Integrated self-management interventions improve oral and metabolic health in community-dwelling adults with type 2 diabetes and periodontitis Authors: Yi Zhang, Nan Zhou, Yijia Chen, Chenchen Wang, Hao Xu, Xin Hong Published: 2026 ABSTRACT To evaluate the effects of integrated self-management interventions on objectively measured clinical periodontal parameters and systemic metabolic health in community-dwelling individuals with type 2 diabetes (T2DM) and periodontitis. From June to August 2022, 784 adults with T2DM and mild-to-moderate periodontitis from Nanjing communities were randomly allocated to a control group (G0), a periodontitis intervention group (G1), a diabetes intervention group (G2), and a comprehensive intervention group (G3), with 196 participants in each group in this randomized controlled trial. From September to November 2022, the respective intervention groups received three-month, structured self-management programs, while G0 received standard care. Outcomes—including oral health behaviors, probing depth (PD), clinical attachment loss (CAL), fasting plasma glucose (FPG), glycated hemoglobin (HbA1c) level, healthy lifestyle score, and body mass index (BMI)—were assessed at baseline (T0), immediately post-intervention (T1), 3 months (T2), and 6 months (T3) post-intervention. Data were analyzed using Generalized Estimating Equations. A total of 747 participants completed the trial (mean age = 65.26 ± 7.72 years). Periodontal improvements were most pronounced at T3, where both PD and CAL were significantly decreased in G3 compared with G0 (PD: mean difference = − 0.253 mm, P = 0.009; CAL: mean difference = − 0.429 mm, P = 0.017). At T2, G3 demonstrated greater FPG reduction than G0 (mean difference = − 0.687 mmol/L, P = 0.003) and lower HbA1c than G1 (mean difference = − 0.387%, P = 0.001). The comprehensive intervention group (G3) exhibited greater improvements in blood glucose level and clinical periodontal health compared to the other two intervention groups. Compared with G0, the comprehensive intervention significantly improved patients’ oral health behaviors (≥ 2 kinds; mean difference = 22.8%, P < 0.001) and health lifestyle scores (mean difference = 0.287 points, P < 0.001) at T1, and significantly reduced BMI at T3 (mean difference = − 1.470 kg/m2; P < 0.001). The integrated self-management interventions significantly improved clinical periodontal parameters (PD and CAL), oral health behaviors, blood glucose levels, health lifestyle scores, and BMI in patients with T2DM and periodontitis.Trial registration: ChiCTR2100052301, registered on October 24, 2021.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-55674-8. Introduction Diabetes mellitus is a chronic metabolic disorder characterized by hyperglycemia, with over 90% of cases classified as type 2 diabetes mellitus (T2DM)1. The global prevalence reached 537 million adults in 2021, and is projected to rise to 783 million by 20452. Diabetes-related deaths increased from 672,000 in 1990 to 1.657 million in 2021, with healthcare expenditure estimated at 966 billion USD in 2021, expected to reach 1.054 trillion USD by 20452–4. Strict glycemic control reduces complications1. Periodontitis, the sixth most common complication of diabetes5, induces systemic inflammation that reduces insulin sensitivity, impairing glycemic control6–8. Despite this correlation, integrated management remains overlooked. Self-management transforms patients into active participants in chronic disease care9. Its effectiveness in diabetes is established, improving knowledge, lifestyle, and glycemic control10–12. However, research on periodontal self-management in diabetic patients is limited. Current studies focus on traditional periodontal therapy, which improves glycemic control and oral health, but high costs and low oral health awareness limit its reach13. A randomized trial in Taiwan showed that combining periodontal therapy with community health programs improved oral health-related quality of life and self-management adherence, benefiting both periodontal health and glycemic control14. Periodontal health education and self-management are thus vital. Building upon our previous report from the same parent trial, which demonstrated the intervention’s efficacy in improving patient-reported oral health symptoms and psychosocial outcomes15, this study focuses on objective clinical periodontal markers—probing depth (PD) and clinical attachment loss (CAL)—the gold standards for diagnosing periodontitis and evaluating treatment. We also analyze fasting plasma glucose (FPG) and glycated hemoglobin (HbA1c) as continuous variables for a granular metabolic assessment. We hypothesize the integrated intervention will improve PD and CAL, alongside glycemic control, lifestyle scores, and body mass index (BMI). Methods Ethical considerations This study employed a randomized controlled trial (RCT) conducted from June 2022 to June 2023 across 12 districts in Nanjing, China, and was registered with the Chinese Clinical Trial Registry (ChiCTR2100052301) on October 24, 2021. The study received ethical approval from the Nanjing Municipal Center for Disease Control and Prevention (PJ2021-A001-02). Informed consent was obtained. Participants and eligibility criteria From June to August 2022, this study employed a multi-stage stratified cluster random sampling method to select 1,600 T2DM patients from communities in Nanjing participating in the national basic public health service program. A cross-sectional survey was conducted, including questionnaires, physical measurements, laboratory tests, and periodontal examinations. T2DM patients with periodontitis (mild to moderate) were invited by phone to participate in a self-management program. Inclusion criteria: (1) Age ≥ 35 years; (2) Residing in Nanjing for at least 6 months in the past 12 months; (3) Diagnosed with T2DM1; (4) Enrolled in the national basic public health service program for T2DM health management; (5) Stable condition without severe complications. Exclusion criteria: (1) History of infectious diseases such as tuberculosis, syphilis, or HIV; (2) Pregnancy; (3) Cognitive/mental health disorders; (4) Use of medications that affect glucose metabolism or cause gingival changes. T2DM diagnosis was based on the 2020 Chinese Guidelines for the Prevention and Treatment of Type 2 Diabetes41 (FPG ≥ 7.0 mmol/L or HbA1c ≥ 6.5%, or a diagnosis by a township/community or higher-level hospital, or use of glucose-lowering medications/insulin). Periodontal disease was diagnosed using the 2018 AAP/EFP World Workshop classification16, and further classified into severe and mild-to-moderate per 2017 Chinese Expert Consensus17. Random allocation Participants were randomly assigned to four groups: control group (G0), periodontitis intervention group (G1), diabetes intervention group (G2), and diabetes and periodontitis comprehensive intervention group (G3). Randomization used a random number table. Sample size Sample size was initially calculated for HbA1c [mean difference = 0.5%, standard deviation σ = 1.4, α = 0.05 (two-sided), β = 0.10 (90% power)]18. A minimum of 180 participants per group was required. Post hoc power analysis for periodontal parameters (PD: clinically meaningful difference 0.5 mm, σ = 0.81; CAL: clinically meaningful difference 0.68 mm, σ = 0.88)19 showed that with approximately 55 participants per group, the achieved power was likewise > 99%. The final 747 participants provided ample power. Interventions From September to November 2022, T2DM patients with mild to moderate periodontitis selected from the baseline were grouped into self-management groups (12–18 patients per group). A 3-month program included bi-weekly sessions (6 total, 1–1.5 h each). The control group (G0) received routine health management. The periodontitis intervention group (G1) received education on oral health knowledge, healthy oral behaviors, the bidirectional relationship between periodontitis and diabetes, periodontal treatment options, regular oral follow-up care, etc. The diabetes intervention group (G2) covered diabetes-related knowledge, self-management skills, psychological adjustment, physical activity, healthy diet, medication use, blood glucose self-monitoring, complication management, lifestyle maintenance, etc. The comprehensive intervention group (G3) integrated all content from both G1 and G2 programs. No clinical periodontal treatment was provided as part of the intervention. The programs focused exclusively on self-management education and behavioral support. Detailed activity plans for each intervention group were provided in Supplementary Tables S1–S3 of our previous publication15. Assessments at baseline (T0), immediately post-intervention (T1), and at 3 months (T2) and 6 months (T3) post-intervention included surveys, physical measures, and periodontal exams. Data were collected using the “Self-Management Model-Based Community Diabetes and Chronic Periodontitis Intervention System” (https://njkq.yzsmart.cn:9592/) via tablet computers through face-to-face interviews. Outcome measures The primary outcomes were oral health behaviors (< 2 kinds; ≥ 2 kinds), PD, and CAL. Secondary outcomes included FPG and HbA1c levels. Tertiary outcomes comprised healthy lifestyle score and BMI. The survey included: (1) sociodemographic information (name, gender, age, education, etc.); (2) diabetes-related information (duration, control status, complications, etc.); (3) diabetes-related behaviors and lifestyle (smoking, alcohol consumption, physical activity, diet, etc.); (4) oral health behaviors. Healthy lifestyle was defined as: (1) non-smoker or ex-smoker; (2) moderate alcohol intake (men < 25 g/d, women < 15 g/d)20; (3) sufficient physical activity (≥ 150 min of moderate-intensity or ≥ 75 min of vigorous-intensity activity per week)21; (4) healthy diet (scored based on frequency of 6 food groups: daily vegetables, fruits, whole grains, weekly fish and legumes, and red meat < 7 times a week). Each criterion met scores 1 point, resulting in a diet score range of 0–6 (≥ 4 points denotes a healthy diet)22. The healthy lifestyle score is the sum of points (range 0–4), with higher scores indicating healthier lifestyles23. Oral health behaviors, based on the Fourth National Oral Health Epidemiological Survey standards, include: (1) measures taken for gum bleeding; (2) brushing ≥ 2times daily; (3) brushing duration ≥ 3 min; (4) use of floss or interdental brushes; (5) annual oral health checkup. Blood samples were collected following the Human Blood Sample Collection and Handling (GB/T 38576-2020) standards, with fasting venous blood drawn after at least 8 h of fasting. FPG and HbA1c were measured using an Olympus AU5400 analyzer. Height and weight were measured according to the Methods for Human Body Measurements in Population Health Monitoring (WS/T 424–2013), and BMI was calculated as weight (kg)/height (m)2. Periodontal measures, including PD and CAL, were assessed by trained specialists from the Periodontology Department of Nanjing University School of Stomatology, following standard protocols for examining and scoring the Ramfjord 6 index teeth24. Statistical analysis This study followed the per protocol Set (PPS) principle, and only included data from participants who completed three intervention evaluations. Differences in baseline demographic characteristics and outcomes between the control and intervention groups were analyzed using ANOVA or Kruskal-Wallis H tests for continuous variables, and chi-squared tests for categorical variables. Generalized Estimating Equations (GEE) with unstructured correlation matrix were used to analyze changes in outcome measures at different follow-up points, assessing differences between the intervention and control groups during the intervention period. Time, group, and group×time interaction effects were evaluated to assess the trend of variables over time. A significant main effect of time indicated that the outcome variable changed over time across the four groups. A significant main effect of group indicated that the outcome differed among groups when time was not considered. A significant group×time interaction effect indicated that the temporal trends of the outcome differed between the intervention and control groups across the four time points. Parameter estimates (β coefficients) for time-by-group contrasts were extracted to identify significant effects. Then pairwise comparisons of marginal means were conducted, with Cohen’s d calculated for continuous variables to assess effect size: 0.2–0.5 (small), 0.5–0.8 (moderate), ≥ 0.8 (large). For categorical variables, the Phi (φ) coefficient was calculated, with values closer to 1 indicating stronger associations. The GEE model adjusted for age, gender, area, education level, marital status, occupation status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history. All statistical analyses were performed using SPSS 23.0 and the statistical significance level was set at P < 0.05 (2-tailed). Results Demographic and outcome variables of participants A total of 1600 participants meeting the inclusion criteria were screened, of whom 104 were excluded for not completing the baseline questionnaire or periodontal examination. Additionally, 411 patients without periodontitis and 105 with severe periodontitis were excluded. After excluding 196 participants who refused to participate, 784 participants were included in the study. During the follow-up period, 24 participants dropped out in G0, 8 in G1, 3 in G2, and 2 in G3. The primary reasons for dropout were loss to follow-up (uncontactable) and withdrawal of consent. To address potential bias from dropouts, we compared baseline characteristics between study completers and dropouts in each group, finding no significant differences (all P > 0.05). Finally, 747 participants completed all three assessments, with 172, 188, 193, and 194 participants in the G0, G1, G2, and G3, respectively (Fig. S1). At baseline, the participants had an average age of 65.26 ± 7.72 years, with 454 females (60.80%). There were no significant differences in sociodemographic characteristics across the four groups (P > 0.05). Baseline measurements for PD, CAL, FPG, HbA1c and BMI were 2.23 ± 0.73 mm, 3.06 ± 1.42 mm 7.64 ± 2.34 mmol/L, 6.96 ± 1.59% and 24.57 ± 3.31 kg/m2 respectively, with no significant differences between groups (P > 0.05). Additionally, 35.2% engaged in two or more kinds of oral health behaviors, and 7.0% of participants had a healthy lifestyle score of 4 (Table 1). Table 1Baseline characteristics of T2DM patients with mild or moderate periodontitis allocated to the control group and intervention groups.VariableTotal(n = 747)G0(n = 172)G1(n = 188)G2(n = 193)G3(n = 194)χ2/FP-valueDemographics Age (year), n (%) 35–59145 (19.4)26 (15.1)38 (20.2)48 (24.9)33 (17.0)6.9320.327 60–69349 (46.7)83 (48.3)89 (47.3)86 (44.6)91 (46.9) ≥ 70253 (33.9)63 (36.6)61 (32.4)59 (30.6)70 (36.1) Gender (female), n (%)454 (60.8)100 (58.1)121 (64.4)120 (62.2)113 (58.3)2.1950.533 Area (urban), n (%)353 (47.3)81 (47.1)89 (47.3)98 (50.8)105 (54.1)2.4800.479 Education level, n (%) Primary school or below289 (38.7)73 (42.4)78 (41.5)78 (40.4)60 (30.9) Secondary/high school390 (52.2)85 (49.4)98 (52.1)95 (49.2)112 (57.7)9.0030.173 College or above68 (9.1)14 (8.1)12 (6.4)20 (10.4)22 (11.3) Marital Status, n (%) Married696 (93.2)157 (91.3)177 (94.2)186 (96.4)176 (90.7)6.1910.103 Unmarried/divorced/ separated/widowed51 (6.8)15 (8.7)11 (5.8)7 (3.6)18 (9.3) Occupational status, n (%) Retired288 (38.6)66 (38.4)72 (38.3)68 (35.2)82 (42.3)4.7470.878 Unemployed213 (28.5)47 (27.3)49 (26.1)60 (31.1)57 (29.4) Physical labor187 (25.0)46 (26.7)52 (27.7)48 (24.9)41 (21.1) Mental labor59 (7.9)13 (7.6)15 (8.0)17 (8.8)14 (7.2) Duration of T2DM (year), n (%) < 5160 (21.4)35 (20.3)35 (18.6)56 (29.0)39 (20.1)7.1710.067 ≥ 5587 (78.6)137 (79.7)153 (81.4)137 (71.0)155 (79.9) Taking medication for diabetes, n (%)610 (81.7)132 (76.7)154 (81.9)166 (86.0)158 (81.4)5.2280.156 Disease history, n (%) Hypertension309 (41.4)72 (41.9)77 (41.0)80 (41.5)80 (41.2)4.2730.640 Dyslipidemia185 (24.8)47 (27.3)38 (20.2)44 (22.8)56 (28.9)11.1730.083 Cardiovascular and cerebrovascular diseases135 (18.1)32 (18.6)30 (16.0)30 (15.5)43 (22.2)3.6290.304 Malignant tumors30 (4.0)7 (4.1)5 (2.7)9 (4.7)9 (4.6)6.6150.358Outcomes Oral health behavior, n (%) < 2 kinds484 (64.8)109 (63.4)123 (65.4)125 (64.8)127 (65.5)0.2240.974 ≥ 2 kinds263 (35.2)63 (36.6)65 (34.6)68 (35.2)67 (34.5) PD (mm), mean ± SD2.23 ± 0.732.22 ± 0.662.26 ± 0.722.24 ± 0.712.20 ± 0.810.2680.848 CAL (mm), mean ± SD3.06 ± 1.423.03 ± 1.293.13 ± 1.423.01 ± 1.553.04 ± 1.410.2700.847 FPG (mmol/L), mean ± SD7.64 ± 2.347.75 ± 2.447.71 ± 2.437.57 ± 2.037.56 ± 2.450.3110.817 HbA1c (%), mean ± SD6.96 ± 1.597.03 ± 1.366.91 ± 1.486.96 ± 1.366.93 ± 2.020.1900.903 Healthy lifestyle score, n (%) 0–2287 (38.4)59 (34.3)77 (41.0)70 (36.3)81 (41.8)10.4910.123 3408 (54.6)107 (62.2)93 (49.5)109 (56.5)99 (51.0) 452 (7.0)6 (3.5)18 (9.6)14 (7.3)14 (7.2) BMI (kg/m2), mean ± SD24.57 ± 3.3124.89 ± 3.8824.47 ± 2.8124.54 ± 3.2624.42 ± 3.280.7100.546T2DM, type 2 diabetes mellitus; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; PD, probing depth; CAL, clinical attachment loss; BMI, body mass index. Baseline characteristics of T2DM patients with mild or moderate periodontitis allocated to the control group and intervention groups. Unmarried/divorced/ separated/widowed Cardiovascular and cerebrovascular diseases T2DM, type 2 diabetes mellitus; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; PD, probing depth; CAL, clinical attachment loss; BMI, body mass index. Of the 26,892 Ramfjord index sites examined across all participants, 29.7% had a PD ≥ 3 mm and 7.9% had a PD ≥ 4 mm. At the participant level, 86.6% (n = 647) had at least one site with PD ≥ 3 mm, 56.2% (n = 420) had at least one site with PD ≥ 4 mm, and 45.4% (n = 339) had two or more sites with PD ≥ 4 mm. The mean number of sites with PD ≥ 3 mm and PD ≥ 4 mm per participant was approximately 10.7 and 2.9, respectively. These findings are consistent with the localized pattern of mild-to-moderate periodontitis, in which a subset of sites exhibit deepened pockets despite a low whole-mouth mean PD. The distribution of participants stratified by the number of sites with PD ≥ 3 mm and PD ≥ 4 mm is shown in Tables S1 and S2. Evaluation of intervention effects on primary outcomes Table 2 showed the results of GEE analysis on outcome differences between groups at different time points. A significant group×time interaction effect was found for oral health behavior (P = 0.043), indicating that the proportion of participants exhibiting ≥ 2 oral health behaviors changed differentially over time among the four groups. Pairwise comparisons of estimated marginal means (Table 3) showed that the proportion of participants with ≥ 2 oral health behaviors increased significantly from T0 to T1 in both G1 and G3. At T1, G1 exhibited an 18.2% increase relative to T0 (Phi(φ) = 0.159, P < 0.001), and G3 exhibited a 24.2% increase (Phi(φ) = 0.234, P < 0.001). Between-group comparisons showed that at T1 and T2, the proportion of individuals in G3 with oral health behaviors was significantly higher than in G0 and G2. At T1, G3 exhibited a 22.8% increase relative to G0 (Phi(φ) = 0.245, P < 0.001) and a 14.7% increase relative to G2 (Phi(φ) = 0.156, P = 0.007). At T2, G3 showed an 11.6% increase compared with G0 (Phi(φ) = 0.201, P = 0.001) and a 17.8% increase compared with G2 (Phi(φ) = 0.182, P = 0.001). The results of the parameter estimates were consistent with the pairwise comparisons of estimated marginal means (Table S3). Trends in oral health behavior proportions over time in the intervention and control groups were shown in Fig. S2. Table 2GEE analysis of differences on outcomes between groups at different time points.OutcomesG0(n = 172)G1(n = 188)G2(n = 193)G3(n = 194)Time effectGroup effectInteraction effectWald χ2P-valueWald χ2P-valueWald χ2P-valuePrimary outcomes Oral health behavior ≥ 2 kindsa, n (%) T063 (36.6)65 (34.6)68 (35.2)67 (34.5)25.983 < 0.00123.944 < 0.00117.3890.043 T157 (33.1)95 (50.5)80 (41.5)112 (57.7) T260 (34.9)86 (45.7)72 (37.3)107 (55.2) T356 (32.6)89 (47.3)76 (39.4)104 (53.6) PD (mm)a, mean ± SD T02.22 ± 0.662.26 ± 0.722.24 ± 0.712.20 ± 0.8110.8670.0124.2180.2399.8920.359 T12.23 ± 0.972.18 ± 0.982.28 ± 0.932.16 ± 0.86 T22.20 ± 1.042.09 ± 0.942.23 ± 0.952.04 ± 0.98 T32.31 ± 0.942.08 ± 0.802.19 ± 0.842.02 ± 0.94 CAL (mm)a, mean ± SD T03.03 ± 1.293.13 ± 1.423.01 ± 1.553.04 ± 1.4114.5020.0023.2790.35114.8180.096 T13.06 ± 1.922.99 ± 1.492.88 ± 1.622.90 ± 1.50 T23.09 ± 1.892.71 ± 1.552.92 ± 1.822.70 ± 1.75 T33.07 ± 1.782.66 ± 1.642.90 ± 1.752.65 ± 1.54Secondary outcomes FPG (mmol/L)a, mean ± SD T07.75 ± 2.447.71 ± 2.437.57 ± 2.037.56 ± 2.4510.7210.0134.1640.24417.1580.046 T17.60 ± 2.207.60 ± 1.587.38 ± 1.877.28 ± 1.69 T27.88 ± 2.397.50 ± 2.597.20 ± 2.247.05 ± 1.87 T37.65 ± 2.127.32 ± 1.807.43 ± 1.917.35 ± 1.97 HbA1c (%)a, mean ± SD T07.03 ± 1.366.91 ± 1.486.96 ± 1.366.93 ± 2.0239.394 < 0.0013.2180.35918.0850.034 T16.97 ± 1.317.05 ± 1.166.89 ± 1.566.80 ± 1.15 T27.01 ± 1.166.70 ± 1.136.66 ± 1.676.54 ± 1.09 T36.93 ± 1.066.86 ± 1.136.80 ± 1.586.72 ± 1.10Tertiary outcomes Healthy lifestyle scoreb, mean ± SD T02.67 ± 0.582.66 ± 0.652.66 ± 0.692.64 ± 0.6423.376 < 0.0016.9410.07417.2580.044 T12.62 ± 0.882.79 ± 0.802.88 ± 0.812.92 ± 0.72 T22.60 ± 0.762.71 ± 0.742.72 ± 0.702.85 ± 0.82 T32.62 ± 0.842.69 ± 0.632.71 ± 0.632.79 ± 0.75 BMI (kg/m2)a, mean ± SD T024.89 ± 3.8824.47 ± 2.8124.54 ± 3.2624.42 ± 3.284.9030.1795.1260.16329.690 < 0.001 T124.88 ± 3.1124.53 ± 3.0724.64 ± 3.4824.37 ± 3.57 T225.06 ± 3.2024.58 ± 2.8724.58 ± 3.6524.33 ± 3.43 T325.10 ± 3.2324.60 ± 2.7724.48 ± 3.1723.71 ± 3.79FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; PD, probing depth; CAL, clinical attachment loss; BMI, body mass index.aAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history.bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, duration of T2DM, disease history and medication history. GEE analysis of differences on outcomes between groups at different time points. FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; PD, probing depth; CAL, clinical attachment loss; BMI, body mass index. aAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history. bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, duration of T2DM, disease history and medication history. Table 3A pairwise comparison of estimated marginal means of oral health behavior, PD and CAL.OutcomesMean difference SE Cohen’s d / Phi(φ)aP-valuecOral health behavior ≥ 2 kinds (%)b Within groups G1(T1-T0)0.1820.0500.159 < 0.001 G1(T2-T0)0.1190.0530.1080.021 G1(T3-T0)0.1480.0520.1330.004 G3(T1-T0)0.2420.0500.234 < 0.001 G3(T2-T0)0.2230.0510.213 < 0.001 G3(T3-T0)0.2110.0560.189 < 0.001 Between groups T1(G1-G0)0.1830.0530.1780.001 T1(G3-G0)0.2280.0530.245 < 0.001 T1(G3-G2)0.1470.0550.1560.007 T2(G3-G0)0.1160.0510.2010.001 T2(G3-G2)0.1780.0520.1820.001PD (mm)b Within groups G1(T3-T0) − 0.1670.076 − 0.2090.023 G3(T2-T0) − 0.1540.075 − 0.1820.039 G3(T3-T0) − 0.1720.084 − 0.2010.043 G3(T2-T1) − 0.1340.055 − 0.1380.022 G3(T3-T1) − 0.1440.062 − 0.1720.023 Between groups T3(G1-G0) − 0.2780.092 − 0.3300.003 T3(G3-G0) − 0.2530.097 − 0.2780.009 T3(G2-G1)0.1720.0860.2170.042CAL (mm)b Within groups G1(T2-T0) − 0.3110.144 − 0.2180.029 G1(T3-T0) − 0.3430.153 − 0.2330.028 G1(T2-T1) − 0.2540.124 − 0.1720.043 G1(T3-T1) − 0.2670.111 − 0.1670.015 G3(T2-T0) − 0.3390.120 − 0.2060.005 G3(T3-T0) − 0.3830.125 − 0.2740.003 G3(T2-T1) − 0.2510.062 − 0.0630.028 G3(T3-T1) − 0.3020.062 − 0.2100.022 Between groups T3(G1-G0) − 0.3650.182 − 0.2170.042 T3(G3-G0) − 0.4290.197 − 0.2530.017 T3(G3-G2) − 0.3740.165 − 0.2320.026PD, probing depth; CAL, clinical attachment loss; SE, standard error.aCohen’s d: effect size for continuous variables; Phi(φ): effect size for categorical.bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history.cP < 0.05 is considered statistically significant. Results presented are those with P < 0.05. A pairwise comparison of estimated marginal means of oral health behavior, PD and CAL. PD, probing depth; CAL, clinical attachment loss; SE, standard error. aCohen’s d: effect size for continuous variables; Phi(φ): effect size for categorical. bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history. cP < 0.05 is considered statistically significant. Results presented are those with P < 0.05. For the objective clinical periodontal parameters, the overall group×time interaction effects were not statistically significant (PD: P = 0.359; CAL: P = 0.096). Nevertheless, significant main effects of time were observed for both outcomes (PD: P = 0.012; CAL: P = 0.002), indicating that PD and CAL changed over time across the four groups (Table 2). The parameter estimates (Table S3) revealed that at T3, PD was significantly lower in both G1 (B = − 0.250, P = 0.010) and G3 (B = − 0.246, P = 0.028) compared to G0. Similarly, CAL was significantly reduced in G1 (T2: B = − 0.465, P = 0.029; T3: B = − 0.507, P = 0.012) and G3 (T2: B = − 0.397, P = 0.039; T3: B = − 0.432, P = 0.020) at both T2 and T3 relative to G0, demonstrating significant differences between groups at specific time points. Pairwise comparisons of estimated marginal means (Table 3) provided further granularity. Within the comprehensive intervention group (G3), PD significantly decreased by 0.154 mm from T0 to T2 (Cohen’s d = − 0.182, P = 0.039) and by 0.172 mm from T0 to T3 (Cohen’s d = − 0.201, P = 0.043). Similarly, CAL significantly decreased by 0.339 mm from T0 to T2 (Cohen’s d = − 0.206, P = 0.005) and by 0.383 mm from T0 to T3 (Cohen’s d = − 0.274, P = 0.003). Notably, the reduction in G3 was greater at T3 than at T2, suggesting a gradual and sustained improvement in clinical attachment levels during the follow-up period. Between-group comparisons at T3 showed that PD decreased by 0.253 mm more in G3 than in G0 (Cohen’s d = − 0.278, P = 0.009), and CAL decreased by 0.429 mm more in G3 than in G0 (Cohen’s d = − 0.253, P = 0.017). Significant reductions relative to G0 were also observed in G1 (PD: − 0.278 mm, P = 0.003; CAL: − 0.365 mm, P = 0.042). Trends in PD and CAL over time in the intervention and control groups are shown in Fig. S3, S4. Taken together, these results provide objective evidence that the periodontal and comprehensive self-management interventions led to clinically meaningful improvements in periodontal tissue health, with the greatest benefits observed at the 6-month follow-up. Evaluation of intervention effects on secondary outcomes The GEE model revealed significant group×time interaction effects for FPG and HbA1c levels (P = 0.046 and P = 0.034, respectively), indicating that glycemic outcomes changed differentially over time among the four groups (Table 2). The results of the parameter estimates (Table S4) indicated that, at T2, FPG and HbA1c levels in G3 were lower than in G0, with significant between-group differences for FPG (B = − 0.593, P = 0.011) and HbA1c (B = − 0.314, P = 0.040). Pairwise comparisons of estimated marginal means (Table 4) showed that within the comprehensive intervention group (G3), FPG significantly decreased by 0.328 mmol/L from T0 to T1 (Cohen’s d = − 0.472, P = 0.046) and by 0.568 mmol/L from T0 to T2 (Cohen’s d = − 0.839, P = 0.001). Between-group comparisons at T2 demonstrated that FPG decreased by 0.687 mmol/L in G3 than in G0 (Cohen’s d = − 0.905, P = 0.003) and by 0.514 mmol/L more in G2 than in G0 (Cohen’s d = − 0.659, P = 0.033). Table 4A pairwise comparison of estimated marginal means of FPG and HbA1c level.OutcomesMean difference SE Cohen’s daP-valuecFPG (mmol/L)b Within groups G3(T1-T0) − 0.3280.164 − 0.4720.046 G3(T2-T0) − 0.5680.176 − 0.8390.001 G3(T3-T2)0.2890.1110.4430.010 Between groups T2(G2-G0) − 0.5140.241 − 0.6590.033 T2(G3-G0) − 0.6870.256 − 0.9050.003HbA1c (%)b Within groups G1(T2-T0) − 0.1980.099 − 0.4590.045 G1(T2-T1) − 0.3580.079 − 0.853 < 0.001 G1(T3-T1) − 0.2160.088 − 0.5040.014 G2(T2-T0) − 0.2900.111 − 0.4480.009 G2(T2-T1) − 0.2190.071 − 0.3350.002 G2(T3-T2)0.1310.0570.2040.023 G3(T2-T0) − 0.4000.122 − 0.8110.001 G3(T2-T1) − 0.2850.071 − 0.577 < 0.001 G3(T3-T2)0.2160.0630.4350.001 Between groups T2(G3-G1) − 0.3870.118 − 0.9240.001FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; SE, standard error.aCohen’s d: effect size for continuous variables.bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history.cP < 0.05 is considered statistically significant. Results presented are those with p-values < 0.05. A pairwise comparison of estimated marginal means of FPG and HbA1c level. FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; SE, standard error. aCohen’s d: effect size for continuous variables. bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history. cP < 0.05 is considered statistically significant. Results presented are those with p-values < 0.05. Pairwise comparisons of estimated marginal means (Table 4) further showed that within G2, HbA1c significantly decreased by 0.290% from T0 to T2 (Cohen’s d = − 0.448, P = 0.009) and by 0.219% from T1 to T2 (Cohen’s d = − 0.335, P = 0.002), but partially rebounded at T3, increasing by 0.131% compared with T2 (Cohen’s d = 0.204, P = 0.023). In G3, HbA1c decreased by 0.400% from T0 to T2 (Cohen’s d = − 0.811, P = 0.001) and by 0.258% from T1 to T2 (Cohen’s d = − 0.577, P < 0.001), with a subsequent increase of 0.216% at T3 relative to T2 (Cohen’s d = 0.435, P = 0.001). Despite this partial rebound, HbA1c at T3 remained below baseline levels in both groups. Between-group comparisons at T2 demonstrated that HbA1c in G3 was 0.387% lower than in G1 (Cohen’s d = − 0.924, P = 0.001). Trends in HbA1c and FPG levels over time in the intervention and control groups are shown in Fig. S5, S6. In all three intervention groups, FPG and HbA1c declined from baseline to T2, with the greatest improvement observed in G3. At T3, FPG and HbA1c levels increased slightly in G2 and G3 but remained below baseline values. Evaluation of intervention effects on tertiary outcomes The GEE results (Table 2) showed significant group×time interaction effects for healthy lifestyle score (P = 0.044) and BMI (P < 0.001), signifying that the intervention groups differed in their trajectories of healthy lifestyle score improvement and BMI change over time. Parameter estimates for healthy lifestyle score and BMI are presented in Table S5. Pairwise comparisons of estimated marginal means (Table 5) revealed that the healthy lifestyle score in G2 and G3 was significantly higher at T1, T2, and T3 compared to T0 (P < 0.05). Between-group comparisons showed that at T1, the scores in G2 and G3 were higher than in G0. The results also showed that G3 had significantly lower BMI at T3 compared to T0, T1, and T2 (P < 0.05). Between-group comparisons indicated that G3 had a lower BMI at T3 than G0, G1, and G2 (P < 0.05). Trends in healthy lifestyle scores and BMI over time are shown in Fig. S7, S8. Table 5A pairwise comparison of estimated marginal means of healthy lifestyle score and BMI.OutcomesMean difference SE Cohen’s daP-valuedHealthy lifestyle scoreb Within groups G2(T1-T0)0.3180.0801.310 < 0.001 G2(T2-T0)0.1390.0680.5710.039 G2(T3-T0)0.1360.0680.5540.044 G2(T2-T1) − 0.1820.068 − 0.7550.008 G2(T3-T1) − 0.1790.065 − 0.7570.006 G3(T1-T0)0.2870.0621.107 < 0.001 G3(T2-T0)0.2210.0630.837 < 0.001 G3(T3-T0)0.1650.0630.6360.009 Between groups T1(G2-G0)0.2860.0880.9560.001 T1(G3-G0)0.2310.0810.7820.004BMI (kg/m2)c Within groups G3(T3-T0) − 0.6520.206 − 0.4270.002 G3(T3-T1) − 0.6390.174 − 0.419 < 0.001 G3(T3-T2) − 0.6040.159 − 0.397 < 0.001 Between groups T3(G3-G0) − 1.4700.357 − 1.044 < 0.001 T3(G3-G1) − 0.8890.331 − 0.6500.007 T3(G3-G2) − 0.9860.343 − 0.6960.004BMI, body mass index; SE, standard error.aCohen’s d: effect size for continuous variables.bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, duration of T2DM, disease history and medication history.cAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history.dP < 0.05 is considered statistically significant. Results presented are those with P < 0.05. A pairwise comparison of estimated marginal means of healthy lifestyle score and BMI. BMI, body mass index; SE, standard error. aCohen’s d: effect size for continuous variables. bAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, duration of T2DM, disease history and medication history. cAdjusted for age, gender, area, education level, marital status, occupational status, sleep duration, healthy lifestyle score, duration of T2DM, disease history and medication history. dP < 0.05 is considered statistically significant. Results presented are those with P < 0.05. Discussion Our findings demonstrate that the integrated intervention yielded significant improvements in clinical periodontal status, as evidenced by reductions in PD and CAL, alongside improvements on glycemic levels, BMI and healthy lifestyle scores. These results complement and extend our previous report which focused on improvements in patient-reported oral health symptoms and psychosocial measures15. The assessment of PD and CAL by dental professionals represents a gold-standard, objective measure of periodontal tissue destruction. While patient-reported outcomes are crucial for understanding the symptomatic burden and impact on quality of life, they can be subjective and influenced by perception. The significant improvements in these objective clinical parameters observed in our study provide stronger, more definitive evidence that the self-management intervention altered the underlying disease state of periodontitis, rather than merely improving symptom awareness. The reductions in PD and CAL objectively indicate a decrease in the anatomical niches that harbor bacterial biofilms and sustain periodontal inflammation—a biological plausibility supported by immunohistochemical evidence that deepened pockets provide an ulcerated epithelial surface area through which pro-inflammatory mediators (e.g., IL-1β, TNF-α, CRP) gain systemic access25,26. It is, therefore, plausible that the structural improvements in PD and CAL contributed to the concurrent glycemic improvements observed in the comprehensive intervention group by reducing the total periodontal inflammatory burden. This finding is consistent with the concept that even modest reductions in the inflamed periodontal surface area can attenuate systemic inflammatory responses and improve insulin sensitivity27. Diabetic patients are more prone to oral health issues, and most T2DM patients lack sufficient knowledge about the relationship between diabetes and periodontal health, leading to poor adherence to oral health behaviors. Existing studies have demonstrated that oral health education or hygiene guidance can effectively improve oral health behaviors and periodontal indicators in diabetic patients19,28–32. Our study extends these findings by showing that, when periodontal self-management is integrated with diabetes self-management, the improvements in oral health behaviors and clinical periodontal parameters are greater than those achieved through single-focus interventions. The combined intervention group showed the most notable improvements, and at 6 months post-intervention, the improvements of periodontal indicators reached their optimal levels. This gradual pattern of improvement suggests that periodontal health improvement through self-management is a cumulative process, consistent with the gradual nature of behavioral change, rather than an immediate therapeutic effect. The sustained effect at 6 months also highlights the durability of the behavioral changes achieved through the structured self-management program, suggesting that the intervention empowered patients with lasting self-care capabilities rather than producing only transient effects. This has particular relevance for community-based settings where access to professional periodontal treatment is limited and self-management becomes the primary vehicle for maintaining periodontal health. Glycemic improvements in the comprehensive group reinforce the bidirectional diabetes–periodontitis relationship, likely mediated by reduced periodontal inflammation enhancing insulin sensitivity. This study extends our earlier findings15 by quantifying HbA1c and FPG reductions: the combined intervention lowered FPG by 0.328–0.568 mmol/L and HbA1c by 0.4%, outperforming single-focus interventions. These findings align with previous studies11,33,34, confirming the positive impact of diabetes self-management on blood glucose control. Although the periodontitis-only intervention (G1) also showed HbA1c decreases (0.198–0.216%), consistent with hospital-based RCTs35,36, our community-based model demonstrates the superiority of integrated self-management. Notably, glycemic levels partially rebounded by 6 months. This attenuation pattern, despite values remaining below baseline, is consistent with behavioral theories which posit that the maintenance of health behavior change requires ongoing motivation, self-monitoring, and environmental support37. Conversely, a 2024 study of the Diabetes Share Care Program in Taiwan demonstrated that Diabetes self-management education (DSME) could sustain glycemic control for at least 2 years when periodic reinforcement was embedded in the care model38. The variability in long-term outcomes may be attributable to multiple contextual factors. A 2023 network meta-analysis highlighted that the effectiveness of self-management interventions is context-dependent, with heterogeneity across studies potentially driven by differences in intervention providers, delivery mode (individual vs. group), session frequency and duration, and the presence or absence of post-intervention reinforcement39. Collectively, the evidence underscores the need to embed sustained support strategies—such as quarterly booster sessions and digital self‑management tools—into routine public health services to preserve the metabolic gains achieved during the active intervention. The American Diabetes Association advocates self-management as a foundational strategy for achieving lifestyle modification in T2DM. Our results align with this guidance, demonstrating that both the diabetes-only (G2) and combined intervention (G3) groups exhibited significant improvements in lifestyle scores, indicating the efficacy of self-management in facilitating short-term behavioral change. These findings align with previous research39–41. Although the periodontitis-only intervention (G1) did not yield statistically significant improvements in the healthy lifestyle score, its temporal trend paralleled those observed in G2 and G3. This likely reflects the intervention’s focus on oral health knowledge and behaviors without addressing broader lifestyle modification, thereby limiting its impact on the composite score. Notably, the integrated intervention group manifested a progressive reduction in body mass index (BMI) throughout the study period. This finding carries clinical importance, as even modest weight reduction improves β-cell function and glycemic control in individuals with type 2 diabetes42. Our findings, consistent with existing literature, suggest that structured self-management interventions enhance patient self-efficacy through multidimensional education and support, thereby concurrently addressing weight management and glycemic targets39,43. This integrated approach represents a viable and cost-effective management paradigm for T2DM patients with comorbid overweight/obesity. Taken together, although the periodontal inflamed surface area (PISA)—a more direct measure of the inflammatory burden—was not quantified in this community‑based trial, converging lines of evidence support the interpretation that the intervention reduced the overall periodontal inflammatory status: (i) the significant structural improvements in PD and CAL; (ii) the previously published reductions in patient‑reported gingival bleeding and swelling15; and (iii) the concurrent metabolic improvements, which are consistent with a pathway of attenuated systemic inflammation. This triangulation allows us to infer, with reasonable scientific confidence, that periodontal inflammatory burden decreased, even in the absence of a direct PISA measurement. Advantages and limitations This study’s strength lies in integrating periodontal intervention into diabetes self-management, making it applicable to primary healthcare settings. This approach effectively slows complications and optimizes primary healthcare resources. By establishing a collaborative mechanism among oral specialty hospitals, primary healthcare institutions, and public health agencies, the study explores an “integrated medical-prevention” model, providing data support and decision-making evidence for diabetes-related periodontitis prevention strategies and health policy formulation. Several limitations warrant consideration. First, although participant attrition occurred during follow-up, comparative analysis revealed no significant differences in baseline characteristics between completers and dropouts, suggesting minimal impact of attrition bias on primary outcomes. Second, the use of the Ramfjord index teeth for partial-mouth periodontal recording, while validated for epidemiological surveys24, may slightly underestimate the prevalence of deep pockets compared with full-mouth examination. Third, the study was conducted within urban Chinese communities with established primary healthcare systems, which may limit generalizability to rural areas or regions with different healthcare infrastructures. External validation through multi-center trials across diverse populations is warranted. Fourth, the reliance on self-reported oral health behaviors may introduce social desirability bias, though we attempted to mitigate this through trained interviewers and standardized questionnaires. Fifth, the 6-month follow-up period, while adequate for detecting short-term improvements in glycemic and periodontal parameters, is insufficient to assess long-term sustainability of behavioral changes, complication prevention, or cost-effectiveness. Future research should prioritize longer-term trials (≥ 12 months) examining hard endpoints (e.g., cardiovascular events, tooth loss) and incorporating comprehensive economic evaluations. Finally, further mechanistic studies should incorporate PISA or bleeding‑on‑probing indices as continuous measures of inflammatory surface area, together with serial serum biomarkers (e.g., hs‑CRP, IL‑6), to formally evaluate whether reduced periodontal inflammation mediates the observed metabolic improvements. Conclusion This first integrated periodontal and diabetes self-management intervention improved oral health behaviors, PD, CAL, glycemic levels, lifestyle, and BMI in community T2DM patients with periodontitis. Effects declined after 6 months, underscoring the need for maintenance strategies like booster sessions or digital tools. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2