| Sign In to gain access to subscriptions and/or personal tools. |
Dental Flossing and Interproximal Caries: a Systematic Review
1 Department of Dental Public Health Sciences and Correspondence: * corresponding author, hujoel{at}u.washington.edu
Our aim was to assess, systematically, the effect of flossing on interproximal caries risk. Six trials involving 808 subjects, ages 4 to 13 years, were identified. There were significant study-to-study differences and a moderate to large potential for bias. Professional flossing performed on school days for 1.7 years on predominantly primary teeth in children was associated with a 40% caries risk reduction (relative risk, 0.60; 95% confidence interval, 0.48–0.76; p-value, < 0.001). Both three-monthly professional flossing for 3 years (relative risk, 0.93; 95% confidence interval, 0.73–1.19; p-value, 0.32) and self-performed flossing in young adolescents for 2 years (relative risk, 1.01; 95% confidence interval, 0.85–1.20; p-value, 0.93) did not reduce caries risk. No flossing trials in adults or under unsupervised conditions could be identified. Professional flossing in children with low fluoride exposures is highly effective in reducing interproximal caries risk. These findings should be extrapolated to more typical floss-users with care, since self-flossing has failed to show an effect.
Key Words: dental caries • interproximal dental caries • dental floss • dental devices • home care • controlled trials • review • meta-analysis
Suggestions regarding the benefits of flossing date back to the early 19th century, where the belief was expressed that irritating matter between teeth is the source of dental diseases (Parmly, 1819). Further microbiological work, both in the late 19th century and in the 20th century, implicated dental plaque as the cause of caries. Since plaque build-up at interproximal sites has been reported to be more acidogenic than in other areas of the mouth (Igarashi et al., 1989), and since dental floss has the ability to disrupt and remove some interproximal plaque (Waerhaug, 1981), it appears plausible that the use of dental floss should reduce interproximal caries risk. The goal of this study was to provide a systematic review of the controlled clinical trial evidence on dental floss and interproximal dental caries.
Inclusion/Exclusion Criteria The hypothesis of this study was that dental flossing reduces interproximal caries incidence. The treatment comparisons of interest included flossing vs. no flossing, or a comparison of different frequencies of flossing use. Studies where the effect of flossing could not be separated from the effects of other treatments were excluded. The primary study outcome was a measure of caries incidence. There were no restrictions with respect to the study population. Study designs included in this synthesis were limited to controlled clinical trials.
Search Strategy
Quality Assessment Topical fluoride exposure was categorized as not recommended (--), recommended to a subgroup or the whole cohort but with no compliance measures (–), recommended and compliance assessed (+), or delivered under supervised conditions (++). Oral hygiene was similarly classified as no instructions provided (--), instructions provided but compliance not measured (–), instructions provided and compliance measured by plaque scores or gingival bleeding scores (+), or provided under supervision (++).
Statistical Analysis Both the relative risk (RR, or a ratio of the surface caries risk in the flossing group to the surface caries risk in the control group) and the risk difference (RD, or a difference in the surface caries risks between the flossing and the control groups) and their respective standard deviations were calculated. The standard errors of the RR and RD were calculated according to standard formulae for 2 x 2 tables (Rothman et al., 1998), multiplied by the square root of the variance inflation factor to account for the dependence of sites within patients. We estimated the variance inflation factors for each trial by dividing the appropriate standard error of a patient-based t test (one- or two-sample t tests as described in the APPENDIX) by the standard error of a surface-based t test. Three split-mouth trials (Granath et al., 1979; Wright et al., 1979, 1980) did not report sufficient information for variances to be calculated. For two split-mouth trials (Wright et al., 1979, 1980), the variance was derived from the McNemar statistic and an estimate of the correlation of matched pairs within patients (Mäkinen et al., 1996). For the remaining split-mouth trial (Granath et al., 1979), the variance was derived from the reported caries rate (Marinho et al., 2004) and an estimate of the split-mouth correlation from an external dataset (Mäkinen et al., 1995). The results of the studies included in the review were summarized by general variance-based methods, with the weight for each study being the inverse of the variance (fixed-effects model) (Petitti, 1994). The summary RR and RD are equal to the sum of the weights times the risk estimate for each study, divided by the sum of the weights (Petitti, 1994). The weighted grand mean difference was calculated, and the I2 statistic was used to test the hypothesis of the homogeneity of the effect. An I2 statistic larger than 50% is considered moderate to high heterogeneity (Higgins et al., 2003) and an indication to use random-effects models (DerSimonian and Laird, 1986). The effect of study characteristics such as fluoride, oral hygiene, or caries risk on flossing effectiveness was estimated by meta-regression (Thompson and Higgins, 2002).
Study Identification and Characteristics (Fig. 1  )The MEDLINE search identified 104 controlled trials, the CENTRAL search identified 17 controlled trials, and the other sources identified 22 trials. One trial could not be located in MEDLINE or in CENTRAL, but was identified by contacting trial investigators (Wright et al., 1980). Trials were excluded from this review for the following reasons: no assessment of the effect of flossing, absence of caries outcomes, no control group that would have allowed flossing effects to be estimated, or preliminary reports of the included studies (Fig. 1 ). Six randomized clinical trials of the effect of flossing on interproximal caries, involving 808 participants followed for 1.7 to 3 yrs, were included in this review (Table 1). The ages of the study participants ranged from 4 to 13 yrs. Flossing was professionally performed on school days in two studies (typically 5 days per week for a 10-month period or approximately 121–150 flossings days per year) (Wright et al., 1979, 1980), professionally performed once every 3 months in two studies (Gisselsson et al., 1988, 1994), supervised on school days in one study (Granath et al., 1979), and unsupervised in one study (Gisselsson et al., 1983).
Study Quality The quality assessment of the controlled trials revealed generally poor reporting of the studies and the presence of a moderate to high risk of bias (Table 2 ). Two trials used the word "random" to describe the allocation method of proximal surfaces within individuals to treatments, but the allocation concealment method was not described (Wright et al., 1979, 1980). One study did not use the word "random" to describe the allocation method, but did describe that children were ranked according to baseline disease severity and systematically (non-randomly) distributed to experimental and control groups (Gisselsson et al., 1983). In three studies, treatments were assigned not to individuals, but to groups of individuals, such as school classes (Granath et al., 1979; Gisselsson et al., 1988, 1994). The method of assignment of flossing to group was based on birth-year (Gisselsson et al., 1994), by lot (Granath et al., 1979), and by allocation (the word "random" not specified) (Gisselsson et al., 1988). In all four studies reporting information on baseline caries rates, the experimental units assigned to flossing were healthier (fewer cavities) at baseline (Granath et al., 1979; Gisselsson et al., 1983, 1988, 1994). When summarized across the four studies, the difference in baseline caries was significant (p-value < 0.05). No statistical comparison of baseline caries rates by treatment group could be obtained in two studies (Wright et al., 1979, 1980), preventing an assessment of the successfulness of the randomization.
The outcome assessors were not blinded in one study (Gisselsson et al., 1983). No studies reported confidence intervals for the outcome point estimates. It was unclear if any of the reported studies provided an intent-to-treat analysis. Eligibility criteria were provided for all six studies. The three split-mouth trials did not report adjustment for within-patient correlation of split-mouths or class clustering (Granath et al., 1979; Wright et al., 1979, 1980). Two studies reported industry support for the conduct of the study (Gisselsson et al., 1988; Wright et al., 1980). For one study, this support included materials, costs, and salaries (Banting, 2005), while for the other study the support was limited to materials (Birkhed, 2005). The four remaining studies reported government or professional organization support (Granath et al., 1979; Wright et al., 1979; Gisselsson et al., 1983, 1994).
Overall Summary (Fig. 2
Effectiveness Modifiers Flossing’s effectiveness depended significantly on whether studies reported the administration of topical fluorides and the assessment of compliance (p-value < 0.001). The four studies reporting delivery of topical fluoride or assessment of fluoride compliance had a slightly increased caries risk associated with flossing (Granath et al., 1979; Gisselsson et al., 1983, 1988, 1994), while the two studies reporting no assessment of topical fluoride exposure or compliance measures showed a highly significant flossing benefit (Wright et al., 1979, 1980). Oral hygiene measures were not significantly related to the effectiveness of flossing (p-value = 0.31). The flossing effectiveness depended significantly on the caries risk in the population, with high caries risk translating into non-effectiveness of flossing (p < 0.002). The interproximal caries risk was approximately 25% in two studies (Granath et al., 1979; Gisselsson et al., 1994), and ranged from 10% to 14% in the four remaining studies.
Subgroup Analyses
Sensitivity Analyses
A systematic review of the evidence on flossing indicates that professional flossing performed in first-grade children on school days reduced caries risk by 40%. This benefit was identified in predominantly the primary teeth in children who, it is assumed, had comparatively poor oral hygiene and minimal exposure to fluoride. When professional flossing was performed on a three-monthly basis, there was no evidence of a benefit, suggesting that infrequent flossing may be ineffective when it comes to caries control. When flossing was self-performed by young adolescents, even under supervision on school days, there was also no evidence of benefit, which may be due to the presence of fluorides, poor flossing techniques, or other reasons. No evidence on the effectiveness of floss in adults or under real-world clinical conditions could be identified. In particular, there was no evidence that flossing is effective in the presence of topical fluorides. Of the six trials that evaluated the effect of flossing on interproximal caries risk, two trials reported that professional flossing reduced caries risk (Wright et al., 1979, 1980). The strengths of these two studies include the large observed relative risk reduction of 40%, the statistical significance of the combined results, and the possibility that flossing benefits were underestimated because there was no professional flossing performed on weekends and the extended summer break, and, possibly, because parental flossing of control teeth diluted the professional flossing effect. Weaknesses include that both studies were not truly independent, since they were conducted by the same investigators, that financial support may have biased study findings (Wright et al., 1980), that minimal data were available on oral hygiene and fluoride exposure, and that a difference of 54 caries lesions in two studies combined is a small number on which to base universal flossing recommendations. If the benefits of flossing clustered within mouths more than we estimated, or if the baseline randomization was biased, the statistical significance of a flossing benefit could come into question. Four studies failed to identify a flossing benefit. In three of the studies, apparent straightforward reasons can be identified to explain the lack of a flossing effect. A sample size of 20 flossers, only eight of whom actually reported using floss more than every other day, doomed, in all likelihood, one study’s ability to find a flossing effect (Gisselsson et al., 1983). Three-monthly flossing in two other studies may have been too infrequent to provide a benefit (Gisselsson et al., 1988, 1994). The enigma is one split-mouth study showing a high caries rate, but no identifiable anti-caries benefit from supervised flossing (Granath et al., 1979). Like the other studies, this study had weaknesses, including no information on dropouts, insufficient details on statistical analyses, and no information on group randomization. The lack of an effect is puzzling, however. Possibly, moving the floss once through the contact point, as opposed to wrapping the floss around the tooth’s proximal surface and then moving it up and down to disrupt or remove the interproximal plaque, was not effective against caries. An alternative explanation is that young adolescents may have been more conscientious regarding their oral hygiene, and the increased frequency of topical fluoride exposure through toothbrushing may have eliminated any benefit of flossing. An important limitation of the current evidence on flossing is the inability to establish whether flossing provides a benefit above and beyond brushing with a fluoridated toothpaste (Rule et al., 1984; Conti et al., 1988). In the two studies where toothbrushing compliance at home was not assessed, and where no toothbrushing was performed under supervised conditions at school, a flossing benefit was observed (Wright et al., 1979, 1980). In the four studies where topical fluoride compliance was assessed or delivered under supervised conditions, flossing was non-effective (Granath et al., 1979; Gisselsson et al., 1983, 1988, 1994). This stark contrast between studies suggests that topical fluoride exposure may attenuate or eliminate the effectiveness of flossing. The effect of flossing was also limited to studies with low caries rates, raising the possibility that flossing, just like other caries-preventive agents, may be less effective in high-risk populations (Seppä et al., 1991; Forgie et al., 2000; Hausen et al., 2000; Kallestal, 2005). Nonetheless, these are leaps of faith, since many other factors, including flossing frequency and technique, also differed between the studies. Several observations suggest that the actual fluoride exposure in the two studies reporting a significant flossing benefit was low. The two studies were conducted on young (5–6 yrs old) children who are typically assumed to be "unable to perform satisfactory oral hygiene themselves" (Poulsen et al., 1976), and for whom "no other oral hygiene procedures or instruction was provided" (Wright et al., 1979, 1980). This assumption of lack-of-oral hygiene appears to be validated, since children using and not using fluoridated toothpaste had approximately the same number of caries lesions in one of the two studies (Wright et al., 1980). While the lack of a fluoride effect may have been a chance finding due to low power, or due to self-selection bias, it may also possibly have been due to lack of adequate brushing. In addition, the children were living in an area where fluoride levels in the water were low. As a result, there is a possibility that flossing may be effective in a situation where oral hygiene is poor and where exposure to fluorides is minimal. This assumption of poor oral hygiene and consequent low-fluoride exposure is plausible and is supported, in part, by the reported data, but remains an assumption nonetheless, since information on the actual oral hygiene levels and toothpaste characteristics in these two studies was not reported (Wright et al., 1979, 1980). Weaknesses of this systematic review relate to the inclusion/exclusion criteria, the focus of the research question, the statistical uncertainties present in calculating summary estimates, and lack of consideration for the caries lesion severity in the analysis. A rigorous pre-analysis definition of study inclusion/exclusion criteria appeared impossible because of the authors’ familiarity with the trials, and the high likelihood that apparently reasonable pre-synthesis established study inclusion criteria would have excluded most, if not all, of the available evidence. Additionally, this systematic review focused on the effectiveness of floss, not flossing and brushing. A systematic review on the latter topic would have included a study where the effect of brushing and flossing on interproximal caries risk was evaluated (Horowitz et al., 1977), and where, in the absence of fluoride, a marginally significant effect of plaque removal on interproximal caries risk was observed. The three split-mouth studies on flossing and caries (Granath et al., 1979; Wright et al., 1979, 1980) failed to report information that would have allowed variances to be estimated. Unless data from these split-mouth trials can be resurrected, we are unable to determine how (in)accurate our imputed values were. Finally, for this synthesis, caries lesions limited to the outer enamel, caries lesions into the dentin, and caries lesions in primary and permanent tooth were all considered to be similar events. The validity of this assumption has been insufficiently evaluated and must be considered when the findings are interpreted. An additional weakness of the systematic review is the lack of power. This has two consequences. First, possibly, real-world flossing has a modest impact on interproximal caries lesions, and only large studies will be able to identify these benefits reliably. The second consequence of the lack of adequately powered clinical trials on self-performed flossing is the lack of safety information. None of the six controlled trials assessed safety. It cannot be excluded that non-professional simple flossing (moving the floss through the interproximal contact point only), such as performed by the young adolescents in one study (Granath et al., 1979), increases the caries risk by 22% (22% is the upper 95% confidence interval observed in the study on self-flossing). Flossing may cause harm by disrupting from 2 to 3.5 mm of the epithelial cuff around the teeth (Waerhaug, 1981), and by damaging both tooth and periodontal structures (Ratcliff, 1966; Gow and Kelleher, 2003). Daily imperfect flossing may select for colonization by floss-resistant cariogenic strains that penetrate the tooth or the white-spot lesion, may transmit infections from one interproximal site to another, or may enhance pathogenic maturation of inaccessible or unremoved plaque (Loesche, 1993). Granath indicated that some have postulated that the viability of bacteria in a plaque follows a gradient, the outermost bacteria being most viable. Ineffective flossing stirs plaque around and might therefore be harmful if the less viable plaque is removed (Granath, 2005). The possibility for harmful effects of non-professional flossing should be assessed in future trials. A scientific double-standard exists in the evaluation of dental drugs and devices. Some dental devices, such as toothbrushes or floss, have largely escaped the rigorous scientific evaluation that is required for drugs. While the Food and Drug Administration and the American Dental Association indicate that dental floss may reduce caries risk (Food and Drug Administration, 2005), there are not two independent, randomized controlled trials demonstrating that self-performed flossing can reduce caries risk. The Council on Scientific Affairs of the American Dental Association suggests (Acceptance Program Guidelines, 2003) that interdental cleaning devices should be evaluated "under unsupervised conditions" and "by the average patient", conditions under which the effect of floss on caries has not been evaluated. There have been no trials showing that flossing prevents caries in adults in real-world clinical situations. The advocacy of flossing as a caries-preventive tool hinges in large part on apparent common sense. Since dental plaque is cariogenic, and since dental floss disrupts and removes some interproximal plaque (Waerhaug, 1981), flossing should reduce caries risk. Such a common-sense argument represents the lowest level of scientific evidence (Sackett et al., 2000). Common sense was wrong in claiming that knee debridement relieves osteoarthritic knee pain (Moseley et al., 2002), that optic nerve decompression prevents vision loss (Ischemic Optic Neuropathy Decompression Trial: twenty-four-month update, 2000), or that internal mammary artery ligation improves cardiovascular outcomes (Cobb et al., 1959). Several trials have also failed to support the common-sense argument that dental plaque removal lowers caries risk (Horowitz et al., 1977; McKee et al., 1977; Silverstein et al., 1977; Agerbaek et al., 1978; Ashley and Sainsbury, 1981), which led to the hypothesis that a mutans streptococci infection cannot be controlled by mechanical means (Loesche, 1993). One should be careful to justify flossing based on common-sense arguments, especially when other caries-preventive interventions are supported by higher levels of evidence. In summary, the controlled trial evidence on flossing and dental caries is challenging to interpret because of the inconsistent results across trials, the difficulty in extrapolating results of two trials conducted in children who differ substantially from typical floss-users, and the poor to moderate scientific quality of some of the reported studies. There have been no evaluations in real-world clinical situations, and, as a result, clinical recommendations have to be based on a level of evidence which would be questionable if flossing were a drug. The current low-level evidence is consistent with the hypothesis that regular and meticulous flossing can drastically lower interproximal caries risk in young children with poor toothbrushing habits and low fluoride exposure. Better toothbrushing and/or enhanced topical fluoride exposure may attenuate or eliminate this flossing effect. The dental professional should determine, on an individual patient basis, whether professional-quality flossing is an achievable goal, and to what extent a recommendation to floss may decrease the exposure time to caries interventions that are supported by better evidence. Factorial designs, where the effects of novel fluoride toothpastes and flossing devices are evaluated simultaneously, may provide a relatively low-cost opportunity to determine what fraction, if any, of interproximal cavities can be prevented by dental floss in a fluoridated world.
A supplemental appendix to this article is published electronically only at http://www.dentalresearch.org. Received for publication April 18, 2005. Accepted for publication November 8, 2005.
Journal of Dental Research, Vol. 85, No. 4,
298-305 (2006) This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
