Data sources: Electronic searches without time or language restrictions were performed in PubMed, Web of Science and the Cochrane Oral Health Group trials Register. A vast manual search was done in many dental implant-related journals. Reference lists were scanned for possible additional studies. Ongoing clinical trials were also searched.
Study selection: Titles and abstracts of the reports identified were read independently by the three authors. Disagreements were resolved by discussion. Rejected studies were recorded with the reasons for exclusion. The inclusion criteria included clinical human studies, either randomised or not, comparing the implant failure/survival rates in any group of patients receiving antibiotic prophylaxis versus not receiving antibiotics prior to implant placement. Case reports and non-human studies were excluded. Implant failure was considered as complete loss of the implant.
Data extraction and synthesis: Data were extracted by the authors. Study risk of bias was assessed. Implant failure and post-operative infection were the outcomes measured, both dichotomous outcomes. Results were expressed using fixed or a random effect model depending on the heterogeneity calculated using an I(2) statistical test. The estimate of relative effect was expressed in risk ratio (RR) with 95% confidence interval. Number needed to treat (NNT) was calculated and sensitivity analysis was performed to detect differences among the studies considered to have high a risk of bias.
Results: Fourteen trials were included in the review and evaluated a total of 14,872 implants. Of the fourteen studies included in the review eight were randomised clinical trials, four were controlled clinical trials and two were retrospective studies. Seven studies had both patients and operators/outcome assessors blinded to the tested intervention. Nine studies had short follow-ups; six of them with a follow-up of four months, one of five months and two of six months.The antibiotic regimen was variable: seven studies did not use post-op antibiotics in all patients. Seven studies used amoxicillin as the antibiotic of choice. Ten studies had patients rinse with antimicrobial solutions as well.Among the fourteen studies, 8603 implants were placed in patients receiving antibiotics, 304 failures (3.53%) were recorded; 6269 implants were placed in patients not receiving antibiotics or receiving placebo, with 396 failures recorded (6.32%).From eight articles that provided the information about post-operative infection there were 25 occurrences of infection in 1000 patients receiving antibiotics (2.5%) and 29 episodes in 770 patients not receiving antibiotics (3.8%).Some studies involved grafting procedures and a portion of the patients (in nine studies) were smokers.The test of overall effect for implant failure rate showed that the difference between the procedures with or without antibiotic is statistically significant: RR 0.55, 95% confidence interval (CI) 0.41 to 0.75, p value 0.0002, a relative risk reduction (RRR) of 45% and the number needed to treat (NNT) of 50 (95% CI 33-100).For the post operative infection outcome, the results of the meta-analysis showed no statistically significant results; RR 0.84, 95 % confidence interval 0.49-1.44, and a p value of 0.52.The results did not differ in the sensitivity analysis when removing the high risk of bias studies. For the implant failure, the results were RR 0.37, 95% CI 0.19-0.72, p value 0.003, and for the 'post operative infection' outcome the results were RR 0.78, 95% CI 0.38-1.39. p value 0.33. The funnel plot to calculate publication bias showed asymmetry for the studies reporting implant failure, while the studies reporting post-operative infection the funnel plot did not show asymmetry.
Conclusions: The evidence from the review suggests that a prophylactic antibiotic regimen reduces failure of dental implants placed under ordinary conditions. However, there are no apparent differences in the occurrence of post-operative infections in patients receiving or not receiving antibiotics.The results have to be interpreted with caution due to the presence of several cofounding factors in the studies.