We aimed to answer the following questions: What is the comparative effect of different
adjunctive antimicrobial agents as an adjunct to SRP in treatment of residual pockets?
3-1. Selection criteria 3-1-1 Studies
Randomized controlled studi, either parallel or split-mouth designed, included subjects with measures of residual pockets and controls.
3-1-2 Population
Non-medically compromised patient with residual pockets after periodontal phase I therapy.
3-1-3 Intervention
Tested one or more methods used as adjuncts to SRP.
3-1-4 Comparisons
Received the same SRP alone or with a placebo as the treatment group.
3-1-5 Outcomes
Changes in PPD reduction and CAL gain.
3-1-6 Exclusion criteria
Trials were excluded if they met the following criteria:
(1)!Patients were with early-onset or aggressive periodontitis.
(2)!Patients who were not under maintenance care of periodontal therapy.
(3)!Assessed effect of an antimicrobial agent without SRP (i.e.: monotherapy).
(4)!Treatment was not focused in residual pockets.
(5)!Treatment has included furcation sites.
(6)!Treatment is not an antimicrobial agent.
(7)!Insufficient data can be collected.
(8)!Design is not a randomized controlled trial.
3-2. Identifying research evidence
We use th keywords including MeSH Terms of (supportive periodontal therapy OR periodontal maintenance OR residual pocket) AND (“Dental Scaling” [Mesh] OR “Root Planing” [Mesh] OR periodontal treatment OR periodontal therapy OR scaling root planing). No additional restrictions in terms of language were imposed. The reference lists of previously published reviews and meta-analyses were also cross-checked for the trials missed in our electronic literature search.
The search was conducted within the electronic databases via Ovid Medline (via PubMed), Embase, Cochrane library, Cochrane Databse of Systemic Reivews up to October 31st, 2017.
Unpublished data were sought by searching a database listing unpublished studies (OpenGray [http:// www.opengrey.eu/], formerly OpenSIGLE). Furthermore, reference lists of previous systematic reviews and included studies were examined and
hand-searched in order to recognize any further articles that could be considered for inclusion.
3-3. Study selection
Study selection was conducted by one independent reviewer in the following steps:
1.! Initial screening of potentially-suitable titles and abastracts against in the inclusion criteria to identify potentially applicarble studies.
2.! Screening of the full-text articles identified in step one as possible eligibility Studies were excluded if not meeting the inclusion criteria. Studies would be cited under one study name if same data were reported in multiple studies. When doubts about including the study or not were raised, a consensus was obtaitned with second reviewer by discussion.
3-4. Data extraction
A standardized data extracation form from eligible studies were recorded by one reviewer and were confirmed by another independent author. In particular, the
following data were recorded: study design, number of patients, treatment protocol, durationof of follow-up period, outcome measurement including changes of PPD, CAL, and their standard deviation. If standard error or confidence intervals instead of standard deviation were reported in the studies, the standard deviation will be derived by formula according to Cochrane Handbook for Systematic Reviews of Interventions(Cochrane,
2011). If the data presented in the studies is calculated based on site level, we will adjust it to a subject(patient) level(Tu et al., 2012a). Changes of PPD is calculated by PPD after treatment minus initial PPD, changes of CAL is also calculated in the same way. Since follow-up period of each studies is different, we split the data of outcome measurement into three time points: short-term (≤ 3 months), medium-term (4 ~ 6 months), long-term (>7 months). Request for missing data through e-mail to authors were performed in studies with insufficient information.
3-5. Quality assessment
Quality assessment of the studies was made by focusing on methodological topics highlighted by Higgins & Green (2011) in the Cochrane Collaboration's tool for
assessing the risk of bias. This tool involves six domains of bias: selection bias (random sequence generation, allocation concealment), performance bias (blinding of
participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other bias. The risk of bias in the included studies was categorized as follows: (a) low risk of bias (plausible bias unlikely to seriously alter the results) – if the item was met, (b) unclear risk of bias (plausible bias that raises some doubt about the results) – if there was not enough information to judge the item, and (c) high risk of bias (plausible bias that seriously weakens confidence in the results) – if the item was not met.
3-6. Pairwise and Network meta-analysis
We did two types of meta-analysis using the frequentist analysis approach. First, we did standard pairwise meta-analysis using the DerSimonian and Laird random effects model. Pooled outcomes were expressed as weighted mean differences (WMD) with their associated 95% confidence intervals(CI). The significance of discrepancies in the estimates from the different studies was assessed by means of the Cochrane test for heterogeneity and the I2 statistic. Potential publication bias was estimated using Egger's linear regression test and funnel plots(Egger et al., 1997, Begg and Mazumdar, 1994).
Second, we did a network meta-analysis with random effects model(White et al., 2012, Ades, 2006). A consistency model was used to compare effects of different treatments.
Inconsistency between direct and indirect evidence in network meta-analysis was also assessed to check whether its direct evidence and its indirect evidence are consistent, loop inconsistency and design inconsistency was also determined by use of inconsistent factors(Ades, 2006) . To rank the treatments for an outcome, we used surface under the cumulative ranking (SUCRA) probabilities, the larger the SUCRA, the more effective an intervention may be. We did our analysis with STATA version 15 (STATACORP, Texas, USA).