ABSTRACT
Aim
The evidence for tooth splinting during periodontal regenerative therapy is limited. We aimed to investigate the adjunctive benefits of tooth splinting on clinical outcomes.
Materials and Methods
In total, 194 intrabony defects in 126 participants were prospectively evaluated over three years following periodontal regenerative therapy with an enamel matrix derivative. Clinical attachment level (CAL), probing depth (PD) and radiographic bone defect depth (RBD) were assessed. Propensity score matching (PSM) was employed to assign surgical sites to splinting and non‐splinting groups, and multilevel regression was used to analyse the impact of tooth splinting on clinical outcomes.
Results
Using PSM, 37 sites each from 32 and 30 participants in the splinting and non‐splinting groups, respectively, were matched using the propensity score. Both groups included 26 and 11 sites with tooth mobility degrees 0 and 1, respectively. Significant improvements in CAL, PD and RBD were observed in both groups compared to baseline over the 3‐year period. However, there were no significant differences among these parameters between the groups at the 1‐ and 3‐year follow‐ups.
Conclusion
No statistically significant adjunctive benefits of tooth splinting were observed in this study. However, further research with a larger sample size may be required to detect smaller but potentially clinically relevant effects.
Keywords: enamel matrix derivative, periodontal regenerative therapy, propensity score, tooth splinting
1. Introduction
Advanced periodontitis increases tooth mobility, consequently impairing masticatory function (Jepsen et al. 2018; Papapanou et al. 2018; Tonetti et al. 2018). Tooth splinting is commonly applied to periodontally involved teeth with secondary occlusal trauma to enhance masticatory function, distribute occlusal forces and prevent pathological tooth migration. However, a recent systematic review highlighted that tooth splinting does not improve the survival of mobile teeth in patients with stage IV periodontitis (Dommisch et al. 2022), making its practical implications controversial.
The benefits of tooth splinting in periodontal regenerative therapy remain unclear. Clinically, tooth splinting is frequently employed during periodontal regenerative therapy as tooth stabilisation is reported to positively influence outcomes by providing optimal wound stability (Cortellini et al. 2001, 2011; Siciliano et al. 2011). However, a systematic review by the American Academy of Periodontology (AAP) in 2015 assigned a B‐level recommendation for tooth splinting, indicating inconsistent or limited‐quality patient‐oriented evidence based on the Strength of Recommendation Taxonomy (SORT) grading system (Kao et al. 2015). Notably, this review included only one non‐randomised comparative trial that reported significant improvements in probing depth (PD) reduction and clinical attachment level (CAL) gain in the splinting group compared to the non‐splinting group (Schulz et al. 2000).
Propensity score matching (PSM) is a statistical method used to reduce bias between intervention and control groups using observational data. This methodology is increasingly applied in clinical research within the medical field (Austin 2011; Kim et al. 2016). PSM enables covariate adjustment by incorporating information on covariates and confounding factors, making it a highly useful approach for estimating the effects of an intervention using observational data when randomised allocation is not feasible. Despite its potential, this method has not been widely utilised in dental research.
In our previous reports (Matsuura et al. 2023; Mikami et al. 2021), a large‐scale prospective cohort study identified prognostic factors for periodontal regenerative therapy with an enamel matrix derivative (EMD). Utilising data from this observational study, in the present study we conducted PSM to minimise selection bias and confounding factors, effectively emulating a pseudo‐randomised controlled trial (RCT). This approach is expected to provide novel clinical insights into the benefits of tooth splinting. This study aimed to examine the effects of tooth splinting on clinical outcomes at 1 and 3 years after periodontal regenerative therapy using EMD for intrabony defects. This study specifically examined the effects of tooth splinting on teeth with mobility degree 1 or 0 to clarify its role in periodontal regenerative therapy.
2. Materials and Methods
2.1. Study Design and Data Source
This study is a secondary analysis of data obtained from a previous cohort study, which was conducted to assess changes in PD, CAL and radiographic bone defect depth (RBD) over a 3‐year follow‐up period following periodontal regenerative therapy using EMD for intrabony defects (Matsuura et al. 2023; Mikami et al. 2021). Data between 2007 and 2015 were collected at the Periodontics Clinic of the Tokyo Medical and Dental University Hospital (currently Institute of Science Tokyo Hospital).
2.2. The Original Cohort Study
The original cohort included participants who met the general criteria for periodontal regenerative therapy using EMD and provided informed consent for surgery and follow‐up examinations. A total of 312 participants who met the inclusion criteria were included in the cohort study. The cohort study was approved by the Dental Research Ethics Committee of Tokyo Medical and Dental University (approval numbers: D2010‐590, D2012‐095‐3 and D2017‐002‐04) and conducted in accordance with the Helsinki Declaration of 1975, as revised in 2013. The study was registered with the University Hospital Medical Information Network (http://www.umin.ac.jp/) (clinical trial number: UMIN000039846).
2.3. The Participants of the Current Study
For the current analysis, we utilised the de‐identified dataset obtained from the original study. The inclusion criteria were (1) patients aged ≥ 20 years; (2) patients with periodontitis who had completed cause‐related therapy with well‐maintained oral hygiene; (3) the presence of sites with a remaining PD ≥ 4 mm after cause‐related therapy; and (4) the presence of interproximal intrabony defects on radiographs, regardless of the degree I or II furcation involvement (Hamp et al. 1975). The exclusion criteria encompassed (1) patients with diabetes mellitus, (2) those currently smoking, (3) those with suspected endo‐periodontal lesions indicating the extension of intrabony defects to the root apex and (4) those with degree III furcation involvement.
2.4. Clinical Examination
Clinical examinations were conducted according to procedures outlined in our previous reports (Matsuura et al. 2023; Mikami et al. 2021; Mizutani et al. 2021). The operator performed clinical assessments, including (1) tooth mobility classified according to Miller's classification (Miller 1943), (2) PD, (3) CAL and (4) bleeding on probing (BOP) at baseline and 1‐ and 3‐years post‐surgery. PD, CAL and BOP were measured at six sites on each tooth using a manual probe (15 UNC Colour‐Coded Probe; Hu‐Friedy, IL, USA), with measurements rounded to the nearest millimetre. Intraoral radiographs were obtained at baseline, 1 and 3 years post‐surgery using the parallel technique. Radiographic assessments were performed as previously described (Matsuura et al. 2023; Mikami et al. 2022; Mizutani et al. 2021). The RBD was quantified in millimetres, representing the linear measurement from the cemento‐enamel junction on the interproximal root surface to the most apical point of the defect on radiographs (Steffensen and Webert 1989).
2.5. Tooth Splinting
Tooth splinting was performed on teeth exhibiting mobility or fremitus after thorough occlusal adjustment was conducted prior to surgery. Additionally, in cases where an excessive occlusal load was identified on a tooth due to diminished occlusal support in a partially edentulous jaw, splinting was applied even if the tooth did not exhibit mobility. It is known that tooth mobility can temporarily increase after periodontal surgery (Persson 1981). In cases where there was an excessive load on the subjected teeth due to reduced occlusal support, it was anticipated that the temporary increase in mobility post‐surgery could affect wound stability, and thus, tooth splinting was performed. This approach was expected to ensure that the teeth subjected to splinting were adequately supported to prevent postoperative complications. Tooth splinting was performed within one month prior to the surgery. Resin cement (Super‐Bond; SUN MEDICAL, Shiga, Japan) was used alone, in combination with a wire or as a provisional restoration for tooth splinting.
2.6. Surgical Procedure
The surgical procedures were conducted as previously described (Matsuura et al. 2023; Mikami et al. 2021; Mizutani et al. 2021). Briefly, after a full‐thickness flap was elevated to debride the intrabony defect, EMD (Emdogain Gel; Straumann, Basel, Switzerland) was applied to the exposed root surface. In cases of non‐containing defect morphology, autologous bone was adjunctively grafted at the operator's discretion. Postoperative care included monthly oral prophylaxis for the first six months, followed by supportive periodontal care every three months, with annual examinations for monitoring. Tooth splinting was removed 6–12 months post‐operatively. Where provisional restorations were used for splinting, a prosthesis was placed after splint removal.
2.7. Statistical Analysis
Categorical variables were expressed as numbers and percentages, while continuous variables were presented as means and standard deviations (SD). A logistic regression model with the presence or absence of tooth splinting (1 = splinting, 0 = non‐splinting) as the outcome variable was employed to calculate propensity scores (PSs) using baseline parameters, including sex, age, tooth type (incisor or molar), tooth mobility, width of keratinised gingiva, PD, CAL, BOP, RBD, furcation involvement (none/0 or degree I/II) and bony wall type (1‐/2‐wall or 3‐wall). PSM was performed at a 1:1 ratio to divide the sites into intervention (splinting) and control (non‐splinting) groups using the nearest‐neighbour matching algorithm, combining exact matching criteria with a permissible age difference (≤ 5 years), tooth mobility, tooth type, sex and CAL (≤ 2 mm). The calliper width was set to 20% of the SD of the PSs (Austin 2011). Baseline imbalances were examined using standardised differences, with absolute values of < 10% considered balanced (Austin et al. 2007). Multilevel linear regression analysis was conducted to evaluate the influence of tooth splinting on changes in CAL, PD and RBD over 1 and 3 years. The changes in CAL, PD and RBD were calculated using the following formula: (baseline CAL, PD and RBD values) − (CAL, PD and RBD values at the 1‐ or 3‐year follow‐up). The data were structured into multilevel models with sites nested within each patient.
Additional statistical procedures are described in Supplementary Methods in S1.
Statistical significance was set at p values < 0.05. All statistical analyses were performed using the Stata software (version 17.0; StataCorp, TX, USA).
3. Results
A total of 74 sites in 62 patients (37 sites in 32 participants with splinting and 37 sites in 30 participants without splinting) were included from an initial 194 sites in 126 participants, according to PSM (Figure 1). Table 1 presents the baseline characteristics of participants and teeth in the splinting and non‐splinting groups before and after matching. After PSM, the standardised difference between the two groups for all variables was < 0.1. Before PSM, 59 out of 194 teeth exhibited a mobility degree of 1 or 2. After matching, among the 74 included teeth, 52 exhibited a mobility degree of 0 and 22 exhibited a mobility degree of 1.
FIGURE 1.
Representative cases of non‐splinting (a–d) and splinting (e–h) in periodontal regenerative therapy with enamel matrix derivative. In both cases, bone grafting materials were not used, and one operator (KM) performed the surgical procedure. (a) A 45‐year‐old woman diagnosed with Stage III, Grade B periodontitis. Pre‐operatively, the probing depth (PD) was 7 mm, and the clinical attachment level (CAL) was 7 mm at the mesial‐buccal site of tooth #41. The tooth mobility was degree 1. (b) A two‐wall intrabony defect with radiographic bone defect depth (RBD) of 5.2 mm was observed in the preoperative radiograph. (c) At 3 years post‐surgery, a gain of 4 mm in CAL was observed, and PD was maintained at 2 mm. The tooth mobility improved to degree 0. (d) The radiograph demonstrated a reduction in the RBD to 1.4 mm at the 3‐year examination. (e) A 23‐year‐old male diagnosed with Stage III, Grade C periodontitis. Prior to the surgical procedure, tooth #42, with a mobility degree of 1, was splinted using wire and resin cement. Pre‐operatively, a PD of 8 mm and a CAL of 8 mm were detected at the mesial‐lingual site. (f) A two‐wall intrabony defect with RBD of 6.4 mm was observed in the preoperative radiograph. (g) The splinting was removed one year after surgery. At 3 years post‐surgery, a gain of 6 mm in CAL was observed, and PD was maintained at 1 mm. The tooth mobility improved to degree 0. (h) The radiograph demonstrated a reduction in the bony defect depth with RBD of 0.9 mm at the 3‐year examination.
TABLE 1.
The characteristics of the participants and subjected tooth at baseline in the splinting group and the non‐splinting group before and after propensity score matching.
| Before PS matching | After PS matching | |||||
|---|---|---|---|---|---|---|
| Non‐splinting | Splinting | Std diff | Non‐splinting | Splinting | Std diff | |
| n = 115 | n = 79 | n = 37 | n = 37 | |||
| Mean (SD) or n (%) | Mean (SD) or n (%) | Mean (SD) or n (%) | Mean (SD) or n (%) | |||
| Sex | ||||||
| Male | 35 (30.4%) | 20 (25.3%) | 0.11 | 7 (18.9%) | 7 (18.9%) | 0.00 |
| Female | 80 (69.6%) | 59 (74.7%) | 30 (81.1%) | 30 (81.1%) | ||
| Age (year) | 53.5 (13.6) | 56.8 (14.02) | −0.23 | 58.1 (11.8) | 59.2 (12.3) | 0.09 |
| Tooth type | ||||||
| Incisor | 19 (16.5%) | 23 (29.1%) | 0.30 | 3 (8.1%) | 3 (8.1%) | 0.00 |
| Molar | 96 (83.5%) | 56 (70.9%) | 34 (91.9%) | 34 (91.9%) | ||
| Tooth mobility | ||||||
| 0 | 94 (81.7%) | 41 (51.9%) | 0.68 | 26 (70.3%) | 26 (70.3%) | 0.00 |
| 1 | 19 (16.5%) | 31 (39.2%) | 11 (29.7%) | 11 (29.7%) | ||
| 2 | 2 (1.7%) | 7 (8.9%) | ||||
| CAL (mm) | 7.0 (2.1) | 7.4 (2.4) | −0.15 | 6.5 (1.5) | 6.4 (1.4) | 0.07 |
| PD (mm) | 6.0 (1.7) | 6.3 (2.1) | −0.17 | 5.6 (1.3) | 5.7 (1.6) | −0.07 |
| RBD (mm) | 4.77 (2.15) | 5.37 (2.04) | −0.29 | 5.05 (1.64) | 5.05 (1.84) | 0.01 |
| BOP positive | 77 (67.0%) | 52 (65.8%) | 0.02 | 24 (64.9%) | 25 (67.6%) | 0.06 |
| Width of keratinized gingiva (mm) | 4.4 (2.1) | 4.6 (2.0) | −0.11 | 4.0 (2.0) | 4.2 (1.8) | −0.099 |
| Furcation involvement | ||||||
| No‐furcation or degree 0 | 82 (71.3%) | 68 (86.1%) | 0.37 | 28 (75.7%) | 29 (78.4%) | 0.06 |
| Degree 1 or 2 | 33 (28.7%) | 11 (13.9%) | 9 (24.3%) | 8 (21.6%) | ||
| Bony wall | ||||||
| 1, 2‐wall | 84 (83.0%) | 55 (69.6%) | 0.08 | 27 (73.0%) | 28 (75.7%) | 0.06 |
| 3‐wall | 31 (27.0%) | 24 (30.4%) | 10 (27.0%) | 9 (24.3%) | ||
Abbreviations: CAL, clinical attachment level; PD, probing depth; RBD, radiographic bone defect depth; SD, standard deviation; Std Diff, standard difference.
Table 2 shows the clinical outcomes at baseline, 1‐ and 3‐year follow‐ups. Both before and after PSM, all parameters (CAL, PD and RBD) showed significant improvements at 1‐ and 3‐year follow‐ups compared to those at baseline in both non‐splinting and splinting groups. Tooth mobility at each time point is shown in Table S1. The changes in tooth mobility at 1‐year post‐surgery are shown in Table S2. The changes in clinical parameters for teeth with mobility degree 1 or 2 before PSM are presented in Table S3. Before PSM, significant improvements in tooth mobility were observed in the splinting group at 1‐ and 3‐year examinations compared to that at baseline. However, after PSM, neither the non‐splinting group nor the splinting group exhibited significant changes in tooth mobility.
TABLE 2.
Comparison of clinical outcomes at each timepoint before and after propensity score matching.
| Before PS matching | After PS matching | ||||||
|---|---|---|---|---|---|---|---|
| Baseline | 1‐year follow‐up | 3‐year follow‐up | Baseline | 1‐year follow‐up | 3‐year follow‐up | ||
| Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | ||
| Non‐splinting group (n = 115) | Non‐splinting group (n = 37) | ||||||
| CAL (mm) | 7.0 (2.1) | 4.7 (2.0)* | 4.6 (2.0)* | CAL (mm) | 6.5 (1.5) | 4.4 (1.5)* | 4.5 (2.0)* |
| PD (mm) | 6.0 (1.7) | 3.3 (1.3)* | 3.1 (1.4)* | PD (mm) | 5.6 (1.3) | 3.2 (1.2)* | 3.0 (1.4)* |
| RBD (mm) | 4.77 (2.15) | 3.22 (2.09)* | 2.72 (1.64)* | RBD (mm) | 5.05 (1.64) | 3.62 (2.22)* | 2.85 (1.56)* |
| Splinting group (n = 79) | Splinting group (n = 37) | ||||||
| CAL (mm) | 7.4 (2.4) | 4.6 (1.9)* | 4.6 (2.2)* | CAL (mm) | 6.4 (1.4) | 3.9 (1.5)* | 4.1 (2.0)* |
| PD (mm) | 6.3 (2.1) | 3.2 (1.2)* | 3.1 (1.5)* | PD (mm) | 5.7 (1.6) | 3.2 (1.2)* | 3.4 (1.7)* |
| RBD (mm) | 5.37 (2.04) | 3.29 (2.07)* | 2.54 (1.76)* | RBD (mm) | 5.05 (1.84) | 3.05 (2.09)* | 2.28 (1.91)* |
Abbreviations: CAL, clinical attachment level; PD, probing depth; RBD, radiographic bone defect depth; SD, standard deviation.
p < 0.001 (compared to baseline, paired t‐test with Bonferroni correction).
Table 3 shows comparisons between the two groups for each outcome before PSM. The multilevel regression model demonstrated that the splinting group had significantly greater RBD gain at 3 years (coefficient: 0.72, 95% confidence interval [CI]: 0.02–1.42, p = 0.042). However, after PSM, there were no significant differences in any clinical parameter at 1 and 3 years post‐operatively (Table 4). The RBD gain was higher by 0.43 and 0.35 mm in the splinting group compared to the non‐splinting group at the 1‐ and 3‐year examinations, respectively, but these differences were not statistically significant (Table 4, multilevel regression model, p = 0.37 and 0.48). To further investigate whether the effect of splinting differed by tooth mobility, we included an interaction term between tooth splinting and mobility (degree 0 or 1/2) in the multivariable linear regression models using the pre‐PSM sample, adjusting for baseline covariates. There was no outcome where splinting was a significant explanatory variable and also showed a significant interaction. Detailed results are presented in Table S4 and Figure S1.
TABLE 3.
Change in each clinical outcome and the impact of splinting on each clinical outcome before propensity score matching.
| Outcomes | Non‐splinting (n = 115) | Splinting (n = 79) | Multilevel regression model (194 sites, 126 patients) | ||
|---|---|---|---|---|---|
| Mean (SD) | Mean (SD) | Coefficient (reference: non‐splinting) | 95% CI | p | |
| CAL gain at 1 year | 2.30 (1.90) | 2.81 (1.98) | 0.46 | −0.11, 1.03 | 0.11 |
| CAL gain at 3 years | 2.46 (1.80) | 2.82 (2.14) | 0.36 | −0.21, 0.93 | 0.22 |
| PD reduction at 1 year | 2.70 (1.54) | 3.13 (2.09) | 0.42 | −0.09, 0.93 | 0.10 |
| PD reduction at 3 years | 2.88 (1.62) | 3.14 (2.23) | 0.26 | −0.28, 0.80 | 0.34 |
| RBD gain at 1 year | 1.55 (1.91) | 2.08 (2.07) | 0.51 | −0.07, 1.10 | 0.086 |
| RBD gain at 3 years | 2.04 (2.27) | 2.83 (2.50) | 0.72 | 0.02, 1.42 | 0.042 |
Abbreviations: CAL, clinical attachment level; PD, probing depth; RBD, radiographic bone defect depth; SD, standard deviation.
TABLE 4.
Change in each clinical outcome and the impact of splinting on each clinical outcome after propensity score matching.
| Outcomes | Non‐splinting (n = 37) | Splinting (n = 37) | Multilevel regression model (74 sites, 62 patients) | ||
|---|---|---|---|---|---|
| Mean (SD) | Mean (SD) | Coefficient (reference: non‐splinting) | 95% CI | p | |
| CAL gain at 1 year | 2.11 (1.66) | 2.46 (1.32) | 0.36 | −0.34, 1.05 | 0.32 |
| CAL gain at 3 years | 2.00 (1.41) | 2.32 (1.36) | 0.29 | −0.36, 0.93 | 0.38 |
| PD reduction at 1 year | 2.38 (1.38) | 2.51 (1.50) | 0.14 | −0.51, 0.78 | 0.68 |
| PD reduction at 3 years | 2.57 (1.57) | 2.27 (1.45) | −0.30 | −0.98, 0.38 | 0.39 |
| RBD gain at 1 year | 1.43 (2.36) | 2.00 (1.72) | 0.43 | −0.51, 1.38 | 0.37 |
| RBD gain at 3 years | 2.20 (2.00) | 2.76 (2.29) | 0.35 | −0.62, 1.32 | 0.48 |
Abbreviations: CAL, clinical attachment level; PD, probing depth; RBD, radiographic bone defect depth; SD, standard deviation.
The results of the sensitivity analysis using the inverse probability weighting (IPW) method are shown in Tables S4 and S5. The standardised difference between splinting and non‐splinting groups for all covariates was reduced to below 0.1 after weighting (Table S5). The estimated average treatment effect (ATE) by tooth splinting for changes in each clinical outcome was shown in Table S6. The effect of tooth splinting on each clinical outcome, that is, the changes in CAL, PD and RBD at 1 and 3 years, was not significant. The results showed a similar trend to the estimated average treatment effect on the treated (ATT) presented in Table 4.
In this study, the differences in CAL gain observed after PSM were 0.36 and 0.29 mm at 1 and 3 years, respectively, with power values of 0.26 and 0.25, respectively. The post hoc power analysis was performed under the assumption of differences in CAL gain of 0.5, 1.0 and 1.5 mm between the splinting and non‐splinting groups. The corresponding power values were 0.41, 0.88 and 1.00, respectively.
4. Discussion
This study is the first to apply PSM to compare the outcomes of periodontal regenerative therapy between groups with and without tooth splinting. Our results showed no statistically significant effects of tooth splinting on clinical outcomes such as CAL, PD and RBD at the 1‐ and 3‐year examinations. The results from the IPW method, conducted with the same sample size as before PSM, corroborated the findings after PSM, indicating that the observed differences before PSM might be attributable to unadjusted confounding factors. These findings suggest that tooth splinting may not provide adjunctive benefits for regenerative therapy in teeth with mobility of one or less degree; however, this interpretation should be made with caution, as the lack of statistical significance does not necessarily indicate the absence of an effect.
Previously, a non‐randomised comparative trial by Schultz et al. on regenerative therapy with bone grafts for intrabony defects indicated that the splinting group exhibited significantly greater PD reduction and CAL gain than did the non‐splinting group (Schulz et al. 2000). However, they did not account for the confounding factors influencing regenerative therapy outcomes, such as baseline PD, bone defect morphology and tooth mobility. To appropriately assess the effect of tooth splinting, we utilised PSM to investigate the adjunctive effects of splinting in periodontal regenerative therapy after adjusting for these confounding factors. Our results revealed that tooth splinting with mobility degree 1 or 0 had no significant impact on clinical outcomes. For instance, following PSM, the 95% CI for RBD gain at the 3‐year follow‐up ranged from −0.62 to 1.32 mm. Although the point estimate indicated a favourable effect of splinting, the wide interval suggests that both a clinically meaningful benefit and a slight detriment remain within the range of possible true effects. Comparable patterns were observed for other clinical parameters, including CAL and PD, where the estimated effects consistently favoured the splinting group but did not reach statistical significance. A possible explanation for these results is that the tooth mobility of all the subjected teeth was in the range of degree one or less after PSM. Our study using PSM specifically targeted teeth with mobility degree one or less, aligning with the previous reports, suggesting that such mobility does not significantly impact periodontal regeneration. Although tooth mobility is considered an important factor in periodontal regeneration (Cortellini et al. 2001; Sanders et al. 1983), a previous study reported that horizontal tooth mobility not exceeding 1 mm could be successfully treated with periodontal regenerative therapy (Trejo and Weltman 2004). A narrative review by Cortellini and Tonetti indicated that tooth with mobility degree one does not significantly impact periodontal regeneration, and tooth splinting is not required (Cortellini and Tonetti 2015). However, in severe conditions, such as bone defects extending to the root apex and teeth exhibiting uncontrollable mobility, tooth splinting may be critical for appropriate surgical debridement and wound stability (Cortellini et al. 2011; Cortellini and Tonetti 2015). Additionally, tooth splinting may alleviate postoperative patient discomfort. Thus, future studies should consider various practical factors to further elucidate the role of tooth splinting in periodontal regenerative therapy. To further assess whether the inclusion of teeth with mobility grade 0 may have influenced the interpretation of treatment effects, we conducted an interaction analysis using the pre‐PSM sample, adjusting for the same covariates used in the PS model. No outcome demonstrated both significant main effects and a statistically significant interaction term. These findings suggest that including non‐mobile teeth did not attenuate or obscure a potential benefit of splinting in more mobile teeth, supporting the validity of combining mobility degrees 0 and 1 in the main analysis. Nonetheless, the interpretation of interaction effects should be approached with caution, and future studies with larger subgroups are warranted to further explore mobility‐specific responses to splinting.
Before PSM, significant improvements in tooth mobility were observed in the splinting group at the 1‐ and 3‐year examinations compared to baseline. However, these results were obtained without adjusting for potential confounding factors. The splinting group likely included teeth with greater initial mobility, which necessitated splinting, thus potentially exaggerating the observed improvements. After PSM, which accounts for these confounding factors, neither the non‐splinting group nor the splinting group exhibited significant changes in tooth mobility. This suggests that the improvements observed before PSM may have been due to unadjusted confounding factors rather than the effect of splinting itself.
The methodological novelty of this study is particularly significant because PSM is relatively uncommon in periodontal research. PSM analysis enables pseudo‐randomisation using observational data, allowing the attainment of results similar to those of RCTs. Although RCTs are effective at mitigating the influence of confounding factors, they also have limitations, such as ethical constraints, feasibility concerns and strict inclusion criteria, which can sometimes compromise external validity. Furthermore, PSM analysis, based on the concepts of Rosenbaum and Rubin (Rosenbaum and Rubin 1983), can be employed to calculate the probability of receiving the intervention from pretreatment variables. This score is then used to assess the similarity between the samples and select corresponding samples for matching. Consequently, PSM can address ‘confounding by indication’, which is a major concern in comparative therapeutic effect analysis using observational data. In comparative research using observational data, the allocation of intervention and non‐intervention groups is non‐random and may be influenced by participants' background factors, potentially affecting both the allocation and outcomes. Notably, in this study, we achieved results with relatively high external validity by adjusting for potential background factors through PSM without imposing the strict inclusion criteria typical of RCTs.
In this study, we estimated the ATT using PSM to evaluate the intervention effects on individuals likely to receive tooth splinting. As an sensitivity analysis, we also examined the effects of administering tooth splinting to the entire population of patients, that is, the ATE, using the IPW method. The estimated ATE exhibited a similar trend to the ATT, with no significant differences observed across all outcomes. This suggests that the results of this study are robust.
This study has some limitations. Firstly, although PSM methods can reduce bias from observed variables, they do not account for biases from unobserved factors, such as socioeconomic status. Secondly, a priori sample size calculation could not be performed because this study was a secondary analysis of a cohort study. No significant differences in any clinical parameters were observed between the splinting and non‐splinting groups after PSM; however, it cannot be ruled out that this may be due to the small sample size. Regarding CAL gain, which is considered the most critical parameter for evaluating the outcomes of periodontal regenerative therapy, the differences observed between the splinting and non‐splinting groups in this study were only 0.36 mm at 1 year and 0.29 mm at 3 years post‐operatively. Even if a larger sample size were to yield statistically significant differences for these point estimates, it warrants considering the clinical benefit of tooth splinting as an intervention, which might not be considerable. Furthermore, the results of the power analysis indicated that the sample size after PSM would have provided sufficient power to detect an intervention effect of tooth splinting if the effect size had been approximately 1 mm. This result suggests that the additional benefit of tooth splinting, if present, is unlikely to be as substantial as reported in the previous study by Schulz et al. (2000). However, the possibility that this study was underpowered to detect small but clinically meaningful effects cannot be excluded. Thirdly, the non‐standardised radiographs without personal stents may affect the reproducibility and accuracy of radiographic assessments. Future studies should consider using standardised radiographs with personal stents or computed tomography imaging to enable three‐dimensional evaluations, which could enhance the precision and reliability of the findings. Fourthly, there was a predominance of sites with 1‐ and 2‐wall defects, which are considered more challenging for regenerative therapy, as well as a high number of molars included. While 1‐ and 2‐wall defects generally present less wound stability compared to 3‐wall defects, tooth splinting might be more beneficial in such cases. However, our findings did not demonstrate significant benefits of splinting even in this challenging cohort, suggesting that the effectiveness of tooth splinting might be limited regardless of defect morphology. Fifthly, differences between the splinting appliances were not assessed in this study, as it is expected that the differences in stability would be minimal depending on the splinting technique used. Lastly, the study focused on the effects of tooth splinting on periodontal regenerative therapy using EMD. Thus, it remains unclear whether these results can be generalised to other regenerative modalities.
5. Conclusion
The PSM analysis revealed no statistically significant difference in clinical outcomes between groups with and without tooth splinting after periodontal regenerative therapy using EMD during a 3‐year follow‐up. While our findings suggest that the additional benefit of tooth splinting may be limited, the true effect remains uncertain. The possibility that this study was underpowered to detect smaller but clinically relevant effects cannot be ruled out. Further investigations with larger sample sizes and teeth with higher tooth mobility are warranted to elucidate the potential role of splinting in periodontal regenerative therapy.
Author Contributions
Risako Mikami, Koji Mizutani, Akira Aoki and Takanori Iwata developed the concept for this study. Yuichi Izumi, Takanori Matsuura, Koji Mizutani, Hidehiro Shioyama, Norio Aoyama, Tomonari Suda, Yukako Kusunoki and Kohei Takeda provided substantial assistance in data collection. Risako Mikami, Miho Ishimaru, Jun Aida, Tatsuhiko Anzai, Kunihiko Takahashi and Koichiro Matsuo performed statistical analyses. Risako Mikami, Miho Ishimaru and Koji Mizutani wrote the initial draft of the manuscript. All authors reviewed the manuscript and approved its final version.
Ethics Statement
This study was approved by the Research Ethics Committee of Tokyo Medical and Dental University (D2010‐590, D2012‐095 and D‐2015‐561, D2017‐002) and was conducted in accordance with the Declaration of Helsinki of 1975, as revised in 2008 and 2013. Individuals included in the study provided informed consent.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1. Supporting Information.
Figure S1. Marginal effect plots of the interaction between tooth splinting and tooth mobility (degree 0 or 1/2) on clinical outcomes in the pre‐propensity score‐matched sample (n = 194). Each panel illustrates the predicted marginal means of clinical outcomes based on multivariable linear regression models including an interaction term between splinting and tooth mobility (Table S4). Circles represent point estimates of the predicted means, and vertical lines indicate the 95% confidence intervals. (a) Clinical attachment level (CAL) gain at 1 year, (b) CAL gain at 3 years, (c) Probing depth (PD) reduction at 1 year, (d) PD reduction at 3 years, (e) Radiographic bone defect depth (RBD) gain at 1 year, (f) RBD gain at 3 years.
Table S1. Comparison of tooth mobility at each timepoint before and after propensity score matching.
Table S2. Progression of tooth mobility over one year following periodontal regenerative therapy.
Table S3. Comparison of clinical outcomes at each timepoint before propensity score matching in teeth with mobility grade 1 or 2 (N = 59).
Table S4. Interaction analysis between tooth mobility (degree 0 or 1/2) and tooth splinting on clinical outcomes in the pre‐PSM samples (n = 194).
Table S5. Balance statistics for the inverse probability weighting (IPW) method for tooth splinting.
Table S6. Estimated average treatment effect (ATE) by tooth splinting using the inverse probability weighting method (IPW) for changes in each clinical outcome (N = 194).
Acknowledgements
The authors thank the staff of the Department of Periodontology of Institute of Science Tokyo for their assistance with data collection.
Mikami, R. , Ishimaru M., Mizutani K., et al. 2025. “Effect of Tooth Splinting on Clinical Outcomes following Periodontal Regenerative Therapy in Teeth with Mobility Degree 1 or 0: A Propensity Score‐Matched Analysis.” Journal of Clinical Periodontology 52, no. 9: 1254–1262. 10.1111/jcpe.14190.
Funding: This study was supported by a Grant‐in‐Aid for Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (grant number 23K16014 to Risako Mikami).
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- Austin, P. C. 2011. “An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies.” Multivariate Behavioral Research 46, no. 3: 399–424. 10.1080/00273171.2011.568786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Austin, P. C. , Grootendorst P., Normand S. L., and Anderson G. M.. 2007. “Conditioning on the Propensity Score Can Result in Biased Estimation of Common Measures of Treatment Effect: A Monte Carlo Study.” Statistics in Medicine 26, no. 4: 754–768. 10.1002/sim.2618. [DOI] [PubMed] [Google Scholar]
- Cortellini, P. , Stalpers G., Mollo A., and Tonetti M. S.. 2011. “Periodontal Regeneration Versus Extraction and Prosthetic Replacement of Teeth Severely Compromised by Attachment Loss to the Apex: 5‐Year Results of an Ongoing Randomized Clinical Trial.” Journal of Clinical Periodontology 38, no. 10: 915–924. 10.1111/j.1600-051X.2011.01768.x. [DOI] [PubMed] [Google Scholar]
- Cortellini, P. , and Tonetti M. S.. 2015. “Clinical Concepts for Regenerative Therapy in Intrabony Defects.” Periodontology 2000 68, no. 1: 282–307. 10.1111/prd.12048. [DOI] [PubMed] [Google Scholar]
- Cortellini, P. , Tonetti M. S., Lang N. P., et al. 2001. “The Simplified Papilla Preservation Flap in the Regenerative Treatment of Deep Intrabony Defects: Clinical Outcomes and Postoperative Morbidity.” Journal of Periodontology 72, no. 12: 1702–1712. 10.1902/jop.2001.72.12.1702. [DOI] [PubMed] [Google Scholar]
- Dommisch, H. , Walter C., Difloe‐Geisert J. C., Gintaute A., Jepsen S., and Zitzmann N. U.. 2022. “Efficacy of Tooth Splinting and Occlusal Adjustment in Patients With Periodontitis Exhibiting Masticatory Dysfunction: A Systematic Review.” Journal of Clinical Periodontology 49, no. Suppl 24: 149–166. 10.1111/jcpe.13563. [DOI] [PubMed] [Google Scholar]
- Hamp, S. E. , Nyman S., and Lindhe J.. 1975. “Periodontal Treatment of Multirooted Teeth. Results After 5 Years.” Journal of Clinical Periodontology 2, no. 3: 126–135. 10.1111/j.1600-051x.1975.tb01734.x. [DOI] [PubMed] [Google Scholar]
- Jepsen, S. , Caton J. G., Albandar J. M., et al. 2018. “Periodontal Manifestations of Systemic Diseases and Developmental and Acquired Conditions: Consensus Report of Workgroup 3 of the 2017 World Workshop on the Classification of Periodontal and Peri‐Implant Diseases and Conditions.” Journal of Clinical Periodontology 45, no. Suppl 20: S219–S229. 10.1111/jcpe.12951. [DOI] [PubMed] [Google Scholar]
- Kao, R. T. , Nares S., and Reynolds M. A.. 2015. “Periodontal Regeneration – Intrabony Defects: A Systematic Review From the AAP Regeneration Workshop.” Journal of Periodontology 86, no. 2 Suppl: S77–S104. 10.1902/jop.2015.130685. [DOI] [PubMed] [Google Scholar]
- Kim, D. H. , Pieper C. F., Ahmed A., and Colón‐Emeric C. S.. 2016. “Use and Interpretation of Propensity Scores in Aging Research: A Guide for Clinical Researchers.” Journal of the American Geriatrics Society 64, no. 10: 2065–2073. 10.1111/jgs.14253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Matsuura, T. , Mikami R., Mizutani K., et al. 2023. “Assessment of Bone Defect Morphology for the Adjunctive Use of Bone Grafting Combined With Enamel Matrix Derivative: A 3‐Year Cohort Study.” Journal of Periodontology 95, no. 9: 809–820. 10.1002/jper.23-0538. [DOI] [PubMed] [Google Scholar]
- Mikami, R. , Mizutani K., Shioyama H., et al. 2021. “Influence of Aging on Periodontal Regenerative Therapy Using Enamel Matrix Derivative: A 3‐Year Prospective Cohort Study.” Journal of Clinical Periodontology 49, no. 2: 123–133. 10.1111/jcpe.13552. [DOI] [PubMed] [Google Scholar]
- Mikami, R. , Sudo T., Fukuba S., et al. 2022. “Prognostic Factors Affecting Periodontal Regenerative Therapy Using Recombinant Human Fibroblast Growth Factor‐2: A 3‐Year Cohort Study.” Regenerative Therapy 21: 271–276. 10.1016/j.reth.2022.07.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miller, S. C. 1943. Textbook of Periodontia (Oral Medicine). Blakiston Company. [Google Scholar]
- Mizutani, K. , Shioyama H., Matsuura T., et al. 2021. “Periodontal Regenerative Therapy in Patients With Type 2 Diabetes Using Minimally Invasive Surgical Technique With Enamel Matrix Derivative Under 3‐Year Observation: A Prospective Cohort Study.” Journal of Periodontology 92, no. 9: 1262–1273. 10.1002/jper.20-0590. [DOI] [PubMed] [Google Scholar]
- Papapanou, P. N. , Sanz M., Buduneli N., et al. 2018. “Periodontitis: Consensus Report of Workgroup 2 of the 2017 World Workshop on the Classification of Periodontal and Peri‐Implant Diseases and Conditions.” Journal of Clinical Periodontology 45, no. Suppl 20: S162–S170. 10.1111/jcpe.12946. [DOI] [PubMed] [Google Scholar]
- Persson, R. 1981. “Assessment of Tooth Mobility Using Small Loads. IV. The Effect of Periodontal Treatment Including Gingivectomy and Flap Procedures.” Journal of Clinical Periodontology 8, no. 2: 88–97. 10.1111/j.1600-051x.1981.tb02348.x. [DOI] [PubMed] [Google Scholar]
- Rosenbaum, P. R. , and Rubin D. B.. 1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 70, no. 1: 41–55. 10.1093/biomet/70.1.41. [DOI] [Google Scholar]
- Sanders, J. J. , Sepe W. W., Bowers G. M., et al. 1983. “Clinical Evaluation of Freeze‐Dried Bone Allografts in Periodontal Osseous Defects. Part III. Composite Freeze‐Dried Bone Allografts With and Without Autogenous Bone Grafts.” Journal of Periodontology 54, no. 1: 1–8. 10.1902/jop.1983.54.1.1. [DOI] [PubMed] [Google Scholar]
- Schulz, A. , Hilgers R. D., and Niedermeier W.. 2000. “The Effect of Splinting of Teeth in Combination With Reconstructive Periodontal Surgery in Humans.” Clinical Oral Investigations 4, no. 2: 98–105. 10.1007/s007840050123. [DOI] [PubMed] [Google Scholar]
- Siciliano, V. I. , Andreuccetti G., Siciliano A. I., Blasi A., Sculean A., and Salvi G. E.. 2011. “Clinical Outcomes After Treatment of Non‐Contained Intrabony Defects With Enamel Matrix Derivative or Guided Tissue Regeneration: A 12‐Month Randomized Controlled Clinical Trial.” Journal of Periodontology 82, no. 1: 62–71. 10.1902/jop.2010.100144. [DOI] [PubMed] [Google Scholar]
- Steffensen, B. , and Webert H. P.. 1989. “Relationship Between the Radiographic Periodontal Defect Angle and Healing After Treatment.” Journal of Periodontology 60, no. 5: 248–254. 10.1902/jop.1989.60.5.248. [DOI] [PubMed] [Google Scholar]
- Tonetti, M. S. , Greenwell H., and Kornman K. S.. 2018. “Staging and Grading of Periodontitis: Framework and Proposal of a New Classification and Case Definition.” Journal of Clinical Periodontology 45, no. Suppl 20: S149–s161. 10.1111/jcpe.12945. [DOI] [PubMed] [Google Scholar]
- Trejo, P. M. , and Weltman R. L.. 2004. “Favorable Periodontal Regenerative Outcomes From Teeth With Presurgical Mobility: A Retrospective Study.” Journal of Periodontology 75, no. 11: 1532–1538. 10.1902/jop.2004.75.11.1532. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supporting Information.
Figure S1. Marginal effect plots of the interaction between tooth splinting and tooth mobility (degree 0 or 1/2) on clinical outcomes in the pre‐propensity score‐matched sample (n = 194). Each panel illustrates the predicted marginal means of clinical outcomes based on multivariable linear regression models including an interaction term between splinting and tooth mobility (Table S4). Circles represent point estimates of the predicted means, and vertical lines indicate the 95% confidence intervals. (a) Clinical attachment level (CAL) gain at 1 year, (b) CAL gain at 3 years, (c) Probing depth (PD) reduction at 1 year, (d) PD reduction at 3 years, (e) Radiographic bone defect depth (RBD) gain at 1 year, (f) RBD gain at 3 years.
Table S1. Comparison of tooth mobility at each timepoint before and after propensity score matching.
Table S2. Progression of tooth mobility over one year following periodontal regenerative therapy.
Table S3. Comparison of clinical outcomes at each timepoint before propensity score matching in teeth with mobility grade 1 or 2 (N = 59).
Table S4. Interaction analysis between tooth mobility (degree 0 or 1/2) and tooth splinting on clinical outcomes in the pre‐PSM samples (n = 194).
Table S5. Balance statistics for the inverse probability weighting (IPW) method for tooth splinting.
Table S6. Estimated average treatment effect (ATE) by tooth splinting using the inverse probability weighting method (IPW) for changes in each clinical outcome (N = 194).
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.