Epidemiologic relationship between periodontitis and type 2 diabetes mellitus

The search strategy was a combination of an electronic search and a manual search. The manual search included the reference lists of the included studies and the following journals: Diabetes Care, Journal of Periodontology, Journal of Clinical Periodontology and Journal of Dental Research. The following electronic databases were searched without language limitations: MEDLINE (OVID, 1948 to December 2018), EMBASE (OVID, 1984 to December 2018), Chinese BioMedical Literature Database (CBM, 1978 to December 2018), and China National Knowledge Infrastructure (CNKI, 1994 to December 2018). MeSH terms with free text words were combined when conducting electronic searches. The MeSH terms used for periodontitis were “periodontal diseases” and “periodontitis”. The free text word was “(periodont$ or gingivitis or gingiva$ or gum$).mp.”. The MeSH term used for T2DM searching was “diabetes mellitus, type 2”. Free text words were “(((non-insulin or noninsulin or type 2 or type II or matur$ or adult) adj4 (DM or diabet$)) or T2DM or DMT2 or NIDDM or MODY).mp.”. The titles and abstracts were initially scanned, and the full texts of the possibly eligible studies were obtained for final judgment.

Observational studies (cross-sectional studies, case-control studies and cohort studies) investigating the relationship between T2DM and PD were included. The criteria for the outcomes for periodontitis were clinical attachment loss (CAL), periodontal pocket depth (PPD), number of teeth (NOT), loss of teeth (LOT), alveolar bone loss and community periodontal index (CPI) score. The criteria for the outcomes for T2DM were oral glucose tolerance test (OGTT), HbA1c and fasting plasma glucose (FBG) results. Disease (periodontitis or T2DM) prevalence and incidence were also included. The participants chosen represent the natural population grouping into periodontitis versus non-periodontitis or T2DM versus non-DM. Comparisons based on periodontitis parameters, such as the T2DM incidence/prevalence between patients with low CAL levels and high CAL levels, were also included. Studies investigating outcomes in selected populations, such as comorbid patients, all periodontitis patients, all T2DM patients or all healthy participants (periodontitis-free and T2DM-free), were excluded. Studies were selected according to the aforementioned periodontitis/T2DM-related parameters, medical records or self-reported medical history.

The study quality of cohort studies and case-control studies were measured by the Newcastle-Ottawa Scale (NOS) scoring system. Studies with scores less than 3 were regarded as low quality and were excluded. For cross-sectional studies, the Agency for Healthcare Research and Quality (AHRQ) scoring system was applied. Studies with scores less than 3 in the AHRQ scoring system were regarded as low quality and were not included.

The extracted data were as follows: 1) investigator, 2) country, 3) number of participants, 4) age and sex of the participants, 5) recruitment of participants, 6) selected outcomes, and 7) NOS/AHRQ score. For cohort studies, the follow-up period and number of incident cases were also extracted.

The software STATA 14.0 was utilized for meta-analysis. Weighted mean differences (WMDs) with 95% confidence intervals (CIs) were calculated for continuous data. Odds ratios (ORs) and risk ratios (RRs) with 95% CIs were calculated for dichotomous data. Generic inverse variance (lnOR or lnRR) was used for meta-analyses that included studies that only reported ORs or RRs. Significance was determined by two-sided α values with a cut-off p value of 0.05. All meta-analyses were performed under the random-effects model. Cochran’s Q test and I² statistic were used to detect statistical heterogeneity among studies. When P > 0.10 and I² < 50%, the study was regarded as having low heterogeneity; otherwise, it was regarded as having high heterogeneity. Meta-regression was utilized for a meta-analysis that included more than 4 studies to investigate possible sources of heterogeneity. The influence test was conducted by deleting every single study in turn to test whether the results were stable. For a meta-analysis that included more than 10 studies, publication bias was detected by Egger’s test and Begg’s test. The publication was excluded when both test results exhibited p > 0.05. If publication bias existed or unstable results were found, the trim and fill method was applied.

A total of 1387 studies were identified from the primary search after removing duplicate studies. After screening the titles and abstracts, 73 studies were identified for further evaluation. After browsing the full text, 50 studies were considered eligible for inclusion. Twenty-three studies were excluded for various reasons. Among these, 16 studies were excluded because of the study type (7 meta-analyses, 6 review articles and 3 case series reporting periodontal treatment for T2DM patients); 4 studies were excluded because the reported outcomes were insufficient; 2 studies were excluded because they included type 1 DM. Additional reference checking revealed 3 studies that were then included. Journal searching did not add any new studies. Finally, a total of 53 studies were included in the present work. Figure 1 shows the search and inclusion process. Appendix Tables S1 and S2 summarize the characteristics of 43 [19‐61] cross-sectional studies and 12 [23, 39, 62‐71] cohort studies, respectively. Appendix Tables S3 and S4 summarize the AHRQ and NOS scores of cross-sectional studies and cohort studies. All included cross-sectional studies and cohort studies had scored higher than 3.

After systematically reviewing the included studies, we found that they answered 3 questions (questions 1–3, Q1–3). Specifically, cross-sectional studies gave the answer “Q1: Are periodontitis and T2DM associated with each other?” Cohort studies gave the answer to the other two questions: “Q2: Does T2DM increase the risk of developing periodontitis?”, and “Q3: Does periodontitis increase the risk of developing T2DM?”

A total of 43 cross-sectional studies were included to answer Q1. Evidence was from some national large-scale population-based studies, such as the SHIP, NHANES and KCIS, and some small-sample studies recruiting participants from communities or hospitals. Among these studies, only 14 studies reported adjusted outcomes (Table 1). Six meta-analyses were conducted as follows.

A total of 15 cross-sectional studies with 17,924 participants reported the unadjusted OR between these two diseases (Table S1). Since the original data were not directionally adjusted, a meta-analysis was not undertaken. Among the 15 studies, except for 4 studies [21, 34, 43, 51] that reported that the presence of periodontitis was not different between T2DM patients and non-T2DM controls, all the other studies acknowledged that there was a strong connection.

A total of 6 cross-sectional studies were included, and all had T2DM prevalence as an outcome. Three studies with 1956 participants were included in a meta-analysis that included a periodontitis diagnosis as an outcome. The included studies had no significant heterogeneity. The results showed that periodontitis patients had significantly higher odds of T2DM prevalence than participants with no periodontitis (OR = 4.04, 95% CI 2.48–6.59, p = 0.000, Fig. 2a). Influence analysis showed that the pooled result was stable (Fig. S1a). Other exposure factors included CAL, PPD, LOT, tooth mobility and alveolar bone loss. The results all proved that T2DM was more prevalent in participants with worse periodontal health (Table 1).

A total of 8 cross-sectional studies were included, and all took T2DM as exposure. Three studies with 11,459 participants were included in a meta-analysis evaluating periodontitis prevalence. No significant heterogeneity was detected. The results showed that T2DM patients had a significantly higher ORs for PD prevalence (OR = 1.58, 95% CI 1.38–1.81, p = 0.000, Fig. 2b). Influence analysis indicated that the pooled result was stable (Fig. S1b). In addition to periodontitis prevalence, other outcomes were divergent. In brief, all studies demonstrated that periodontitis-related parameters were more prevalent in T2DM patients, although some of the differences were not statistically significant. The results are summarized in Table 1.

Eighteen cross-sectional studies with 9571 participants were included. Significant heterogeneity was detected (p = 0.000; I² = 92.5%). Pooled results showed that T2DM patients had a 0.89 mm higher CAL than controls (WMD = 0.89, 95% CI 0.64–1.15, p = 0.000, Fig. 2c). Influence analysis demonstrated that the pooled result was stable (Fig. S1c). Publication bias was detected by Egger’s and Begg’s tests (Egger, p = 0.003; Begg, p = 0.015). Then, we employed the trim and fill method to further evaluate publication bias and found that the results were still significantly positive after adding the hypothesized studies (Table S5).

Seventeen cross-sectional studies with 8982 participants were included. Significant heterogeneity was detected (P = 0.000; I² = 94.5%). Pooled results showed that the periodontal pockets of T2DM patients were 0.61 mm deeper than those of controls (WMD = 0.61, 95% CI 0.42–0.79, p = 0.000, Fig. 2d). Influence analysis demonstrated that the pooled result was stable (Fig. S1d). Publication bias was detected by Egger’s and Begg’s test (Egger, p = 0.015; Begg, p = 0.006). However, adding hypothesized studies by the trim and fill method still resulted in strong significance (Table S5).

Nine cross-sectional studies with 4415 participants were included. Significant heterogeneity was detected (p = 0.000; I² = 86.6%). Pooled results showed that T2DM patients had, on average, 2.01 fewer teeth remaining than controls. (WMD = -2.01, 95% CI -3.20--0.82, p = 0.000, Fig. 2e). Influence analysis demonstrated that the pooled result was stable (Fig. S1e). No publication bias was detected (Egger, p = 0.723; Begg, p = 0.917).

Eleven cross-sectional studies with 3405 participants were included. Significant heterogeneity was detected (P = 0.000; I² = 90.7%). Pooled results showed that T2DM patients had, on average, lost 2.22 more teeth than controls (WMD = 2.22, 95% CI 0.94–3.49, p = 0.000, Fig. 2f). Influence analysis demonstrated that the pooled result was stable (Fig. S1f). No publication bias was detected (Egger, p = 0.230; Begg, p = 0.755).

Huge statistical heterogeneity existed in the above 4 meta-analyses, and the I² ranged from 86.3 to 94.5%; thus, we performed meta-regression to find the possible sources of heterogeneity. The available covariates included the number of participants, mean age, sex composition of the participants, geographic area and AHRQ scores. However, single variable regression did not find any significant covariates; multiple regression of these covariates only explained approximately 10% of the heterogeneity of all meta-analyses (data not shown). The significant heterogeneity might be caused by statistical heterogeneity or other potential clinical diversity not included in the meta-regression.

A meta-analysis on this topic showed that mild periodontitis led to a 28% elevated risk for incident T2DM (RR = 1.28, 95% CI 1.07–1.54, p = 0.007, Fig. 4a). No significant heterogeneity (I² = 20.4%, p = 0.27) or publication bias (Egger, p = 0.133; Begg, p = 0.133) among studies was detected. Influence analysis found that this result was unstable (Fig. S2b). Deleting Demmer’s study [63] would reduce the effect size and obtain a marginally significant result (RR = 1.17, 95% CI 0.99–1.39, p > 0.05). Due to this unstable result, we used the trim and fill method. After adding 3 hypothetical studies, the results became significant (RR = 1.14, 95% CI 0.92–1.41, p = 0.23, Table S5). The above results indicate that the effect of mild periodontitis on T2DM incidence was not very robust.

Pooled results showed that severe periodontitis increased the risk of T2DM incidence by 53% (RR = 1.53, 95% CI 1.27–1.83, p = 0.000, Fig. 4b). The heterogeneity was very low (I² = 0%, p = 0.649). No publication bias (Egger, p = 0.104; Begg, p = 0.230) was detected. In contrast to mild periodontitis, influence analysis found that the impact of severe periodontitis was very stable (Fig. S2c). To further confirm this, we used the trim and fill method. After adding 2 hypothetical studies, the results were still significant (RR = 1.46, 95% CI 1.23–1.73, p = 0.000, Table S5). The above results indicated that the effect of severe periodontitis on T2DM incidence was strong.