Impact of Interdental Brushing on Pregnancy‐Associated Gingivitis: A Secondary Analysis of a Randomised Controlled Trial

ABSTRACT

Aim

To assess whether daily use of calibrated interdental brushes can reduce gingival bleeding in pregnant women at high risk of preeclampsia.

Materials and Methods

In this multi‐centre randomised controlled trial, 323 nulliparous women at risk of preeclampsia and with an intact periodontium were enrolled at 12 weeks of gestation. Participants were randomly allocated (1:1) to either: (i) the intervention group receiving individualised interdental hygiene instructions and calibrated interdental brushes for daily use, or (ii) the control group: receiving routine antenatal care. All participants were recalled at 1 week after baseline, 4, 5 and 8 months of pregnancy for assessments of bleeding on probing (BOP) and other periodontal indices. Linear mixed models were used to assess changes in bleeding over time, and logistic regression was used to evaluate predictors of bleeding resolution at the final follow‐up.

Results

At 8 months of pregnancy, the intervention group showed a significant reduction in gingival bleeding, from 56% at baseline to 12% (−79.9%, p < 0.001), whereas the control group remained stable. Most of the improvement occurred during the first week (−56.8%). Among participants with severe baseline bleeding (> 81%), the reduction reached 84.3% (p < 0.05). Longitudinal analysis confirmed sustained reductions at each follow‐up. Interdental brushing was the strongest independent predictor of bleeding reduction (OR = 3.14; 95% CI: 2.01–4.90).

Conclusions

Daily use of calibrated interdental brushes, introduced early in pregnancy, significantly reduced gingival bleeding with rapid and sustained effects. These findings support the feasibility and clinical relevance of individualised interdental hygiene as a preventive strategy within standard antenatal oral health care.

Trial Registration

ClinicalTrials.gov NCT04989075 (https://clinicaltrials.gov/study/NCT04989075)

1. Introduction

Pregnancy‐associated gingivitis affects up to 70% of pregnant women and is clinically significant because of the heightened immune response driven by hormonal and immunological changes during gestation (Gare et al. 2023; Chen et al. 2022; Cornejo Ulloa et al. 2024). These endocrine fluctuations promote vascular alterations and oral microbiota shifts, increasing gingival bleeding despite stable plaque levels (Saadaoui et al. 2021; Ye and Kapila 2021). According to the 2018 EFP/AAP classification, pregnancy‐associated gingivitis is defined as a plaque‐induced condition modified by systemic factors (Chapple et al. 2018). Hormonal changes amplify the host immune response to dental biofilm (Zhao et al. 2025; Carrouel et al. 2016), leading to dysregulation characterised by excessive neutrophil recruitment, impaired clearance and sustained release of proinflammatory cytokines such as interleukin (IL)‐1β, IL‐6 and TNF‐α (Carrouel et al. 2016; Ng and Lim 2019). Subclinical immune imbalance between pro‐ and anti‐inflammatory pathways, particularly within the Th17/Treg axis, and elevated matrix metalloproteinase activity contribute to persistent gingival inflammation during pregnancy, even when plaque levels remain stable (Zhao et al. 2025). These mechanisms explain the increased gingival bleeding and inflammation observed during pregnancy and emphasise the relevance of interdental cleaning in this specific context, given the potential implications of oral inflammation for adverse pregnancy outcomes (Saadaoui et al. 2021; Ye and Kapila 2021; Zhao et al. 2025).

Interdental sites are particularly susceptible to periodontal inflammation because of their anatomical complexity and colonisation by keystone pathogens such as Porphyromonas gingivalis , Treponema denticola and Tannerella forsythia (Carrouel et al. 2016). These microorganisms promote local dysbiosis and contribute to gingival inflammation by triggering the host's immune responses. Calibrated interdental brushes (IDBs), defined according to ISO 16409:2016, have been shown to outperform floss in reducing interdental plaque, bleeding on probing (BOP) and gingival inflammation (Ng and Lim 2019; Chang et al. 2023). Mechanical disruption of interdental biofilm with IDBs is also associated with reduced bacterial load as well as local expression of proinflammatory mediators (Conti‐Ramsden et al. 2024; Tonetti et al. 2018). However, the clinical and immunological consequences of these mechanisms during pregnancy remain poorly understood.

To date, no randomised controlled trial (RCT) has evaluated whether early initiation of individualised interdental cleaning during gestation can reduce gingival inflammation and modulate the immune response. Addressing this knowledge gap is particularly relevant because pregnancy is a period of increased inflammatory sensitivity and susceptibility to periodontal disease progression, with potential implications for adverse pregnancy outcomes (Trombelli et al. 2018; Heitz‐Mayfield 2024). This trial specifically targeted pregnant women at high risk of preeclampsia, with a longitudinal follow‐up from 12 weeks of gestation to delivery, a population for whom preventive oral care strategies may be especially beneficial.

To explore this question, a secondary analysis of a multi‐centre RCT in Senegal assessed the impact of personalised interdental hygiene instruction combined with daily use of calibrated IDBs on gingival inflammation during pregnancy. The primary gingival outcome was BOP at the eighth month of gestation, with longitudinal changes as secondary gingival outcomes. The underlying hypothesis was that early and sustained mechanical disruption of interdental biofilm would reduce gingival bleeding throughout pregnancy.

2. Methods

2.1. Study Design and Setting

This multi‐centre, single‐blind RCT with two parallel groups (1:1) was conducted from March 2022 to July 2023 in six antenatal care clinics in Dakar, Senegal. Ethical approval was obtained (protocol 000086/MSAS/CNERS/SP, approved on 8 June 2021). All participants provided written informed consent. The study was registered at ClinicalTrials.gov (NCT04989075) and followed the Declaration of Helsinki and CONSORT guidelines. This analysis concerns a secondary objective of the trial which was initially designed to evaluate the effect of interdental hygiene on preeclampsia risk.

2.2. Participants

Women were screened at their first prenatal visit by a gynaecologist. Risk of preeclampsia was defined by nulliparity and sub‐Saharan African origin, both recognised epidemiological risk factors (Chang et al. 2023; Conti‐Ramsden et al. 2024). Eligible women were invited to participate, and after consent, a calibrated periodontologist performed an oral examination.

2.2.1. Inclusion Criteria

The participants were (i) pregnant women, (ii) aged 18–40 years, (iii) of sub‐Saharan African origin, (iv) nulliparous, (v) attending their first antenatal consultation at 12 weeks pregnancy and (v) able and willing to provide written informed consent and comply with procedures.

2.2.2. Exclusion Criteria

Women were excluded for (i) signs of foetal distress, (ii) known congenital anomalies of the uterus or vagina, (iii) systemic or infectious diseases, (iv) medically indicated early termination of pregnancy, (v) diagnosis of periodontitis, defined as interproximal CAL ≥ 1 mm at ≥ 2 non‐adjacent teeth, or buccal/oral CAL ≥ 3 mm with probing depth > 3 mm at ≥ 2 teeth, not attributable to non‐periodontitis‐related causes (Tonetti et al. 2018), (v) a history or treatment of periodontitis, (vi) current dental or orthodontic care, absence of the four premolar–molar pairs, < 20 natural teeth (excluding third molars), (vii) ongoing use of medications known to affect gingival or oral mucosal health, (viii) regular use of interdental cleaning tools (such as IDBs or floss) or mouthwashes or (ix) cognitive or behavioural limitations impairing informed participation.

2.3. Randomisation and Masking

Participants were randomly assigned (1:1) to the intervention (calibrated IDB) or the control (routine oral hygiene) group. The sequence was generated by an independent investigator using e‐CRF Voozalyon 1.3 (Voozanoo, Caluire, France), stratified by centre and implemented through a centralised system with randomly permuted blocks. Group allocation was masked to investigators, outcome assessors and statisticians.

2.4. Intervention

The intervention group received a personalised kit of calibrated IDBs (Curaprox CPS, Curaden, Switzerland), selected according to the interdental space. Participants were instructed to perform daily evening cleaning, and demonstrations were given at baseline. At each visit, new, appropriately sized brushes were provided, and compliance was monitored with a standardised booklet. Controls maintained routine oral hygiene, defined as toothbrushing with toothpaste, without interdental devices or mouthrinses. Both groups attended follow‐up visits (Figure S1).

At T1 (3 months +1 week), T2 (4 months), T3 (5 months) and T4 (8 months), periodontal assessments were carried out, which included bleeding on probing (BOP), clinical attachment loss (CAL), probing depth (PD), gingival index (GI) and plaque index (PI). At T0, T3 and T4, body mass index (BMI), pregnancy status, fundal height and blood pressure were also recorded.

Attrition was defined as participants lost to follow‐up with no data after baseline or final visit.

2.5. Clinical and Contextual Assessments

2.5.1. Clinical Definitions

Plaque‐induced gingivitis was defined as BOP at ≥ 10% of sites, with no site showing both BOP and PD ≥ 4 mm, per Trombelli et al. (2018).

2.5.2. Clinical Assessments

Six calibrated examiners from the Department of Periodontology, Dakar, performed the examinations after a calibration process (File S1). Assessments included full‐mouth PD, BOP and CAL at six sites per tooth (mesio‐buccal, buccal, disto‐buccal, mesio‐lingual, lingual, disto‐lingual) using a sterile US Williams DT sensor probe (Zila‐Pro‐Dentec Inc., Batesville, AR, USA) with standardised 20 g pressure. Measurements were recorded at five time points (T0–T4). For each participant, BOP was calculated as the percentage of bleeding sites. BOP was the pre‐specified gingival outcome, consistent with the 2018 EFP/AAP Classification, and was considered the most relevant marker of gingival inflammation (Heitz‐Mayfield 2024), particularly in preventive settings. GI and PI were also recorded at baseline, and the extent of gingivitis was classified according to the 2018 EFP/AAP criteria: healthy (< 10% BOP), localised (10%–30%) or generalised (> 30%).

2.6. Covariates

Covariates were selected a priori from the literature for their clinical relevance to periodontal and pregnancy outcomes. Age (years) and education were recorded at baseline. Obstetricians assessed gestational age (weeks), systolic and diastolic blood pressure (mmHg), height and weight (BMI kg/m²) and fundal height (mm) at 3, 5 and 8 months. Toothbrushing frequency (times/day) was self‐reported.

2.7. Outcomes

The primary gingival outcome was the proportion of sites with gingival bleeding at the eighth month of pregnancy (T4). Secondary gingival outcomes were changes in gingival bleeding over time (T1, T2 and T3) relative to baseline (T0).

2.8. Sample Size

Although the trial was powered for preeclampsia, a theoretical post hoc calculation was also performed for the BOP outcome to confirm sample adequacy. Assuming a 70‐point difference between groups, with α = 0.05 (two‐sided), 90% power and 20% attrition, at least 144 participants per group were required. The final sample (n = 323) provided adequate power for exploratory analyses of BOP differences and adjusted models.

2.9. Statistical Analysis

Analyses were performed in R (version 4.4.0). The prespecified gingival endpoint was BOP at 8 months (intention to treat; ITT). Longitudinal and subgroup analyses were exploratory and not adjusted for multiplicity. Group comparisons used nonparametric tests, and logistic regression assessed predictors of improvement (details in File S1).

A post hoc sensitivity analysis in the per‐protocol population (complete bleeding data T0–T4) was performed consistency with ITT findings. Statistical significance was set at two‐sided p < 0.05.

3. Results

Figure 1 shows the participant inclusion flowchart.

FIGURE 1.

Flowchart of the study: LAUS, logistical absence unrelated to the study.

3.1. Baseline Subject Characteristics

The 323 participants were evenly distributed between the active (n = 162) and control (n = 161) groups (Table 1). Baseline characteristics, analysed according to the ITT method, were comparable across groups, indicating balanced allocation. However, the plaque index was significantly greater (0.73 ± 0.62 vs. 1.04 ± 0.72; p < 0.001). No smoking subjects were registered. At T4, crude attrition reached 36.4% in the intervention group and 43.5% in the control group (p = 0.20). After excluding pregnancy interruptions, the adjusted attrition was 20.8% and 34.1%, respectively (p = 0.015), reflecting higher retention in the intervention arm. The AARs for intermediate visits are presented in Table S1. The overall adherence to IDB use, defined as the proportion of study days with recorded use, was 87.0% in the active group. No adverse effects were recorded.

TABLE 1.

Baseline characteristics of the study groups (intention‐to‐treat analysis).

	Active	Control	p
Age (years)	n = 162	n = 161	0.098
Mean ± SD	24.22 ± 4.63	23.48 ± 4.56
Median; range (IQR)	23.5 (20.25–27.0)	23.0 (20.0–25.0)
BMI (kg/m2)	n = 156	n = 146	0.231
Mean ± SD	24.2 ± 6.45	23.47 ± 5.97
Median; range (IQR)	22.69 (20.49–25.76)	22.31 (19.83–25.1)
Gestational age (weeks)	n = 157	n = 150	0.010
Mean ± SD	12.47 ± 0.76	12.38 ± 1.85
Median; range (IQR)	12.0 (12.0–13.0)	12.0 (12.0–12.0)
Fundal height (mm)	n = 154	n = 104	0.367
Mean ± SD	10.67 ± 1.65	11.08 ± 2.23
Median; range (IQR)	11.0 (10.0–12.0)	11.0 (10.0–12.0)
SBP (mmHg)	n = 158	n = 150	0.579
Mean ± SD	112.75 ± 11.77	113.45 ± 11.78
Median; range (IQR)	112.0 (105.0–119.0)	112.0 (106.25–120.0)
DBP (mmHg)	n = 158	n = 150	0.201
Mean ± SD	72.8 ± 7.72	71.7 ± 8.67
Median; range (IQR)	72.0 (68.0–78.0)	71.0 (66.0–78.0)
Bleeding (%)	n = 162	n = 161	0.223
Mean ± SD	0.55 ± 0.28	0.59 ± 0.28
Median; range (IQR)	0.56 (0.36–0.81)	0.67 (0.41–0.81)
Gingival index	n = 158	n = 151	0.930
Mean ± SD	0.48 ± 0.54	0.50 ± 0.63
Median; range (IQR)	0.32 (0.0–0.67)	0.32 (0.0–0.75)
Plaque index	n = 158	n = 151	< 0.001
Mean ± SD	0.73 ± 0.62	1.04 ± 0.72
Median; range (IQR)	0.62 (0.25–1.04)	1.0 (0.46–1.39)
PD (mm)	n = 159	n = 153	0.074
Mean ± SD	2.26 ± 0.53	2.36 ± 0.56
Median; range (IQR)	2.26 (2.36–2.37)	2.5 (2.0–2.78)
CAL (mm)	n = 158	n = 151	0.169
Mean ± SD	2.31 ± 0.5	2.31 ± 0.74
Median; range (IQR)	2.28 (1.89–2.68)	2.43 (2.0–2.84)
Gingival diagnosis a	n = 162	n = 161	0.589
Healthy	15 (9.3%)	10 (6.2%)
Localised gingivitis	18 (11.1%)	19 (11.8%)
Generalised gingivitis	129 (79.6%)	132 (82.0%)
Education level	n = 162	n = 161	0.226
< 7 years	50 (30.9%)	61 (37.9%)
≥ 7 years	112 (69.1%)	100 (62.1%)
Toothbrushing frequency	n = 162	n = 161	0.26
1/day	29/162 (17.9%)	38/161 (23.6%)
≥ 2/days	133/162 (82.1%)	123/161 (76.4%)

Abbreviations: BMI, body mass index; CAL, clinical attachment loss; DBP, diastolic blood pressure; PD, probing depth; SBP, systolic blood pressure.

^a Gingival diagnosis was defined according to the 2018 EFP/AAP classification: healthy (< 10% of sites with BoP), localised gingivitis (10%–30%), and generalised gingivitis (> 30%).

3.2. Evolution of Gingival Bleeding From T0 to T4

Table 2 presents the evolution of gingival bleeding. At baseline, the median degree of gingival bleeding was 0.56 (0.36–0.81) in the active group and 0.67 (0.41–0.81) in the control group. In the active group, bleeding decreased rapidly: at T1, the median decreased to 0.26 (0.15–0.37), corresponding to a −56.8% reduction (p < 0.001). Further reductions were observed at T2 (−28.7%), T3 (−23.1%) and T4 (−18.2%). Overall, the median change from baseline to T4 was −0.40 (−0.63 to −0.26), corresponding to a total decrease of −79.9% (−92.8% to −65.6%) (p < 0.001).

TABLE 2.

Evolution of gingival bleeding (ITT analysis).

	Active (n = 162)	Control (n = 161)	p a
T0
Median (IQR)	0.56 (0.36–0.81)	0.67 (0.41–0.81)	0.202
T1
Median (IQR)	0.26 (0.15–0.37)	0.63 (0.48–0.85)	< 0.001
T1–T0 Median diff (IQR)	−0.30 (−0.41 to −0.14)	0.04 (0.02–0.10)
p b	< 0.001	< 0.001
% Δ T1–T0 median (IQR)	56.8% (−69.4% to −37.9%)	5.3% (−4.0% to 17.6%)
T2
Median (IQR)	0.19 (0.07–0.30)	0.64 (0.44–0.85)	< 0.001
T2–T1 Median diff (IQR)	−0.06 (−0.15 to 0.00)	0.00 (−0.04 to 0.04)
p b	< 0.001	0.767
% Δ T2–T1 median (IQR)	−28.7% (−50.0% to 0.0%)	0.0% (−11.0% to 9.9%)
T3
Median (IQR)	0.11 (0.04–0.22)	0.64 (0.48–0.89)	< 0.001
T3–T2 Median diff (IQR)	−0.04 (−0.08 to 0.00)	0.00 (−0.09 to 0.07)
p b	< 0.001	0.988
% Δ T3–T2 median (IQR)	−23.1% (−54.5% to 0.0%)	0.0% (−17.6% to 15.7%)
T4
Median (IQR)	0.12 (0.04–0.17)	0.66 (0.00–0.07)	< 0.001
T4–T3 Median diff (IQR)	−0.03 (−0.07 to 0.04)	0.00 (−22.9 to 0.04)
p b	< 0.004	0.331
% Δ T4–T3 median (IQR)	−18.2% (−50.1% to +50.0%)	0.0% (−13.3% to 14.3%)
T4–T0
Median diff (IQR)	−0.40 (−0.63 to −0.26)	0.04 (−0.17 to 0.13)	< 0.001
p b	< 0.001	0.685
% Δ T4–T0 median (IQR)	−79.9% (−92.8% to −65.6%)	5.3% (−31.1% to 25.8%)
Mean difference (95% CI)	–0.43 (–0.49 to –0.36)
MLM p value c	< 0.001

Note: Median gingival bleeding values, median differences (with IQRs), percentage changes and p‐values for within‐ and between‐group comparisons at each visit.

^a Group comparison used the Mann–Whitney U test.

^b Within‐group changes used the Wilcoxon signed‐rank test.

^c MLM, mixed linear model.

In the control group, bleeding levels remained stable throughout the study. The median values at T1 (0.63 [0.48–0.85]) and T4 (0.66 [0.34–0.84]) were not significantly different from those at baseline. The overall change from baseline to T4 was +0.04 (−0.17 to +0.13), corresponding to a non‐significant variation of +5.3% (−31.1% to +25.8%) (p = 0.685).

The between‐group difference was significant at T1 and remained significant at each subsequent visit (p < 0.001). The mean difference in change from baseline to T4 between groups was −0.43 (−0.49 to −0.36). A mixed linear model confirmed the interaction between treatment and time (p < 0.001), indicating a stronger effect in the active group.

Figure 2 summarises the intervention effect over time. The active group showed a rapid decline in gingival bleeding as early as T1, accounting for most of the total reduction observed by T4. The IQR narrowed progressively, indicating both a lower mean and a more homogeneous response among participants. In contrast, the control group showed no substantial variation across visits, with medians oscillating around 0.67. From T1 onwards, the separation between groups became increasingly pronounced, reaching a median difference of −0.54 at T4 (p < 0.001). This pattern underscores the rapid onset and sustained effect of the intervention while highlighting the absence of spontaneous improvement in the control arm.

FIGURE 2.

Change in gingival bleeding from baseline (T0) to the final follow‐up (T4). Box plots show the median, interquartile range (IQR) and distribution of bleeding percentage at each time point, separately for the active group (red) and the control group (orange).

The post hoc subgroup analysis of participants with severe baseline bleeding (BOP > Q3) is presented in Tables S2 and S3 and Figure S2. ‘Severe bleeding’ refers to participants in the upper quartile (> Q3) of the baseline BOP distribution and is not a formal disease classification. Among women with high baseline bleeding (BOP > 0.81), the active group (n = 40) showed a significant 84.3% reduction (p < 0.05).

Figure  3 shows the distribution of gingival bleeding over time, stratified by baseline clinical extent (healthy, localised, generalised). In the healthy category, violin plots remained narrow and low in both groups across all time points, indicating clinical stability and minimal variation. In the localised category, the active group displays a progressive downward shift in bleeding values, accompanied by a visible thinning of the violins, a sign that most participants converged towards lower bleeding levels. The control group, by contrast, showed minimal shift and persistent heterogeneity.

FIGURE 3.

Distribution of gingival bleeding over time, stratified by baseline clinical severity: Clinically healthy (bleeding on probing [BOP] < 10% of sites), localised gingivitis (10%–30% of sites) and generalised gingivitis (> 30% of sites). Violin plots show the full distribution of bleeding percentages at each timepoint (T0–T4), with quartiles indicated by inner lines. Comparisons are shown separately for the active group (red) and the control group (orange). Above each pair of violins, the number of participants (n) and the median bleeding percentage are reported. The thickness of each violin reflects the density of data points: wider areas indicate a higher concentration of values at that level.

In the generalised category, the intervention effect appeared both rapid and consistent. The active group exhibited an important decrease in violin height and width: by T1, the distribution compressed significantly, and by T4, the violins were not only lower but also markedly thinner. This reduction in thickness reflected a strong homogenisation of responses, suggesting that nearly all participants benefited from the intervention, regardless of their initial extent. In contrast, the control group maintained wide, high violins throughout the follow‐up, indicating persistent inflammation and greater heterogeneity of response.

The evolution in violin shape, particularly the narrowing over time in the active group, visually reinforces the notion that baseline extent modulates responsiveness and that the intervention yields both effectiveness and uniformity in clinical outcomes.

3.3. Logistic Regression Analysis of Clinical Response at T4

Figure 4 displays the results of multivariate logistic regression models estimating the probability of a clinical response at T4 (bleeding less than threshold), adjusted for baseline variables. According to the multivariable logistic regression model adjusted for relevant clinical and demographic covariates, the intervention group remained significantly associated with bleeding resolution at T4 (OR = 3.84; 95% CI: 2.01–7.32; p < 0.001). Although additional variables such as gingival inflammation reached statistical significance, they did not confound or reduce the magnitude of the intervention effect. Logistic regression analysis of the clinical response at intermediate visits is detailed in File S1.

FIGURE 4.

Multivariate linear regression model assessing factors associated with gingival bleeding at baseline (T0). Multivariate logistic regression model estimating the odds of achieving a clinical response in the full study population. Odds ratios (ORs), 95% confidence intervals (CIs) and p‐values are reported for each baseline covariate, rounded to two decimal places. The variable ‘Treatment group’ was used to compare the active and control groups. The dashed vertical line indicates the null value (OR = 1). The baseline median (IQR) values for each covariate are shown to the right of each plot.

This pattern was consistent in the high‐risk subgroup (> Q3). According to the univariate analysis, the intervention had a strong and statistically significant effect (OR = 14.8, 95% CI: (4.9–44.7), p < 0.000001). In contrast, the effect of treatment was attenuated in the penalised model and became non‐significant, likely due to the overshrinkage of the coefficients. As such, the univariate model was retained for primary interpretation, and multivariable models were explored but not retained for presentation owing to instability and penalisation bias (see File S1).

These results reinforce the strong and independent effect of the intervention, regardless of sociodemographic or clinical characteristics.

3.4. Sensitivity Analyses

The primary analysis was conducted in the ITT population. To assess robustness, complementary sensitivity analyses were performed. In the per‐protocol population, gingival bleeding decreased from 0.56 (0.36–0.81) to 0.04 (0.00–0.22) in the intervention group, whereas the control group showed no substantial change (0.67 [0.41–0.81] to 0.72 [0.19–0.89]; p < 0.001). The between‐group mean difference in change from baseline to T4 was −0.42 [−0.50 to −0.33] (MLM p < 0.001; Table S4). These findings are consistent with the ITT analysis (−0.43 [−0.49 to −0.36]; MLM p < 0.001), indicating that the conclusions are robust to the analysis population.

To address the significant baseline imbalance in PI, an ITT sensitivity analysis stratified baseline PI into three categories (< 1, 1–2, ≥ 2). The intervention effect at T4 remained significant across all categories (p < 0.001), with no PI × treatment interaction (all p > 0.09; Table S5 and Figure S3).

4. Discussion

Unlike previous studies, our trial targeted women at increased risk of preeclampsia, integrating individualised interdental brushing into antenatal care. Calibrated IDBs significantly reduced gingival bleeding by the eighth month of pregnancy, particularly among participants with severe baseline inflammation. Most improvement occurred in the first week, suggesting that prevention may be especially beneficial during the hormonally sensitive first trimester. This early response suggests that gingival inflammation during pregnancy is reversible through regular mechanical biofilm disruption.

In contrast, gingival bleeding in the control group remained stable. While some studies reported trimester‐dependent gingival inflammation surges with postpartum resolution (Figuero et al. 2013), our data indicated persistent pregnancy‐related gingivitis without oral hygiene intervention, suggesting that hormonal effects amplify baseline inflammation (Jang et al. 2021), a hypothesis requiring further investigation.

These results support implementing targeted oral hygiene strategies early in pregnancy (Adeniyi et al. 2021), although the added benefits remain to be confirmed (Carrouel et al. 2016; Salhi et al. 2022). The significant bleeding reduction with calibrated IDBs supports mechanical biofilm disruption's role in managing gingival inflammation (Bourgeois et al. 2019), including hormonally and immunologically altered contexts. However, this study did not assess hormonal parameters directly.

The association between interdental brushing and gingival improvement was confirmed by multivariate analysis, with participants in the intervention group being > 3 times more likely to achieve bleeding resolution at follow‐up. Education level and baseline inflammation showed some associations with gingival outcomes at specific visits, but their effects were less robust. Among women with severe baseline inflammation, the intervention effect appeared nearly 15 times stronger, although model instability limited multivariate interpretation in this subgroup.

The significant reduction in bleeding with calibrated interdental brushing supports the role of mechanical biofilm disruption in managing gingival inflammation, both during pregnancy, as shown in recent prenatal trials (Parry et al. 2023), and in non‐pregnant adults (Tsikouras et al. 2024; Han et al. 2021), even under a hormonally and immunologically altered context.

Our findings align with previous evidence in non‐pregnant populations, where interdental cleaning offered additional benefit over toothbrushing alone (Holtfreter et al. 2024). They extend this to pregnant women with limited access to preventive oral care and followed throughout gestation, a group underrepresented in clinical research despite increased susceptibility to inflammation‐related complications. The potential link between local inflammation and systemic pregnancy outcomes further reinforces the relevance of preventive strategies targeting interdental biofilm.

Although microbiological or immunological data were not assessed, the clinical improvements observed with calibrated IDBs suggest that mechanical biofilm disruption may attenuate gingival inflammation, even in hormonally and immunologically altered contexts. This hypothesis, namely reduction of pathogenic load or modulation of inflammatory mediators, is biologically plausible (Abdulkareem et al. 2023) but unproven in pregnancy. Particularly, cytokine dysregulation, neutrophil hyperactivation and hormonally amplified inflammatory signalling warrant investigation. Future studies incorporating salivary or crevicular biomarker profiles could determine whether such clinical improvements reflect shifts in immune activity.

Caution is warranted in interpreting these findings as evidence of a preventive effect on disease progression. Gingival bleeding is a surrogate of inflammation, not a direct indicator of tissue destruction or long‐term periodontal risk. While its reduction is clinically relevant, further mechanistic and longitudinal studies are needed to assess whether these effects alter disease trajectories.

This trial was deliberately designed as a primary prevention intervention in pregnant women at high risk of preeclampsia but with an intact periodontium. Including women with periodontitis would have shifted the aim from prevention to treatment, which was beyond the study scope. In this high‐risk cohort, interdental brushing, which started at 12 weeks and continued until the eighth month of pregnancy, did not significantly reduce composite obstetric outcomes (preeclampsia, preterm birth, low birth weight or small for gestational age); these complete obstetric outcomes will be reported separately (Bourgeois et al. 2026). This finding aligns with a previous trial evaluating the effect of periodontal treatment on preterm birth (Michalowicz et al. 2006), although our study specifically targeted primary prevention in women with intact periodontium. Clinically, this study provides novel evidence on individualised interdental hygiene in pregnant women, a population rarely targeted in preventive periodontal trials. To our knowledge, it is the first RCT to evaluate calibrated interdental brushing throughout the entire gestational period in women at increased risk of preeclampsia, enhancing both scientific and clinical relevance.

Although some potential confounders (diet or systemic inflammation) could not be fully addressed, multivariable models included key covariates selected a priori (age, education, baseline bleeding). These findings support the feasibility of integrating calibrated interdental hygiene into antenatal care, but this should be confirmed in broader populations with long‐term clinical impact.

Several limitations should be considered. First, the trial was powered for preeclampsia; gingival bleeding was a prespecified secondary outcome. Robustness derives from observed ITT and per‐protocol analyses, not post hoc calculations, ensuring reliable results across follow‐up visits. Nevertheless, the results should be interpreted with caution. Second, attrition (~40%) at T4 was balanced between groups, avoiding bias. Consistent ITT/per‐protocol findings and excluding pregnancy interruptions confirm robustness, typical in perinatal trials. Third, periodontal assessments relied on clinical parameters, without microbiological or biomarker data, limiting exploration of biological mechanisms such as immune modulation or host–microbiome interactions. Mechanistic studies on a participant subsample will address this gap. Fourth, the study focused on preventive oral care during pregnancy, with follow‐up limited to gestation; thus, postpartum effects remain unknown. The selective Dakar population improved internal validity but restricted generalisability, especially to women with periodontal disease or other backgrounds. Fifth, site‐level analysis and healing indicators such as pocket closure were not assessed, as no sites with PD ≥ 4 mm were detected; analyses were therefore conducted at the subject level, consistent with the predefined outcome (BOP). Sixth, temporal dynamics of biomarkers were not monitored, limiting mechanistic insights into how interdental brushing influences inflammation. Seventh, although gingival bleeding was reduced, its impact on obstetric or systemic outcomes was not evaluated, and no mechanistic or translational endpoints were included; future trials should incorporate obstetric endpoints to assess broader benefits. Eighth, to avoid overfitting, stepwise regression was excluded; multivariate models used a priori covariates with mixed effects and stratification. Only prespecified T4 analysis supports the main conclusion; other analyses were exploratory, unadjusted, hypothesis‐generating or unstable in small strata. Univariate results are reported for transparency but do not affect the robustness of the overall ITT findings. Finally, participants could not be blinded to group allocation. However, outcome assessors and statisticians were masked and compliance was monitored.

5. Conclusion

This RCT provides evidence that early and consistent use of calibrated IDBs significantly reduces gingival bleeding during pregnancy. Findings highlight the potential clinical efficacy and feasibility of individualised interdental hygiene as a preventive antenatal care strategy. Although immunological or microbiological endpoints were not included, future research should investigate how mechanical interdental biofilm disruption influences local inflammation, particularly during early‐stage gingival disease in pregnancy.

Author Contributions

F.C. and D.B. conceived and designed the study and coordinated the international collaboration. A.K. and D.F. supervised the recruitment and clinical assessments in Dakar and ensured compliance with the protocol approved by the Senegalese National Ethics Committee. D.B., F.C. and A.K. monitored the implementation of procedures on site, with multiple visits between France and Senegal to support fieldwork and ensure data quality. R.A. performed the statistical analyses, contributed to the interpretation of statistical results and critically revised the manuscript. F.C., A.K. and D.B. drafted the first version of the manuscript. M.M., F.V., H.P., C.C. and R.L. contributed to the methodological design, interpretation of results and critical revision of the manuscript. All authors read and approved the final version of the manuscript.

Funding

This study was supported by Curaden AG, Switzerland.

Ethics Statement

Ethical approval was obtained from the National Ethics Committee for Health Research (protocol 000086/MSAS/CNERS/SP, approved on June 8, 2021). All participants provided written informed consent. The study was registered at ClinicalTrials.gov (NCT04989075) and complied with the Declaration of Helsinki and CONSORT reporting guidelines.

Conflicts of Interest

D.B. reports receiving personal fees from Curaden AG, outside the submitted work. The other authors declare no conflicts of interest.

Supporting information

Data S1: jcpe70085‐sup‐0001‐supinfo.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.