Periodontal disease treatment and risk of preterm birth : a systematic review and meta-analysis

The events leading to preterm birth are still not completely understood. A quantitative systematic review was performed to estimate the effects of periodontal care during pregnancy on preventing preterm birth and low birth weight. The meta-analysis included randomized trials with pregnant women with a diagnosis of periodontal disease before 20 weeks of gestation. Relative risk (RR) with 95% confidence intervals (95%CI) was calculated. We evaluated the reduction in preterm and low birth weight. Thirteen trials were included, comparing 3,576 women in intervention groups with 3,412 women receiving usual care. The meta-analysis of the effects of periodontal disease treatment during pregnancy indicated a non-significant reduction in preterm births (RR = 0.90; 95%CI: 0.68-1.19) and low birth weights (RR = 0.92; 95%CI: 0.71-1.20). The creation and examination of a funnel plot revealed clear evidence of publication bias. In summary, primary periodontal care during pregnancy cannot be considered an efficient way of reducing the incidence of preterm birth. Periodontal Disease; Premature Birth; Low Birth Weight Infant; Review; Meta-Analysis Introduction Preterm labor complicates 10 to 15% of all pregnancies, is the leading cause of neonatal morbidity and mortality and causes 75% of neonatal deaths that are not due to congenital anomalies 1. Preterm birth, defined as childbirth occurring at less than 37 completed weeks (or 259 days) of gestation, is a major determinant of neonatal morbidity and mortality and has long-term, adverse health consequences 2,3. Estimates indicate that, in the United States alone, costs associated with preterm birth, in terms of medical, educational expenditure and lost productivity, were more than US$ 26.2 billion in 2005 2. Although the etiology is thought to be multifactorial, the events leading to preterm birth are still not completely understood. It is unclear whether preterm birth results from the interaction of several pathways or from the independent effects of each pathway 2. There is convincing evidence that infections in pregnant women may alter normal cytokine and hormone-regulated gestation, which could result in preterm labor, premature rupture of membranes and preterm birth 4. Periodontitis has been associated with pregnancy outcomes such as preterm birth, low birth weight deliveries and premature rupture of membranes 4. The development of periodontitis involves the invasion of primarily Gram-negative bacteria through the periodontium, stimulating a chronic inflammaREVISÃO REVIEW

Periodontal disease treatment and risk of preterm birth: a systematic review and meta-analysis Tratamento de doença periodontal e risco de parto prematuro: revisão sistemática e metanálise Introduction Preterm labor complicates 10 to 15% of all pregnancies, is the leading cause of neonatal morbidity and mortality and causes 75% of neonatal deaths that are not due to congenital anomalies 1 .Preterm birth, defined as childbirth occurring at less than 37 completed weeks (or 259 days) of gestation, is a major determinant of neonatal morbidity and mortality and has long-term, adverse health consequences 2,3 .Estimates indicate that, in the United States alone, costs associated with preterm birth, in terms of medical, educational expenditure and lost productivity, were more than US$ 26.2 billion in 2005 2 .Although the etiology is thought to be multifactorial, the events leading to preterm birth are still not completely understood.It is unclear whether preterm birth results from the interaction of several pathways or from the independent effects of each pathway 2 .
There is convincing evidence that infections in pregnant women may alter normal cytokine and hormone-regulated gestation, which could result in preterm labor, premature rupture of membranes and preterm birth 4 .Periodontitis has been associated with pregnancy outcomes such as preterm birth, low birth weight deliveries and premature rupture of membranes 4 .The development of periodontitis involves the invasion of primarily Gram-negative bacteria through the periodontium, stimulating a chronic inflamma-Cad.Saúde Pública, Rio de Janeiro, 28 (10):1823-1833, out, 2012 tory response and forming pockets that become infected 5,6 .However, according to Oliveira et al. 7 , non-surgical periodontal treatment during the second semester of gestation did not reduce the risk of preterm birth, low birth weight, or preterm low birth weight.
The link between maternal periodontal disease and preterm birth or low birth weight is a grey area and it remains unclear whether adverse pregnancy outcomes have a causal relationship with periodontal disease or if they are a surrogate for another maternal factor.
The objective of the current work was to conduct a systematic review and meta-analysis of randomized, controlled trials to quantify the relationship between periodontal disease, preterm birth and low birth weight in order to explore the reasons for the ongoing controversies surrounding this issue.

Methods
The PRISMA guidelines were followed for the meta-analysis of randomized trials 8 .

Inclusion and exclusion criteria
To be eligible for inclusion in our systematic review, studies had to examine specific treatments for periodontal disease during pregnancy, compare the results of usual care ("intensified") and specific treatment ("less intensified") and report on at least one outcome of interest (preterm birth, low birth weight, and/or preterm low birth weight).We included only randomized trials whose participants met the following criteria: women over the age of 18 with a single gestation at 22 weeks or less, who had gingival inflammation with ≥ 25% of sites bleeding upon probing and sites with clinical attachment loss > 2mm.The exclusion criteria included the following: having fewer than 18 natural teeth, indication of prophylactic antibiotics for invasive procedures, occurrence of diabetes prior to pregnancy, and the intention of giving birth at a hospital outside this study.

Search strategy
We performed a literature search of studies published between 1980 and March 2012 using MEDLINE, Embase, BIOSIS, LILACS, Scopus, the Cochrane Central Register of Controlled Trials, the ISI Web of Science and IBECS.First, we derived three themes that were combined with the Boolean operator "AND".Each theme was created with the Boolean operator "OR" to search for terms appearing as either exploded medical headings (MeSH) or text words.The first theme was "randomized control trial".The second theme was "preterm birth", "low birth weight", "preterm delivery", "neonatal deaths", "intrauterine infection", "chorioamnionitis", "perinatal care", "LBW", and "PTB", and the third theme was "gingivitis", "periodontal treatment", "periodontitis" and "periodontal status".We then manually scanned the reference lists of all identified articles.No restrictions were placed on the language of the publications.All randomized controlled trials that allocated pregnant women to receive treatment with scaling and root planning, versus no treatment or prophylaxis, were eligible for inclusion.Trials were considered eligible if they included patients with documented periodontal disease (periodontitis or gingivitis), as defined by the International Workshop for Classification of Periodontal Diseases and Conditions in 1999.All trials were eligible regardless of the depth and the severity of the periodontal disease.We based our classification of severity of periodontal disease on the conclusions of the 2003 working group of the Centers for Disease Control and Prevention and the American Academy of Periodontology.Based on these classifications, we defined moderate and severe periodontitis in terms of probing depth and clinical attachment loss to enhance the case definitions and provide distinct categories.We excluded randomized trials that included patients with threatened preterm delivery who received tocolytic agents, non-randomized trials and pseudorandomized trials.

Data extraction
Multiple teams consisting of four reviewers (M.I.R, L.R.M, M.I.E. and P.D.S.P) independently screened the title, abstract and key words of each reference identified by the search and applied the inclusion and exclusion criteria.The same procedure was applied to full text articles and potentially eligible references.Differences in reviewers' opinions were resolved by discussion or a by a fifth reviewer.Data on quality, patient characteristics, interventions and relevant outcomes were independently abstracted by two reviewers (J.M.-M.and M.L.S).
Risk of bias was assessed by considering the adequacy of randomization and allocation concealment, and the comparability of the women in the different study groups.Although periodontal disease is treated with complex interventions, outcome assessors could be blinded to the treatment status of participants, thus minimizing bias.

Outcomes of interest
The effects of interventions were compared with respect to the following outcomes: preterm birth (< 37 weeks), low birth weight (< 2,500g).

Statistical analysis
We measured the inter-rater agreement for study inclusion and methodological quality assessment (weighted κ) 9 and outcomes were reported using relative risk (RR) with 95% confidence intervals (95%CI).Pooled-effect estimates were derived using a random effects model with Mantel-Haenszel statistics 10 .Study heterogeneity was determined using the I 2 statistic, in which numbers greater than 75% suggest considerable heterogeneity 9 , and p-values from the χ 2 test.In cases of considerable heterogeneity, no pooled-effect estimate was provided.A sensitivity analysis was planned a priori to compare study results and designs, and to report on study quality, focusing on those defined as "good quality" studies 11 .
Meta-analysis was performed using version 5.0.17 of the Review Manager software (Nordic Cochrane Center, Copenhagen, Denmark).

Characteristics of trials, participants and interventions
The 13 trials were conducted in seven countries from five different continents and included 6,988 patients (3,576 in the periodontal work group and 3,412 in the control group).
We excluded four randomized trials for the following reasons: one included patients hospitalized for threatened preterm birth that received tocolytic agents 21 , another 22 was a duplicate of Michalowicz et al. 6 , and in another 23 patients were randomized based on the efficacy of periodontal treatment that differed from other included studies.
All reviewed studies were published in English.Table 1 shows the details of the study populations, interventions, outcome assessments and data reporting.Table 2 presents the risk of bias for each trial.Inter-rater agreement of assessments of methodological quality ranged from 0.58 to 1.00 for the eight categories, with an overall agreement of 0.75.The lowest agreement was in the category other bias, whereas perfect agreement was achieved in the areas of adequate sequence generation and allocation concealment.Although all studies described randomization, nine did not adequately describe the random sequence generation methods 3,5,7,12,15,16,17,18,19,20 and nine did not adequately describe allocation concealment 3,5,7,12,15,16,17,18,19,20 .Twelve studies did not use blinding methods for adjudication outcomes 3,5,6,7,12,13,14,15,16,17,18,19,20 .Four trials described a modified intention-to-treat analysis 3,6,13,15 .Only two papers 3,6 distinguished between gingivitis and periodontitis (such as periodontal disease) and the remaining authors defined only periodontitis.Unfortunately, there is no universally accepted standard of periodontal disease 3 .Unlike gingivitis, periodontitis cannot be assessed by visual examination alone.The disease is diagnosed with the use of a probe that is inserted into the gingival crevice between the teeth and gums 6 .Therefore, the other authors establish the depth of each pocket (the insertion loss of the structure of the periodontium, i.e., the cementum, ligament and alveolar bone) according to their own criteria.None of the authors discussed the effectiveness of periodontal treatment in the treatment group.
Heterogeneity was confirmed by the χ² test, which produced a p-value of 0.00001 and 0.02 for preterm birth and low birth weight, respectively.

Assessment of publication bias
An examination of the funnel plot for our data suggests strong evidence of publication bias for the preterm birth and low birth weight outcomes in the meta-analysis.This bias was confirmed by the results of tests proposed by Steichen 24 , which produced p-values of 0.001 and 0.072 for preterm birth and low birth weight, respectively (Figure 4).Meta-analysis plot for preterm birth < 37 weeks of gestation.Figure 3 Meta-analysis plot for low birth weight < 2,500g.

Sensitivity analysis
A sensitivity analysis was performed to test the robustness of the findings 24,25,26,27 .Pooled treatment effect estimates from studies with adequate sequence generation, randomization and allocation concealment were analyzed 6,17,18 .

Discussion
The World Health Organization (WHO) reported an estimated 12.9 million annual worldwide preterm births between 1997 and 2007, representing an incidence of 9.6% 28 .This meta-analysis showed that periodontal disease treatment during pregnancy does not confer a general protec- tion against preterm birth and low birth weight.
Our systematic review identified 13 randomized 3,5,6,7,12,13,14,15,16,17,18,19,20 , relevant studies published between 2002 13 and 2012 20 .These studies provided evidence from 6,988 pregnant women with periodontal disease who experienced a total of 790 preterm births.A second and equally important finding is that most of the studies included found no association between the subjects of study.In fact, the studies that rejected the alternative hypothesis were the most controlled and had the largest sample sizes.

Strengths and limitations of this review
Many aspects of this review lead us to believe that our results are valid.Firstly, we formulated a clinical question and then performed comprehensive searches that encompassed multiple online databases and searched the reference sections of relevant studies.No language restrictions were imposed and we used broad search terms to avoid making our question too specific to be adequately sensitive.Although we did not seek any unpublished data, there is a risk of publication bias.However, we included only randomized, controlled trials to minimize selection bias.Many of the I 2 esti-mates calculated in this meta-analysis may be considered high.This is a drawback, and may be the result of considerable heterogeneity between studies 9 .

Comparisons with other studies
Jeffcoat et al. 23 performed a randomized, controlled trial but included an intervention analysis based on the efficacy of periodontal treatment that differed from other studies.They found that successful, routine periodontal treatment was associated with decreased incidence of spontaneous preterm birth in the study population.We did not include this study in our systematic review because the intervention group included two subgroups (successful and unsuccessful) that were not found in other studies.According to Di Mario et al. 29 , the applied study design of Jeffcoat's 23 study does not permit the control of confounders and bias and, therefore, the lack of randomization and control for previous preterm birth affects the validity of this study.
Polyzos et al. 30 conducted a meta-analysis of randomized clinical trials to determine whether periodontal treatment during pregnancy reduced preterm birth.After examining eleven trials that included 6,558 patients, they found that periodontal treatment during pregnancy had no significant effect on the overall rate of preterm birth (OR = 1.15, 95%CI: 0.95-1.40;p = 0.15).In 2011, two meta-analyses were published by Chambrone et al. 31 and Fogacci et al. 32 , including 11 and 10 trials, respectively, neither of which supported the hypothesis that periodontal therapy reduces preterm birth and low birth weight indices.
The differences reported can be explained by the different strategies employed in the meta-analysis.George et al. 33 pooled all the trials together, while Polizos et al. 30 conducted the review by analyzing the high and low-quality trials.
Our study included 6,988 patients, all items covered by Polyzos et al. 30 and two other clinical trials published in subsequent years 20,29 .Our results corroborate Polyzos's and Chambrone's findings 30,31 .Two other systematic reviews were published in 2006 and 2007, but used case-control, cohort and controlled trials 34,35 .

Conclusions
Primary periodontal care during pregnancy cannot reduce the rate of preterm birth or low birth weight.Although this review found no benefits of periodontal care during pregnancy, future research is needed to address this important question.

Figure 1 Flow
Figure 1Flow chart summarizing the literature search process.

Figure 4 A
Figure 4A funnel plot of publication bias of the studies included in this systematic review.

Table 1
Characteristics of included studies: treatment for periodontal disease versus usual care.

Table 2
Assessment of study quality.