ABSTRACT
Aim
To evaluate the association between diabetes (types 1 and 2) and peri‐implantitis through a register‐based cohort study.
Methods
Four groups of individuals with dental implants were identified using multiple Swedish nationwide registers—two groups with diabetes (type 1, T1D; type 2, T2D) and two groups without diabetes (non‐T1D, non‐T2D). Longitudinal data from 2010 to 2020 were analysed. Peri‐implantitis was defined as any registered treatment of peri‐implantitis. Prevalence of peri‐implantitis (n = 18,975) was evaluated using covariate‐adjusted logistic regression analyses. Incidence (n = 2030) was compared using survival analyses across groups matched by age, gender, education, income and the number of implants, through propensity scores.
Results
Peri‐implantitis was more frequent among those with T1D compared to non‐T1D (21.1% vs. 15.2%; OR 1.46, 95% CI: 1.05–2.04), whereas the prevalence was similar in T2D and non‐T2D (20.5% vs. 18.2%; OR 1.06, 95% CI: 0.98–1.16). The hazard ratios for incident peri‐implantitis were 1.52 (95% CI: 0.96–2.42) and 1.36 (95% CI: 1.02–1.82) for T1D and T2D, respectively.
Conclusions
T1D and T2D were associated with a higher risk for peri‐implantitis. While the elevated risk for peri‐implantitis in T1D was particularly apparent in prevalence estimates, the association for T2D was evident mainly in terms of incidence.
Keywords: dental implants, diabetes mellitus, peri‐Implantitis, risk assessment
1. Introduction
Diabetes is a major global public health challenge, posing a significant burden on healthcare systems worldwide (Zhou et al. 2024). Epidemiological studies and consensus reports from European and World Workshops on Periodontology have highlighted diabetes as one of the major risk factors for periodontitis (e.g., Papapanou et al. 2018). In a nationwide register‐based study, we demonstrated that type 2 diabetes (T2D) is associated with both periodontitis and tooth loss, as is poorly controlled type 1 diabetes (T1D; Trullenque‐Eriksson et al. 2024).
Rehabilitation with dental implants is a standard treatment alternative following tooth loss. Although the use of implant‐supported restorations is a well‐established method, approximately one in five patients develop peri‐implantitis over time (e.g., Lee et al. 2017). Peri‐implantitis is an inflammatory condition affecting tissues surrounding dental implants. If left untreated, the disease leads to further loss of supporting bone and, ultimately, implant loss (Berglundh et al. 2018). Patients with a history of periodontitis are at particularly high risk of developing peri‐implantitis (Schwarz et al. 2018).
The potential association between diabetes and peri‐implantitis has been widely discussed. While a link is plausible, given the impact of diabetes on inflammatory responses and its association to periodontitis (Nibali et al. 2022), evidence on an association between diabetes and peri‐implantitis remains limited (Monje et al. 2017; Schwarz et al. 2018). Conflicting results have been reported in cross‐sectional studies including small groups of individuals with diabetes. Some studies supported an association between the two conditions (Ferreira et al. 2006), while others did not (Dalago et al. 2017; Derks et al. 2016). Notably, most of these data derived from patients with T2D or unspecified diabetes type, whereas information on T1D is scarce.
As implant therapy is a common restorative approach in Sweden (SKaPa 2024), the population identified in our previous study on the association between diabetes and periodontitis (Trullenque‐Eriksson et al. 2024) also included patients with dental implants. Given the difficulties in recruiting adequate patient samples, particularly for T1D, we believed a register‐based approach might circumvent some of the limitations of traditional study designs. Here we report data from multiple nationwide registers, with the aim to evaluate the association between T1D/T2D and peri‐implantitis.
2. Material and Methods
2.1. Study Design
This register‐based retrospective cohort study included dental implant–carrying individuals with and without diabetes. The study protocol was approved by the Swedish Ethical Review Authority (registration number: 2019–04140) and was reported in accordance with STROBE guidelines. Initially, four cohorts were identified by cross‐referencing the Swedish National Diabetes Register (NDR), the Swedish Quality Registry for Caries and Periodontal disease (SKaPa) and the Swedish Total Population Register (Statistics Sweden) for the period 2010–2020. Additional data on socioeconomic status, emigration and death were obtained from Statistics Sweden and the National Board of Health and Welfare, as described elsewhere (Trullenque‐Eriksson et al. 2024). Adults with dental implants were eligible for inclusion. Specific selection criteria are outlined in Figure 1.
FIGURE 1.
Selection criteria and matching process for the study participants.
2.2. Diabetes (Exposure)
Individuals with T1D and T2D were identified based on clinical diagnosis recorded in NDR (2012/2013 or earlier). When available, blood haemoglobin A1c (HbA1c) data spanning at least 5 years (2010–2020) were used to categorise participants by glycaemic control (good: HbA1c < 52 mmol/mol at ≥ 75% of measurements; poor: HbA1c > 62 mmol/mol at ≥ 75% of measurements).
2.3. Peri‐Implantitis and Implant Loss (Outcomes)
The primary outcome (i.e., peri‐implantitis) was defined based on treatment codes, that is, first documentation of non‐surgical/surgical treatment/implant removal with peri‐implantitis as the associated diagnostic code. Implant removal due to peri‐implantitis (‘implant loss’) was analysed separately as a secondary outcome. Treatment codes were retrieved annually from SKaPa for the time period 2010–2020. Repeated records were not considered.
2.4. Covariates
We considered the following covariates: age in 2012, gender (legal sex), education and income (latest available by 2012), as well as the number of implants and the presence of periodontitis. Education was categorised as follows: up to lower secondary education, upper secondary to post‐secondary education < 2 years and post‐secondary ≥ 2 years to tertiary education. Annual income included salary, pension and capital gains (continuous variable). Periodontitis was defined as the presence of three or more teeth with probing depths ≥ 6 mm, based on annual data from routine clinical examinations. Data on covariates were retrieved from SKaPa, Statistics Sweden and the National Board of Health and Welfare (Trullenque‐Eriksson et al. 2024).
2.5. Study Population and Statistical Analyses
Statistical analyses were conducted in two steps.
2.5.1. Prevalent Peri‐Implantitis and Peri‐Implantitis‐Associated Implant Loss
In the first step, we assessed the association between diabetes (diagnosis in or before 2012) and peri‐implantitis/implant loss (2010–2020). Individuals with one or more implants (in or before 2012) from the four diabetes/non‐diabetes cohorts were included. The sample consisted of 361 individuals with T1D (mean age 55.6 ± 17.6 years, 52.1% female, mean number of implants: 2.8 ± 2.5), 791 without diabetes (non‐T1D; mean age 56.7 ± 18.4 years, 47.0% female, mean number of implants: 2.6 ± 2.2), 4597 individuals with T2D (mean age 72.0 ± 9.7 years, 45.4% female, mean number of implants: 3.8 ± 2.9) and 13,226 without diabetes (non‐T2D; mean age 69.6 ± 11.5 years, 48.4% female, mean number of implants: 3.3 ± 2.6; Table 1). Standardised differences between diabetes and non‐diabetes groups were calculated for continuous and categorical baseline characteristics using the stddiff command (Bayoumi 2016), following the method by Yang and Dalton (2012). Glycaemic control was categorised as poor for 101 participants with T1D (28.0%) and 347 with T2D (7.6%), and as good for 18 individuals with T1D (5.0%) and 890 with T2D (19.4%).
TABLE 1.
Baseline characteristics (2010–2012) for individuals contributing to the analysis of prevalent peri‐implantitis.
| No diabetes (n = 791) | Type 1 diabetes (n = 361) | Standardised difference | |||
|---|---|---|---|---|---|
| Age | 56.7 | (18.4) | 55.6 | (17.6) | 0.06 |
| Gender | |||||
| Female | 372 | 47.0% | 188 | 52.1% | |
| Male | 419 | 53.0% | 173 | 47.9% | 0.10 |
| Place of birth | |||||
| Sweden | 677 | 85.6% | 322 | 89.2% | |
| Scandinavia (excl. Sweden) | 32 | 4.0% | 12 | 3.3% | 0.11 |
| Europe (excl. Scandinavia) | 51 | 6.4% | 16 | 4.4% | |
| Other | 31 | 3.9% | 11 | 3.0% | |
| National area according to NUTS1 | |||||
| East Sweden | 233 | 29.5% | 117 | 32.4% | 0.08 |
| South Sweden | 356 | 45.0% | 162 | 44.9% | |
| North Sweden | 202 | 25.5% | 82 | 22.7% | |
| Civil status | |||||
| Single | 226 | 28.6% | 114 | 31.6% | 0.07 |
| Married/registered partnership | 379 | 47.9% | 165 | 45.7% | |
| Divorced/separated partner | 113 | 14.3% | 50 | 13.9% | |
| Widowed/surviving partner | 73 | 9.2% | 32 | 8.9% | |
| Education | |||||
| Up to lower secondary education | 160 | 20.2% | 71 | 19.7% | |
| Upper secondary to post‐secondary education < 2 years | 394 | 49.8% | 168 | 46.5% | 0.08 |
| Post‐secondary ≥ 2 years to tertiary education | 237 | 30.0% | 122 | 33.8% | |
| Annual income (SEK) | 191,300 | [201,900] | 191,800 | [193,300] | −0.11 |
| Number of implants | 2.6 | (2.2) | 2.8 | (2.5) | −0.05 |
| Number of teeth | 19.4 | (7.6) | 18.6 | (8.5) | 0.10 |
| Years of follow‐up | 7.2 | (3.1) | 7.1 | (3.1) | 0.02 |
| Years with ≥ 1 examination | 3.7 | (2.4) | 3.6 | (2.4) | 0.02 |
| Periodontitis (2010–2012) | |||||
| No | 717 | 91.3% | 324 | 90.8% | |
| Yes | 68 | 8.7% | 33 | 9.2% | 0.02 |
| Periodontitis (2010–2020) | |||||
| No | 666 | 84.2% | 295 | 82.2% | |
| Yes | 125 | 15.8% | 64 | 17.8% | 0.05 |
| No diabetes (n = 13,226) | Type 2 diabetes (n = 4597) | Standardised difference | |||
|---|---|---|---|---|---|
| Age | 69.6 | (11.5) | 72.0 | (9.7) | −0.22 |
| Gender | |||||
| Female | 6404 | 48.4% | 2086 | 45.4% | |
| Male | 6822 | 51.6% | 2511 | 54.6% | 0.06 |
| Place of birth | |||||
| Sweden | 11,293 | 85.4% | 3852 | 83.8% | |
| Scandinavia (excl. Sweden) | 710 | 5.4% | 246 | 5.4% | 0.06 |
| Europe (excl. Scandinavia) | 803 | 6.1% | 315 | 6.9% | |
| Other | 420 | 3.2% | 184 | 4.0% | |
| National area according to NUTS1 | |||||
| East Sweden | 3890 | 29.4% | 1213 | 26.4% | 0.08 |
| South Sweden | 5779 | 43.7% | 2170 | 47.2% | |
| North Sweden | 3557 | 26.9% | 1214 | 26.4% | |
| Civil status | |||||
| Single | 1425 | 10.8% | 454 | 9.9% | 0.09 |
| Married/registered partnership | 7514 | 56.8% | 2508 | 54.6% | |
| Divorced/separated partner | 2289 | 17.3% | 796 | 17.3% | |
| Widowed/surviving partner | 1998 | 15.1% | 839 | 18.3% | |
| Education | |||||
| Up to lower secondary education | 4014 | 30.3% | 1696 | 36.9% | |
| Upper secondary to post‐secondary education < 2 years | 6005 | 45.4% | 2131 | 46.4% | 0.20 |
| Post‐secondary ≥ 2 years to tertiary education | 3207 | 24.2% | 770 | 16.8% | |
| Annual income (SEK) | 168,650 | [127,000] | 153,000 | [93,000] | 0.14 |
| Number of implants | 3.3 | (2.6) | 3.8 | (2.9) | −0.21 |
| Number of teeth | 16.0 | (8.0) | 13.7 | (7.9) | 0.31 |
| Years of follow‐up | 6.9 | (3.1) | 6.3 | (3.2) | 0.20 |
| Years with ≥ 1 examination | 3.7 | (2.5) | 3.2 | (2.4) | 0.19 |
| Periodontitis (2010–2012) | |||||
| No | 11,683 | 89.6% | 3963 | 87.5% | |
| Yes | 1357 | 10.4% | 566 | 12.5% | 0.07 |
| Periodontitis (2010–2020) | |||||
| No | 10,446 | 79.3% | 3539 | 77.3% | |
| Yes | 2731 | 20.7% | 1041 | 22.7% | 0.05 |
Note: Categorical data are presented as frequencies and percentages. The continuous variables age and number of years in the lowest fifth percentile of income are presented as mean (SD); income is presented as median [IQR].
Logistic regression analysis was used, adjusted for age, gender, education, income and the number of implants. Outcomes were expressed as adjusted odds ratios (ORs) accompanied by 95% confidence intervals (95% CI). Analyses were repeated including periodontitis (recorded at least once between 2010 and 2020) as a covariate. Additionally, subgroups of T1D/T2D with good versus poor glycaemic control were considered. A sensitivity analysis was conducted using probing depth (PD ≥ 6 mm at one or more implants) instead of treatment codes to define peri‐implantitis.
2.5.2. Incident Peri‐Implantitis and Peri‐Implantitis‐Associated Implant Loss
In the second step, we evaluated the incidence of peri‐implantitis/implant loss in the diabetes and non‐diabetes cohorts. Different analytical approaches were employed for T1D and T2D due to sample size restrictions. Timelines and matching processes are illustrated in Figure 1. Theoretical assumptions are depicted in the directed acyclic graph provided in Figure A1 of Supporting Information.
Among T1D/non‐T1D, individuals with implants not affected by peri‐implantitis at baseline were eligible (one or more implants registered in 2010–2012, no peri‐implantitis by 2012). Individuals with T1D (by 2012) were matched to controls from the non‐T1D cohort using propensity scores based on age, gender, education, income and the number of implants (in or earlier than 2012; average treatment effects on the treated [ATT]; nearest neighbour 1:1 without replacement; command kmatch ps; Jann 2017). Hence, 290 individuals with T1D (mean age 54.0 ± 17.4 years, 52.8% female, mean number of implants: 2.4 ± 2.3) and 290 without T1D (mean age 55.3 ± 17.9 years, 53.4% female, mean number of implants: 2.5 ± 2.1) were included (Table 2). Incident peri‐implantitis/implant loss up to 2020 was then compared through flexible parametric survival analysis, using restricted cubic splines (command stpm2; Lambert 2010; baseline: 2013).
TABLE 2.
Baseline a characteristics for individuals contributing to the analysis of incident peri‐implantitis (propensity score–matched cohorts).
| Non‐T1D (n = 290) | Type 1 diabetes (n = 290) | Standardised difference | |||
|---|---|---|---|---|---|
| Age (mean) | 55.3 | (17.9) | 54.0 | (17.4) | 0.07 |
| Gender | |||||
| Female | 155 | 53.4% | 153 | 52.8% | |
| Male | 135 | 46.6% | 137 | 47.2% | 0.01 |
| Place of birth | |||||
| Sweden | 251 | 86.6% | 257 | 88.6% | |
| Scandinavia (excl. Sweden) | 11 | 3.8% | 10 | 3.4% | 0.06 |
| Europe (excl. Scandinavia) | 17 | 5.9% | 14 | 4.8% | |
| Other | 11 | 3.8% | 9 | 3.1% | |
| National area according to NUTS1 | |||||
| East Sweden | 80 | 27.6% | 97 | 33.4% | 0.14 |
| South Sweden | 133 | 45.9% | 128 | 44.1% | |
| North Sweden | 77 | 26.6% | 65 | 22.4% | |
| Civil status | |||||
| Single | 81 | 27.9% | 96 | 33.1% | 0.15 |
| Married/registered partnership | 150 | 51.7% | 132 | 45.5% | |
| Divorced/separated partner | 33 | 11.4% | 39 | 13.4% | |
| Widowed/surviving partner | 26 | 9.0% | 23 | 7.9% | |
| Education | |||||
| Up to lower secondary education | 44 | 15.2% | 45 | 15.5% | |
| Upper secondary to post‐secondary education < 2 years | 134 | 46.2% | 139 | 47.9% | 0.04 |
| Post‐secondary ≥ 2 years to tertiary education | 112 | 38.6% | 106 | 36.6% | |
| Annual Income (SEK) | 205,300 | [210,200] | 216,200 | [202,900] | −0.06 |
| Number of implants by 2012 | 2.5 | (2.1) | 2.4 | (2.3) | 0.01 |
| Number of teeth | 20.4 | (7.1) | 19.8 | (7.8) | 0.09 |
| Years of follow‐up | 8.0 | (2.2) | 7.9 | (2.3) | 0.03 |
| Years with ≥ 1 examination | 4.1 | (2.1) | 4.1 | (2.2) | −0.01 |
| Periodontitis (2010–2012) | |||||
| No | 269 | 94.1% | 262 | 91.3% | |
| Yes | 17 | 5.9% | 25 | 8.7% | 0.11 |
| Non‐T2D (n = 725) | Type 2 diabetes (n = 725) | Standardised difference | |||
|---|---|---|---|---|---|
| Age | 67.7 | (10.0) | 67.3 | (9.2) | 0.04 |
| Gender | |||||
| Female | 310 | 42.8% | 321 | 44.3% | |
| Male | 415 | 57.2% | 404 | 55.7% | 0.03 |
| Place of birth | |||||
| Sweden | 608 | 83.9% | 581 | 80.1% | |
| Scandinavia (excl. Sweden) | 38 | 5.2% | 47 | 6.5% | 0.15 |
| Europe (excl. Scandinavia) | 49 | 6.8% | 44 | 6.1% | |
| Other | 30 | 4.1% | 53 | 7.3% | |
| National area according to NUTS1 | |||||
| East Sweden | 208 | 28.7% | 210 | 29.0% | 0.06 |
| South Sweden | 329 | 45.4% | 344 | 47.4% | |
| North Sweden | 188 | 25.9% | 171 | 23.6% | |
| Civil status | |||||
| Single | 83 | 11.4% | 86 | 11.9% | 0.03 |
| Married/registered partnership | 440 | 60.7% | 431 | 59.4% | |
| Divorced/separated partner | 123 | 17.0% | 127 | 17.5% | |
| Widowed/surviving partner | 79 | 10.9% | 81 | 11.2% | |
| Education | |||||
| Up to lower secondary education | 224 | 30.9% | 222 | 30.6% | |
| Upper secondary to post‐secondary education < 2 years | 389 | 53.7% | 379 | 52.3% | 0.05 |
| Post‐secondary ≥ 2 years to tertiary education | 112 | 15.4% | 124 | 17.1% | |
| Annual income (SEK) | 168,000 | [128,900] | 161,900 | [137,200] | 0.04 |
| Number of implants by 2013 | 2.7 | (2.3) | 2.7 | (2.3) | 0.01 |
| Number of teeth by 2013 | 17.0 | (7.6) | 16.2 | (7.3) | 0.11 |
| Years of follow‐up | 8.4 | (1.8) | 8.3 | (1.9) | 0.07 |
| Years with ≥ 1 examination | 4.9 | (2.0) | 4.7 | (2.0) | 0.10 |
| Periodontitis (2010–2013) | |||||
| No | 618 | 85.5% | 582 | 81.1% | |
| Yes | 105 | 14.5% | 136 | 18.9% | 0.12 |
Note: Categorical data are presented as frequencies and percentages. The continuous variables age and number of years in the lowest fifth percentile of income are presented as mean (SD); income is presented as median [IQR].
2010–2012 unless stated otherwise.
The larger sample of T2D individuals allowed for a specific focus on newly installed implants (no implants in 2010–2011, implants first registered in 2012–2013, no peri‐implantitis by 2013). Individuals with T2D (in or before 2013) were matched to individuals from the non‐T2D cohort ATT propensity score matching based on age, gender, education, income (in or before 2012) and the number of implants (in or before 2013). In all, 725 individuals with T2D (mean age 67.3 ± 9.2 years, 44.3% female, mean number of implants: 2.7 ± 2.3) and 725 without T2D (mean age 67.7 ± 10.0 years, 42.8% female, mean number of implants: 2.7 ± 2.3) were included (Table 2). Incident peri‐implantitis/implant loss followed the same approach outlined above (flexible parametric survival analysis; baseline: 2014).
Individuals without events (peri‐implantitis/implant loss) were censored at their last examination, emigration or death. In case of incomplete data on covariates, individuals were excluded from analysis (Figure 1). If peri‐implantitis was recorded within the first year (2013 for T1D; 2014 for T2D), a time value of 0.5 years was assigned. Estimates are presented as hazard ratios (HRs) with 95% CI. Additionally, analyses were repeated by adjusting for periodontitis (in or before 2012 for T1D; in or before 2013 for T2D). Balance diagnostics following matching are provided in Figures A2 and A3 of Supporting Information, and standardised differences are provided in Table 2.
The unit of all analyses was the patient, and statistical significance was set to p < 0.05. Calculations were carried out using Stata/SE v.18.0; full statistical models are provided in Supporting Information.
3. Results
3.1. Prevalent Peri‐Implantitis and Peri‐Implantitis‐Associated Implant Loss
Peri‐implantitis was more common in T1D compared to non‐T1D (OR 1.46, 95% CI: 1.05–2.04, p = 0.024), as was implant loss (OR 2.39, 95% CI: 1.10–5.20, p = 0.027). The prevalence of peri‐implantitis was similar among T2D and non‐T2D (OR 1.06, 95% CI: 0.98–1.16, p = 0.158), whereas implant loss was more frequent in T2D (OR 1.33, 95% CI: 1.06–1.66, p = 0.015; Figures 2 and 3, Tables A1 and A2 of Supporting Information).
FIGURE 2.
Prevalent peri‐implantitis and peri‐implantitis‐associated implant loss (2010–2020) in T1D (type 1 diabetes)/non‐T1D and T2D (type 2 diabetes)/non‐T2D.
FIGURE 3.
Odds ratio for prevalence of peri‐implantitis. Estimates originate from adjusted logistic regression models, with and without periodontitis included as a covariate, as well as models using the alternative definition of peri‐implantitis based on probing depth (PD).
ORs remained consistent after adjusting for periodontitis, both for prevalent peri‐implantitis (T1D: OR 1.43, 95% CI: 1.02–2.00, p = 0.037; T2D: OR 1.03, 95% CI: 0.94–1.13, p = 0.469; Figure 3) and implant loss (T1D: OR 2.33, 95% CI: 1.07–5.07, p = 0.033; T2D: OR 1.26, 95% CI: 1.01–1.59, p = 0.044; Tables A1 and A2 of Supporting Information). The odds for peri‐implantitis were 2.54 (T1D/non‐T1D; 95% CI: 1.75–3.69, p < 0.001) and 3.67 (T2D/non‐T2D; 95% CI: 3.37–4.00, p < 0.001) times higher for individuals presenting with periodontitis compared to those without (Figure 3).
Within T1D, peri‐implantitis was found to be most frequent among those with poor glycaemic control (poor: 31.7%, OR 2.46, 95% CI: 1.52–4.00, p < 0.001; good: 5.6%, OR 0.43, 95% CI: 0.06–3.30, p = 0.415; reference: non‐T1D, Table A3 of Supporting Information), as was implant loss (poor: 6.9%, OR 4.24, 95% CI: 1.61–11.19, p = 0.003; good: none; reference: non‐T1D, Table A3 of Supporting Information). In T2D, no differences were detected in terms of prevalent peri‐implantitis by glycaemic control (poor: 23.3%, OR 1.19, 95% CI: 0.92–1.55, p = 0.185; good: 21.2%, OR 1.13, 95% CI: 0.95–1.34, p = 0.158; reference: non‐T2D, Table A3 of Supporting Information). Implant loss, however, was more frequent among T2D with poor glycaemic control (poor: 3.8%, OR 1.84, 95% CI: 1.04–3.27, p = 0.037; good: 2.7%, OR 1.39, 95% CI: 0.91–2.13, p = 0.131; reference: non‐T2D, Table A3 of Supporting Information).
The sensitivity analysis with alternative criteria for peri‐implantitis (PD instead of treatment codes) resulted in roughly similar estimates (Figure 3, Table A3 of Supporting Information).
3.2. Incident Peri‐Implantitis and Peri‐Implantitis‐Associated Implant Loss
Overall, peri‐implantitis occurred in 44 individuals with T1D (15.2%), 30 non‐T1D (10.3%), 105 with T2D (14.5%) and 81 non‐T2D (11.2%). Implants were lost in 9 individuals with T1D (3.1%), 3 non‐T1D (1.0%), 11 with T2D (1.5%) and 13 non‐T2D (1.8%).
Both T1D and T2D were associated with a higher incidence of peri‐implantitis, with HRs of 1.52 (95% CI: 0.96–2.42, p = 0.077) for T1D and 1.36 (95% CI: 1.02–1.82, p = 0.035) for T2D (Figure 4). Different trends were observed for implant loss in T1D and T2D, with a tendency towards a higher hazard in T1D (T1D: HR 3.11, 95% CI: 0.84–11.49, p = 0.089; T2D: HR 0.88, 95% CI: 0.39–1.95, p = 0.745). When periodontitis was included in the models, HRs were consistent for both incident peri‐implantitis (T1D: n = 573, HR 1.47, 95% CI: 0.92–2.34, p = 0.104; T2D: n = 1441, HR 1.31, 95% CI: 0.97–1.75, p = 0.074) and implant loss (T1D: 3.21, 95% CI: 0.87–11.85, p = 0.080; T2D: 0.84, 95% CI: 0.38–1.88, p = 0.675). Hazard for peri‐implantitis was 1.92 (T1D/non‐T1D; 95% CI: 0.96–3.87, p = 0.067) and 1.52 (T2D/non‐T2D; 95% CI: 1.07–2.14, p = 0.018) times higher for individuals presenting with periodontitis at baseline (2010–2012) compared to those without.
FIGURE 4.
Cumulative incidence of peri‐implantitis in T1D (type 1 diabetes)/non‐T1D (n = 580) and T2D (type 2 diabetes)/non‐T2D (n = 1450) following propensity score matching.
4. Discussion
In this register‐based study, both T1D and T2D were associated with an increased risk for peri‐implantitis. T1D was associated with both a higher prevalence of peri‐implantitis and implant loss due to peri‐implantitis, whereas for T2D the link was evident in terms of incidence of peri‐implantitis and prevalence of implant loss. In addition, analysis of prevalence data indicated that the association with T1D was modulated by glycaemic control.
Previous studies on the link between diabetes and peri‐implantitis were mainly based on small sample sizes and the reported findings were inconclusive. For instance, Ferreira et al. (2006), in a cross‐sectional study on 29 individuals with and 183 without diabetes, reported that the odds for peri‐implantitis were 1.9 times higher in patients with diabetes (type not specified). In contrast, de Araujo Nobre et al. (2014), in a case–control study on 1350 individuals, including 67 with diabetes (type not specified), reported no such association. Monje et al. (2017) synthesised the results from these and five other studies in a systematic review and reported a pooled OR of 1.9 when comparing patients with and without diabetes (type not specified). While the problem with small sample sizes persisted also in more recent studies (e.g., Apaza‐Bedoya et al. 2024; Kissa et al. 2021), the population‐wide approach used in the present study resulted in a sample of over 18,000 individuals, including 361 with T1D and 4597 with T2D. We believe the sample to be representative of patients with diabetes in Sweden and, thus, the generalisability should be high. The observed ratio of T1D to T2D in our dataset was consistent with the distribution of diabetes types in Sweden (NDR 2024). Despite this large number of individuals, the concern regarding sufficient statistical power in the evaluation of infrequent events remains. Our estimates for implant loss, for instance, were accompanied by large confidence intervals.
One of the novel findings in the present study was the association between T1D and peri‐implantitis, as evidenced by both prevalence and incidence estimates. For T2D, we observed higher odds for implant loss as well as an increased incidence of peri‐implantitis, whereas prevalence estimates did not differ significantly between T2D and non‐T2D. These seemingly inconsistent findings could be explained by differences in terms of exposure time (i.e., time at risk) or methodological aspects (i.e., covariate adjustment vs. propensity score matching).
One of the critical aspects in the interplay between oral diseases and diabetes is glycaemic control, as illustrated in our earlier evaluation focusing on periodontitis (Trullenque‐Eriksson et al. 2024). The present findings suggest that glycaemic control may also play a key role in peri‐implantitis. The data on glycaemic control and risk for peri‐implantitis, however, must be interpreted with caution. Longitudinal fluctuations in HbA1c levels, combined with a limited sample size, prevented us from establishing clear temporal relationships or deriving precise risk estimates.
We based our primary outcome, namely peri‐implantitis, on treatment codes submitted together with the diagnosis ‘peri‐implantitis’. We also presented an alternative definition based on probing depths. Unfortunately, we did not have data on marginal bone loss, as no related codes are recorded in SKaPa. It is unclear to what extent clinicians applied established diagnostic thresholds (e.g., Berglundh et al. 2018). While our approach captures everyday dental care in Sweden, underestimation of disease remains a possibility, as non‐treatment does not necessarily indicate absence of disease. A potential misclassification bias would most likely have been non‐differential. Furthermore, the overall prevalence of peri‐implantitis in our study (roughly 18%) is consistent with rates reported in dedicated surveys (Derks and Tomasi 2015; Lee et al. 2017). A strength of the present study is the fact that diabetes diagnoses and categorisation of glycaemic control were based on professional assessments. This is in contrast to prior studies in which diabetes was typically self‐reported (de Araujo Nobre et al. 2014; Kissa et al. 2021).
As expected, periodontitis was a critical risk factor for peri‐implantitis (e.g., Dalago et al. 2017; Derks et al. 2016; Roccuzzo et al. 2014) also in the present dataset. It is noteworthy that the relationship between T1D/T2D and peri‐implantitis was independent of periodontitis.
In addition to the shortcomings inherent to register‐based research, a limitation specific to the present study was the lack of information on the exact timing of implant installation. While our evaluation of incident peri‐implantitis in T2D focused on newly registered implants, the smaller sample size in the T1D/non‐T1D cohorts precluded a similarly refined approach. Consequently, these analyses may be subject to bias due to an unclear time at risk. Another limitation is the absence of data on other risk factors for peri‐implantitis such as smoking and local factors (e.g., implant placement, prosthetic construction, soft‐tissue dimensions) (Monje et al. 2023; Reis et al. 2023). Still, we believe that the risk of critical imbalances between groups was low due to a solid matching process.
In conclusion, persons with T1D or T2D were at higher risk for peri‐implantitis, as indicated by higher prevalence rates among the former and higher incidence among the latter. The association between peri‐implantitis and T1D was particularly strong in cases of poor glycaemic control.
Author Contributions
A.T.E. substantially contributed to data analysis and interpretation, as well as drafting of the work and revising it critically for important intellectual content. F.V.B. and C.T. substantially contributed to data interpretation, as well as critically revising the work for important intellectual content. T.B. substantially contributed to the conception of the work, data interpretation and critically revising the work for important intellectual content. J.D. substantially contributed to study conception and design, data acquisition, analysis and interpretation, as well as drafting and revising the work for important intellectual content. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Ethics Statement
This study was approved by the Swedish Ethical Review Authority (registration number: 2019–04140).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Appendix S1: STROBE Statement—checklist of items that should be included in reports of observational studies.
Acknowledgements
We are grateful to the Eklund Foundation and TUA Research Funding for their financial support. We thank The Swedish Quality Registry for Caries and Periodontal Diseases (SKaPa) and the Swedish National Diabetes Register (NDR) for providing access to the register data. F.V.B. is supported by the Danish Diabetes and Endocrine Academy, which is funded by the Novo Nordisk Foundation.
Trullenque‐Eriksson, A. , Bitencourt F. V., Tomasi C., Berglundh T., and Derks J.. 2025. “Association Between Diabetes and Peri‐Implantitis: Evidence From a Swedish Register‐Based Study.” Journal of Clinical Periodontology 52, no. 12: 1650–1661. 10.1111/jcpe.70023.
Funding: This work was supported by TUA Research Funding (TUAGBG‐919531 and TUAGBG‐979382) and Eklund Foundation (2018‐132).
A comprehensive literature search has been carried out prior to manuscript submission (14 March 2025), and no similar manuscripts using the same datasets and addressing the same scientific question has been published. The biological plausibility of the association between diabetes and peri‐implantitis is supported by the following published articles (PMID: 35913467, 28346753).
Data Availability Statement
Data are available upon reasonable request. The de‐identified participant data that underlie the results reported in this article, as well as the statistical code are available from the corresponding author upon reasonable request and upon a signed data access agreement.
References
- Apaza‐Bedoya, K. , Galarraga‐Vinueza M. E., Correa B. B., Schwarz F., Bianchini M. A., and Magalhaes Benfatti C. A.. 2024. “Prevalence, Risk Indicators, and Clinical Characteristics of Peri‐Implant Mucositis and Peri‐Implantitis for an Internal Conical Connection Implant System: A Multicenter Cross‐Sectional Study.” Journal of Periodontology 95, no. 6: 582–593. 10.1002/JPER.23-0355. [DOI] [PubMed] [Google Scholar]
- Bayoumi, A. 2016. STDDIFF: Stata Module to Compute Standardized Differences for Continuous and Categorical Variables. Statistical Software Components, Boston College Department of Economics. [Google Scholar]
- Berglundh, T. , Armitage G., Araujo M. G., et al. 2018. “Peri‐Implant Diseases and Conditions: Consensus Report of Workgroup 4 of the 2017 World Workshop on the Classification of Periodontal and Peri‐Implant Diseases and Conditions.” Journal of Clinical Periodontology 45, no. Suppl 20: S286–S291. 10.1111/jcpe.12957. [DOI] [PubMed] [Google Scholar]
- Dalago, H. R. , Schuldt Filho G., Rodrigues M. A., Renvert S., and Bianchini M. A.. 2017. “Risk Indicators for Peri‐Implantitis. A Cross‐Sectional Study With 916 Implants.” Clinical Oral Implants Research 28, no. 2: 144–150. 10.1111/clr.12772. [DOI] [PubMed] [Google Scholar]
- de Araujo Nobre, M. , Malo P., and Antune E.. 2014. “Influence of Systemic Conditions on the Incidence of Periimplant Pathology: A Case–Control Study.” Implant Dentistry 23, no. 3: 305–310. 10.1097/ID.0000000000000071. [DOI] [PubMed] [Google Scholar]
- Derks, J. , Schaller D., Hakansson J., Wennstrom J. L., Tomasi C., and Berglundh T.. 2016. “Effectiveness of Implant Therapy Analyzed in a Swedish Population: Prevalence of Peri‐Implantitis.” Journal of Dental Research 95, no. 1: 43–49. 10.1177/0022034515608832. [DOI] [PubMed] [Google Scholar]
- Derks, J. , and Tomasi C.. 2015. “Peri‐Implant Health and Disease. A Systematic Review of Current Epidemiology.” Journal of Clinical Periodontology 42 Suppl 16: S158–S171. 10.1111/jcpe.12334. [DOI] [PubMed] [Google Scholar]
- Ferreira, S. D. , Silva G. L., Cortelli J. R., Costa J. E., and Costa F. O.. 2006. “Prevalence and Risk Variables for Peri‐Implant Disease in Brazilian Subjects.” Journal of Clinical Periodontology 33, no. 12: 929–935. 10.1111/j.1600-051X.2006.01001.x. [DOI] [PubMed] [Google Scholar]
- Jann, B. 2017. KMATCH: Stata Module Module for Multivariate‐Distance and Propensity‐Score Matching, Including Entropy Balancing, Inverse Probability Weighting, (Coarsened) Exact Matching, and Regression Adjustment. Statistical Software Components, Boston College Department of Economics. [Google Scholar]
- Kissa, J. , El Kholti W., Chemlali S., Kawtari H., Laalou Y., and Albandar J. M.. 2021. “Prevalence and Risk Indicators of Peri‐Implant Diseases in a Group of Moroccan Patients.” Journal of Periodontology 92, no. 8: 1096–1106. 10.1002/JPER.20-0549. [DOI] [PubMed] [Google Scholar]
- Lambert, P. 2010. STPM2: Stata Module to Estimate Flexible Parametric Survival Models. Statistical Software Components, Boston College Department of Economics. [Google Scholar]
- Lee, C. T. , Huang Y. W., Zhu L., and Weltman R.. 2017. “Prevalences of Peri‐Implantitis and Peri‐Implant Mucositis: Systematic Review and Meta‐Analysis.” Journal of Dentistry 62: 1–12. 10.1016/j.jdent.2017.04.011. [DOI] [PubMed] [Google Scholar]
- Monje, A. , Catena A., and Borgnakke W. S.. 2017. “Association Between Diabetes Mellitus/Hyperglycaemia and Peri‐Implant Diseases: Systematic Review and Meta‐Analysis.” Journal of Clinical Periodontology 44, no. 6: 636–648. 10.1111/jcpe.12724. [DOI] [PubMed] [Google Scholar]
- Monje, A. , Kan J. Y., and Borgnakke W.. 2023. “Impact of Local Predisposing/Precipitating Factors and Systemic Drivers on Peri‐Implant Diseases.” Clinical Implant Dentistry and Related Research 25, no. 4: 640–660. 10.1111/cid.13155. [DOI] [PubMed] [Google Scholar]
- NDR . 2024. Nationella Diabetesregistret, årsrapport 2023.
- Nibali, L. , Gkranias N., Mainas G., and Di Pino A.. 2022. “Periodontitis and Implant Complications in Diabetes.” Periodontology 2000 90, no. 1: 88–105. [DOI] [PMC free article] [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]
- Reis, I. , do Amaral G., Hassan M. A., et al. 2023. “The Influence of Smoking on the Incidence of Peri‐Implantitis: A Systematic Review and Meta‐Analysis.” Clinical Oral Implants Research 34, no. 6: 543–554. 10.1111/clr.14066. [DOI] [PubMed] [Google Scholar]
- Roccuzzo, M. , Bonino L., Dalmasso P., and Aglietta M.. 2014. “Long‐Term Results of a Three Arms Prospective Cohort Study on Implants in Periodontally Compromised Patients: 10‐Year Data Around Sandblasted and Acid‐Etched (SLA) Surface.” Clinical Oral Implants Research 25, no. 10: 1105–1112. 10.1111/clr.12227. [DOI] [PubMed] [Google Scholar]
- Schwarz, F. , Derks J., Monje A., and Wang H. L.. 2018. “Peri‐Implantitis.” Journal of Clinical Periodontology 45, no. Suppl 20: S246–S266. 10.1111/jcpe.12954. [DOI] [PubMed] [Google Scholar]
- SKaPa . 2024. “Svenskt Kvalitetsregister för Karies och Parodontit, Årsrapport 2023.” ISSN 2001‐4295. http://www.skapareg.se/wp‐content/uploads/2024/08/SKaPa‐%C3%85rsrapport‐2023.pdf.
- Trullenque‐Eriksson, A. , Tomasi C., Eeg‐Olofsson K., Berglundh T., Petzold M., and Derks J.. 2024. “Periodontitis in Patients With Diabetes and Its Association With Diabetes‐Related Complications. A Register‐Based Cohort Study.” BMJ Open 14, no. 7: e087557. 10.1136/bmjopen-2024-087557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yang, D. , and Dalton J. E.. 2012. “A Unified Approach to Measuring the Effect Size Between Two Groups Using SAS.” In SAS Global Forum, vol. 335, 1–6. [Google Scholar]
- Zhou, B. , Rayner A. W., Gregg E. W., et al. 2024. “Worldwide Trends in Diabetes Prevalence and Treatment From 1990 to 2022: A Pooled Analysis of 1108 Population‐Representative Studies With 141 Million Participants.” Lancet 404, no. 10467: 2077–2093. 10.1016/S0140-6736(24)02317-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1: STROBE Statement—checklist of items that should be included in reports of observational studies.
Data Availability Statement
Data are available upon reasonable request. The de‐identified participant data that underlie the results reported in this article, as well as the statistical code are available from the corresponding author upon reasonable request and upon a signed data access agreement.