A Systematic Review and Meta-Analysis Evaluating Antibiotic Prophylaxis in Dental Implants and Extraction Procedures

Background and objectives: The use of antibiotic prophylaxis in extraction and implant dentistry is still controversial, with varying opinions regarding their necessity. The overuse of antibiotics has led to widespread antimicrobial resistance and the emergence of multi drug resistant strains of bacteria. The main aim of this work was to determine whether there is a genuine need for antibiotic prophylaxis in two common dental procedures; dental implants and tooth extractions. Methods: Electronic searches were conducted across databases such as Cochrane Register of Controlled Trials, the UK National Health Service, Centre for reviews, Science Direct, PubMed and the British Dental Journal to identify clinical trials of either dental implants or tooth extractions, whereby the independent variable was systemic prophylactic antibiotics used as part of treatment in order to prevent postoperative complications such as implant failure or infection. Primary outcomes of interest were implant failure, and postoperative infections which include systemic bacteraemia and localised infections. The secondary outcome of interest was adverse events due to antibiotics. The Critical Appraisal Skills Programme tool was used to assess the risk of bias, extract outcomes of interest and to identify studies for inclusion in the meta-analysis. Results: Seven randomised clinical trials (RCTs) were included in the final review comprising n = 1368 patients requiring either tooth extraction(s) or dental implant(s). No statistically significant evidence was found to support the routine use of prophylactic antibiotics in reducing the risk of implant failure (p = 0.09, RR 0.43; 95% CI 0.16–1.14) or post-operative complications (p = 0.47, RR: 0.74; 95% CI 0.34–1.65) under normal conditions. Approximately 33 patients undergoing dental implant surgery need to receive antibiotics in order to prevent one implant failure from occurring. Conclusions: There is little conclusive evidence to suggest the routine use of antibiotic prophylaxis for third molar extractive surgery in healthy young adults. There was no statistical evidence for adverse events experienced for antibiotics vs. placebo. Based on our analysis, even if financially feasible, clinicians must carefully consider the appropriate use of antibiotics in dental implants and extraction procedures due to the risk of allergic reactions and the development of microbial drug resistance.


Introduction
Since the accidental discovery of antibiotics by Sir Alexander Fleming of Scotland in 1929, they have been the greatest contribution to the 20th century world of therapeutics [1]. Initially the first systemic antibiotics (penicillin and sulphonamides) were reserved only for military use during World War II due to the expense and complicated manufacturing processes. As these processes were simplified, new formulations were developed, access to antibiotics increased and widespread use Table 1. Summary of commonly prescribed antimicrobial drugs in dentistry in the UK. Adapted from Ramu & Padmanabhan (2012) and the British National Formulary (BNF, 2013) [21,22].

Antibiotic Resistance
According to the World Health Organisation (WHO), antibiotics are the most misused of all medicines due to ease of access, being inexpensive, familiar and with generally good safety profiles. This has led to the growing problem of antimicrobial resistance (AMR) which is becoming a global threat that could cause an eventual loss of antibiotic efficacy [23]. The Global Antimicrobial surveillance (GLASS) programme runs by WHO revealed 500,000 people across 22 countries with suspected infections becoming antibiotic resistant with microorganisms such as Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, Streptococcus pneumoniae, and Salmonella spp. showing high rates of antibiotic resistance [24]. The European Union (EU) fact sheet on AMR estimates that antibiotic resistance results in approximately 25,000 deaths per year and in excess of €1.5 billion in related healthcare costs and productivity losses leading to resistance against different classes of antibiotics discovered to date [3,25]. Alanis (2005) reports that infections caused by the new strains of antibiotic resistant bacteria are not only difficult to treat but require longer courses of antibiotics and more complex therapy [2].
The new EU 'One health action plan against AMR' primarily aims to reduce the emergence and spread of AMR, thereby preserving the efficacy of existing antimicrobial agents for use in both humans and animals. In addition, greater monitoring and surveillance, strengthening infection prevention and control measures, promoting the appropriate use of antimicrobial agents followed by developing new alternative treatments and increasing awareness and understanding of AMR among both public and healthcare professionals were advocated by the EU action plan [26]. WHO has published a global strategy for the containment of resistance. The guidelines identified 68 recommendations calling for governments and health systems to produce their local guidelines [25,27].

Clinical Guidelines
Infective endocarditis (IE) is a severe infection causing inflammation of the endocardium due to a range of infectious agents including Staphylococci, Streptococci, fungi and Pseudomonas aerunginosa [28] and has a high mortality rate. Since many dental procedures cause bacteraemia, this may lead to invasive endocarditis in susceptible individuals. Previously, various national and international guidelines recommended that prior to invasive orthodontic procedures; those individuals at heightened risk of developing IE should be administered prophylactic antibiotics [17]. In general, prophylactic antibiotics are only recommended in surgery for:  [29][30][31] on antibiotic prophylaxis against infective endocarditis (IE), recommending that antibiotics for the purpose of preventing the development of IE should not be given to adults and children at risk of IE who are undergoing dental procedures. Prior to this, preventative antimicrobials were prescribed for prevention of IE for many decades. The evidence for this decision was that a consistent association between a patient having an interventional procedure and the risk of developing IE could not be found [28]. Therefore, it is on this basis that the clinical effectiveness of prophylactic antimicrobials is not considered to be proven [31][32][33]. These guidelines further suggest prophylactic antibiotics used against IE for dental procedures are not cost effective [29,30]. According to NICE regular tooth brushing almost certainly presents a greater risk of IE than a single dental procedure because of repetitive exposure to bacteraemia with oral flora [29].
Antibiotics are however appropriate for oral infections where there is evidence of spreading infection (swelling, lymph node involvement and cellulitis) or systemic involvement such as malaise and fever according to the Scottish Dental Clinical Effectiveness Programme [13].
The National Health Service (NHS) dentists are required to observe the guidance of NICE whilst prescribing. Clinicians who work privately may not have the same contractual obligation to follow this guidance. They would however require strong justification to their local clinical commissioning groups (CCGs) for choosing not to do so [30].
Implementing a change in clinical practice has own set of problems, even if the proposed changes are evidence based via national guidelines, because it not only involves studying new evidence but also abandoning the old evidence [31]. Cottingham (2012) reports that much needs to be done in order to improve the understanding of NICE guidelines among the dental profession as only 62% of dental trainers and 69.7% of dental trainees have read the CG No. 64 guideline and 55.7% trainers and 77.6% trainees applied it [32].

Study Aim
The study aim was designed using Process, Intervention, Comparator and Outcomes (PICO) guidelines. The purpose of this systematic review was to determine whether there is a genuine need for antibiotic prophylaxis in two common dental procedures; dental implants and tooth extractions, for which antibiotics are still currently being prescribed as part of therapy.

Search Strategy
Initial electronic searches were conducted using the Cochrane Register of Controlled Trials, the UK National Health Service (NHS) Centre for reviews, Science Direct, PubMed and the British Dental Journal to identify clinical trials of either dental implants or tooth extractions, whereby the independent variable was systemic prophylactic antibiotics used as part of treatment in order to prevent postoperative complications (implant failure, infection).

Study Criteria
The study was designed based on the PRISMA guidelines to produce systematic review and perform metanalysis. Potential studies identified in the initial search were required to meet inclusion criteria; clinical randomised control trials investigating dental implant or tooth extraction for any indication using prophylactic antibiotics as part of treatment to prevent postoperative complications such as implant failure or infection. Clinical trials were also required to be published in English and from 2000 until 2013. Studies which did not contain a control group and were not randomised were also excluded from this review ( Table 2). A double-blind RCT is of significant importance to eliminate the Hawthorn effect where patients may report fewer or more adverse events depending on personal beliefs or interpretation of the medication used [33].

Quality Assessment and Risk of Bias
Each study was critically appraised using the critical appraisal skills programme (CASP) tool checklist for clinical trials. To assess for risk of bias the RCTs were checked against four main quality criteria by the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (study details for each criterion can be found in Appendices A and B):
Participant compliance with follow-up

Results
A total of 1469 articles were identified by the electronic searches conducted on the specified databases. Titles and abstracts were analysed for relevancy to this work resulting in 1434 articles being irrelevant and subsequently rejected. Thirty-five full text articles were reviewed, however 14 were irretrievable and 8 had a lack of compliance with inclusion criteria or inappropriate interventions and therefore rejected. Thirteen full text articles then underwent detailed analysis resulting in 6 further studies being excluded from this review. Seven randomised clinical trials (RCTs) were included in the final review comprising of a total of 1368 patients (657 extraction patients and 711 implant patients) requiring either tooth extraction(s) or dental implant(s). All included studies were published in English and complied with the inclusion criteria. All of these studies compared at least one type of antibiotic regimen against placebo in patients undergoing either dental extraction or implant placement ( Figure 1).

Results
A total of 1469 articles were identified by the electronic searches conducted on the specified databases. Titles and abstracts were analysed for relevancy to this work resulting in 1434 articles being irrelevant and subsequently rejected. Thirty-five full text articles were reviewed, however 14 were irretrievable and 8 had a lack of compliance with inclusion criteria or inappropriate interventions and therefore rejected. Thirteen full text articles then underwent detailed analysis resulting in 6 further studies being excluded from this review. Seven randomised clinical trials (RCTs) were included in the final review comprising of a total of 1368 patients (657 extraction patients and 711 implant patients) requiring either tooth extraction(s) or dental implant(s). All included studies were published in English and complied with the inclusion criteria. All of these studies compared at least one type of antibiotic regimen against placebo in patients undergoing either dental extraction or implant placement ( Figure 1). Each of the randomised clinical trials used in this review were then categorised according to the level of bias as determined by the above specified criteria (Tables 2 and 3). Each of the randomised clinical trials used in this review were then categorised according to the level of bias as determined by the above specified criteria (Tables 2 and 3). Table 2. Rationale for each level of bias.

Level of Bias Description
High risk Possible bias seriously affecting the reliability of the results and high risk of bias if one or more of the criteria were not met Low risk Possible bias not seriously affecting the reliability of the results and low risk of bias if all criterion met  Sekhar et al. (2001) was the only extraction study where multiple extractions per patient were allowed [12]. All participants were ≥18 years of age with various form of edentulism, however only one clinical implants study [34] included patients that required a single implant supported crown. The greatest difference in ratio of males to females in any arm of any study was seen in the Kaczmarzyk et al. (2007) placebo group [35]. Implant studies have a greater mean age than extraction studies. No study included elderly, young children or immune compromised patients (Table 4).

Extraction Studies
Four of the seven studies involved prophylactic antibiotics for dental extraction. All four extraction studies were multi-arm randomised control trials comprising of a total of 657 patients requiring single/multiple dental extractions (835 extractions) for various indications (impacted wisdom teeth, abscess etc). Three studies were conducted in Europe [34][35][36], whilst the fourth was conducted in India [12]. All extraction studies involved patients being treated at referral centres by oral surgery specialists rather than general dental practitioners. All extractions studies used local anaesthesia to perform dental extractions and each study contained compared at least one antibiotic regimen against placebo. Interestingly, the most common indications for dental extraction are caries or periodontal disease, yet no trials were identified which assessed the effect of prophylactic antibiotics in patients requiring dental extraction for these indications (Table 4).

Dios et al. (2006) Trial
The authors performed microbiological analysis on post-operative bacteraemia present in blood cultures as an outcome measure to determine the effectiveness of prophylactic antibiotics in dental extraction. Dios et al. (2006) found Streptococcus spp. were the most commonly identified bacteria in all groups ranging from 44% to 68% with the lowest percentage being detected from the amoxicillin group (p < 0.0001). Amoxicillin and moxifloxacin prophylaxis showed high efficacies (p < 0.001 and p < 0.05 respectively) in reducing prevalence and duration of bacteraemia following dental extraction. Clindamycin prophylaxis was seen to be non-effective (p < 0.9). The results of the study therefore implicate that amoxicillin and moxifloxacin would be highly likely to reduce post-operate infections following dental extraction [36].

Lacasa et al. (2007) Trial
The authors conducted a phase III comparative study evaluating the efficacy of two schedules of a sustained release amoxicillin/clavulanate preparation in order to reduce infection after third molar surgery. A total of 225 patients were randomised equally into three groups: placebo, prophylaxis using single pre-op dose of amoxicillin/clavulanate 2000/125 mg, and a pre-emptive therapy group given a matching placebo dose (2000/125 mg) pre-op followed by amoxicillin /clavulanate 2000/125 mg twice daily for 5 days. A statistically significant higher rate of infection was seen amongst the placebo group: 16% (12/75) vs. single dose prophylaxis: 5.3% (4/75) vs. 5-day pre-emptive therapy: 2.7% (2/72) (p = 0.006). A linear correlation was found between the length of procedure and rate of incidence (p < 0.027) probably due to the length of exposure associated with more lengthy and complex bone removal procedures (ostectomy). Both therapeutic and prophylactic regimes vs. placebo had achieved greater reduction of pain postoperatively (p = 0.0001). However, prophylaxis was seen to be more beneficial in cases where ostectomy is not performed [34]. Overall results favoured the use of pre-emptive antibiotic therapy to reduce the rate of subsequent infection in patients subjected to ostectomy and a single prophylactic dose to be useful in simpler extraction procedures. Out of 8 planned outcomes that were listed, only one was explained fully whilst pain was reported as a mean for each arm of the trial without estimate variance [37].

Kaczmarzyk et al. (2007) Trial
This study involved 86 patients to evaluate the efficacy of a single and multi-dose clindamycin 5-day therapy to prevent inflammatory complications after third molar extractive surgery requiring bone removal. Clindamycin was chosen as it exerts strong antimicrobial action towards isolated strains from odontic infections as well as reaching high tissue concentrations. The only statistically significant result for any of the outcome measures (trismus, facial swelling, body temperature, pain, alveolar osteitis and lymphadenopathy) was a variation in body temperature was reported on the 7th day post-op (p = 0.03, Kruskal-Wallis rank test). All other outcome measure results in the study were p > 0.05, indicating a lack of statistical significance regarding efficacy in prophylaxis and pre-emptive therapy in any examined group. Results do not support the use of prophylactic antibiotics using clindamycin for preventing inflammatory complications in those requiring third molar extraction with bone removal under normal conditions [35].  (Table 5). They used random allocation into three groups: placebo vs. prophylactic antibiotics pre-op vs. antibiotic treatment post-op for 5 days. Metronidazole was the antibiotic of choice but was not justified. Pain score, swelling and wound state were all assessed on day 2 and 6 postoperatively. Results showed no significant differences in the outcome between the three groups (p = 0.09). Between individual variables assessed (swelling, pain, wound discharge), the degree of swelling was significantly less in the 5-day antibiotic post-op group (p = 0.03). The study concludes that in this case, results failed to show advantage in any group. Prophylactic antibiotics did not reduce morbidity after the removal of impact third molars [12].

Dental Implant Studies
Three implant studies were included in the final review comprising of 711 patients (1225 implants). All three implant studies were randomised and double-blinded. Two studies were multicentre parallel studies [6,15] conducted in Italy whilst the third study [31] was conducted in Spain. All three multicentre trials were conducted in private dental practices. Only one trial was supported by the implant manufacturer [31,34]. One clinical study [6] used placebo and antibiotics which were donated from a generic drug manufacturing company (Table 6). vs. n = 4 respectively). Immediate post-extractive implants were more likely to fail in comparison to delayed implants (9% vs. 2% respectively). Although trends clearly favoured the antibiotic group, no statistically significant differences were observed for outcome measures and no adverse events were reported. The authors conclude that sample size was insufficient to show a statistically significant difference [6].  [34].
Overall, trends favour the use of antibiotics in implant, but results are not statistically significant in order to support the use of prophylactic antibiotics in single implant placement for any of the outcome measures.

Caiazzo et al. (2011)
This study (high risk of bias) compared 4 interventions (n = 25 for each group): single dose 2 g amoxicillin 1-h pre-op vs. 2 g amoxicillin 1-h pre-op + 1 g twice daily for 7 days vs. 1 g amoxicillin post-op twice daily for 7 days vs. no antibiotic. The duration of follow up was 3 months after placement. No patients dropped out at any time. Two implant failures occurred in the placebo group vs. no failures in any of the 3 antibiotic groups (n = 75 patients). No statistically significant differences were observed for between groups (p > 0.05). No postoperative complications were reported in any group at weeks 1, 2, 4 and 8. No adverse events had been reported. Overall the authors concluded the lack of statistically significant evidence was perhaps due to the limited number of samples but still believe that implant placement may be one of the limited oral surgical procedures requiring routine antimicrobial prophylaxis [15].

Implant Failure
Data was pooled using REVMAN 5.0 software (Appendix A). Overall, results show more than twice the number of implant failures occurred in the placebo/no antibiotic group (4.8%) vs. antibiotic group (1.8%); RR 0.43; 95% CI 0.16 to 1.14.
The forest plot is a graphical representation of effect estimates and confidence intervals for each study using risk ratio (RR) and % weight as representation of event data. The blue box corresponds to the risk ratio point estimate and the % weight of each study is represented by the size of the box. Esposito et al. (2010) contributed the largest weighting (66%) for this particular outcome measure and is therefore represented by the largest blue box. Each horizontal line passing through a blue box depicts the 95% confidence interval (CI) range of intervention effects compatible with the study's result. This indicates whether each effect was individually statistically significant for that particular study. The line of no effect is seen passing vertically through 1 (when using RR). The overall combined data (overall effect estimate) is graphically represented on the forest plot by a black diamond box. This provides a meta-analytic summary of all data for an outcome to provide the best possible estimate of the effect of the intervention with confidence interval. The height of the black box represents the RR (0.64), whilst the width represents the 95% CI (0.43; 0.16-1.14) A risk ratio describes the multiplication of the risk which occurs due to experimental (antibiotic) intervention. Results show a risk ratio of 0.43 (95% CI 0.16-1.14), implying that antibiotics probably reduce the risk of implant failure by 57% (100 × (1 − RR)%) [35] based on these 711 patients under normal conditions. This is also known as the relative risk reduction. However according to the results of this review, prophylactic antibiotics were not statistically beneficial in those undergoing implant surgeries since p value (p = 0.09) for overall effect is greater than 0.05. This can also be seen in the forest plot ( Figure 2) as each individual 95% CI passes through the line of no effect and the overall effect black box is in contact with the line of no effect. According to the Cochrane guidelines for systematic reviews of interventions 5.1.0, small study effects are difficult to identify with less than 10 studies and so a funnel plot may not be useful in this instance.

Implant Failure
Data was pooled using REVMAN 5.0 software (Appendix A). Overall, results show more than twice the number of implant failures occurred in the placebo/no antibiotic group (4.8%) vs. antibiotic group (1.8%); RR 0.43; 95% CI 0.16 to 1.14.
The forest plot is a graphical representation of effect estimates and confidence intervals for each study using risk ratio (RR) and % weight as representation of event data. The blue box corresponds to the risk ratio point estimate and the % weight of each study is represented by the size of the box. Esposito et al. (2010) contributed the largest weighting (66%) for this particular outcome measure and is therefore represented by the largest blue box. Each horizontal line passing through a blue box depicts the 95% confidence interval (CI) range of intervention effects compatible with the study's result. This indicates whether each effect was individually statistically significant for that particular study. The line of no effect is seen passing vertically through 1 (when using RR). The overall combined data (overall effect estimate) is graphically represented on the forest plot by a black diamond box. This provides a meta-analytic summary of all data for an outcome to provide the best possible estimate of the effect of the intervention with confidence interval. The height of the black box represents the RR (0.64), whilst the width represents the 95% CI (0.43; 0.16-1.14) A risk ratio describes the multiplication of the risk which occurs due to experimental (antibiotic) intervention. Results show a risk ratio of 0.43 (95% CI 0.16-1.14), implying that antibiotics probably reduce the risk of implant failure by 57% (100 × (1 − RR)%) [35] based on these 711 patients under normal conditions. This is also known as the relative risk reduction. However according to the results of this review, prophylactic antibiotics were not statistically beneficial in those undergoing implant surgeries since p value (p = 0.09) for overall effect is greater than 0.05. This can also be seen in the forest plot ( Figure 2) as each individual 95% CI passes through the line of no effect and the overall effect black box is in contact with the line of no effect. According to the Cochrane guidelines for systematic reviews of interventions 5.1.0, small study effects are difficult to identify with less than 10 studies and so a funnel plot may not be useful in this instance.

Adverse Events
Adverse events were reported in only 2 out of 7 studies [32,34], of which only one study saw adverse events occur in the placebo/ no antibiotic group. Overall results (Figure 3) show that there is no statistical significance for adverse events (P = 0.30). The risk ratio of 1.84 (95% CI 0.59 to 5.77) implies (based on these results) that a patient is 1.84 times more likely to experience adverse events with treatment than adverse events without treatment. Alternatively, this can also be expressed as: 100 × (RR − 1)% = 100 × (1.84 − 1) = 100 × (0.84) = 84%. Treatment increases the risk of adverse events by 84% in the antibiotic group. Calculating the numbers needed to treat (NNT) helps understand the impact of an intervention on an outcome. Based on the results of this review, evidence suggests that approximately 33 patients undergoing dental implant surgery need to receive antibiotics in order to prevent one implant failure from occurring ( Figure 2 and Table 7).

Adverse Events
Adverse events were reported in only 2 out of 7 studies [32,34], of which only one study saw adverse events occur in the placebo/ no antibiotic group. Overall results (Figure 3) show that there is no statistical significance for adverse events (p = 0.30). The risk ratio of 1.84 (95% CI 0.59 to 5.77) implies (based on these results) that a patient is 1.84 times more likely to experience adverse events with treatment than adverse events without treatment. Alternatively, this can also be expressed as: 100 × (RR − 1)% = 100 × (1.84 − 1) = 100 × (0.84) = 84%. Treatment increases the risk of adverse events by 84% in the antibiotic group. Calculating the numbers needed to treat (NNT) helps understand the impact of an intervention on an outcome. Based on the results of this review, evidence suggests that approximately 33 patients undergoing dental implant surgery need to receive antibiotics in order to prevent one implant failure from occurring ( Figure 2 and Table 7).

Numbers Need to Treat (NNT)
Control group event rate (CER): proportion of outcomes that occur in control group.

Dental Extraction
In general, few statistically significant results were seen across all included extraction studies for all the listed outcomes. Two out of the four extraction studies were regarded as low risk of bias [35][36][37], both of which reported that preoperative antibiotics were associated with reduced prevalence of bacteraemia levels and reduced risk of infection. Amoxicillin given preoperatively

Numbers Need to Treat (NNT)
Control group event rate (CER): proportion of outcomes that occur in control group.

Numbers Need to Treat (NNT)
Control group event rate (CER): proportion of outcomes that occur in control group.

Dental Extraction
In general, few statistically significant results were seen across all included extraction studies for all the listed outcomes. Two out of the four extraction studies were regarded as low risk of bias [35][36][37], both of which reported that preoperative antibiotics were associated with reduced prevalence of bacteraemia levels and reduced risk of infection. Amoxicillin given preoperatively

Dental Extraction
In general, few statistically significant results were seen across all included extraction studies for all the listed outcomes. Two out of the four extraction studies were regarded as low risk of bias [35][36][37], both of which reported that preoperative antibiotics were associated with reduced prevalence of bacteraemia levels and reduced risk of infection. Amoxicillin given preoperatively showed higher efficacy than moxifloxacin and clindamycin (p < 0.001 and p < 0.05 respectively). Amoxicillin was more favourable in reducing serum bacteraemia levels in comparison to placebo (p < 0.0001) [33]. Lacasa et al. (2007) found a statistically significant linear correlation between an increase in length of procedure and incidence of infection (p < 0.027). Dental extraction performed with ostectomy (bone removal) takes longer to perform and so this was evident in all arms of the trial vs. rate of infection without ostectomy.
Adverse events were reported in only two studies [35,37], of which only one study saw events occur in the placebo/ no antibiotic group. Only two minor adverse events had been reported in antibiotic groups for included studies (diarrhoea and itching). This indicates that these antibiotic regimens seem to have been well tolerated but due to the small sample size of patients included in this review it is not possible to assess the occurrence of other rare adverse events associated with antibiotic use such as anaphylactic shock.
There is a minor trend showing adverse events occurring more so in the antibiotic group across the only two trials which reported any adverse events (2.43% in antibiotic group vs. 0.71% in placebo group). However, results of the present review show there is no statistical significant results for adverse events between groups (p = 0.30). Based on these results, a risk ratio of 1.84 (95% CI 0.59-5.77) implies a patient is 84% more likely to experience adverse events with treatment than adverse events without treatment.
Overall, 2 out of 3 extraction studies [12,35] which report postoperative inflammatory complications as an outcome measure do not support the use of prophylactic antibiotics after third molar extraction (n = 237). Lacasa et al. (2006) found pre-emptive antibiotics are more beneficial than prophylactic antibiotics for complex extractive surgery requiring ostectomy (p = 0.006). The authors recommend single dose prophylaxis in simpler extractive procedures where ostectomy is not performed to reduce postoperative complications. Two of the lead authors of this phase III clinical study are employees of the drug company funding the phase III trial.

Dental Implants
Using three extraction studies (n = 711), the present study has found no statistically significant evidence to support the use of routine prophylactic antibiotics in reducing the risk of implant failure (p = 0.09, RR 0.43; 95% CI 0.16 to 1.14) and post-operative complications (p = 0.47, RR: 0.74; 95% CI 0.34-1.65). The only trial which used 3 various prophylactic antibiotic durations [14] failed to reveal conclusive evidence as not a single outcome event occurred in any arm of the trial (postoperative infection, prosthesis/implant failure, adverse events).
A risk ratio of 0.43 for implant failure implies that prophylactic antibiotics probably reduce the risk of implant failure by 57% based on these 711 patients under normal conditions. Antibiotics probably reduce the risk of post-operative complications following implant placement by 26% (RR: 0.74; 95% CI 0.34-1.65). Approximately 33 patients undergoing dental implant surgery need to receive antibiotics in order to prevent one implant failure from occurring (NNT, Table 7). This will cost £120.00 for 33 patients to receive a 2 g amoxicillin prophylactic dose [22] in order to prevent one implant failure. Although this may seem financially feasible, clinicians must carefully consider the increase in rate of antibiotic resistance and the chance of allergic/toxic reactions occurring before deciding to treat 33 patients in order to prevent one implant failure from occurring.

Other Views
A systematic review conducted by Esposito et al. (2013) assessed the beneficial or harmful effects of systemic prophylactic antibiotics at dental implant placement vs. placebo. Six randomised controlled clinical trials (RCTs) with a follow-up of at least three months were analysed which compared the administration of various prophylactic antibiotic regimens vs. placebo to patients undergoing dental implant placement. The authors conclude their evidence suggests that, in general, antibiotics are beneficial for reducing the failure of dental implants placed in ordinary conditions vs. placebo (p = 0.002, OR 0.33; 95% CI 0.19-1.00). No statistically significant differences were seen between groups regarding postoperative complications (p = 0.28, RR 0.60: 95% CI 0. 36-1.35) or adverse events (p = 1.0, RR 1.0; 95% CI 0.006-15.85). The authors remain unsure whether postoperative antibiotics at dental implant are beneficial, and which antibiotic would be the most effective. Schwartz & Larson (2007) conducted a detailed literature review to assess 'antibiotic prophylaxis and postoperative complications following tooth extraction and implant placement'. Eight randomised clinical trials, one retrospective study and four additional randomised interventions were studied. In general sample sizes were small and provided insufficient statistical power to avoid type II, or false negative errors. The work emphasises how methodological differences in many of the dental clinical trials pose a problem regarding quality of evidence. The study concludes that there is a lack of evidence regarding the use of prophylactic antibiotics in general dentistry, including tooth extraction and implant placement [10].
A recent Cochrane systematic review conducted by Lodi et al. (2012) investigated the benefits of antibiotic prophylaxis in patients undergoing tooth extraction(s) for any indication. The authors found statistically significant evidence for a reduction in dry socket, pain, fever and trismus (p < 0.05). An increase in mild and transient adverse effects was observed in antibiotic groups vs. placebo. They conclude there is evidence that antibiotic prophylaxis reduces the risk of dry socket pain and infection following third molar extractive surgery. The main limitation of the review was almost half of the assessed trials were conducted before 1992 (6 of which were conducted in the 1980's) and many of which used selective reporting, incomplete outcome data and wide variations in methodological approaches. All studies included were either high risk of bias or had an unclear risk of bias, and therefore this may question the reliability of results [38,39]. Martin et al. (2005) assessed the appropriateness of antibiotic prophylaxis for third molar extractive surgery. The body of evidence examined showed that antibiotics may provide certain benefits in certain circumstances and little or no benefit in other circumstances. For example, it was found prophylactic antibiotics may be beneficial in extractive surgery requiring bone removal. Despite this, the authors believe the body of evidence questions the benefit of routine prophylactic antimicrobial therapy which does not appear to overcome risk of undesirable outcomes after dental extraction of third molars [9].

Limitations
One of the main limitations of the present review is that only seven studies were identified and available for review. Although the clinical extraction studies used in this review were conducted in various countries worldwide, many of the patients were young healthy patients in their early twenties and so the results of the review were more applicable to healthy young adults undergoing surgical tooth extraction. In contrast, older participants were recruited across all included implant studies (varying between 42-52 years). No trials were identified which included young children, elderly patients or immune compromised patients requiring dental extractions, therefore the results of this review may not be applicable to this group although they would be expected benefit more from prophylactic antibiotics due to increased risk of infection. Indeed, the NNT for outcomes would be likely to decrease if this group of patients were to be included, however it may not be possible or ethical to conduct clinical studies using this group of patients. Extraction studies identified involved patients being treated by oral surgery specialists by referral and so again, it is unclear whether these results are relatable to general dental practice. No clinical extraction studies were found evaluating the use of prophylactic antibiotics for patients with periodontal complications or severe caries as these are the most common indications for dental extractive surgery of third molars.

Implications for Dentists
A clinician's awareness about correct antibiotic choice is key to reducing 'blind prescribing', a factor which has contributed to the increase in antibiotic resistant microorganisms. Changes required in antibiotic prescribing habits presents its own set of problems as described by Soheilipour et al. (2011) whereby a qualitative study regarding the views of healthcare professionals on NICE guidelines revealed that prescribers experienced difficulty in explaining to their patients the change in clinical practice necessitated by adherence to the NICE guidance [40]. Concerns were also raised about the legal position of a clinician who did not follow the guidance. Further monitoring of antibiotic prescriptions among dentists is needed in order to effectively audit this controversial therapy. It is also recommended for continuing education of practitioners regarding the growing public health risks related to antibiotic prescriptions.

Implications for Further Work
There are varying opinions regarding the prophylactic use of antibiotics in dentistry [8][9][10]. More large scale randomised, double blind clinical studies need to be conducted. There is also a lack of clinical studies which have evaluated and defined the most appropriate and effective antibiotic regimen for dental procedures and so further work is recommended based on these proposals.

Conclusions
No statistically significant evidence was found to support the routine use of prophylactic antibiotics in reducing the risk of implant failure or post-operative complications under normal conditions. Approximately 33 patients undergoing dental implant surgery need to receive antibiotics in order to prevent one implant failure from occurring. Prophylactic antibiotics probably reduce the risk of implant failure by 57% based on 711 patients under normal conditions. There is also little conclusive evidence favouring the routine use of prophylactic antibiotics for third molar extractive surgery requiring bone removal in healthy young adults.
No trials were identified with the group of patients that would most likely benefit from the use of prophylactic antibiotics; elderly, young and immunocompromised patients. The results of the present study may therefore not be applicable to this group of patients. No trials were identified for the most common indications for dental extraction; dental caries or periodontal disease.
Much remains to be achieved in dental research including further large scale randomised, double-blind clinical studies using patients with infective complications such as infective endocarditis or immuno-compromised patients for various dental procedures.
Based on the articles analysed in this review it is recommended that clinicians carefully consider the appropriate use of antibiotics in dental implants and extraction procedures even if it is financially feasible due to risk of allergic/toxic reactions and the development of antibiotic resistance. Further monitoring of antibiotic prescribing in dentistry is required in addition to continuing education for dentists concerning the public health risks associated with antibiotic misuse.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Outcome Measures and Statistical Analysis Using REVMAN 5.0 Software
Data was pooled using REVMAN 5.0 software. The relative risk (RR) with 95% confidence intervals (CIs) for dichotomous data (data with two mutually exclusive groups, i.e., occurrence of implant failure or not, presence or absence of adverse events and presence and absence of postoperative complications was calculated for each study to quantitatively measure the probability of an event occurring. Where possible, based on the reviewed study design, the numbers needed to treat (NNT) were calculated to measure the number of patients who need to be treated to prevent one additional adverse outcome.

CASP Analysis of Kaczmarzyk et al. (2007)
Section (A) Are the results of the review valid? Screening Questions 1. Did the trial address a clearly focused issue? Yes. The authors attempted to evaluate the efficacy of two antibiotic regimens (single dose clindamycin and multidose clindamycin (5 days) for the reduction of inflammatory complications in patients undergoing third molar surgery with bone removal (extraction). Outcomes of interest were trismus, facial swelling, body temperature, pain, submandibular lynphadenopathy and alveolar osteitis.

Was the assignment of patients to treatments randomised?
A random number table was used for to determine group assignment for each patient in advance. Furthermore, allocation concealment involved opaque and sequentially numbered envelopes. 3. Were all of the patients who entered the trial properly accounted for at its conclusion? Nine out of 100 patients did not check in for follow up. Three had been disqualified due to various described complications and two had resigned during the trail without stating any reasons. Therefore 86 patients in total were analysed statistically at the end of the trial. Detailed questions 4. Were patients, health workers and study personnel 'blind' to treatment? Patients, clinicians and statisticians were all blinded.

Were the groups similar at the start of the trial?
Sizes of all three groups were similar (n = 27-31) The composition of each group was very similar except one group composed of a higher female population. Clindamycin 600 mg preoperatively then 300 mg placebo for 5 days: 8 (M)/23 (F), mean age = 23.4 years Clindamycin 600 mg preoperatively then 300 mg placebo for 5 days post-op: 9 (M)/19 (F), mean age = 23.5 years placebo: 6 (M)/21 (F), mean age = 24.6 years Total extractions between groups ranged from 27-31. 6. Aside from the experimental intervention were the groups treated equally? Yes. All groups were treated at one treatment center in Poland and subject to the same inclusion/exclusion criteria. Section (B) What are the results? 7. How large was the treatment effect? There was no statistically significant differences observed between groups for postoperative alveolar osteitis, pain scores, postoperative trismus and facial swelling (p > 0.05). Regarding a change in body temperature, a statistically significant difference between groups was recorded (p = 0.03). Despite this one statistically significant result for this study, the authors conclude that there is not enough evidence to suggest that clindamycin used prophylactically by itself or with subsequent 5-day therapy fails to demonstrate significant efficacy for prevention of inflammatory complications after third molar surgery. 8. How precise was the estimate of the treatment effect? The X 2 test was used to evaluate trismus, facial swelling, lymphadenopathy and alveolar osteitis using an odds ratio with a confidence interval of 95%. The Krukal-Wallis rank test was used to analyse surgery duration, analgesic intake, body temperature and level of pain experienced. Section (C) Will the results help locally? 9. Can the results be applied in your context? (or to the local population?) Results may not be applied to a whole general population. Patients were not treated in a general dental surgery, where most extractions take place. Patients did not comprise of the elderly, young children, immunocompromised patients or those at risk of IE. The mean age of patients across all three groups was in the mid 20's, and so they may be more applicable to younger adults, however even then, larger sample sizes than this study would also require to represent a typical population of patients. 10. Were all clinically important outcomes considered? No. All-important outcomes were considered except adverse events were not reported. 11. Are the benefits worth the harms and costs? Results described above are not statistically significant. In the case of this study, adverse events were not reported. With regards to cost, clindamycin is relatively cheap in the UK but the risk of other complications such as microbial resistance may outweigh the benefit if it is to be routinely prescribed for an extraction procedure which is uncomplicated and of short duration. Overall, the study shows there is little benefit from taking prophylactic antibiotics for the reduction in inflammatory complications following third molar surgery.