The efficacy of baking soda dentifrice in controlling plaque and gingivitis: A systematic review

Abstract Objective To test the efficacy of a dentifrice containing baking soda (BS), compared with dentifrice without BS for controlling plaque and gingivitis. Materials and methods MEDLINE‐PubMed and Cochrane‐CENTRAL were searched. The inclusion criteria were randomized controlled clinical trials including healthy participants aged 18 years or older. Studies were selected that compared the effect of toothbrushing with a dentifrice with and without BS on the clinical parameters of plaque and gingivitis. Data were extracted from the selected studies, and a meta‐analysis was performed. Results The search retrieved 21 eligible publications. Among these papers, 43 comparisons were provided, with 23 involving a single‐use design and 20 being evaluations with a follow‐up. Negative controls were found, or positive controls for which various active ingredients had been used. The included studies showed a moderate overall potential risk of bias and considerable heterogeneity. The meta‐analysis of plaque scores from the single‐brushing experiments showed that BS dentifrice (BS‐DF) was associated with significantly better outcomes than the negative control dentifrices (DiffM −0.20; P < 0.0001; 95% CI: [−0.27; −0.12]) or the positive control dentifrices (DiffM −0.18; P < 0.0001; 95% CI: [−0.24; −0.12]). This finding was only confirmed in studies that used a follow‐up design as compared to a negative control (DiffM −0.19; P = 0.01; 95% CI: [−0.34; −0.04]). The indices of gingival bleeding also improved when the comparison was a negative control (DiffM −0.08; P = 0.02; 95% CI: [−0.16; −0.01] and (DiffM −0.13; P < 0.001; 95% CI: [−0.18; −0.08]. However, for the gingival index scores, the meta‐analysis did not reveal any significant differences. Conclusion BS‐DF showed promising results with respect to plaque removal in single‐use studies. However, the finding was partially substantiated in follow‐up studies. Studies that assessed bleeding scores indicated that a small reduction can be expected from BS, relative to a control product.


| INTRODUC TI ON
Dentifrice, also known as toothpaste, is used in conjunction with a toothbrush to help maintain oral health. The most common components of dentifrice are an abrasive agent, a binder, a surfactant and a humectant. The main intention of the use of paste is to help remove debris and plaque but it also has secondary functions such as breath freshening and tooth whitening, which are widely marketed. There is an almost universal recommendation that people should brush their teeth twice a day with a fluoridated dentifrice. 1 Several dentifrice manufacturers have incorporated sodium bicarbonate, commonly known as baking soda (BS), into their formulas. This is a salt composed of sodium ions and bicarbonate ions. BS is nontoxic and is mild on the soft tissues of the gums and oral mucosa.
In commercial dentifrices, BS mainly serves the purpose of an abrasive. Relative dentin abrasion tests have shown that the abrasiveness of sodium bicarbonate has low abrasivity of the tooth surface. It is an alkaline substance capable of neutralizing acids. As such, it potentially can prevent tooth decay by neutralizing the acids produced by bacteria in the mouth. 2 BS also neutralizes acidic components of common tooth-staining chemicals, such as the chromogens in tea, and red wine, 3 thereby lessening their staining potential.
The current widespread use of BS in dentifrices and home oral hygiene regimens is largely attributable to the impact of Dr Paul H.
Keyes. 4 In the 1970s, he was among the first to employ anti-infective agents and microbiological testing in non-surgical periodontal therapy, including patient home irrigation with BS or salt solutions, and brushing with a mix of BS and hydrogen peroxide. This approach is known as "the Keyes technique," popularly referred to as the "salt-and-soda" method. The method became widely integrated into people's oral hygiene routines. However, it was critically evaluated by the American Academy of Periodontology from which it was concluded that the benefits of the technique are almost exclusively derived from the detailed oral hygiene procedures and root planning. 4 Nowadays, BS is found in many dentifrices. In an era with upcoming preference for "assumed" naturally based products, 5 it is important to investigate the associated oral health benefits. Until this study, no systematic evaluation had been conducted on the adjuvant effect of sodium bicarbonate in dentifrices. The aim of this systematic review (SR) was to establish the effect of BS on plaque removal and gingivitis.

| MATERIAL S AND ME THODS
This SR was prepared and described in accordance with the Cochrane handbook for systematic reviews of interventions 6 and the guidelines in Transparent Reporting of Systematic Reviews and Meta-analysis (PRISMA-statement). 7 The protocol for this review was developed "a priori" and registered with the International Prospective Register of Systematic Reviews 8 under the registration number CRD42018080649. All post hoc changes were appropriately noted (see Appendix S1).

| Focused question
In healthy individuals, what is the efficacy of toothbrushing with a dentifrice that contains BS compared to a dentifrice without BS on clinical indices of plaque and gingivitis?  Table 1.

| Screening and selection
The titles and abstracts of the studies obtained from the searches were screened independently by three reviewers (AD, YK and CV) to select studies that potentially met the inclusion criteria. No language restrictions were imposed. Based on the title and abstract, the full-text versions of potentially relevant papers were obtained.
These papers were categorized (by CV and DES) as definitely eligible, definitely not eligible or questionable. Disagreements concerning eligibility were resolved by consensus, and if disagreement persisted, the decision was resolved through arbitration by another reviewer (GAW). Papers that fulfilled all the inclusion criteria were processed for data extraction.
The included full report studies were considered to meet the following criteria: (a) the study design was either a randomized controlled clinical trial (RCT) or a controlled clinical trial (CCT), (b) the studies were conducted with healthy participants, who were not institutionalized and were 18 years of age or older, (c) the studies included participants without orthodontic treatment and/or removable prostheses, (d) as an intervention, a dentifrice with BS was evaluated in comparison with a dentifrice without this ingredient, (e) chlorhexidine was not an ingredient incorporated in a dentifrice, baking soda, bleeding, dentifrice, gingivitis, plaque, sodium bicarbonate, systematic review, toothpaste (f) rinsing with an additional antiseptic was not a part of the intervention or control regimen and (g) the studies evaluated plaque and gingivitis scores. For details, see Appendix S2.

| Assessment of heterogeneity
The following factors were used to evaluate the heterogeneity of the outcomes of the different studies: study design, participant characteristics, study group details and regimens. In addition, side effects and industry funding were evaluated.

| Assessment of methodological quality and risk of bias
All included studies were independently scored for their methodological quality by three reviewers (AD, CV and YK). Disagreement was resolved by consensus, and if disagreement persisted, the decision was resolved through arbitration by a fourth reviewer (DES).
The assessed items are detailed in Appendix S3. 9

| Data extraction
The characteristics of the population, intervention, comparison and outcomes were extracted from all studies independently by two reviewers (AD and YK) using a specially designed data extraction form. A third reviewer (CV) also read the full texts of the included trials and, independently from the two others, checked the data extracted. Disagreement between the reviewers was resolved through discussion and consensus. If this was not satisfactory, the judgement of another reviewer (GAW) was decisive. Means and standard deviations (SDs) were extracted. Some studies provided standard errors (SEs) of the means. Where possible, the current authors calculated SD based on the sample size (SE = SD/√N). For those papers that provided insufficient data to be included in the analysis, the first and/or corresponding author was contacted to request additional data.

| Data analysis
Studies were categorized as single-brushing designs that were selected to evaluate a change in plaque scores. Studies with a followup were selected to evaluate plaque as well as gingivitis scores. The dentifrices without BS were separated into negative and positive controls. As a positive control, dentifrices containing stannous fluoride (SnF) or triclosan (Tcs) as ingredients were considered. 10,11 All the other dentifrices without BS were considered as negative controls. As a summary, a descriptive data presentation was used for all studies. Where feasible, a meta-analysis (MA) was performed with at least two included experiments evaluating the same outcome parameter.
When a study had multiple non-BS dentifrice treatment arms, and data from the BS-DF were used in more than one comparison, the number of participants (n) in that group was divided by the number of comparisons. The difference of means (DiffM) between the test and control groups was calculated using a "random effects" model with an "inverse variance" method as proposed by DerSimonian and Laird. 12 The primary method of calculating all pooled estimates and a sub-analysis was performed with the Knapp- Hartung adjustment 13 in cases of at least five eligible studies. 14 For meta-analyses with more than two comparisons, 95% predictive intervals were calculated to quantify treatment effects in a future clinical setting. 15 Heterogeneity was tested using the chi-square test and the I 2 statistic. A chi-squared test resulting in P < 0.1 was considered to be an indication of significant statistical heterogeneity. If possible, the formal testing for publication bias using the minimum amount of 10 comparisons was applied, as proposed by Egger et al 16  A sub-analysis was performed using a network meta-analysis (NMA). [18][19][20][21][22] Treatments were ranked 23,24 through a frequentist weighted least squares approach, as described by Rücker. 25,26 The direct evidence proportion as described in König et al 27 was used to calculate the indirect evidence. 26 A decomposition of heterogeneity within designs and between designs was provided, 28 and a net heat plot graphical tool, as proposed by Krahn et al, 29 was used to locate inconsistency in the NMA. 29 For the transitivity assumption, 11,30 the ingredients were analysed. All computations were performed using R (https://www.r-project.org) with the packages meta, 31 metafor 32 and netmeta. 26

| Grading the "body of evidence"
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system was used to rank the evidence. 33 Two reviewers (CV and DES) rated the quality of the evidence and the strength and direction of the recommendations 34 according to the following aspects: risk of bias, consistency of results, directness of evidence, precision and publication bias and magnitude of the effect.
Any disagreement between the two reviewers was resolved after additional discussion with a third reviewer (GAW).

| Heterogeneity
The included studies exhibited considerable heterogeneity with respect to the study design, participant characteristics, study group details and the regimens used. Information regarding the study outline and characteristics is shown in detail in Appendix S2.

| Adverse effects
Twelve papers 5,[36][37][38]44,45,48,50 mentioned evaluation for possible adverse effects. Only in one study did four participants discontinue the study because of disliking the dentifrice taste. 45 In another study, the participants complained of an unpleasant taste in the initial period when using a BS-DF. 36 Ulcerations were reported in one study, but they appeared unrelated to the trial and eventually. 5 In Winer et al, 39 two persons were dropped out of the experimental group, which was suggested to be product-related. In one study, a participant experienced a mild burning sensation and moderate dental hypersensitivity. 54

| Industry funding
Most of the 21 included studies reported on the use of commercially available dentifrices and toothbrushes. For three studies, it was unclear whether the dentifrices were marketed products. 35,39,40 Five other studies used non-marketed experimental dentifrices. 38,42,45,52,54 Fifteen studies had industry involvement, with seven different companies acting either as a study initiator or where the authors were employees; companies also provided products, funding or financial grants. Five studies did not mention industry connections and one study included a disclosure statement of no financial interest. 39

| Methodological quality and assessment of bias
To estimate the potential risk of bias, the methodological qualities of the included studies were used, as assessed in the checklist presented in Appendix S3 (methodological quality and potential risk of bias scores of the individual included studies). Based on a summary of the proposed criteria, the estimated potential risk of bias was low for nine studies, 5,37,38,45,48,51,53,54 moderate for six studies [42][43][44]47,50 and high for six studies. 35,[39][40][41]46

| Study outcome results
Appendix S4 presents the results of the data extraction. Baseline scores, end scores and incremental changes within each intervention group are presented. Table 2 provides a descriptive summary of the significant differences between toothbrushing with a BS-DF and without BS as reported by the original authors. In all but one of the 23 comparisons that presented results using the single-brushing design, when BS-DF was compared to either a negative control or a positive control, it was found to be significantly more effective for plaque removal (Table 2A).

| Descriptive analysis
Plaque index scores obtained from the majority of studies with a follow-up showed a pattern of no difference between BS-DF and their controls. An inconsistent pattern was noted for the gingival index scores and for bleeding scores when a negative control was used. In two of the three comparisons that used Tcs-DF as a positive control, BS-DF showed a significant improvement in plaque removal.
This could not be confirmed with respect to gingival health scores.
The only study that used SnF-DF as a positive control showed that it was significantly more effective than BS-DF regarding the gingival index and bleeding scores (Table 2B).

| Meta-analysis
The obtained data allowed for several meta-analyses, which could be performed separately for plaque, gingivitis and bleeding index scores. The results are summarized in Table 3. The forest plots and corresponding funnel plots that illustrate these outcomes are shown in Appendix S6-S12. For the studies that evaluated a single-use de- The findings of the MA were supported by the NMA when the heterogeneity and the inconsistency across networks were accounted for. For details of the NMA results, see Appendix S11. Table 4 presents a summary of the various factors used to rate the quality of evidence and to appraise the strength and direction of recommendations according to GRADE. 33 There is evidence from singlebrushing studies to support the use of BS as an ingredient for improving plaque removal. However, because of the fact that this design does not replicate home use, it is considered indirect evidence. With a moderate precision, the strength and direction of the recommendation based on single-use studies were estimated to be "weakly in favour."

| E VIDEN CE PROFILE
No difference was determined for plaque scores and gingivitis index scores in studies with a follow-up. However, based on the statistically significant difference in means and the prediction intervals, in future studies, a small difference in bleeding scores between controls and experimental participants can be expected. Given the strength of the recommendation, there is a "moderate" certainty that the BS-DF did not provide an additional benefit in the studies with a follow-up. The efficacy of BS-DF is comparable to that of other commercially available dentifrices.

| D ISCUSS I ON
From a previously published meta-review on the evidence for dentifrices, it appeared that there was a lack of a systematic appraisal of the evidence concerning the efficacy of BS-DF. 74 Therefore, the aim of this systematic review was to assess the effect of toothbrushing with a BS-DF on plaque and the clinical parameters of gingivitis. Data were extracted from 21 studies which included 2517 participants. The present SR shows, based on the single-brushing experiments, a small but significant improvement of plaque removal when toothbrushing is performed with a BS dentifrice. However, no favourable effect of BS on plaque scores was found in studies with a follow-up when the prediction interval was considered. On the other hand, follow-up studies have shown that on bleeding scores a small effect with a 95% prediction interval including the null or opposite direction can be expected from the use of BS.
The MA in this review differentiated between single-use brushing exercises and the longer-term effect of brushing, in order to eliminate design-related differences. Also, it distinguished between negative control dentifrices and proven positive control dentifrices such as those containing Tcs and SnF. Additionally, indirect and direct evidence was combined in a NMA to provide a more precise estimates of treatment effects. 19,75,76 However, in the NMA of studies with a follow-up, problems of heterogeneity and potential inconsistency are present which emphasizes that conclusions about ranking should be carefully interpreted.
Interestingly, in the descriptive summary (see Table 2) of the studies with a follow-up, the results were not in favour of Tcs or SnF, with the only exception in one comparison when the control was TA B L E 2 A descriptive summary of the statistical significance of individual study outcomes for the single-brushing and long-term studies. (A) Descriptive summary of the single-brushing dentifrice comparisons; (B) Descriptive summary follow-up dentifrice comparisons SnF-DF. 46 In systematic reviews evaluating the efficacy of Tcs or SnF, these active ingredients generally showed better results for plaque and gingival index scores than conventional dentifrices. 10,11,79,80 In ranking the treatments in studies with a follow-up according to the NMA (see Appendix S11), the efficacy of Tcs or SnF was in line with the findings of the systematic reviews discussed above.
In addition to the difference of means (DiffM) and 95% confidence intervals, we calculated 95% prediction intervals. The advantage of prediction intervals is they reflect the variation in treatment effects across different settings, including what effect is to be expected in future patients. 15 The prediction interval of the singlebrushing studies indicated that in a future single-brushing setting, the difference in means for end plaque scores would likely be between −0.28 and −0.11 (Table 3A) if compared with a negative control as recorded on a five-point scale according to TQ&H. Compared to a positive control, this will most likely be between −0.33 and −0.03 (Table 3A). Consequently, the probability that in future studies the effect size is less than the threshold 0 is quite certain for both negative and positive controls. 15 The favourable effect of BS on plaque was not substantiated in studies with a follow-up.  The baseline, end and difference data evaluating a dentifrice with BS compared to a negative either a positive control dentifrice; (B) Follow-up studies compared to a negative or positive control dentifrice. (B1) Meta-analysis for follow-up brushing design studies for the baseline, end and difference data evaluating a dentifrice with BS compared to a negative either a positive control dentifrice according to plaque indices; (B2) Meta-analysis for follow-up brushing design studies for the baseline, end and difference data evaluating a dentifrice with BS compared to a negative either a positive control dentifrice according to GINGIVAL Index; (B3) Meta-analysis for follow-up brushing design studies for the baseline, end and difference data evaluating a dentifrice with BS compared to a negative control dentifrice according to BLEEDING indices; (C) Sub-metaanalysis on ingredients for follow-up brushing design dentifrice comparisons end data evaluating a dentifrice with BS compared to a negative control dentifrice. plaque indices for the Turesky

P-value
Negative control Yankell et al (1993) 42 Mullally et al (1995) 5 Saxer et al (1995) 45 Triratana et al (2015) 51 Hosadurga et al (2017) 36 Jose et al (2018) 54 Akwagyiram et al (2018)  Yankell et al (1993) 42 Mullally et al (1995) 5 Saxer et al (1995) 45 Beiswanger et al (1997) 46 Al   39 Yankell and Emling A (1988) 40 Yankell (1993) 42 Saxer (1995)   Putt et al 49 investigated the effect of varying concentrations of BS. The results suggested a possible positive relationship between the concentration of BS and plaque reduction. In the ranking of the NMA results (Online Appendix S11-C), such a pattern is clearly visible. An almost inverse relationship exists between the percentage of BS in a dentifrice and its abrasiveness. 86 However, a key difference between BS and common abrasives is the size of the particles. 49 In the case of BS crystals, the particles are notably larger, softer and potentially less damaging to tooth mineral than the conventional abrasive particles in other dentifrices. This size could play a role in disturbing the adhesion of plaque to the tooth surface, in addition to the force exerted by the toothbrush. 49 The dissolved bicarbonate ions in BS are thought to bind with calcium ions, disrupting the mutual bond between bacteria and disrupting the attachment of bacteria to the tooth surface. These bicarbonate ions are also thought to charge the tooth surface negatively, which enhances the detachment of bacteria. 50 Furthermore, BS is an alkali, which boosts the cleansing activity of the surfactant in the dentifrice. 50 Although these proposed mechanisms are promising, BS is known to be easily soluble and slow-acting. Therefore, it is unlikely to reside long enough in the mouth to actually inhibit plaque growth. 49,86,87 A recent systematic review found, with moderate certainty, that the adjunctive use of a standard fluoride dentifrice with toothbrushing did not contribute to the effectiveness of mechanical removal of dental plaque, in single-brushing experiments. 74 Given that in single-use studies, BS showed a positive effect in instant plaque removal, incorporating this ingredient into novel dentifrices seems therefore an interesting approach to improve a product.
Other results indicate that BS in dentifrice is an effective buffering agent through its ability to increase pH to a safe, neutral level. 2,89 A long-term in situ crossover study showed that BS did not significantly enhance the ability of fluoride dentifrice to reduce demineralization and increase remineralization of the enamel. Most BS-based dentifrices contain fluoride, which is compatible with BS. 90,91 Findings from in vitro studies suggest, however, that adding BS to a dentifrice may interfere with the reactivity of fluoride with enamel, reducing mainly the concentration of CaF2 formed. 92,93 This indicates that adding BS to dentifrice requires careful formulation.

| Limitations
An important limitation of this review was the variability of dentifrice formulations in the included papers. The composition of the studied dentifrices was often not clear. The choice of a control dentifrice with which to compare dentifrices formulated for plaque control is also important and could affect conclusions drawn from clinical trials of such products. 94 In most studies, great effort was invested in creating neutral packaging. However, the unique properties of BS make blinding relatively hard. BS has the reputable property of an odd taste and texture. 49 This makes a BS-DF easily distinguishable, especially to participants who are accustomed to a regular fluoride dentifrice.
Only 6 out of 21 studies provided information about allocation concealment, a critical design feature to minimize bias.
The majority of the studies were published between 1986 and 2011, and in most cases, the manner of reporting did not follow current standards, such as TIDieR 2014. 95 Also, more recently published studies contain data from over 5 years ago. 53 This limitation is also reflected in the results of the risk of bias assessment. However, all groups seem to have been treated equally and in most of the studies seem to have been well balanced. Unclear in the included studies were the instructions on brushing duration and brushing frequency, details concerning the toothbrushes and the study procedures.

| CON CLUS ION
BS-DF showed promising results with respect to plaque removal in single-use studies. However, the finding was partially substantiated in follow-up studies. Studies that assessed bleeding scores indicated that a small reduction can be expected from BS, relative to a control product.

| Scientific rationale for the study
Twice daily toothbrushing with a fluoride dentifrice is a universal recommendation for personal oral care.

| Principal findings
With moderate certainty, a dentifrice containing BS is comparable to other commercially available dentifrices for controlling plaque and gingivitis.

| Practical implications
In order to remove plaque and improve gingival health, toothbrushing can be combined with a baking soda dentifrice.

ACK N OWLED G EM ENTS
The authors gratefully acknowledge the support of Joost Bouwman,

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest. Van der Weijden, Slot and their research team at ACTA have previously received either external advisor fees, lecturer fees or research grants from toothbrush and dentifrice manufacturers. Those manufacturers included: Colgate, Dentaid, GABA, GSK, Lactona, Oral-B, Procter & Gamble, Sara Lee, Sunstar and Unilever. Ethical approval was not required.