Personal oral hygiene and dental caries: A systematic review of randomised controlled trials

OBJECTIVE
To conduct a systematic review of randomised trials assessing the association between personal oral hygiene and dental caries in the absence of the confounding effects of fluoride.


BACKGROUND
Dental caries continues to affect close to 100% of the global population. There is a century-old conflict on whether dental caries is caused by poor oral hygiene or poorly formed teeth (ie, teeth with dental defects). Resolving this conflict is of significant public health importance as these two hypotheses on dental caries aetiology can lead to different prevention strategies.


METHODS
A systematic search for randomised trials was conducted using predefined criteria in 3 databases. The impact of personal oral hygiene interventions on coronal dental caries incidence was evaluated using random-effects models.


RESULTS
Three randomised studies involving a total of 743 participants were included. Personal oral hygiene interventions failed to influence the incidence of dental caries (Δ Decayed, Missing and Filled Surfaces (DFMS) = -0.11; 95% confidence interval: (-0.91, 0.69; P-value < .79)) despite meticulous deplaquing of teeth. There was no significant heterogeneity in the trial results (heterogeneity chi-squared = 1.88, P = .39). The findings were robust to sensitivity analyses, including consideration of the results of nonrandomised studies.


CONCLUSION
Personal oral hygiene in the absence of fluorides has failed to show a benefit in terms of reducing the incidence of dental caries.

The dental defect hypothesis posits that dental caries starts in microscopic cracks or crevices in teeth, and not on defect-free or sound enamel. The biofilm within the dental defects is also thought to become cariogenic in the presence of dietary carbohydrates.
However, the biofilm cannot be removed with a toothbrush or interproximal cleaning devices; hence, oral hygiene is believed to be ineffective. Under this alternative hypothesis, prevention of dental caries must focus on preventing the formation of dental defects during odontogenesis, by repairing the defects from the pulpal side, or by sealing or surgically eliminating surface defects in the enamel.
Historically, the motto of the proponents for the dental defect hypothesis was that "sound teeth do not decay". 4 A key step in resolving the conflict between these two hypotheses is to determine the value of personal oral hygiene without the confounding effects of fluoride, the latter having strong evidence in support of its anticariogeneic effect. 5 The aim of this study was to report on a systematic review of randomised trials assessing the association between personal oral hygiene and coronal dental caries.

| Search strategy
The systematic search was performed following the guidelines set out by the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA). 6 Three databases (PubMed, Web of Science and the Cochrane Central Register of Controlled Trials) F I G U R E 1 Mediation framework for evaluating the aetiology of dental caries. Arrows 1-3 illustrate the "oral hygiene" or the "clean tooth" hypothesis of dental caries. Arrows 4 and 5 illustrate the dental defect or "sound tooth" hypothesis of dental caries. Both hypotheses were widely debated in the early 20th century TA B L E 1 The design, interventions and outcome quality of included and excluded trials used to measure the effectiveness of personal oral hygiene in controlled dental caries

| Inclusion criteria
Controlled clinical trials (CCTs) of personal oral hygiene interventions were included if the primary aim was to assess the impact of toothbrushes with or without interproximal cleaning devices on dental caries, if the trial design mentioned the word "random" and was prospective, if dental caries scores (as a count) were reported at baseline and at the end of the follow-up, and if there was a concurrent control group. There were no restrictions on method of treatment assignment, participant characteristics or on primary vs permanent teeth. Nonrandomised studies which met all other inclusion criteria were accepted to assess the robustness of the primary analysis.

| Exclusion criteria
Trials were excluded in which the effect of personal oral hygiene interventions was combined with fluoride products or dietary interventions and the control group had no such interventions. Trials were excluded also if they focused on dental caries surrogates (eg, bacterial counts), professional prophylaxis or where oral hygiene assignment was based on patient behaviour. Trials with a primary aim of assessing chemotherapeutics (e.g., fluoride, chlorhexidine) were excluded. Retrospective studies, cross-sectional studies and letters to the editors were not considered.

| Quality assessment
The CCT quality was quantified using a modified Jadad scale 8 and risk of bias measures for randomised controlled trials. 9 The Jadad score and our modification of it is a score between 0 and 5 where 0 indicates poor quality and 5 highly rigorous quality. The Jadad modification consisted of changing the word "double-blind" to "single-blind" as it was considered impossible to blind trial participants towards self-performed tasks such as brushing and flossing (

| Data synthesis
Summary DMFS estimates were based on random-effects models.
The heterogeneity of the studies was evaluated using the heterogeneity chi-squared statistic. Effective sample sizes for the trials were calculated assuming an intracluster coefficient (ρ) of 0.02. 10 This effective sample size was calculated as the total sample size divided by the design effect (1+ρ (m-1)) where m is the average class or school size when the cluster size was unavailable, otherwise m was calculated as the sum of the cluster sizes squared divided by the number of participants in that group when cluster sizes were reported. 11 For studies which assigned classes to treatments, the number of classes in public schools was calculated assuming each class had on average about 20 pupils. Radiographic caries increment scores combined with clinical scores were selected over clinical caries increments alone when both outcomes were available. 12 Exploratory analyses were conducted to assess the robustness of the findings. Included in these secondary analyses was an assessment of the impact of nonrandomised studies. These nonrandomised studies included different dental outcome scores (such as Decayed, Missing or Filled Teeth or DMFT) for caries and were standardised using the Glass's Δ method. 13 Highly significant heterogeneity was taken as an indication to analyse the data using a random-effects model, not fixed-effects models. 14,15 All analyses were completed using SAS 9.4 (including the forest and the metaanal macros), STATA 11.1 meta-analysis software and R 3.3.2 (meta).

| Selection of controlled clinical trials
A total of 984 unique citations from 3 sources were identified ( Figure 2). Thirteen references of interest were identified for fulltext review, which included 12 published articles and one abstract for which we obtained the NIH grant report. 12,16-21 After full-text review, three randomised trials were included ( Figure 2). Four nonrandomised trials were retained for the purpose of sensitivity analyses.

| Characteristics of randomised controlled trials assessing dental plaque removal
Three randomised trials on 743 participants were identified (Table 1); two trials were conducted in the United States (US), and one in the United Kingdom (UK). 12 The enrolled participants were girls aged 11 and 12, 12 and boys and girls aged 10 and 13, 18 and 12 and 13. 16 All three trials reported DMFS scores; two trials continued for 3 years 12,18 and one trial for 29 months. 16 The units of randomisation were participants 16,18 and classes. 12 The two US trials were Included conducted in nonfluoridated communities. The level of water fluoridation was not reported for the UK trial. There were no significant differences in the baseline score for caries between the intervention and control groups. The Jadad score was "4" for all 3 trials. None were funded by commercial companies. Drop-out rates ranged between 15% and 39% (Table 1). Details on standard deviation calculations are presented as a footnote in Figure 3.

| Synthesis of results
There was no heterogeneity between trials (Heterogeneity chi-

| Risk of bias across controlled clinical trials
Meta-regression or statistical assessment of publication bias was not performed due to the limited number of trials and minimal variability in terms of duration or quality of the randomised trials.

| Sensitivity analyses
One of the three randomised trials used clustered randomisation, which we accounted for in our main analysis. Results did not change appreciably when the clustered randomisation of this trial was ig-

| D ISCUSS I ON
A meta-analysis of controlled clinical trials does not support the hypothesis that improved personal oral hygiene without fluoride decreases the risk of coronal caries. This lack of effectiveness cannot be attributed to the presence of water fluoridation dwarfing the benefits of oral hygiene because two of the three randomised trials were conducted in nonfluoridated communities. Neither can the lack of effectiveness be attributed to non-supervision of the oral hygiene intervention. Two randomised trials with strict protocols involving daily supervised plaque staining and removal on school days reported significant reduction in gingivitis but not of dental caries. 16,18 These randomised trial findings are robust towards sensitivity analyses. The randomised trial findings are also robust when considering nonrandomised trials with the exception of one industry-funded, nonrandomised study which was the only study to report highly significant benefits of personal oral hygiene.
The primary conclusion of this systematic review, that oral hygiene has no impact on dental caries rates, is strengthened by other independent lines of evidence. A network meta-analysis of six systematic reviews on 130 controlled trials selected under the auspices of the Cochrane Oral Health Group compared the effectiveness of four fluoride products (gels, varnishes, toothpaste and rinses) in over 60 000 participants. The findings of this network meta-analysis showed that fluoride toothpaste, and fluoride rinse were similarly effective in preventing coronal dental caries. So, in effect, personal oral hygiene was shown to be an inactive delivery mechanism for an active pharmaceutical, fluoride. Our findings are furthermore consistent with systematic reviews on the ineffectiveness of selfperformed flossing, [23][24][25] professional dental plaque removal 26,27 and removal of dental plaque prior to fluoride applications for preventing coronal dental caries. 28 Controlled trials on topical chemotherapeutics provide additional insights on the validity of the oral hygiene hypothesis. Biofilm on smooth non-dental surfaces, but not in cracks and crevices, can be removed by mechanical cleansing and chemotherapeutics. 29 Chemotherapeutics are therefore more likely to be effective if dental caries starts on smooth, defect-free surfaces. A century of controlled trials provides insights into this assumption. A hexylresorcinol rinse was the first major antimicrobial advertised commercially as an effective caries control, but in one of the first controlled trials, it increased the caries rates. 30 This finding was followed by trials involving about one thousand children who were assigned to a placebo-or a penicillin-containing toothpaste. The penicillin-containing toothpastes were found to be unsuccessful at controlling caries. 31,32 Chlorhexidine was described in 1994 as the most potent chemotherapeutic agent against dental caries surrogates, 33 but it too was deemed ineffective following a meta-analyses of controlled trials on the endpoint of frank cavitation. 10 Thus, trials on chemotherapeutics support the dental defect hypothesis on dental caries etiology.
Roughly one century ago, Mellanby predicted the failure of the oral hygiene hypothesis and had proposed instead, based on experimental animal research, that dental caries started in dental defects. 34 This dental defect hypothesis was consistent with the epidemiology of dental caries and widely accepted in the medical and dental professions. 35 The dental defect hypothesis ended up explaining dramatic historical and geographical trends in dental caries prevalence in ways that the oral hygiene hypothesis could not. [36][37][38] The ubiquity of dental caries in the early 20th century was attributed to sugar, rickets and paediatric fevers, and the latter two identified as causes of dental defects. 39 According to the dental defect hypothesis, the dramatic decline of caries in wealthier countries that occurred in the (hydroxy-apatite) has been shown to offer no resistance against demineralisation attacks. 43 Even enamel fluorapatite (30 000 ppm fluoride) was shown to offer minimal resistance to demineralisation 44 and to offer a questionable barrier to dental caries. 45 Further investigations should determine to what extent fluoride and vitamin D prevent dental caries through pulpal side mechanisms. 46 This review does not question whether clinicians should provide advice on the potential benefits of oral hygiene for preventing dental caries. There is still substance to the arguments that small therapeutic effects of personal oral hygiene remained undetected in statistically underpowered trials, that findings in healthy paediatric populations with no exposed cementum do not extrapolate to adults with exposed cementum or decreased saliva flow, or that adults with other systemic diseases or disorders may benefit from personal oral hygiene in terms of dental caries prevention. Indeed, oral hygiene can be a pleasant and cost-effective way to deliver fluoride, reduce gingivitis, remove food impactions, or to help patients in their recovery from oral surgical procedures. The dangers in the unqualified promotion of oral hygiene for dental caries prevention are that it may lead individuals to select fluoride-free toothpastes, to sacrifice effective fluoride exposure for interproximal cleaning without fluor, or to forego effective therapeutics such as fluoride rinses. Such beliefs which reduce fluoride exposure increase dental caries risk and are most dangerous when used as a justification to promote sugar consumption, and to perpetuate the myth that sugar is safe to eat as long as one brushes their teeth. 47

| CON CLUS IONS
A large body of controlled trials has failed to provide convincing evidence in support of the efficacy of personal oral hygiene in preventing or controlling coronal dental caries.