A commentary on

Romandini M, Tullio I D, Congedi F et al.

Antibiotic prophylaxis at dental implant placement: which is the best protocol? A systematic review and network meta-analysis. J Clin Periodontol 2019; 46: 382395. DOI: 10.1111/jcpe.130.

figure 1

GRADE rating

Commentary and analysis

Efficacy of antibiotic prophylaxis for implant placement in reducing implant failure has been demonstrated by a Cochrane systematic review published in 2013.1 It was claimed that antibiotic prophylaxis reduced the risk of implant failure by 67%. However, their use in systemically healthy patients, requiring straightforward implant surgeries, has been prohibited to decrease the risk of enhancing antibiotic resistant strains of bacteria. This decision was based on the 2015 consensus conference of the EAO.2 Considering that the risk of antimicrobial resistance increases with longer regimens, it would be particularly important to determine, if shorter protocols are sufficient to prevent early implant failures, or if longer courses are preferred.3 This systematic review aimed to explore which dose, timing and type of antibiotic prophylaxis should be used at time of implant placement using network meta-analysis. The latter allows for direct and indirect comparisons among the different regimens. The quality of the review is very good, however, it might be at some risk of bias. It appears that the authors of the review were very thorough in finding RCTs covering the topic. Two authors worked in duplicate, searching four electronic databases, in addition to hand-searching of relevant journals, reference lists and grey literature with no restrictions on language or date of publication. They provided a detailed search strategy for all databases. However, they restricted the included RCTs to those treating ≥20 participants and those with a minimum of three month follow up period post-implant placement without justifications. The restrictions seem to be unnecessary, since early implant failure might happen before three months following implant placement. Besides, when intending to pool data in a meta-analysis, the number of participants included in each trial does not seem to be important, because the meta-analysis per se aims to increase the sample size, thereby decreasing the possibility of β error. The unnecessary restriction might place the review at a risk of selection bias. The review suggests that adverse events were insufficiently investigated or reported in the primary studies. Assessment of risk of bias seems to have been satisfactorily performed, since it was carried out by two reviewers independently within and across the studies at the outcome. Unfortunately, the review is based on seven RCTs which were of high risk of bias and only two which were of low risk of bias. Most of the included RCTs were underpowered or did not report a sample size calculation. The statistical methods used were appropriate. However, the network meta-analysis, which allowed for 18 indirect and ten direct pairwise comparisons, increased the heterogeneity and hence widened the confidence interval, thereby decreasing the precision and the reliability of the results. The risk of publication bias was not assessed in the review. This might be related to the insufficient number of RCTs involved in the direct comparisons. This should have been clarified in the review.

Accordingly, dose and duration of antibiotic prophylaxis are still debatable, however, the reviewers recommend a single dose of 3 g oral amoxicillin administered one hour before implant surgery but they do so with caution. Antibiotic cover is especially important, where a complex surgical procedure is expected, although the systemic condition of the patient should be considered along with whether he is allergic to penicillin or not.

Further RCTs with a calculated sample size, a follow up period not less than one year and a direct comparison among the different protocols, especially in complex cases as immediate implants, are still required. Besides, the added benefit of clavulanic acid in amoxicillin needs to be verified.