcome data, however, no data regarding confounder. Authors commented on the populations being equal at baseline but did not collect data directly from participants regarding the confounder. Measurement of the outcome. No blinding regarding assessment of pit and fissure lesions. For aproximal caries: "The radiographs made in Tiel and Culemborg were put into unlabelled envelopes, and examined at random". Each examiner evaluated the same number of radiographs without knowledge of the origin of the films. Selection of the reported result. Outcome of interest reported, however, SDs missing
Notes

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: China Geographic location: 3 cities (Harbin, Mudanjiang, Zhaodong) and 3 rural areas (Zhaoyuan, Shuangcheng, Linkou) in the Heilongjiang province Year of study: not stated Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: 12‐year‐old children in Heilongjiang Exclusion criteria: not reported Other sources of fluoride: not reported SES: 396 (198 male; 198 female) from cities; 396 (198 male; 198 female) from rural areas Ethnicity: Chinese Residential history: not reported Other confounding factors: not reported
Interventions	All natural fluoridation Group 1 (Linkou): 0.29 ppm Group 2 (Mudanjiang): 0.40 ppm Group 3 (Shuangcheng): 0.68 ppm Group 4 (Harbin): 0.77 ppm Group 5 (Zhaoyuan): 0.80 ppm Group 6 (Zhaodong): 1.14 ppm
Outcomes	Dental fluorosis (CFI); caries data evaluated in study, but excluded from review due to study design Age at assessment: 12 years
Funding	Research Fund of Bureau of Health of Heilongjiang Province (grant no.2005[122])
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes	Translation from Chinese
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	Quote: "Representative samples were selected by multi‐stage, stratified and random sampling" "For each site, 66 12‐year‐old boys and 66 12‐year‐old girls were randomly chosen".
Confounding	High risk	3 groups were from cities and 3 groups were from rural areas. The study authors did not record/report or adjust for other confounding factors (e.g. other fluoride sources, diet, residential history)
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The study authors did not report any information on loss of follow‐up or exclusion of participants. Judging by the number of people they chose randomly (792), and the number of people (792) with results of caries examination, there was no loss of follow‐up or exclusion of participants.
Selective reporting (reporting bias)	High risk	Quote: "Dean’s Index was used to classify fluorosis." Comment: data not presented in a format that allowed for further evaluation. The study authors did not report the number of affected people for each Dean's Index category. They did not report the prevalence fluorosis (number of affected people/number of people examined)
Other bias	Low risk	No other apparent bias

*Study characteristics*
Methods	CARIES STUDY Country of study: England Geographic location: Group 1: Balsall Heath and Northfield, Birmingham (F); Group 2: Dudley (non‐F). Total population sizes for each location not stated Year study started: 1967 Year study ended: 1970 Year of change in fluoridation status: 1964 Study design: CBA. A different sample of children was assessed at baseline and time points throughout the study, according to age at the time of assessment.
Participants	Inclusion criteria: children aged 5 years attending schools that participated in each year of the study Exclusion criteria: none stated Sample size at baseline: Group 1: 297 children; Group 2: 217 children Sample size at final assessment: Group 1: 314 children; Group 2: 229 children SES: quote: "The socio‐economic composition of the districts has been described previously ...". Balsall Heath is a poor area of the city with high proportion of immigrants; Northfield and Dudley are both industrial areas with comparable populations, but there were more immigrants in Dudley. Co‐interventions: not stated Ethnicity: all areas have some proportion of immigrants Gender: not stated, but study authors describe results for boys and girls as "not significantly different". Residential history: no attempt was made to select continuously resident children from the samples.
Interventions	Initiation of water fluoridation Group 1 and Group 2: 1 ppm (artificial fluoridation) Group 3: < 0.1 ppm (natural fluoridation)
Outcomes	dmft; % caries‐free children Age at baseline assessment: 5 years (primary dentition) Age at final assessment: 5 years (primary dentition)
Funding	MRC grant‐funded trial
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries	See Table for ROBINS‐I assessment Confounding. Study assessed as at critical risk for confounding due to one area being reported as a poor area of the city with a higher proportion of immigrants in the population. No further assessment Classification of interventions. Intervention status classified correctly Selection of participants into the study/analysis. Insufficient information Deviations from intended interventions. No deviations from intended intervention reported Missing data. Given lack of information on sampling, proportion of missing data at each time point unknown Measurement of the outcome. Examination undertaken in mobile clinic at each school; blinding unlikely Selection of the reported result. No apparent selective reporting
Notes	Quote: "The children, who were 5 years old in 1967, were aged about 3 years when the fluoride in their drinking water reached the recommended level; they had erupted all their deciduous, and these would be expected to have derived only slight benefit at this time. These children do not represent a true baseline; any dental advantage that this group had received, compared with the true but unexamined baseline before fluoride was added would have the effect of decreasing the observed reduction, if any, over subsequent years."

*Study characteristics*
Methods	CARIES STUDY Country of study: England Geographic location: Group 1: Scunthorpe (F) ‐ population size of 70,000 residents; Group 2: Corby (non‐F) ‐ population size of 52,000. Study authors state that despite differences in size, both towns are similar in other respects. Year study started: 1969 Year study ended: 1975 Year of change in fluoridation status: 1968 Study design: CBA. A different sample of children was assessed at baseline and time points throughout the study, according to age at the time of assessment.
Participants	Inclusion criteria: lifetime residents in study areas; children aged 5, 8 and 12 years Exclusion criteria: teeth extracted for orthodontic purposes Sample size at baseline: Group 1: 196 children; Group 2: 205 children Sample size at final assessment: Group 1: 170 children; Group 2: 180 children SES: both areas had iron/steel as main industry‐socioeconomic; composition of the 2 areas was similar Co‐interventions: not stated Ethnicity: not stated Gender: not stated Residential history: lifetime residents
Interventions	Fluoride initiation  Group 1: 0.9 ppm (artificial fluoridation) Group 2: 0.35 ppm (natural fluoridation)
Outcomes	dmft; DMFT; % caries‐free children (primary teeth); % caries‐free subjects (permanent teeth) Age at baseline assessment: 5, 8 and 12 years Age at final assessment: 5, 8 and 12 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries	See Table for ROBINS‐I assessment Confounding. The SES composition of the 2 towns was stated as being similar, although no details on how SES was measured or distributed Classification of interventions. Intervention status classified correctly Selection of participants into the study/analysis. Schools were chosen by random selection and every child of eligible age in these schools was examined. Deviations from intended interventions. No deviations from intended intervention reported Missing data. The study reports that "every child of eligible age in these schools was examined", suggesting outcome data for all participants are presented although not explicitly stated. No data on confounding variable for participants reported Measurement of the outcome. Insufficient information regarding blinding of assessors Selection of the reported result. No apparent selection of reporting
Notes

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: USA Geographic location: not stated Year of study: 1986 Year study ended: 1987 Year of change in fluoridation status: not stated Study design: cross‐sectional
Participants	Inclusion criteria: aged 12 to 14 years; availability of data on type of water system and fluorosis; having residences served by the same type of public water system with respect to fluoride status; determinable date of public water system fluoridation initiation and residence at area before initiation of water fluoridation; availability of continuous residence history if more than 1 residence; fewer than 5 residences; ascertainable exposure to fluoride drops or tables; served by public water systems with ascertainable fluoride status in residences Other fluoride sources: tablets = 623 (14.9%); drops = 627 (14.5%); tablets and drops = 317 (8.4%). Suboptimal fluoride: drops only = 507 (23.0); tablets only = 512 (22.5); tablets and drops = 279 (13.2). Optimal fluoride:drops only = 103 (6.8); tablets only = 98 (6.0); tablets and drops = 32 (2.2) Natural fluoride: drops only = 13 (5.5); tablets only = 17 (7.5); tablets and drops = 6 (2.5) Exclusion criteria: any criterion in discord with the inclusion criteria SES: not stated Ethnicity: not stated Residential history: all the children were continuous residents of areas with the reported water systems Other confounding factors: not stated
Interventions	Group 1: < 0.7 ppm (natural fluoridation) Group 2: 0.7 to 1.2 ppm (artificial fluoridation) Group 3: 0.7 to 4 ppm (natural fluoridation)
Outcomes	Dental fluorosis (Dean's Index) Age at assessment: 12 to 14 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Low risk	The sampling frame was specified, and the sample represented 41 percent of all 12‐ to 14‐year‐olds and > 4 million schoolchildren. There is no evidence that any eligible children were excluded
Confounding	High risk	The use of other fluoride sources was similar in those who consumed water with optimal and natural fluoride, but very different from those in the suboptimal fluoride group. Did not account for SES
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Children with missing outcome data were excluded. It is not clear whether there was an imbalance across groups in excluded children
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	High risk	There is an overlap in fluoride concentration between the exposure groups (0.7 to 1.2 ppm and 0.7 to 4.0 ppm) which is likely to dilute the observable effect of exposure to intervention across groups. It is unclear whether the examiners were calibrated as the paper provides insufficient information, and we were unable to access associated reports which may have contained examination protocols.

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Namibia Geographic location: Ombili, Ondera, Vryheid, Kakuse Year of study: October 2004 Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: aged 8 to 21 years Other fluoride sources: 47 (39.3%) reported oral hygiene practice with fluoridated toothpaste (1400 ppm); 8 (6.7%) used traditional 'natural' toothbrush. Different ethnic groups differed markedly in their oral hygiene behaviour (P value 0.02) Exclusion criteria: not stated SES: not stated Ethnicity: !Kung (45%); Heikum (35%); Damara (13%); Bantu (7%) Residential history: residents of Ombili had been resident since 1991 and the residents of the other farms were lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 0.28 ppm Group 2: 0.38 ppm Group 3: 1.06 ppm Group 4: 1.43 ppm
Outcomes	Dental fluorosis (Dean's Index; CFI) Age at assessment: 8 to 21 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	Children selected from Ombili Primary School and divided into groups according to place of birth and ethnicity
Confounding	High risk	Imbalance in oral health behaviour and duration of residency between ethnic groups
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants accounted for in analysis
Selective reporting (reporting bias)	Low risk	Outcome data fully reported
Other bias	Low risk	No other apparent bias

*Study characteristics*
Methods	FLUOROSIS STUDY Country of study: Sudan Geographic location: Triet el Biga, Abu Delaig and Abu Groon Year of study: not stated Year of change in fluoridation status: NA Study design: cross‐sectional
Participants	Inclusion criteria: residence in the village from the age of 1 year Exclusion criteria: not stated Other fluoride sources: not stated SES: similar socioeconomic conditions Ethnicity: similar ethnicity Residential history: lifetime residents Other confounding factors: not stated
Interventions	All natural fluoridation Group 1: 0.3 to 1.4 ppm Group 2: 0.8 to 2.2 ppm Group 3: 2 to 4.2 ppm
Outcomes	Dental fluorosis (TF Index) Age at assessment: 11 to 13 years
Funding	Not stated
ROBINS‐I comments for studies evaluating initiation or cessation of CWF for prevention of dental caries
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sampling	Unclear risk	The schools were selected from an unspecified sampling frame and insufficient detail was reported to determine how selection of schools took place. However, children were selected at random from the schools.
Confounding	High risk	No details were reported on the use of fluoride from other sources.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data presented for all participants
Selective reporting (reporting bias)	Low risk	Outcome of interest reported
Other bias	High risk	There is inconsistency in the number of water samples tested (Triet el Biga = 6, Abu Delaig = 11, Abu Groon = 8) and an overlap in range of fluoride concentrations between the 3 study areas. Also, examinations were done by a dental assistant and it is not clear whether reliability testing was carried out.