An Umbrella Review of The Periodontal Disease—Diabetes Mellitus Association

1. Introduction

Diabetes mellitus (DM) is a common metabolic disease resulting from a defect in insulin secretion, a defect in insulin action, or a combination of both [1,2]. This chronic disease is one of the most common in the world and the prevalence is expected to increase by almost 50% in the coming years. It is estimated that there will be around 700 million people living with diabetes by 2045 and it will be within the five main causes of death [3]. Mortality is due to its complications in microvascular changes, neuropathy, retinopathy and nephropathy, delayed healing, and the sixth important problem of people with diabetes is periodontitis [4,5].

Periodontitis is a disease characterized by chronic inflammation of the entire periodontium that can irreparably destroy the tissue surrounding the tooth, leading to gingival bleeding, increased tooth mobility and tooth loss [6]. It is estimated that approximately between 20 and 60% of the world's population suffers from periodontal disease [7,8].

DM and periodontal disease (PD) are associated with each other, despite not having a common pathophysiology [6]. However, glucose alteration in these patients shows a positive association, which is evident in patients with poorly controlled diabetes having an 86% higher risk of developing periodontitis than non-diabetics or those with well-controlled diabetes [4]. Furthermore, there are indications that PD is a risk factor for diabetes mellitus, which shows a possible bidirectional association [3]. The proposed link between the two diseases is based on the characteristics of the immune response, neutrophil function, and cytokine biology [9,10].

Given the need for dentists to know the association of these diseases and the short- and long-term consequences, there is a need for a better understanding of the scientific basis using the highest quality evidence available for clinical decision making [11]. The synthesis of knowledge facilitates clinical decisions, finding information in a single document, thus facilitating the reading and understanding of a particular topic. To date, no comprehensive synthesis or evaluation of all systematic reviews, including those from recent years, has been performed. Therefore, the aim of this overview was to consolidate the available evidence and address the following question: “What is our current understanding of the link between periodontal disease and diabetes mellitus?”, in addition to assessing the general confidence in systematic reviews that analyze this issue.

2. Materials and Methods

2.1. Protocol and Registration

Based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) [12], the procedure was executed and subsequently registered in the Prospective Registry of Systematic Reviews (PROSPERO) [13]. The registry can be accessed by the general public with the number CRD42023464043. Furthermore, the PRIO-harms (Preferred Reporting Items for Overview of Systematic Reviews Checklist) form the basis of this study´s report [14]. Notably, our methodology was based on Pauletto et al. [15], given its similarity to our approach. Ethical approval was not required for this umbrella review.

2.2. Eligibility Requirements and Notewhorthy Results

All included articles were systematic reviews (SR) that assessed primary research assessing the relationship between PD and DM, whether or not it has meta-analysis, and without regard to time or language constraints.

Exclusions included abstracts, remarks, case reports, procedures, personal opinions, letters, posters, fast reviews, intervention studies, observational studies, preclinical and fundamental research, and literature or narrative reviews.

2.3. Information Sources, Search Tactics and Further Searches for Primary Studies

August 20, 2023 saw the completion of an electronic search across four databases: Pubmed, Cochrane database, Scielo, and Scopus. We also looked up gray literature using OpenGrey and Google Scholar. The included studies´reference lists were also examined. Duplicate articles were removed from the found articles and exported to Zotero®, a reference management program developed by the Center for History and New Media in Virginia, USA. Table 1 lists the search approach used for each database.

2.4. Data Management and Selection Process

The identified articles were inserted into Rayyan® Online Software (Qatar Research Institute of Computing, Qatar). The selection of the studies was carried out in 2 phases: in phase 1, two reviewers (F.C.O. and E.I.) independently selected the studies by reading the title and abstract; subsequently, phase 2 was carried out, which consisted of reading the full text, carried out independently by the same two reviewers. A third reviewer (F.C.Z.) was consulted in case of disagreement.

2.5. Data Collection Process

Data from the study were independently collected in duplicate using a table previously formulated by two reviewers (E.B. and W.T.). Data were cross-checked and disagreements resolved by the third review author (P.H.). The following information was extracted from the selected articles: authors, year of publication, study design, design of primary studies included, number of studies included in the qualitative and quantitative analysis, country, type of diabetes, results, main conclusions, and the mention of the following were used or executed: Mata-analysis, PRISMA, PROSPERO, and Grading of Recommendations Assessment, Development and Evaluation (GRADE).

2.6. Evaluation of Meta-Bias, Evidence Quality, and Methodological Quality

Two reviewers (O.S. and R.O.) independently and in duplicate assessed the methodological quality of the included SRs using the AMSTAR-2 checklist (A MeaSurement Tool to Assess Systemic Reviews), which was calibrated (Kappa 0.85) [16]. With 16 questions that have three alternative answers - "yes", "no" or "partially yes"- the AMSTAR-2 assesses the methodological quality of the SR. Shea et al. [16] rated the study´s overall confidence level, which was classified as high, moderate, low, and severely low.

2.7. Measurement Summary

We took into account the study´s findings in the event of an SR meta-analysis. The results that were displayed with a mean difference, normalized mean difference, relative risk, or odds ratio are taken into consideration if the SR offers meta-analysis.

2.8. Results Summary

The primary findings of the included SRs were compiled and categorized into the following categories: number of teeth, clinical attachment level, prevalence, general association, plaque index, gingival index, probing depth, and bleeding on probing.

3. Results

3.1. Examining and Choosing Original Research

After duplicates were removed, 519 references were left from 572 that the electronic database search had originally yielded. Phase 1 involved evaluating the titles and abstracts of the selected studies and taking into consideration 12 papers that could be read in full. After two more articles from other umbrella reviews and four articles added through various means, there were only eighteen SRs left for the quality synthesis. Table 2 provides the rationale behind the article´s exclusion. Figure 1 depicts the entire identification and selection procedure for the research.

3.2. Review and Characteristics of Included Studies

The SRs included were published between 2009 and 2023 and were published in English. They were carried out in Portugal [10], Peru [24], Malaysia [25], China [26,27], Germany [3], Italy [28,29,30], Australia [31], Japan [32], Denmark [33], Netherlands [34], Spain [35], Brazil [36,37] and United States [38,39]. More information on SR characteristics can be found in Table 3.

3.3. Assessment of Methodological Quality and Quality of Evidence:

13 SRs [3,10,24,25,26,30,33,34,35,36,37,38,39] were considered to have high confidence, 1 SR [32] to have low confidence and 4 SRs [27,28,29,31] had critically low confidence (Table 4).

3.4. Overlapping

A total of 296 primary studies were identified within the SRs. Of these, approximately 17.26% of the primary studies were included in more than one SR. Thirty-five studies were included twice, eight were included three times, and seven were included four times. More information on the overlap and characteristics of the primary studies is available in Table 5.

3.5. Synthesis of Results

The syntheses of the results are presented in Table 3.

3.6. General Association

Eleven SRs [3,10,24,27,28,30,31,33,34,38,39] included reported that there were an association between PD and DM, while 2 SRs [32,36] reported that there was no association. Six SRs [3,28,33,34,36,38] meta-analyzed the results, where they found that the relative risk ratio ranged from 1.26 (CI: 1.12 to 1.41) [3] to 1.86 (CI: 1.25 to 2.77) [33] and the odds ratio ranged from 1.69 (CI: 0.68 to 4.21) [36] to 2.59 (CI: 2.12 to 3.15) [34]. Costa et al. [10], León-Ríos et al. [24], Nguyen et al. [31], Graziani et al. [30], Ismail et al. [27] and Borgnakke et al. [39] they presented the results descriptively and reported an association between PD and DM; while Tanaka et al. [32] reported that there is no association between PD and DM.

3.7. Plaque Index

Three SRs [25,26,29] included reported that there is an association between PD and DM. All meta-analyzed the results, where they found that the mean difference was 0.20 (CI: 0.18 to 0.23) [26] and standardized mean difference ranged from 054 (CI: 0.20 to 0.87) [25] 0.71 (CI: 0.19 to 1.22) [29].

3.8. Gingival Index

Two SR [25,29] included reported that there is an association between PD and DM. This study meta-analyzed its results, where they found that the standardized mean difference ranged from 0.46 (CI: 0.08 to 0.84) [29] to 0.63 (CI: 0.39 to 0.87) [25].

3.9. Clinical Attachment Level

Four SRs [25,26,28,29] included reported that there is an association between PD and DM, while 1 SR [37] reported that there is an association between PD and type 2 diabetes mellitus (T2DM) but not with type 1 diabetes mellitus (T1DM). All meta-analyzed the results, where they found that the mean difference ranged from 0.26 (CI: 0.00 to 0.53) to 1.00 (CI: 0.15 to 1.84) [37] and the standardized mean difference range from 0.47 (CI: 0.37 to 0.57) [28] to 0.82 (CI: 0.59 to 1.04) [29].

3.10. Number of Teeth

One SR [26] included reported that there is an association between PD and DM. The mean difference, according to the study's meta-analysis, was -2.14 (CI: -2.87 to -1.40).

3.11. Prevalence

Three SRs [26,28,35] included reported that there is an association between PD and DM. Two SRs [26,28] meta-analyzed the results, where they found that the odds ratio ranged from 0.19 (CI: 0.08 to 0.37) [28] to 1.85 (CI: 1.61 to 2.11) [26]. Mauri-Obradors et al. [35] presented the results descriptively and reported an association between PD and DM.

3.12. Probing Depth

Three SRs [25,26,29] included reported that there is an association between PD and DM, while 1 SR [37] reported that there is an association between PD and T2DM but not with T1DM. All meta-analyzed the results, where they found that the mean difference ranged from 0.11 (CI: -0.03 to 0.25) to 0.46 (CI: 0.01 to 0.91) [37] and the standardized mean difference ranged from 0.36 (CI: 0.16 to 0.55) [29] to 0.67 (CI: 0.23 to 1.11) [25].

3.13. Bleeding on Probing

Three SRs [25,26,29] included reported that there is an association between PD and DM. The two studies meta-analyzed the results, where they found that the mean difference was 7.90 (CI: 4.24 to 11.56) [26] and standardized mean difference ranged from 0.32 (CI: 0.07 to 0.58) [25] to 0.65 (CI: 0.08 to 1.23) [29].

4. Discussion

The evaluation and analysis of the association between PD and systemic diseases, such as DM, has been of interest in recent years. Numerous RCTs have investigated this topic and report evidence supporting the association between these 2 diseases.

Some studies have evaluated this association in a general way, while other studies have evaluated it according to periodontal clinical parameters. To date, several SRs have been published that analyze the association between PD and DM, this has made it necessary to compile the data from these studies and assess the methodological quality of each one.

Currently, three umbrella reviews have been conducted, all three [11,93,94] reviewed the most current evidence on the nature of the relationship/association between PD and DM, however, they included studies that analyzed whether periodontal therapy was effective in people with DM; causing the results and conclusions of these studies to be taken with caution.

The present study conducted an exhaustive literature search to summarize the available SRs that deal with the association between PD and DM (T1DM, T2DM, and GDM), obtaining 18 SRs that matched the inclusion criteria and were taken into consideration for the analysis. SRs represent the highest level of the scientific evidence pyramid, but their results should be evaluated cautiously because they are likely biased.

The SRs included in this study showed certain limitations related to the selected primary studies, such as: the existence of different types of studies included, selection criteria for the studies to be included, the inclusion of different population groups (children, adolescents and adults), different diagnostic criteria for periodontal diseases and evaluation of different types of diabetes mellitus (T1DM, T2DM and GDM).

On the other hand, more than 50% of the included studies revealed a high level of confidence, which could increase the level of evidence for the results and conclusions offered in this study. However, the fact that SRs continue to be obtained without a high level of confidence points to the need for greater rigor in the development of this group of studies on this topic.

The current instrument, the AMSTAR-2, was used to assess the methodological quality of the included SRs. A factor that deserves to be highlighted in relation to the result of methodological quality is related to the critical domains 2, 4, 7, 9, 13, and 15 of AMSTAR - 2, since some SRs did not explicitly address their methods, they did not use a search strategy was exhaustive, they did not provide a list of excluded studies justifying their exclusion, they did not use a satisfactory technique to assess the risk of bias, they did not consider the risk of bias of the included studies to interpret or discuss their results, and they did not report the bias of publication; highlighting the need to include these elements in the development of future SRs.

Additionally, caution should be taken as some studies in the SRs were included more than once, leading to data from the included studies being reevaluated multiple times, which could increase the perception of what has been done. Although it would be interesting to carry out new SRs to overcome the methodological limitations recommended by Moher [95], due to the high rate of overlap, it is more important to carry out well-conducted RCTs, with long-term follow-up and coming from different research groups to expand knowledge on this topic.

4.1. Evidence Summary

The present umbrella review was carried out seeking to assist in the association between PD and DM with the purpose of minimizing biases and random errors in SRs and meta-analyses on the subject. The material below can be summed up and discussed while keeping in mind the limitations of SRs used in this study.

The SRs included in this study suggested a general association between PD and DM. This result is similar to that reported by the Consensus Report of the European Federation of Periodontics [17] where they point out that periodontitis is associated with diabetes mellitus. However, two of the included studies [32,36] report no association between these 2 diseases, this was due to the fact that one [32] of them focused only on the Japanese population, concluding that there was little evidence on this topic in the said population group, while, the another study [36] performed a subgroup analysis by type of primary study included, concluding that in the cross-sectional studies the association did exist but in the case-control studies the association did not exist.

In terms of plaque index, gingival index, number of missing teeth, prevalence, and bleeding on probing; studies indicate that patients with DM have a greater probability of presenting with an increase in all these clinical parameters, indicating that there is an association between PD and DM.

Regarding the depth of catheterization and the clinical attachment level, studies also indicate that patients with DM have a greater probability of presenting with an increase in all these clinical parameters. However, one study [37] indicated that there is an association between PD and T2DM but not with T1DM.

Furthermore, it should be noted that of the included studies: four [10,27,28,29] analyze T1DM, with only one [10] having high overall confidence; three [24,36,38] analyzing GDM where all have high overall confidence; and ten [3,25,26,30,31,32,33,35,37,39] analyzing DM in general, eight [3,25,26,30,33,35,37,39] of which They have high general confidence.

4.2. Implications for Clinical Practice

There is an association between PD and DM, therefore, a correct anamnesis and a correct diagnosis must be made to be able to present the best viable, simple, and less invasive treatment options to this group of patients.

4.3. Implications for Research

Despite the high number of SRs, it was evident from this assessment that their reporting has to be improved immediately. The observation of instruments that direct their evolution, including quality evaluation tools, is the author´s recommendation for upcoming SRs. In order to get more reliable results, additional SRs on the relationship between PD and MGD should be conducted with strong methodological rigor. It is also recommended for future primary studies to use the latest definition related to the diagnosis of periodontal diseases.

5. Conclusions

According to the findings of this umbrella review, there is strong evidence supporting the association between periodontal disease and various forms of diabetes mellitus, including type 1 and type 2, and gestational diabetes. The studies reviewed demonstrate a consistent relationship across different clinical parameters, such as plaque index, gingival index, clinical attachment level, probing depth, and bleeding on probing, with diabetes patients showing a higher likelihood of presenting worsened periodontal conditions.