Introduction

Infective endocarditis (IE) remains one of the most serious conditions in medicine in the 21st century [1, 2]. The introduction of antibiotics in its treatment has decreased the mortality rates of the disease; however, these can reach up to 20 % in hospital and can be even higher on 1-year follow up (up to 40 %) [3]. Fatality rates of IE remain high and 2.7% of patients suffer an unexpected sudden death at 6 months during or after treatment [4]. After the near eradication of rheumatic fever in the western world, infective endocarditis accounts for the vast majority of cases of endocarditis with an incidence ranging between 1.7 and 11.2 cases per 100.000 people/year [5]. Overall, the male to female ratio is estimated at around 2 but outcomes tend to be worse in women [6].

Incidence of IE is higher in elderly patients (reaching 14.5 cases per 100.000 person/years) [7] as well as among specific sub-populations such as injectable drug user(s) (IDUs) [8], where clinical characteristics of the disease also differ from the general population [9]. Diabetes mellitus, invasive techniques, haemodialysis [6, 10] and implantable cardiac devices [11] have caused a shift in the most commonly identified infective microorganisms in the past decades, significantly increasing the prevalence of staphylococcal infections [12] which generally had shown higher mortality rates when compared to streptococcal infections [6, 10, 13]. As a result, many studies have emphasized the need to better understand the mechanism of infection in cardiac devices and reinforce preventive measures of health-care associated staphylococcal bacteraemia in an effort to decrease IE incidence [11, 13].

Limited understanding of the disease pathogenesis and progression is reflected upon different guidelines for prophylaxis among countries in recent years [14]. Identifying the most common microbial agents in IE is of particular current interest, following publication of new research at the American Heart Association meeting in Chicago in November 2014 showing increase in the incidence of infective endocarditis and the subsequent launch of a review of the guidance on Prophylaxis for Infective Endocarditis by the National Institute for Health and Care Excellence, in the UK (http://www.nice.org.uk/news/press-and-media/nice-to-review-its-guidance-on-the-use-of-antibiotics-to-prevent-infective-endocarditis). As far as treatment is concerned, IE remains a therapeutic challenge to date primarily due to the changing epidemiology of the causative pathogens together with the lack of knowledge on the exact mechanism of the disease and the insufficiency of diagnostic and therapeutic methods [1]. Early surgical therapy is increasingly becoming more popular with studies showing superior outcomes when compared to conventional treatment [15].

In this context, the objective of this study was to systematically review the literature to identify and prioritize the most common microbial factors causing IE in recent years and provide insight on special subpopulations. The rationale was that if more definitive conclusions on microbial associations could be drawn, these could inform prevention and treatment strategies. In this context, the study followed an inclusive approach to record study results predicting the most common microbial diagnosis for the patients presenting with IE.

Methods

Search strategy

This systematic review adopted the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [16]. Eligible articles were identified by a systematic search of the Pubmed bibliographical database for studies published from January 1, 2003 to March 31, 2013 by three investigators (CV, NV, LP) working independently. All articles derived from a search with the term “infective endocarditis” were assessed. In addition, all references of eligible articles retrieved by the search were scanned. Whenever full text was not available or additional data information was required, an e-mail was sent to the corresponding authors.

Study eligibility was based on the following inclusion criteria: (1) observational studies in patients with infective endocarditis and explicit mention of the pathogens related to the disease, and (2) more than five patients with infective endocarditis reported in the article. Discrepancies were discussed until complete agreement was reached; an additional reviewer (AK) gave input where required to reach consensus. The selection process excluded: (1) reviews and secondary research, (2) case reports, (3) missing full text or key data, also excluding articles that only commented on a single pathogen, (4) overlapping populations, (5) irrelevant articles, i.e. articles not reporting on the etiologic agents of IE, and (6) articles published in languages other than English. In the interest of not excluding clinically important case series, no formal quality assessment of eligible articles was undertaken.

Data extraction

The following variables were extracted and tabulated for each eligible study: first author’s name, year of publication, study design, country of origin, study site, patients’ age (mean, standard deviation, median, and age range or interquartile range if available), reported use of Duke’s diagnostic criteria, number of cases, number of unidentified microbes, numbers of cases reported per microbe, as well as special patient subpopulations. Due to the anticipated high heterogeneity of the included studies and the inclusive character of the search strategy, the five most frequent microbes associated with each study were recorded (with polymicrobial and unspecified cases included when being numerically among the five main causes) and sensitivity analyses were performed. A purely descriptive approach was adopted (i.e. data expressed as non-weighted means whenever possible) concerning continuous variables. Microbial causes results and all descriptive variables were treated in numerical values of totals or means. No further statistical analysis was undertaken.

The species was recorded where available, otherwise the genus was used.

Results

Eligible articles

The initial search of the Pubmed database yielded 3,397 potentially relevant articles, 2,313 of which were excluded as irrelevant and 477 as not written in English. For the remaining 607 articles, the full-text was studied. Further exclusion reasons comprised 345 case reports, and 107 secondary research papers. Fourteen were excluded on account of overlapping populations (by majority relating to the International Collaboration on Endocarditis Prospective Cohort Study), results from 23 studies could not be included because they only described one pathogen, while for 13 studies, though requested through repeat correspondence, data could not be retrieved, in part, because there was no response from the corresponding authors. Four authors provided additional data on their studies [17–20].

Eventually, 105 studies were included in this review (Fig. 1) availing data for a total of 33,214 cases of infective endocarditis. Although the mean age of patients was not available in all studies, virtually all age groups were represented, with age ranging from 6 days to 100 years. However, paediatric patients were under-represented compared to adults. Ninety-two studies (87.6 %) used the Duke’s criteria or modified Duke’s criteria for case selection.

Fig. 1
figure 1

Literature search PRISMA flow chart (from Moher et al, PLoS Med 6(6): e1000097)

Full size image

As shown in the Appendix, the majority of included studies (92.4 %, N = 97) used a cohort design (prospective: 30 / retrospective: 67), and only five were cross-sectional (4.8 %) and three case–control (2.9 %).

Most common microorganisms

Overall Staphylococcus aureus was found to be the most common microorganism causing infective endocarditis being isolated as one of the five most common microorganisms in 99 out of 105 studies (94.3 %), and being the most frequent in more than half of the studies (N = 57, 54.3 %), and second in a further third of the studies (N = 32, 30.5 %). Thirty-five of the above-mentioned 99 studies report data regarding MRSA strains; MRSA was isolated in the 4.5 % to 51.1 % of S. aureus infective endocarditis cases, while the median percentage was 28.1 Viridans group streptococci (VGS) were the second most commonly reported microorganisms, being the primary cause in over a fifth (21.9 %) of all studies (N = 23), and second in 24 studies (22.9 %) or third in 10. Coagulase-negative staphylococci (CoNS) were the most common not otherwise specified group recorded, isolated as the most frequent cause in 8.6 % of studies (N = 9), and reported as one of the five most common in 73.3 % (N = 77) of studies. Enterococcus spp was only rarely the most common microbe reported (N = 3), but it appeared consistently among the five most frequent in the majority of all studies (78.1 %, N = 82). Cases of Streptococcus bovis were also frequently reported, being the most common cause in three studies. The top decile of our studies (N = 11) comprised 18,065 cases. Of these studies, when using Duke's criteria, all but one had S. aureus as the most commonly diagnosed microbe. Only Day et al. [88] included a special sub-group: pediatric patients. This study contributes 61.6 % of the sample of this sub-group, substantially shifting the most common result towards S. aureus, as 4/7 studies report other microbes as most common. However, absolute sum of the results still points towards S. aureus, accounting for 45.4 % of all cases.

Table 1 summarises the most common pathogens per population category.

Table 1 Most common microbes in special populations
Full size table

Special populations

Congenital heart defects

Five cohort studies (2 from Asia, 3 from Europe) examined 672 patients with congenital heart defects. S. aureus and VGS were almost equally identified among the most common pathogens in all five studies. Alpha hemolytic streptococci [21] and S. bovis [22] were also reported. All patients were included irrespective of the type of their congenital heart disease and whether they had undergone surgical correction or not. Patients were not stratified by age and effect of surgery.

Prosthetic valve

Three cohort studies reported exclusively 994 patients with prosthetic valve endocarditis. CoNS were the most commonly isolated microorganisms being the primary cause in one study and among the three most common microorganisms in the other two studies. S. aureus was the primary cause in one study and among the four most common pathogens in the other two studies. VGS were the main pathogen in the study authored by Lalani et al. [23]. Enterococcus spp. was the third most common microorganism in two out of three studies. Finally, Propionibacterium acnes appeared as the second most common causative agent in one study but not among the five most common microorganisms in the other two studies [24].

Paediatric patients

Seven retrospective cohort studies (3 from North America, 3 from Asia and 1 from Oceania) referred to 1,026 paediatric patients, their ages ranging from 3 days to 20 years. S. aureus was again found to be the most commonly isolated microorganism responsible for the disease (first in 3 studies and second in 4), followed by VGS (first in 3 studies and second in 2). CoNS ranked as the third most common pathogen in five out of seven studies. Two studies found Candida albicans to be the second and third most common pathogen, respectively [25, 26], while cases of Enterococcus spp were also reported.

Injectable drug user(s) (IDUs)

Four retrospective cohort studies (North America: 2, Asia: 1, Europe: 1) comprising a total of 580 cases examined the characteristics of IE among populations of IDUs. As expected, S. aureus was by far the most commonly isolated microorganism (first in 4 out of 5 studies). Three studies report MRSA strains, which were isolated in 11.4 %, 45.2 %, and 46.4 % of S. aureus infective endocarditis cases, respectively; VGS followed (second in 2 studies and third in 1 more study) and CoNS completed the triad of the most commonly isolated microorganisms (third in 2 studies). Enterococcus spp (fourth in 2 studies) and C. albicans (second and fifth in 2 studies respectively) were also reported.

Implantable heart devices

Six cohort studies examined 505 cases in patients with implantable heart devices. S. aureus and CoNS were isolated with equal frequency as the primary causative agents of the disease and were consistently found in the top five microorganisms causing IE in this special population. Enterococcus spp and Gram-negative bacteria, though not usually found among the three most common pathogens, comprised a big number of cases accounting as the fourth or fifth most common microbial agent isolated.

Fungal endocarditis

Sixteen studies (2,145 cases), 15 cohort and one cross-sectional, reported fungi as one of the five most frequent microbes causing infective endocarditis. These studies came primarily from Europe (N = 9) and Asia (N = 5), with one additional in the United States and one in Latin America. Fungi were reported as the third cause in one study, fourth in 2 studies and fifth in 2 studies. Candida species are reported in two studies as the fourth most common cause of endocarditis. C. albicans was the most frequently isolated fungus, as it is found in 6 % of studies (N = 7), thrice as the second cause, twice as the third and once as the fourth and fifth cause. Finally, Aspergillus fumigatus and Candida parapsilosis were isolated in only one study.

Using Duke’s criteria

A sensitivity analysis was performed among the 92 studies reporting use of the Duke’s criteria for diagnosis of IE and accounting for 22,081 cases of IE. The five most common pathogens were, in order of decreasing frequence, S. aureus, VGS, CoNS, Enterococcus spp and Streptococcus spp. S.aureus was found in a total of 5,546 cases and was the most common pathogen in 51 of these studies and second in 29. VGS were the second most common cause, found in a total of 2,694 cases, and it was the first cause in 24 studies and second in 22. CoNS followed, accounting for 1,765 cases of IE in total, being first in five studies and second in nine. Enterococcus spp was the fourth most common cause, isolated in 1,543 cases, and being the second common pathogen in fourth studies. Finally, Streptococcus spp was the fifth leading cause of IE, isolated in 1,145 patients in total, being the most common cause in seven studies and second in 6.

Results per continent

Forty-nine studies (46.7 %, N = 12,752 patients) originate from European countries, 29 from Asia (27.6 %, N = 12,105), 14 from North America (13.3 %, N = 1,918), four from Africa (3.8 %, N = 602), four from Oceania (3.8 %, N = 1,681 patients), two from South America (1.9 %, N = 452), and three were multi-continent (2.9 %, N = 3,704 patients). Table 2 summarizes the main microbial causes per continent.

Table 2 Most common microbes per continent
Full size table

Discussion

General population

As presented above, our study concluded that the five most common pathogens causing infective endocarditis in the 21st century are by order of frequency: S. aureus, viridans group streptococci (VGS), coagulase-negative staphylococci (CoNS), Enterococcus spp and S. bovis. S. aureus was the most common microorganism in over half of the studies included in our research.

Our results highlight the change in the epidemiology of the causative agents of IE throughout time, as they contrast studies from the 1990s reporting streptococcal infections outnumbering staphylococcal [27, 28], but are consistent with more recently published literature; Slipczuk et al. report a significant increase in S. aureus IE which may associate with increasing numbers of IDUs in North America [29]. Also, in a review published in 2006 streptococci and staphylococci were found to account for the vast majority of infective endocarditis cases [30]. Enterococci have been found to be the third leading cause of IE in other studies as well [13, 31]. On the other hand, quite surprisingly, a recent study carried out across 11 years in France identified S. bovis to be responsible for 149 of 847 cases (17.6 %) [4]. These changes in the pattern of the pathogens of IE can be attributed to several factors, such as modern cardiology invasive techniques, non-nosocomial health care acquisition, the growing importance of the central line associated blood stream infections (CLABSIs), the percentage of colonization by MRSA in cardiac surgery patients, the aging population and the rise of enterococci and S. bovis. Of note, frequency of identified microbes was not identified in the same pattern in Asia, strongly emphasizing that a global one-size-fits-all approach to the management of IE patients is not appropriate.

Special groups

This is the first study that has systematically assessed the cause of IE in several special patient groups of importance. Our study found that those with congenital heart defects did not differ significantly compared to the general IE population as far as causative microorganisms are concerned. S. aureus and VGS were the most commonly isolated microorganisms. Of interest, patients with bicuspid aortic valve were more prone to perivalvular abscess formation particularly when S. aureus was the causative microorganism of IE [22] and surgical intervention was required in a big number of cases—ranging between 26 and 72 %—to treat infective endocarditis [21, 22, 32, 33]. Taking into consideration that oral hygiene and dental prophylaxis seem to be of vital importance among these patients [32, 34] and that incidence of IE after reparative surgery has been increasing [34], revision of the current guidelines regarding patients with CHD is crucial. Similarly, among paediatric patients S. aureus was the most common pathogen, followed by VGS, CoNS, C. albicans and Enterococcus spp. This is in accordance with a recent study which reported increasing frequencies of S. aureus, CoNS and fungal IE among children [35]. A large proportion of the affected children had underlying heart disease as described in all the included studies. Of interest, Marom et al. noted that children with no predisposing factors exhibited a more aggressive form of the disease with S. aureus and S. pneumoniae being significantly more frequent among these patients [25]. On the other hand, patients with prosthetic valves were found positive for CoNS in the majority of cases, while S. aureus appeared second and VGS third. Of interest, CoNS infection conferred increased risk for perivalvular abscess development and heart failure compared to S. aureus and VGS IE [23]. Also, proportion of patients with early death was higher among patients with CoNS IE [24]. Thus, prompt attention should be given to antimicrobial prophylaxis against CoNS in patients with prosthetic valves. Our study identified that S. aureus was the most commonly reported cause of IE in IDUs and patients with implantable heart devices. Cases of CoNS and fungal endocarditis need to be explored in these patients. The extensive use of invasive methods and implantable devices in the past decades has changed the spectrum of microorganisms that tend to colonise the heart valves causing infective endocarditis. Cabell et al. described a 42 % increase in the use of implantable heart devices between 1990 and 1999, especially depicting the increased frequency of the use of permanent pacemaker, and leading to a 124 % increase in device infections and 50 % increase in infective endocarditis prevalence [11]. Thanavaro et al. suggested that the increased mortality and morbidity due to implantable cardiac devices infection can be partly attributed to the increased age of the patients and comorbidities [36]. Taking this evidence into account, revisions in guidelines regarding chemoprophylaxis after such procedures could be explored.

Limitations and strengths

Though inclusive, significant heterogeneity and a number of poorly designed studies among the included is the primary limitation of this review [37]. For example, a large cohort study from Thailand that reported non-fermentative Gram-negative rods as the leading cause is subject to various limitations, such as lack of use of the modified Duke’s criteria for patient selection and failure to identify a causative agent in 86.59 % of cases [38]. Also, the included studies used different ways to record the pathogens; some agents are reported by genuses, others by subgenuses and others by species making presentation of data inconsistent in terms of nomenclature. Notably, evidence of MRSA infection was only recorded in 35 out of 99 studies in which S. aureus was identified as one of the five most common microorganisms, and evidence of vancomycin-resistant S. epidermidis (VRSE) only in one study implicating enterococcal infections. A large number of studies routinely did not specify the species of fungi or Gram-negative microorganisms. Another possible limitation of this study is that the world population is represented disproportionally with the cases included in this review, as cases from Europe and Asia represent a 74.28 % of the total cases displayed. Also, excluding case reports and case series may lead to negative reporting bias for emerging pathogens and excluding non English literature may lead to selection bias.

Finally, in one in four of our included cases (8,835/33,214; 26.6 %) the causative agent was not identified. The above most likely reflect resource limited settings (or in a few cases poor laboratory techniques) and highlight the importance of the use of newer techniques, such as PCR, which may lead to an increase in the detection of the etiologic agents of infective endocarditis.

The review benefits from its strict methodology and the large number of studies and cases analysed. Sensitivity analyses were performed throughout. Notably, all five continents and various special subgroups are represented.

Conclusion

The present review is the largest epidemiological study regarding causative agents in IE including a collective cohort of 33,214 infective endocarditis cases. The results document the rapidly changing profile of IE etiology, especially among special sub-groups of patients, as well as the predominance of S. aureus as the leading cause for infective endocarditis in the 21st century. However, the most common agent differed among several special groups of patients (most notably implantable heart devices) as well as patients from Asia, showing the need to tailor patient prophylaxis and treatment. Also, our study showed that in an important percentage of IE cases the responsible agent remained unidentified, even in high-level reference laboratories; this stresses how newest techniques have the capacity to change the profile of IE diagnosis and prevention. The results further emphasize the need to revisit IE prophylaxis and management strategies, as well as improving the reporting of the causative agents in future studies.