Does Nonsurgical Periodontal Treatment Improve Systemic Health?

Abstract

Clinicians frequently stress the importance of maintaining good oral health for multiple reasons, including its link to systemic health. Because periodontal treatment reduces inflammation in oral tissues, some hypothesize it may positively affect systemic outcomes by reducing inflammation in the body. A significant number of systematic reviews (SRs) and meta-analyses (MAs) have evaluated the effect of periodontal treatment on systemic outcomes. However, inconsistent findings and questionable methodological rigor make drawing conclusions difficult. We conducted a systematic review of reviews that studied the effect of nonsurgical periodontal treatment on systemic disease outcomes. We report on outcomes evaluated, categorizing them as biomarkers, and surrogate or clinical endpoints. In addition, we used A MeaSurement Tool to Access systematic Reviews 2 (AMSTAR 2) to evaluate the methodological quality of the reviews. Of the 52 studies included in our review, 21 focused on diabetes, 15 on adverse birth outcomes, 8 on cardiovascular disease, 3 each on obesity and rheumatoid arthritis, and 2 on chronic kidney disease. Across all studies, surrogate endpoints predominated as outcomes, followed by biomarkers and, rarely, actual disease endpoints. Ninety-two percent of studies had “low” or “critically low” AMSTAR 2 confidence ratings. Criteria not met most frequently included advance registration of the protocol, justification for excluding individual studies, risk of bias from individual studies being included in the review, and appropriateness of meta-analytical methods. There is a dearth of robust evidence on whether nonsurgical periodontal treatment improves systemic disease outcomes. Future reviews should adhere more closely to methodological guidelines for conducting and reporting SRs/MAs than has been the case to date. Beyond improved reviews, additional rigorous research on whether periodontal treatment affects systemic health is needed. We highlight the potential of large-scale databases containing matched medical and dental record data to inform and complement future clinical research studying the effect of periodontal treatment on systemic outcomes.

Introduction

Globally, the prevalence of periodontitis ranges from 20% to 50% among adults (Nazir et al. 2020). Periodontitis is associated with several chronic conditions, including cardiovascular disease, diabetes mellitus, chronic kidney disease, rheumatoid arthritis, and obesity (Martinez-Herrera et al. 2017; Sanz et al. 2018; Bui et al. 2019; Kapellas et al. 2019; Liccardo et al. 2019; Mankia et al. 2019; Rodríguez-Lozano et al. 2019). Inflammation plays a pivotal role in the initiation and progression of many of these diseases, and sustained low-grade inflammation in the body is suspected to be an antecedent of obesity (Wiebe et al. 2019), diabetes (Pradhan et al. 2001), and cardiovascular disease (Libby et al. 2002; Ridker et al. 2005; Ridker et al. 2017). Consequently, preventing and managing inflammatory processes may improve the prognosis of these diseases (Ridker et al. 2005; Ridker et al. 2017). Because periodontal treatment reduces inflammation in oral tissues, it may positively affect systemic disease outcomes by reducing inflammation in the body (Teeuw et al. 2014; Simpson et al. 2015; Calderaro et al. 2017). However, this hypothesis has yet to be validated.

In the past decade, a large number of systematic reviews (SRs) and meta-analyses (MAs) have examined the relationship between periodontal treatment and disease-specific systemic outcomes in over 175 primary studies. However, 3 major limitations across these SRs and MAs constrain the degree to which they can inform clinical care, policy, and research. First, their findings are inconsistent. While some SRs and MAs found positive systemic effects related to the periodontal treatment (Silvestre et al. 2016; Teshome and Yitayeh 2016; Calderaro et al. 2017), others have reported either no effect (Janket et al. 2003; Chambrone et al. 2011; Boutin et al. 2013) or insufficient evidence to make any claims (Mauri-Obradors et al. 2015; Pérez-Losada et al. 2016; Liu et al. 2019). Second, methodological issues such as the inclusion of studies with high risk of bias, differences in how periodontal disease is defined and measured, low sample sizes, and wide variation in study duration limit validity and generalizability (Faggion et al. 2016; Hasuike et al. 2017; Beck et al. 2019). Third and last, each review focused on only 1 systemic disease, making it difficult, if not impossible, to draw inferences regarding the relationship of periodontal treatment to overall systemic health across diseases. In consequence, it remains unclear whether periodontal treatment should be recommended as part of best practices in the management of systemic diseases and outcomes. Given that host inflammatory responses are similar in the pathogenesis of periodontitis and many chronic conditions, it is conceivable that reducing the systemic burden of inflammation in the body benefits the clinical management of these comorbidities. Since periodontal treatment is recommended to reduce the potential risk and complications associated with other chronic diseases, a comprehensive review of the effect of periodontal treatment on systemic health (clinical outcomes and biomarkers) is needed.

The purpose of this systematic review of reviews was to 1) summarize information about the effect of nonsurgical periodontal treatment on multiple systemic diseases and conditions, with emphasis on findings from high-quality SRs/MAs, and 2) assess the methodological quality of the underlying SRs and MAs. In addition, we offer recommendations for future research to address gaps identified by this review. Our findings are likely to be of interest to medical and dental providers who share in the care and management of patients, particularly those with chronic conditions, as well as policymakers and researchers interested in the link between oral and systemic health.

Methods

Protocol and Registration

We developed an a priori protocol for our study, which was registered in the International Prospective Register of Systematic Reviews (PROSPERO; #CRD42018087787). Furthermore, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for this article (Liberati et al. 2009).

Search Strategy

With the assistance of an academic librarian, we conducted a preliminary search in Ovid and EMBASE to identify SRs and MAs synthesizing primary literature on the relationship between periodontal treatment and systemic conditions. SRs and MAs synthesize the best available evidence from a body of work (typically randomized control trials [RCTs] and prospective cohort studies for intervention studies) to support evidence-based clinical decision making. The initial search focused on periodontal treatment and Alzheimer disease, asthma, cardiovascular disease, cerebrovascular disease, chronic kidney disease, diabetes mellitus, metabolic syndrome, obesity, pneumonia, pregnancy complications, respiratory tract diseases, and rheumatoid arthritis. We found SR and MAs for 6 diseases and conditions: adverse birth outcomes, cardiovascular disease, chronic kidney disease, diabetes mellitus, obesity, and rheumatoid arthritis, all of which were included in the study.

Medline via Ovid, EMBASE, and Cochrane Central databases were searched from inception to December 30, 2019. Keywords included in our search were periodontics, dental prophylaxis, pregnancy, cardiovascular diseases, kidney disease, obesity, rheumatoid arthritis, and diabetes mellitus. These search terms were modified and adapted based on the database searched (Appendix A). We did not search the gray literature because our focus was on high-quality, peer-reviewed evidence.

Inclusion Criteria

To be included, a study had to be 1) a peer-reviewed systematic review or meta-analysis, 2) written in English, and 3) focused on the effect of nonsurgical periodontal treatment on either adverse birth outcomes, cardiovascular disease, chronic kidney disease, diabetes mellitus, obesity, or rheumatoid arthritis. We included all outcomes associated with a systemic disease or condition, except for periodontal outcomes (such as probing depth or clinical attachment level). We excluded any study that evaluated the effect of nonsurgical periodontal treatment on only periodontal outcomes of patients with systemic diseases. For our intervention, we chose nonsurgical mechanical periodontal treatment, which uses scaling and root debridement to remove plaque, stain, and calculus. Nonsurgical periodontal treatment (root debridement) is the most common approach to treat periodontal disease (Sanz et al. 2012; Smiley et al. 2015). While there are other methods, such as surgical approaches, laser treatment, and antimicrobial therapy, they are less widely used than nonsurgical periodontal treatment. Studies were excluded if they only focused on the effect of antimicrobial therapy, laser treatment, or surgical periodontal treatments.

Study Selection

Article titles and abstracts were combined from all database searches, and duplicates were removed. Two reviewers (HLT and TKS) independently examined the title and abstract of all articles for potential inclusion. The full text of these articles was reviewed in detail against the inclusion/exclusion criteria. Discordant opinions on whether an article qualified for inclusion were discussed and resolved.

Quality Assessment

Three reviewers (HLT, SR, and TJT) independently appraised the methodological quality of each study using the A MeaSurement Tool to Access systematic Reviews 2 (AMSTAR 2) (Shea et al. 2017). This critical appraisal tool was developed specifically to evaluate the quality of reviews and meta-analyses that summarize RCTs, nonrandomized intervention studies, or both (Shea et al. 2017; Pieper et al. 2019). Since a significant number of SRs and MAs studying the effect of periodontal treatment on systemic outcomes draw on both RCTs and prospective cohort studies, AMSTAR 2 was deemed more appropriate to assess this body of work than AMSTAR, which should be used to assess quality of SRs and MAs based on RCTs only (Shea et al. 2017).

AMSTAR 2 provides a scheme for rating overall confidence in the quality of a SR or MA. The ratings range from “high” (indicating the study has no or 1 noncritical weakness) to “critically low” (indicating the study has more than 1 critical flaw with or without noncritical weaknesses). A full explanation of the criteria and rating scheme has been previously reported (Shea et al. 2017). AMSTAR 2 domains or items should not be used to calculate an overall numeric score, as this may hide critical weaknesses. Briefly, AMSTAR 2 lists 7 domains that can “critically affect the validity of a review and its conclusions” (Shea et al. 2017). These domains include consideration for a protocol being registered, adequacy of the literature search, justification for excluded studies, assessment of the risk of bias, the appropriateness of meta-analytical methods, whether risk of bias was considered in the interpretation of results, and whether publication bias was assessed (Shea et al. 2017). A study that is flawed in at least 1 of these critical 7 domains weakens confidence in the review. Thus, studies rated as having “low” or “critically low” confidence may not provide an accurate representation of the effect of periodontal treatment on systemic outcomes and should not be relied upon for clinical decision making or policy implications (Shea et al. 2017).

Data Extraction

Three reviewers (HLT, SR, and TJT) individually categorized the article’s systemic disease focus, the outcome(s) evaluated in each SR/MA, and the empirical result of each MA (negative, null, or positive). Empirical results were coded as “positive” if a study found a statistically significant positive association for positive outcomes (e.g., increase in birthweight) and negative association for negative outcomes (e.g., decline in HbA1c), “negative” if a study found a statistically negative association for positive outcomes (e.g., decline in birthweight) or positive associations for negative outcomes (e.g., increase in perinatal mortality), and “null” if no statistically significant associations were found. We also created a categorical variable to describe each outcome as a biomarker, surrogate endpoint, or clinical endpoint in accordance with a definition of chronic disease outcomes by the National Academy of Medicine (formerly the Institute of Medicine) (Institute of Medicine Committee on Qualification of Biomarkers and Surrogate Endpoints in Chronic Disease 2010). A biomarker is an “indicator of normal biological processes, pathogenic processes, or pharmacologic responses,” such as cholesterol level. A surrogate endpoint is an indicator believed to accurately predict a clinical endpoint. An example is blood pressure as a surrogate for myocardial infarction. Finally, a clinical endpoint is a variable that represents how a patient “feels, functions, or survives.” Clinical endpoints include mortality, improvements in quality of life, and relief of disease symptoms. All 3 reviewers met frequently to discuss findings and calibrate data extraction methods. Differences among reviewers were resolved by discussion.

Analyses

We used Covidence (2019) to combine search results, remove duplicates, and select articles for inclusion. Covidence is a SR automation software package that supports multiphase review workflow processes and includes features for data abstraction, screening, reviewer conflict resolution, and PRISMA flow diagrams (Kellermeyer et al. 2018). It is also the standard platform used in Cochrane reviews (Kellermeyer et al. 2018). Data extracted from the full text of studies were combined and prepared for analysis in Microsoft Excel (2016). Summary descriptive statistics were calculated for all meta-analyses and stratified by study quality using Stata/SE16 (2019). Overall conclusions for SR/MAs rated as having “moderate” or “high” confidence levels according to the AMSTAR 2 quality assessment were described.

Results

The initial search retrieved 926 articles (see Fig.). After duplicates were removed, 733 titles and abstracts were screened. Subsequently, 85 full-text articles were reviewed against inclusion/exclusion criteria. Articles excluded during full-text review and reasons for their exclusion are listed in Appendix B. In total, 52 articles met our inclusion criteria. Of these, 21 were focused on diabetes, 15 on adverse birth outcomes, 8 on cardiovascular disease, 3 each on obesity and rheumatoid arthritis, and 2 on chronic kidney disease. Most included studies defined the disease or condition of interest broadly, aiming to be more inclusive than exclusive of underlying studies. For instance, “CVD” encompassed heart disease, atherosclerosis, coronary artery disease, congestive heart failure, and hypertension. “Diabetes” was defined as type 1 diabetes mellitus, type 2 diabetes mellitus, hyperglycemia, and poor glycemic control. Last, “chronic kidney disease” definitions included renal failure and kidney damage.

Figure.

Study identification and selection process. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram illustrating article screening process. Of 926 articles retrieved from the search strategy, a total of 52 were selected for inclusion in this study.

Forty-four of the 52 studies were meta-analyses, and the remaining 8 systematic reviews did not include quantitative analyses. About half of the studies (53.8%; n = 28) restricted their inclusion criteria to RCTs only. The MAs comprised a total of 174 unique analyses of 41 outcomes. Table 1 displays outcomes evaluated in at least 3 separate analyses. Results of these analyses are grouped by study quality as determined by the AMSTAR 2 tool (see below). Across all studies, analyses of surrogate endpoints (n = 123; 70.7%) predominated by a large margin, followed by biomarkers (n = 40; 23.0%). True clinical endpoints, such as abortions/stillbirths and perinatal mortality, were evaluated only rarely (n = 11; 6.3%). The types of outcomes evaluated within the 174 unique analyses were also grouped by the condition or disease (Table 1). No meta-analysis has been conducted on clinical endpoints associated with cardiovascular disease, chronic kidney disease, diabetes, or obesity. Clinical endpoints evaluated thus far include adverse birth outcomes (such as perinatal mortality, abortion/stillbirth, and congenital malformations) and rheumatoid arthritis outcomes (such as tender joint count, swollen joint count, and disease activity score).

Table 1.

Outcomes Evaluated in Meta-Analyses Determining the Effect of Periodontal Treatment on Systemic Outcomes.

			High Qualitya		Low/Critically Low Qualityb
Systemic Outcome	Outcome Type	Total Number of Meta-Analyses	Number of Analyses with Positive Results	Numberof Analyseswith Null Results	Number of Analyses with Positive Results	Number of Analyses with Null Results	Chronic Condition Focus of Study
Preterm birth	Surrogate endpoint	38	1	5	9	23	ABO
HbA1c	Surrogate endpoint	25	1	1	19	4	Diabetes,c obesity
Low birthweight	Surrogate endpoint	25	2	4	1	18	ABO
C-reactive protein	Biomarker	8	0	0	5	3	CVD, diabetes,c obesity, RA
Fasting blood glucose	Surrogate endpoint	7	0	0	3	4	Diabetesc
High-density lipoprotein	Biomarker	5	0	0	1	4	CVD, diabetesc
Total cholesterol	Biomarker	5	0	0	2	3	CVD, diabetesc
Birthweight	Surrogate endpoint	5	0	3	2	0	ABO
Low-density lipoprotein	Biomarker	4	0	0	0	4	CVD, diabetesc
Triglycerides	Biomarker	4	0	0	1	3	CVD, diabetesc
Abortion/stillbirths	Clinical endpoint	4	0	0	1	3	ABO
Interleukin-6	Biomarker	3	0	0	1	2	CVD, obesity
Perinatal mortality	Clinical endpoint	3	0	1	1	1	ABO
Tumor necrosis factor–α	Biomarker	3	0	0	3	0	CVD, diabetes,c obesity
Otherd	Biomarkers and surrogate endpoints	35	0	1	7	27	ABO, CKD, CVD, diabetes,c obesity, RA
	Total	174	4	15	56	99

ABO, adverse birth outcomes (perinatal outcomes); CKD, chronic kidney disease (renal failure, kidney damage); CVD, cardiovascular disease (heart disease, atherosclerosis, coronary heart disease, congestive heart failure, hypertension); RA, rheumatoid arthritis.

^aHigh quality rating as determined according to the AMSTAR 2 (Shea et al. 2017) tool.

^bLow/Critically low quality rating according to the AMSTAR 2 (Shea et al. 2017) tool.

^cDiabetes includes type 1, type 2, and hyperglycemia.

^dIncludes all other biomarkers or surrogate endpoints evaluated in 2 or fewer unique meta-analyses. Examples include disease activity score, estimated globular filtration rate, creatinine, fibrinogen, leukocytes, flow-mediated dilation, adiponectin, body mass index, leptin, and so on.

As shown in Appendix C, 78.8% of studies were found to be of critically low quality (n = 41), 13.5% of low quality (n = 7), 0% of moderate quality (n = 0), and 7.7% of high quality (n = 4). Among studies that were rated as having “critically low” confidence (n = 41), 17 focused on diabetes, 10 on adverse birth outcomes, 7 on cardiovascular disease, 3 each on obesity and rheumatoid arthritis, and 1 on chronic kidney disease. Among studies that were rated as having “low” confidence (n = 7), 3 each focused adverse birth outcomes and diabetes and 1 on chronic kidney disease. Among studies identified as having a “high” overall confidence level (n = 4), 2 studies focused on adverse birth outcomes, 1 on cardiovascular disease, and 1 on diabetes.

Within the studies rated as having “high” confidence levels, 19 meta-analyses were conducted, with 4 analyses finding positive effects and 15 finding null effects. Simpson et al. (2015) reported that periodontal treatment reduced HbA1c levels within 3 to 4 mo of treatment but found insufficient evidence of reductions in months thereafter. Kim et al. (2012) reported a positive effect of periodontal treatment on low birthweight (<2,500 g) and preterm birth (<37 wk) among high-risk pregnant women and null effects for birthweight, low birthweight, and preterm birth among typical pregnancies. Iheozor-Ejiofor et al. (2017) reported periodontal treatment had a positive effect on low birthweight (<2,500 g) and null effects in other meta-analyses of birth outcomes, including preterm birth (<37 and <35 wk), perinatal mortality, preeclampsia, and low weight (<1,500 g). Liu et al. (2019) reported insufficient evidence for the effect of periodontal treatment on primary and secondary prevention of cardiovascular disease and were not able to conduct any meta-analyses.

Table 2 presents the number of SRs/MAs that adhere to each AMSTAR 2 critical domain. Critical domains in which the overwhelming majority of articles were in compliance included the adequacy of the literature search (n = 46; 88.5%), consideration of risk of bias when interpreting results (n = 36; 69.2%), the assessment of risk of bias from included studies (n = 35; 66.0%), the application of appropriate meta-analytical methods (n = 35; 66.0%), and assessing the presence and impact of publication bias (n = 31; 59.6%). The most frequent critical flaws included not registering a protocol prior to the review (n = 41; 78.8%) and not justifying excluded studies (n = 26; 50%). Detailed AMSTAR 2 ratings for each SR and MA can be found in Appendix D.

Table 2.

Number of Systematic Reviews and Meta-Analyses Evaluating the Effect of Periodontal Treatment on Systemic Disease That Did Not Meet AMSTAR 2 Critical Domains (n = 52).

AMSTAR 2 Critical Domains	Criteria Not Met, n (%)
Protocol registered before commencement of the review (item 2)	41 (78.8)
Justification for excluding individual studies (item 7)	26 (50.0)
Assessment of presence and likely impact of publication bias (item 15)	21 (40.4)
Risk of bias from individual studies being included in the review (item 9)	18 (34.0)
Appropriateness of meta-analytical methods (item 11)	18 (34.0)
Consideration of risk of bias when interpreting the results of the review (item 13)	16 (30.8)
Adequacy of the literature search (item 4)	6 (11.5)

AMSTAR 2, A MeaSurement Tool to Access systematic Reviews 2 (Shea et al. 2017).

Discussion

Our systematic review of reviews evaluated the methodological quality of SRs/MAs that study the effect of periodontal treatment on systemic disease outcomes. The review yielded 2 major findings: 1) within the 4 (7.7%) high quality reviews, 4 analyses found positive effects and 15 analyses found null effects of periodontal treatment on a systemic outcome, and 2) the overwhelming majority of SRs/MAs (92.3%) have “low” or “critically low” confidence ratings as determined by the AMSTAR 2 instrument. Just over a half (53.8%) of the studies were based exclusively on RCTs, including the 4 studies rated as high quality. The remaining 46.2% of studies included prospective cohort studies, which, given their design, may contribute to bias and unreliable results within SRs/MAs. Among the 4 meta-analyses determined to be of high quality, periodontal treatment was found to have a positive effect on HbA1c among diabetics, low birthweight (<2,500 g), and preterm birth in high-risk pregnancies (<37 wk) (Kim et al. 2012; Simpson et al. 2015; Iheozor-Ejiofor et al. 2017). Yet, even the interpretations of these positive findings were framed with caution due to the low-quality studies that underpin the analyses (Kim et al. 2012; Simpson et al. 2015; Iheozor-Ejiofor et al. 2017). Given the paucity of positive effects as determined by high quality studies and the low quality of the remaining studies, there is a dearth of robust evidence as to whether periodontal interventions improve systemic outcomes.

Surrogate endpoints were, by far, the most frequently analyzed outcome, followed by biomarkers and, rarely, clinical endpoints. Many surrogate endpoints and biomarkers are variables that can be obtained fairly easily (for instance, through a laboratory test) and/or in a timely manner (for instance, low birthweight in the context of pregnancy). Surrogate endpoints and biomarkers may be useful proxies for clinical outcomes if they reliably predict clinical endpoints (FDA-NIH Biomarker Working Group 2016). For instance, HbA1c is considered a validated surrogate indicator for microvascular complications and can be used to demonstrate direct clinical benefits (FDA-NIH Biomarker Working Group 2016). Since a number of biomarkers may be indicators for several diseases, particularly those related to inflammation, we need to broadly understand how periodontal treatment affects these markers. For instance, many studies use C-reactive protein (CRP), a systemic indicator of inflammation in multiple diseases, as a primary outcome of interest (Koromantzos et al. 2012; Bharti et al. 2013; Okada et al. 2013). However, a change in biomarker levels following an intervention does not necessarily indicate a change in disease risk (Krumholz 2015). At the same time, clinical endpoints for many chronic diseases, such as myocardial infarction or death in cardiovascular disease, progression to end-stage renal disease in chronic kidney disease, and diabetic peripheral neuropathy in diabetes, may not be observable within typical prospective study periods and thus difficult to capture. In these cases, surrogate endpoints are useful study outcomes.

As a result of this review, we recommend 4 measures to help advance our knowledge on whether and how periodontal treatment affects systemic disease:

1. The quality of SRs and MAs should be improved. High standards in methods are a “pre-requisite for valid interpretation and application of review findings” (Shea et al. 2009). In 2008, the Cochrane Collaboration provided guidance on conducting high quality reviews (Higgins et al. 2019). While AMSTAR and AMSTAR 2 have been available for some time, poor quality in systematic reviews is common (Tian et al. 2017; Pieper et al. 2018). Regardless of the quality of the primary literature, methodologically sound SRs are expected to provide the highest level of evidence for research, policy, and clinical care decisions.

2. Future reviews should adhere more closely to methodological guidelines for conducting and reporting SRs/MAs, such as the Cochrane Collaboration Handbook (Higgins et al. 2019) and PRISMA (Moher et al. 2009), than has been the case to date. Requiring measures such as the use of an appropriate tool to assess the methodological quality of a review and protocol registration in advance may be practically implemented by journals to which reviews are submitted (Shea et al. 2017).

3. We need to understand the quality of the primary literature. While a poor-quality SR/MA does not necessarily mean that the included studies are also of low quality, concerns regarding their rigor have been raised (Faggion et al. 2016; Shaqman et al. 2020). Examining the methodological rigor of individual studies is critical to understanding the true nature of the quality of literature in the entire domain. Methodological issues that already have been identified include how periodontal disease is defined and measured, low sample sizes, and wide variation in study duration (Simpson et al. 2015; Faggion et al. 2016; Beck et al. 2019). These inconsistencies affect systematic review and meta-analyses outcomes, and they make comparisons across studies subject to bias (Beck et al. 2019). Quantifying these methodological issues may help guide future research on best practices for determining whether and how periodontal treatment affects systemic health.

4. Additional, rigorous research that examines the relationship between periodontal treatment and health outcomes longitudinally is needed. Multiple circumstances make it difficult to study disease endpoints through prospective clinical studies and RCTs, such as ethical implications and issues with feasibility, costs, and the statistical power needed to evaluate systemic effects of periodontal treatment. Furthermore, while RCTs often have high internal validity, they may suffer from low external validity, potentially limiting the generalizability of their findings. One notable opportunity for research complementing RCTs is studying medical and dental health record data of the same patients and using self-reported medical history information contained within an electronic dental record (EDR) (Song et al. 2013; Merchant and Virani 2017; Thyvalikakath et al. 2020). With the increasing use of EDR data to determine dental treatment effectiveness (Siddiqui et al. 2017; Thyvalikakath et al. 2020), researchers can conduct large-scale observational studies on available electronic health and dental data in a potentially ethical, feasible, and cost-efficient way (Merchant et al. 2016; Merchant and Virani 2017). Given the large sample sizes, high granularity, and high dimensionality of data obtained by linking of EHR and EDR data (St Sauver et al. 2017; Patel et al. 2018), robust analytic approaches and quasi-experimental designs can be used to generate high quality, causal evidence. Characterizing outcomes of the care provided in real-world clinical settings could produce more generalizable findings than RCTs.

This study is the first to assess the methodological quality of SRs/MAs evaluating the effect of periodontal treatment on systemic health outcomes across various diseases and conditions. Our study was limited to 6 conditions because we found no SRs or MAs focused on the effect of nonsurgical periodontal treatment on any others. Given the large number of diseases associated with periodontal disease (Monsarrat et al. 2016), we attempted to minimize the chance of missing relevant SRs/MAs focused on periodontal treatment by enlisting the aid of an academic librarian and by searching 3 different biomedical literature databases. Some primary studies have evaluated the effect of periodontal treatment on diseases such as asthma and chronic obstructive pulmonary disease. However, since no SRs/MAs were available for these diseases, they were not included in this review. Nonetheless, the 6 conditions included in this study constitute a significant burden of morbidity and as a result are relevant from a clinical care and public health perspective. Finally, we endeavored to reduce bias due to subjectivity in AMSTAR 2 ratings by obtaining consensus between 3 independent reviewers for the quality assessment.

Conclusion

Weak evidence and the limited number of methodologically rigorous SRs/MAs underscore how little we know about the effect of periodontal interventions on chronic disease outcomes. While consistent evidence shows periodontal disease is associated with chronic diseases, more research is needed to validate whether periodontal treatment improves systemic health. Given the quality of existing studies, the dental community should be cautious not to interpret associations of oral and systemic disease as causal (Pihlstrom et al. 2018).

Author Contributions

H.L. Taylor, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; S. Rahurkar, contributed to conception, design, data acquisition, and interpretation, critically revised the manuscript; T.J. Treat, contributed to design, data acquisition, and interpretation, critically revised the manuscript; T.P. Thyvalikakath, contributed to design and data interpretation, critically revised the manuscript; T.K. Schleyer, contributed to data conception, design, data acquisition, and interpretation, drafted and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.