Article Text

Antibiotic prophylaxis for infective endocarditis: a systematic review and meta-analysis
  1. Thomas J Cahill1,
  2. James L Harrison2,
  3. Paul Jewell1,
  4. Igho Onakpoya3,
  5. John B Chambers2,
  6. Mark Dayer4,
  7. Peter Lockhart5,
  8. Nia Roberts6,
  9. David Shanson7,
  10. Martin Thornhill8,
  11. Carl J Heneghan9,
  12. Bernard D Prendergast2
  1. 1 Oxford Heart Centre, John Radcliffe Hospital, Oxford, UK
  2. 2 Department of Cardiology, St Thomas Hospital, London, UK
  3. 3 Centre for Evidence-Based Medicine, Nuffield Department of Primary Care Health Sciences, University of Oxford, Oxford, UK
  4. 4 Department of Cardiology, Taunton and Somerset NHS Foundation Trust, Taunton, UK
  5. 5 Department of Oral Medicine, Carolinas Medical Center, Charlotte, North Carolina, USA
  6. 6 Outreach Librarian Knowledge Centre, Bodleian Health Care Libraries, Oxford, UK
  7. 7 Department of Microbiology, Great Ormond Street Children's Hospital, London, UK
  8. 8 Unit of Oral & Maxillofacial Surgery & Medicine, University of Sheffield School of Clinical Dentistry, Sheffield, UK
  9. 9 Department of Primary Care Health Sciences, University of Oxford, Oxford, UK
  1. Correspondence to Dr. Bernard D Prendergast, Department of Cardiology, St Thomas’ Hospital, Westminster Bridge Rd, London SE1 7EH, UK; bernard.prendergast{at}


Objective The use of antibiotic prophylaxis (AP) for prevention of infective endocarditis (IE) is controversial. In recent years, guidelines to cardiologists and dentists have advised restriction of AP to high-risk groups (in Europe and the USA) or against its use at all (in the UK). The objective of this systematic review was to appraise the evidence for use of AP for prevention of bacteraemia or IE in patients undergoing dental procedures.

Methods We conducted electronic searches in Medline, Embase, Cochrane Library and ISI Web of Science. We assessed the methodological characteristics of included studies using the Strengthening the Reporting of Observational Studies in Epidemiology criteria for observational studies and the Cochrane Risk of Bias Tool for trials. Two reviewers independently determined the eligibility of studies, assessed the methodology of included studies and extracted the data.

Results We identified 178 eligible studies, of which 36 were included in the review. This included 10 time-trend studies, 5 observational studies and 21 trials. All trials identified used bacteraemia as an endpoint rather than IE. One time-trend study suggests that total AP restriction may be associated with a rising incidence of IE, while data on the consequences of relative AP restriction are conflicting. Meta-analysis of trials indicates that AP is effective in reducing the incidence of bacteraemia (risk ratio 0.53, 95% CI 0.49 to 0.57, p<0.01), but case–control studies suggest this may not translate to a statistically significant protective effect against IE in patients at low risk of disease.

Conclusions The evidence base for the use of AP is limited, heterogeneous and the methodological quality of many studies is poor. Postprocedural bacteraemia is not a good surrogate endpoint for IE. Given the logistical challenges of a randomised trial, high-quality case–control studies would help to evaluate the role of dental procedures in causing IE and the efficacy of AP in its prevention.

Statistics from

Key messages

What is already known about this subject?

Antibiotic prophylaxis (AP) was recommended prior to dental procedures in patients at risk of infective endocarditis (IE) for over half a century. Recent international guidelines have restricted (Europe and the USA) or recommended against (UK) this practice due to lack of evidence.

What does this study add?

This is the first study to systematically appraise the entire evidence base for AP, including time-trend studies of the effect of changing guidelines on the incidence of IE, observational studies and trials using bacteraemia as a surrogate endpoint for the development of IE.

Meta-analysis of trials indicates that AP is effective in reducing bacteraemia, but case–control studies suggest that this may not translate into a significant benefit for low-risk patients. One time-trend study suggests that total antibiotic restriction may be associated with a rising incidence of IE, but data on relative restriction are conflicting.

How might this impact on clinical practice?

The methodological quality of many studies of AP is poor and demonstrates the need for high-quality studies. Despite the limited evidence base, guidelines advising AP for patients at highest risk provide a pragmatic and justified approach.


Infective endocarditis (IE) is a rare but life-threatening disease.1 Despite trends towards multidisciplinary ‘heart team’ care and early surgery, 1-year mortality approaches 30%.2 In patients with prosthetic heart valves, rheumatic and congenital heart disease, the risk of acquiring IE is thought to be 10- to 50-fold higher than that of the general population.3 Effective strategies for prevention of both community and healthcare-acquired IE in at-risk groups are required.4

The oral cavity was identified as a major portal of entry for bacteria in 1909 by Thomas Horder.5 Oral streptococci are commensal flora of the oropharynx and account for 10%–30% of cases of IE, depending on the location, risk factor profile and socio-demographic characteristics of the population studied.6 7 Transient bacteraemia, which occurs in the setting of poor oral hygiene and periodontal diseases, dental procedures or in the course of normal daily activities (eg, tooth brushing), is thought to be a precursor to the development of some cases of IE.8

For over 50 years, oral antibiotic prophylaxis (AP) was given to patients at risk of IE undergoing dental procedures. Between 2007 and 2009, however, the European Society for Cardiology (ESC), American Heart Association/American College of Cardiology (AHA/ACC) and the National Institute for Health and Care Excellence (NICE) recommended restriction of AP to varying degrees.9–11 In Europe and the USA, there was relative AP restriction to those at highest risk (eg, patients with previous IE, congenital heart disease and rheumatic heart disease and selected heart transplant recipients) undergoing high-risk dental procedures. In the UK, NICE advised against use of prophylaxis entirely (total AP restriction) in 2008 but softened this stance in July 2016 to state that antibiotics should not routinely be recommended as prophylaxis for dental procedures.12

The rationale for relative or total AP restriction was threefold. First, as medicine shifted towards evidence-based practice, there was (and remains) no randomised controlled trial assessing the efficacy of AP for prevention of IE. Second, the relative importance of dental procedures as a cause of IE remained in doubt, compared with other portals of entry or low-grade recurrent bacteraemia occurring in the course of daily life.8 13 Third, in moderate risk (and high risk in England) groups, the overall hazards of antibiotic use (particularly anaphylaxis and the development of antibiotic resistance) were felt to weigh against use of AP. The NICE guideline committee also deemed that AP was not cost-effective as a result of lack of efficacy and the perceived risks of anaphylaxis.

The primary object of this study was to provide a systematic review and synthesis of evidence that directly or indirectly informs clinical use of AP for at-risk patients undergoing dental procedures. The evidence base comprises three types of study: (1) trials examining the effect of AP on the incidence of bacteraemia following dental procedures; (2) observational studies assessing the efficacy of AP for prevention of IE and (3) time-trend studies which examine the effect of changing national or international AP guidelines on the population incidence of IE.


Eligibility and search strategy

We searched the following databases from inception until 25 February 2016 to identify studies of the efficacy of AP for the prevention of bacteraemia or IE in patients undergoing dental procedures: Medline and Medline In-Process (OvidSP) (1946-present), Embase (OvidSP) (1974 to 08 February 2016), Cochrane Central Register of Controlled Trials (Cochrane Library, Wiley) (Issue 1 of 12 January 2016), Cochrane Database of Systematic Reviews (Cochrane Library, Wiley) (Issue 2 of 12 February 2016), Database of Abstracts of Reviews of Effects (Cochrane Library, Wiley) (Issue 2 of 4 April 2015), Science Citation Index Expanded and Conference Proceedings Citation Index—Science (Web of Science Core Collection) (1945 to present), ( and the WHO International Clinical Trials Registry Platform ( Search terms used included subject headings and title/abstract keywords for bacterial endocarditis, antibiotics and prophylaxis (see search strategy in online supplementary appendix 1). We also searched the reference lists of all included articles. The following categories of study were excluded: studies conducted prior to 1960, studies of AP in patients undergoing cardiac surgery or implantation of cardiac electronic devices, topical therapies and comparative antibiotic trials with no placebo/control arm.

Data abstraction

We assessed methodological quality of studies using the Cochrane Risk of Bias tool14 (for trials) or a checklist adapted from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria (for observational studies).15 Two reviewers (TJC and JLH) independently adjudicated the eligibility of studies, assessed the methodology of included studies and performed data extraction. Disagreement was resolved through consensus.

We extracted data on the study design: for case–control studies, we extracted baseline characteristics on the cases and the controls; for time-trend studies, we extracted study population characteristics, the study time period, relevant guideline changes and effects on incidence of IE per 100 000 population. The primary outcome of interest was the incidence of IE, incidence of (any) bacteraemia, or for time-trend studies, population-adjusted incidence of IE. Where total incidence of bacteraemia was not reported, the time point at which the highest incidence of bacteraemia was observed in the placebo group was used for comparison.

Data analysis

We derived summary tables to report methodological quality and main results of the included studies according to study design. For pooled effects, we used a fixed-effects model to generate Forest plots and used ORs as the summary measure. We assessed heterogeneity using I-square values, with 25%, 50% and 75% representing mild, moderate and substantial heterogeneity, respectively.16 Forest plots and data summary graphs were compiled using RevMan (Cochrane, UK) and SPSS, respectively.


The electronic search identified 3830 articles, after removal of case reports, editorials, animal studies and duplicates (figure 1). After screening of the title and/or abstract of these, 178 articles were deemed eligible for full-text assessment. In total, 36 studies were deemed suitable for inclusion (see online supplementary table 1 for excluded studies), comprising 10 time-trend studies, 5 observational studies (4 case–control studies and 1 retrospective cohort study) and 21 trials. All identified trials used bacteraemia as a surrogate endpoint for IE. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart for study inclusion is shown in figure 1.

Time-trend population studies

We identified 10 studies assessing the effect of changing national and international guidelines concerning the use of AP on the population incidence of IE. These included nine studies of relative AP restriction (from the USA and Europe) and one study examining the effect of total AP restriction (from the UK) (table 1). Changes in the guidelines between 2007 and 2009 by the ESC, ACC/AHA and NICE greatly reduced the use of AP. Annual incidence was reported in two studies17 18 and obtained from the authors for two studies.19 20 Figure 2 shows the incidence of IE per 100 000 population before and after changes in ACC/AHA and NICE guidelines. While only one study identified a significant rise in the incidence trend of IE, it is important to note that this change was observed in the only population with total AP restriction.

Figure 2

Annual incidence of infective endocarditis (IE) reported in time-trend analyses. The data for annual incidence or prevalence were reported in three studies17 18 30 and obtained from the authors for two studies.19 20 The incidence of viridans streptococcal IE in DeSimone et al was 0 in 2009 and 2011. The incidence values for Pant et al 17 were higher than other studies due to the inclusion of IE as both a primary and secondary diagnosis (included solely as a primary diagnosis in the other studies).

Table 1

Time-trend studies examining the effect of antibiotic prophylaxis guideline change on the incidence of IE

Observational studies

We identified five observational studies for inclusion, including four case–control studies and one retrospective cohort study (table 2). Data extracted included characteristics of cases and controls (or the two cohorts which were compared),21 exposures and interventions (ie, invasive procedures, use of AP) and where possible, the numbers of patients specifically undergoing dental procedures (table 2). All studies were assessed to be at high risk of intrinsic methodological bias (online supplementary table 3). Meta-analysis was conducted on three studies with available data concerning the numbers of cardiac patients exposed to dental procedures, use of AP and IE outcome (figure 3). Overall, the OR for use of AP in patients with IE was 0.59 (95% CI 0.27 to 1.30, p=0.14, I2=48%), suggesting no statistically significant difference in exposure to AP between cases (patients with IE) and controls. In Van Der Meer et al cases and controls were analysed up to 30 days postprocedure and subgroups combined (first-time and recurrent IE; definite and possible indications for AP). If patients without a definite AP indication were excluded, this study provided an overall OR of 0.63 (95% CI 0.17 to 2.36) for AP, modifying the overall meta-analysis to an OR of 0.47 (95% CI 0.21 to 1.06, p=0.07).

Figure 3

Meta-analysis of case–control studies testing the association between AP and IE after dental procedures. In these studies, cases are patients with IE and controls are matched patients at risk (table 2). The number of ‘events’ is the use of AP in each group as a proportion of the total number of procedures. Overall, the OR of AP comparing patients with IE to those without is 0.59 (95% CI 0.27 to 1.30, p=0.14), suggesting no statistically significant difference in AP exposure between cases and controls. AP , antibiotic prophylaxis; IE, infective endocarditis.

Table 2

Observational studies of AP and IE

Bacteraemia trials

We included 21 studies investigating the effect of AP on the incidence of bacteraemia (as a surrogate for IE) following a dental procedure. All studies reported the incidence of bacteraemia in a placebo group compared with an AP intervention group after a dental procedure. Some studies tested multiple antibiotic regimens (detailed in online supplementary table 3), and some compared additional endpoints such as the duration or magnitude of bacteraemia, a breakdown of specific organisms grown or antibiotic sensitivity patterns. A forest plot summarising a total of 35 antibiotic arms against control or placebo is shown in figure 4. AP was associated with a risk ratio of 0.53 (95% CI 0.49 to 0.57, p<0.01, I2=90%) for bacteraemia in patients following dental procedures.

Figure 4

Meta-analysis of trials of AP for prevention of bacteraemia after dental procedures. Where an individual study tested multiple antibiotic regimens, these are represented as (a), (b) and so on, and compared against the control/placebo arm. Details of the dental procedure and antibiotic regimen are shown in online supplementary table 5. Overall, use of AP was associated with a risk ratio for bacteraemia of 0.53 (95% CI 0.49 to 0.57, p<0.01, I2=90%). AP, antibiotic prophylaxis.


In this study, we have systematically reviewed the evidence base for the use of AP for prevention of IE. This comprises (1) population time-trend analyses of the effect of changing national and international guidelines on the incidence of IE, (2) focused observational studies, including four case–control studies and a retrospective cohort and (3) trials of antibiotics after dental procedures, using bacteraemia as a surrogate endpoint for the development of IE. No randomised controlled trial (RCT) of AP has been undertaken.

This is the first study to systematically appraise the total evidence base for AP across a range of study designs. We have conducted a comprehensive search and extensively reviewed studies that either directly or indirectly address the question of AP efficacy. All studies have been quality assessed, with risk of bias assessed in a systematic manner. However, our study has some limitations. Our conclusions are limited by the poor methodological quality of included studies (and their heterogeneity) and the lack of randomised trials. Furthermore, we have excluded studies prior to 1960 (to maintain relevance to current antimicrobial practice) and have not reviewed the data on use of AP to prevent IE in animal models, where some evidence suggests that single-dose amoxicillin prophylaxis is effective in preventing streptococcal IE.

In total, we identified 10 studies assessing the effect of national and international guideline change on the incidence of IE. In all countries where AP is still recommended, there has been no significant change in the overall rate of increase of IE, although in several it is claimed that there has been an increase in the number of streptococcal cases. However, IE rates have increased overall in the only study of total AP cessation from the UK.19 Although this study was unable to ascertain whether this increase was driven by a rising incidence of streptococcal IE, a further study is underway to identify the microbiological aetiology of these additional cases. These studies are intrinsically at high risk of methodological bias (as determined by the STROBE criteria) due to their observational study design and cannot fully account for confounding variables. Studies relying solely on discharge coding may not adequately account for readmissions or recoding of historical diagnoses. Pant et al 17 included secondary diagnoses of IE in their analysis, leading to higher estimates than other studies. Several smaller population studies with validated diagnoses have provided lower estimates for the incidence of IE of fewer than five cases per 100 000 per year.7 22 23 Finally, even the larger time-trend studies may remain underpowered to detect a significant change in IE incidence given the limited duration of follow-up. In particular, detection of any small change in incidence in studies of relative AP restriction requires a large population or prolonged duration of follow-up.

We identified five observational studies assessing the efficacy of AP. These were retrospective, of poor methodological quality, varying design (four case control, one retrospective cohort) and small sample size. Accordingly, they are at high risk of methodological bias and conclusions should be drawn with caution. With this major caveat, our meta-analysis of three observational studies did not show a statistically significant difference in exposure to AP in cases (patients with IE after dental procedures) compared with controls. There was a trend towards a protective effect of AP, however, and the lack of statistical significance may reflect the small sample sizes in the primary studies. Furthermore, most of the patients in these studies did not have replacement valves or other high-risk pathology so would not have been considered for AP even according to US or European guidelines. Two studies (Duval and Horstkotte) not included in our meta-analysis examined the protective effect of AP in patients with replacement valves. Both suggested a protective effect from AP. In the study by Horskotte et al 229 patients with replacement heart valves were followed after 287 diagnostic interventions (including some dental) requiring AP.21 A group of 304 patients who had undergone invasive procedures without AP was used for comparison. Six cases of IE occurred in the group with no AP, compared with 0 in the AP group. In a population study by Duval et al approximately 14 times more IE occurred after unprotected than protected dental procedures in people with replacement valves.24 The study by Horstkotte et al was excluded because the number of dental procedures was not stated and the study by Duval et al was excluded due to use of extrapolated rather than absolute numbers.

We identified 21 trials assessing the efficacy of AP in reducing the incidence of bacteraemia after dental procedures. Overall, AP is effective at reducing the incidence of bacteraemia. Other surrogate measures addressed in some studies include the nature of isolated bacteria, the duration of bacteraemia and its magnitude. However, the relationship between bacteraemia and IE is not straightforward. In particular, the relative importance of bacteraemia following dental extraction remains debated, and low level bacteraemia occurs commonly in association with daily activities such as tooth brushing, especially in the setting of periodontal disease.25 As such, its validity as a surrogate endpoint for IE is uncertain.

An RCT of AP has been debated for several decades but is unlikely to be performed for several reasons. Using IE as the primary outcome, such a trial would require several hundred thousand participants with a prolonged duration of recruitment and follow-up. In addition, there may be a lack of equipoise in an RCT, given that the standard of care for high-risk individuals (based on ESC and ACC/AHA guidelines) is to give AP. The size, scale and cost of a government-sponsored trial have been deemed unacceptable to national funding bodies in both the UK and USA.26

There is general acceptance that the majority of cases of IE caused by oral bacterial species are likely to result from frequent bacteraemia arising from routine daily activities, but this does not exclude the possibility that some cases result from infrequent invasive dental procedures.13 The focus of clinical research on IE prevention has therefore shifted in recent years from surrogate bacteraemia studies to those examining the role played by inflammation and ulceration of gingival tissues. A large multicentre case–control study assessing the associations between poor oral hygiene, dental disease and IE is currently underway and may provide the necessary data to permanently shift the focus away from AP as the best strategy to prevent IE.

As the debate continues, IE is changing. Oral streptococci—the target of AP—account for a falling proportion of cases in developed world series.7 In the absence of high-quality evidence and with significant barriers to an RCT, uncertainty is likely to prevail. For cardiologists and dental practitioners faced with high-risk individuals, AP remains a low-risk, inexpensive approach that may have benefit.27 28 We have previously described a framework for discussion of the risk:benefit balance for high-risk patient groups based on current ESC guidelines.29 Despite the low-quality and limited evidence base, these guidelines (and their counterpart from the ACC/AHA) advising AP for patients at highest risk provide a pragmatic and justified approach.

View Abstract


  • Contributors Data extraction and analysis was performed by TJC, JLH, PJ and IO. Electronic searches were performed by NR. Input on methodology, study identification, appraisal and analysis was provided by JC, MJD, PL, DS, MHT, CH and BDP. The manuscript was drafted by TJC, JLH and BDP and was approved by all authors.

  • Competing interests MJD was a non-voting member of the NICE panel who reviewed guidance concerning the use of antibiotic prophylaxis to prevent infective endocarditis in 2015.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.