Network meta-analysis for evaluating interventions in implant dentistry: The case of peri-implantitis treatment

Background/Aim: Evidence from head-to-head comparison trials on peri-implantitis treatment is limited, and it is therefore impossible to conduct a direct meta-analysis. We propose an alternative statistical method, network meta-analysis, for evidence synthesis, which enables to compare the results of multiple treatments. Methods: We searched, in triplicate, for randomized controlled trials (RCTs) and controlled trials in the PubMed, Cochrane Central Register of Controlled Trials, Clinicaltrials.gov, and Latin American and Caribbean Health Sciences Literature databases up to and including August 2010. We also conducted a manual search of the reference lists regarding published systematic reviews and searched for gray literature in OpenSIGLE. We assessed changes in clinical attachment level (CAL) and pocket probing depth (PPD) after nonsurgical and surgical treatments of peri-implantitis. The risk of bias of selected studies was determined by the use of specific criteria, and it was performed in triplicate and independently. We used multilevel mixed modeling to perform the network meta-analysis and Markov Chain Monte Carlo simulation to obtain confidence intervals for the fixed and random effects. Results: Eleven studies were included in the review. All RCTs are at unclear or high risk of bias. Surgical therapy in conjunction with bone grafts and non-resorbable membranes achieved 3.52mm greater PPD reduction than nonsurgical therapy alone, 95% high-probability density (HPD) intervals: -0.19, 6.81. Surgical treatment in conjunction with bone grafts and resorbable membranes achieved 2.80mm greater CAL gain than nonsurgical therapy alone, 95% HPD intervals: -0.18, 5.59. Conclusion: Surgical procedures in peri-implantitis treatment achieve more PPD reduction and CAL gain than nonsurgical approaches. Nevertheless, these results should be interpreted with caution because of the limited number of studies included and their low methodological quality. Network meta-analysis is a useful statistical methodology for evidence synthesis and to summarize the strength and limitation in the current evidence.