Antibiotic prophylaxis (AP) to prevent infective endocarditis (IE) is established practice and recommended in most countries around the world at present, the UK being the notable exception. The first recorded use of AP was in 1941 . The American Heart Association (AHA) published the first guidelines in 1955 ; the first European consensus document was published in 1995 . Since then, there has been gradual refinement of the guidelines. The most recent European Society of Cardiology (ESC) guidelines were published in 2015 .
The current ESC guidelines recommend that AP should be considered for individuals thought to be at “high risk” of IE when undergoing “high-risk” dental procedures.
High-risk dental procedures involve “manipulation of the gingival or periapical region of the teeth or perforation of the oral mucosa”. The AHA guidelines are very similar.
Other procedures, such as respiratory or urological procedures, for which AP was recommended in earlier versions of the guidelines, are no longer regarded as high-risk, and AP is no longer advised, unless procedures are being undertaken in the context of active infection.
The ESC defines three patient groups as being at “high risk” of IE. These groups are:
- Patients with a prosthetic valve, or where prosthetic material has been used to repair a valve. This group includes patients with a percutaneous aortic or pulmonary valve.
- Patients with previous IE.
- Patients with congenital heart disease who are cyanotic and those who have had palliative shunts/conduits/other prostheses. Those who have had complete repair are considered high-risk for the first six months after the procedure only.
The magnitude of the increased risk has recently been confirmed by a number of studies [5,6]. In the past, “moderate-risk” patients, such as those with native valve disease or less severe congenital heart disease, have been covered by AP, but this is no longer the case.
The class of evidence is given as IIa, the level of evidence as C. Class IIa is defined as “The weight of evidence/opinion is in favour of usefulness/efficacy”. Level of evidence C is defined as “Consensus opinion of the experts and/or small studies, retrospective studies, registries”.
There are no randomised controlled trials to support the use of AP, and the recommendation to use AP is based upon the consensus view of experts. This article will summarise the evidence that does exist which underpins this consensus.
Dental procedures were first linked with IE in 1923 . Lewis and Grant suggested that bacteria released during dental procedures might cause IE in some susceptible individuals. This hypothesis was proven by Okell and Elliott in 1935, who noted that 61% of patients had positive blood cultures for oral viridans group streptococci (OVGS) following a dental extraction . The concept that antibiotics given prior to a dental procedure might reduce the risk of bacteraemia and hence the risk of developing IE followed and, in 1941, Thomas et al were the first to report the use of AP . In 1955, the AHA published the first set of guidelines, entitled “Prevention of rheumatic fever and bacterial endocarditis through control of streptococcal infections” .
Durack and Petersdorf developed an experimental model of IE in rabbits. A polyethylene catheter was introduced via the internal jugular vein and passed across the tricuspid valve into the right side of the heart. The following day bacteria were injected intravenously. A variety of antibiotic regimens was tested and some were successful at preventing this experimentally induced IE . Many similar studies using rabbit and other animal models also showed that antibiotics given prior to injections of bacteria could prevent IE, supporting the use of AP.
If AP is of benefit, a number of observations must be true. Data have been gathered in an attempt to answer all of these questions.
- Dental procedures should cause a bacteraemia.
- Antibiotics should reduce the duration and/or magnitude of that bacteraemia.
- It should be possible to demonstrate a link between dental procedures and the development of IE.
- When patients thought to benefit from AP do not receive it, then they should be at higher risk of developing IE than those who do receive AP.
Do dental procedures cause a bacteraemia and do antibiotics affect that?
The observation that invasive dental procedures can release bacteria into the bloodstream has been repeated on many occasions with increasingly stringent protocols. For example, Lockhart et al  clearly demonstrated that just over half of patients who had a tooth extracted developed a brief bacteraemia with an organism recognised to cause IE.
Furthermore, they demonstrated that amoxicillin given as AP significantly reduced the proportion with a positive blood culture to around one in five.
However, the study also demonstrated that around one in ten individuals would develop a bacteraemia with simple tooth brushing. Many other studies have demonstrated bacteraemias after everyday activities, including the simple act of chewing food.
This latter observation has caused many to question the value of AP. What is the point of covering one activity which on average occurs every two years, when there will be literally thousands of activities in that time which can cause a bacteraemia and will not be covered?
In any event, it is possible that both scenarios can cause IE; neither argument precludes the other from being true.
Are dental procedures associated with endocarditis?
If it is argued that AP is effective at preventing IE, then it follows that there should be evidence that dental procedures cause IE. Thus far, those data have proved elusive.
Van der Meer et al looked at whether or not dental procedures preceded cases of IE across the Netherlands and published their results in 1992 . They interviewed patients with IE about procedures undergone in the 180 days before the onset of symptoms of IE. Eighty-nine patients (20.8%) had a relevant procedure in the previous six months which could have resulted in IE.
In 1995, Lacassin et al compared 171 IE cases and matched controls in France . Forty (40) cases had undergone extraction (11), scaling (14), or root canal treatment (15) in the previous three months, whereas only 26 controls had undergone similar procedures. This difference was not statistically significant.
In 1998, Strom et al undertook a case control study in the USA . Two hundred and seventy-three cases were interviewed. In the three months prior to treatment, 63 patients (23.1%) and 64 controls (23.4%) underwent a dental procedure. This difference was not significant.
Porat Ben-Amy et al  used a case-crossover design to look at the frequency of dental procedures in the three months prior to admission compared with other three-month periods in 170 patients with endocarditis. They found no evidence that dental procedures significantly increased the risk of IE.
In 2015, Chen et al published data looking retrospectively at 739 patients hospitalised with IE in Taiwan . They used a case-crossover design (where patients serve as their own controls). They looked back 12 weeks prior to admission with IE and used three 12-week blocks prior to that as control periods. The chance of having a dental procedure in the 12 weeks prior to developing IE was not significantly different to the chances of having a dental procedure during a control period.
Tubiana et al looked at the risk of developing IE after a dental procedure . They used administrative data from France for the period from January 2006 to December 2014. In total, they identified 138,076 individuals with a prosthetic valve and followed them for a total of 285,034 person-years. A total of 69,303 individuals underwent 103,463 dental procedures which had an indication for AP. For each case, the presence or absence of invasive dental procedures during the three months immediately preceding oral streptococcal IE was compared to the presence or absence of exposure to invasive dental procedures during earlier control periods in the same patient. When no dental procedure had taken place within three months, the incidence rate of oral streptococcal IE was 94.6 per 100,000 person-years (95% CI: 82.5-106.6). In the three months following an invasive dental procedure, the rate of oral streptococcal IE rose to 118.5 per 100,000 person-years (95% CI: 56.4-180.6). The case-crossover analysis demonstrated a statistically significant association between invasive dental procedures and oral streptococcal IE (5.1% vs. 3.2%; odds ratio (OR) 1.66, 95% CI: 1.05-2.63; p-value=0.03). Importantly, during the three months preceding oral streptococcal IE, only 5.1% of patients had an invasive dental procedure, emphasising that most cases of oral streptococcal IE are not related to dental procedures.
The most recent study to look at the link is by Chen et al (a different group to the above) . They found no link between invasive dental procedures and the development of IE; however, there were a number of important methodological issues with this paper which are beyond the scope of this article.
It is clear from these studies that IE caused by dental procedures can, at most, only cause a small number of cases, and that studies have to enrol many thousands of patients to detect a small difference. The data that exist do not provide a clear answer as to whether or not dental procedures can result in IE; many have interpreted the data to say that no link exists.
Do other procedures cause infective endocarditis?
It should not be forgotten that, prior to the iterations of the ESC guidelines in 2009 , many other procedures were considered to increase the risk of IE and AP was correspondingly recommended.
NICE re-reviewed the data in 2015 , when they re-examined their guidance first published in 2008. They found only one study which they considered showed evidence of a link between a procedure and the development of IE. In 2014, Mohee et al  demonstrated a link between urological procedures and the development of enterococcal IE, when they examined causative factors in 384 patients, retrospectively, from their institution. No other studies were considered methodologically robust enough to warrant inclusion and consideration.
Does antibiotic prophylaxis prevent the development of infective endocarditis?
Studies have been undertaken in an effort to understand whether AP might be effective, despite there being no proof that dental procedures lead to IE.
In 1986, Horstkotte et al compared 229 patients with prosthetic heart valves in whom 287 procedures were performed and who had AP, with 304 patients with prosthetic heart valves in whom 390 similar interventions were performed and who did not have AP . In the first group no patient developed IE. In the second group, six developed IE within 14 days.
In 1990, Imperiale and Horwitz published a very small case control study . They enrolled eight patients with “high-risk” lesions who had IE for the first time on a native valve within 12 weeks of a dental procedure. They were each matched with three patients who had also undergone a dental procedure and who had a similar valve lesion and age. AP was used by 1/8 patients and by 15/24 controls. They concluded that AP offered protection from IE.
In 1992, Van der Meer et al published two linked studies from the Netherlands. The first study has already been described. The second study was a case-control study that examined the efficacy of AP to prevent IE in patients with native valve disease . Forty-eight patients who developed IE within 180 days of a medical or dental procedure requiring AP were compared with 200 age-matched controls who had a relevant procedure but did not develop IE. AP was given to 8/48 cases and 26/200 controls. It was estimated that AP reduced the risk of developing IE within 30 days by 49%.
In the Lacassin study referred to previously , forty-eight subjects with known heart disease underwent a dental procedure (26 cases and 22 controls). Six cases and six controls received AP. For cases due to Streptococcus viridans and those with negative blood cultures, 3/18 cases received AP whereas 6/22 controls did. This difference did not reach statistical significance.
In recent times data collected at a national level have become available for analysis and have enabled researchers to assess the impact of guideline changes on the rates of IE.
In 2007, the American Heart Association released new guidelines for AP . The previous iteration in 1997  had recommended AP for patients at high risk and moderate risk of developing IE. The 2007 guidance restricted AP to those at high risk of developing or suffering an adverse outcome from IE. In March 2008, the National Institute for Health and Care Excellence published their recommendations regarding the use of AP in England and Wales . They recommended that AP should no longer be used, citing the lack of strong evidence for its efficacy, and expressing concerns about potential side effects from the use of AP, the potential development of antibiotic resistance, and the cost. The UK went from prescribing AP widely to patients at moderate and high risk of IE to not using AP at all . The European Society of Cardiology updated its 2004 guidelines  for the prevention of IE in 2009 . They adopted a similar approach to the Americans, moving from advising that patients at moderate or high risk should have AP, to just recommending it for patients at high risk.
We have recently reviewed this literature , and further studies have been published since. The data are complex to interpret and conflicting in their conclusions, with as many studies finding changes as not. These studies have not answered the question as to whether AP is effective.
How should we understand the variation in the conclusions of these studies? The first thing to appreciate is that in most countries AP was still recommended for high-risk patients, although not in the UK. Furthermore, the coding used between studies to identify cases of IE differed. It is also important to realise that there are no ICD-9 or 10 codes that identify OVGS specifically. Codes used to identify cases of streptococcal, and particularly OVGS IE, vary markedly between studies, making comparison difficult.
These studies are observational and cannot explain the changes observed. Over the time periods studied there have been many changes other than to guidelines for AP. Some of these, such as a growing and ageing society, better diagnostic techniques and the increasing use of new medical technologies such as percutaneous prosthetic valve insertion, may naturally tend to increase rates of IE. Other changes, such as a focus on practices to reduce healthcare-associated infection, may tend to reduce rates.
It is clear from studies that not all patients recommended for AP by guidelines are given AP in real life. This inconsistency has been exploited by Tubiana et al . Of the 69,303 individuals with a prosthetic valve who underwent 103,463 dental procedures which had an indication for AP, AP was given in only 50.5% of these cases. A total 267 patients developed IE likely to have been caused by OVGS during the follow-up period. Of these, a total of four patients developed IE within three months of an invasive dental procedure after receiving AP, whereas 10 who did not receive AP developed IE within three months of the procedure. Oral streptococcal IE incidence rates were 78.1 per 100,000 person-years (95% CI: 1.6-154.6) during the three months following an invasive dental procedure when AP was given, whereas they were 149.5 per 100,000 person-years (95% CI: 56.8-242.2) during the three months following an invasive dental procedure without AP. This difference approached, but did not reach, significance (p=0.08).
It is imperative that this study is replicated in other countries where such data are available.
It remains unclear as to whether AP is effective. It is a subject which divides clinicians. A quote by Stuart Chase, an American economist, is apt: “For those who believe, no proof is necessary. For those who don’t believe, no proof is possible.”.
At present, the ESC guideline committee believes that it is reasonable to give AP to patients at high risk of IE, as the risks and costs of AP are small and the potential consequences to a patient are devastating.
Until more evidence is available, this is a pragmatic and reasonable stance to take.